1 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
3 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
4 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
5 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
6 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
7 (rtl_tidy_fallthru_edge): Likewise.
8 (rtl_verify_fallthru): Likewise.
9 (rtl_verify_bb_layout): Likewise.
10 (skip_insns_after_block): Likewise.
11 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
12 * dwarf2out.c: Include print-rtl.h.
13 (dwarf2out_next_real_insn): New.
14 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
15 Dump debug binds in asm comments.
16 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
17 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
18 callers to use gsi_start_nondebug_bb instead.
19 (gsi_after_labels): Skip gimple debug stmts.
20 (gsi_start_nondebug): New.
21 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
22 (proper_loop_form_for_interchange): Adjust.
23 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
24 * gimple.h (gimple_seq_last_nondebug_stmt): New.
25 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
26 (collect_fallthrough_labels): Likewise.
27 (should_warn_for_implicit_fallthrough): Likewise.
28 (warn_implicit_fallthrough_r): Likewise.
29 (expand_FALLTHROUGH_r): Likewise.
30 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
31 (graphite_copy_stmts_from_block): Skip nonbind markers.
32 * haifa-sched.c (sched_extend_bb): Skip debug insns.
33 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
34 * jump.c (clean_barriers): Skip debug insns.
35 * omp-expand.c (expand_parallel_call): Skip debug insns.
36 (expand_task_call): Likewise.
37 (remove_exit_barrier): Likewise.
38 (expand_omp_taskreg): Likewise.
39 (expand_omp_for_init_counts): Likewise.
40 (expand_omp_for_generic): Likewise.
41 (expand_omp_for_static_nochunk): Likewise.
42 (expand_omp_for_static_chunk): Likewise.
43 (expand_omp_simd): Likewise.
44 (expand_omp_taskloop_for_outer): Likewise.
45 (expand_omp_taskloop_for_inner): Likewise.
46 (expand_oacc_for): Likewise.
47 (expand_omp_sections): Likewise.
48 (expand_omp_single): Likewise.
49 (expand_omp_synch): Likewise.
50 (expand_omp_atomic_load): Likewise.
51 (expand_omp_atomic_store): Likewise.
52 (expand_omp_atomic_fetch_op): Likewise.
53 (expand_omp_atomic_pipeline): Likewise.
54 (expand_omp_atomic_mutex): Likewise.
55 (expand_omp_target): Likewise.
56 (grid_expand_omp_for_loop): Likewise.
57 (grid_expand_target_grid_body): Likewise.
58 (build_omp_regions_1): Likewise.
59 * omp-low.c (check_combined_parallel): Skip debug stmts.
60 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
61 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
62 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
64 * tree-cfg.c (make_blobs_1): Skip debug stmts.
65 (make_edges): Likewise.
66 (cleanup_dead_labels): Likewise.
67 (gimple_can_merge_blocks_p): Likewise.
68 (stmt_starts_bb_p): Likewise.
69 (gimple_block_label): Likewise.
70 (gimple_redirect_edge_and_branch): Likewise.
71 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
73 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
75 * tree-pretty-print (print_declaration): Omit initializer in slim
77 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
79 (eliminate_unnecessary_stmts): Stabilize block removal order.
80 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
81 * var-tracking.c (get_first_insn): New.
82 (vt_emit_notes): Call it.
83 (vt_initialize): Walk any insns before the first BB.
84 (delete_debug_insns): Likewise.
86 * gimple.h (enum gimple_debug_subcode): Add
87 GIMPLE_DEBUG_BEGIN_STMT.
88 (gimple_debug_begin_stmt_p): New.
89 (gimple_debug_nonbind_marker_p): New.
90 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
91 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
92 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
93 * insn-notes.def (BEGIN_STMT): New.
94 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
95 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
96 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
97 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
98 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
99 (INSN_DEBUG_MARKER_KIND): New.
100 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
101 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
102 (INSN_VAR_LOCATION_PTR): New.
103 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
105 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
106 deps only if debug bind insns are enabled.
107 (pass_expand::execute): Avoid deep TER and expand
108 debug locations for debug bind insns only.
109 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
110 debug stmts special handling down to debug bind stmts.
111 * combine.c (try_combine): Narrow debug insns special handling
112 down to debug bind insns.
113 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
114 Narrow debug insns preexisting special handling down to debug
116 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
117 handling down to debug bind insns.
118 * function.c (instantiate_virtual_regs): Skip debug markers,
119 adjust handling of debug binds.
120 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
121 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
122 * haifa-sched.c (schedule_insn): Narrow special handling of debug
123 insns to debug bind insns.
124 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
125 special handling of debug stmts to debug bind stmts.
126 * ipa-split.c (split_function): Likewise.
127 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
128 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
130 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
131 * regrename.c (build_def_use): Likewise.
132 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
133 (pass_cprop_hardreg): Narrow special casing of debug insns to
135 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
136 * reload1.c (reload): Likewise.
137 * sese.c (sese_insert_phis_for_liveouts): Narrow special
138 casing of debug stmts to debug bind stmts.
139 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
140 * ssa-iterators.h (num_imm_uses): Likewise.
141 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
142 debug stmts to debug bind stmts.
143 * tree-inline.c (tree_function_versioning): Narrow special casing
144 of debug stmts to debug bind stmts.
145 * tree-loop-distribution.c (generate_loops_for_partition):
146 Narrow special casing of debug stmts to debug bind stmts.
147 * tree-sra.c (analyze_access_subtree): Narrow special casing
148 of debug stmts to debug bind stmts.
149 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
150 stmts to debug bind stmts.
151 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
152 casing of debug stmts to debug bind stmts.
153 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
154 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
155 casing of debug stmts to debug bind stmts.
156 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
158 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
159 debug stmts to debug bind stmts.
160 (gimple_replace_ssa_lhs): Likewise.
161 (insert_debug_temp_for_var_def): Likewise.
162 (insert_debug_temps_for_defs): Likewise.
163 (reset_debug_uses): Likewise.
164 * tree-ssanames.c (release_ssa_name_fn): Likewise.
165 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
166 (adjust_debug_stmts): Likewise.
167 (adjust_phi_and_debug_stmts): Likewise.
168 (vect_do_peeling): Likewise.
169 * tree-vect-loop.c (vect_transform_loop): Likewise.
170 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
171 * var-tracking.c (adjust_mems): Narrow special casing of debug
172 insns to debug bind insns.
173 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
174 (compute_bb_dataflow, vt_find_locations): Likewise.
175 (vt_expand_loc, emit_notes_for_changes): Likewise.
176 (vt_init_cfa_base): Likewise.
177 (vt_emit_notes): Likewise.
178 (vt_initialize): Likewise.
179 (vt_finalize): Likewise.
181 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
182 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
183 (next_nonnote_nondebug_insn_bb): New.
184 (prev_nonnote_nondebug_insn_bb): New.
185 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
186 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
187 (prev_nonnote_nondebug_insn_bb): Declare.
188 (next_nonnote_nondebug_insn_bb): Declare.
189 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
190 * cfgrtl.c (get_last_bb_insn): Likewise.
191 * lra.c (push_insns): Likewise.
193 2017-12-11 David Malcolm <dmalcolm@redhat.com>
196 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
198 * selftest.c (selftest::test_runner::test_runner): New ctor.
199 (selftest::test_runner::~test_runner): New dtor.
200 * selftest.h (class selftest::test_runner): New class.
202 2017-12-11 Carl Love <cel@us.ibm.com>
204 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
205 vec_extract_fp32_from_shortl]): Add #defines.
206 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
207 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
208 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
209 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
210 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
211 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
212 * doc/extend.texi: Add documentation for the added builtins.
214 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
216 PR rtl-optimization/80693
217 PR rtl-optimization/81019
218 PR rtl-optimization/81020
219 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
220 are not mentioned in i3. Place the REG_UNUSED note on i2,
221 possibly modified to REG_DEAD, if it did not originate in i3.
223 2017-12-11 Jakub Jelinek <jakub@redhat.com>
225 * recog.c (store_data_bypass_p_1): New function.
226 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
227 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
229 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
231 PR rtl-optimization/83361
232 * ifcvt.c (if_convert): Call fixup_partitions.
234 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
236 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
237 early folding of splat_u{8,16,32}.
239 2017-12-11 Jakub Jelinek <jakub@redhat.com>
241 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
242 output_operand_lossage first argument with capital letter.
243 (aarch64_override_options): Don't start error and sorry first argument
246 2017-12-11 Andi Kleen <ak@linux.intel.com>
248 PR gcov-profile/83355
249 * auto-profile.c (string_table::get_index_by_decl): Don't
250 recurse when abstract origin points to itself.
252 2017-12-11 Bin Cheng <bin.cheng@arm.com>
254 PR tree-optimization/83320
255 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
256 (prune_datarefs_not_in_loop): Ditto.
258 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
260 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
262 2017-12-10 Jakub Jelinek <jakub@redhat.com>
264 PR tree-optimization/83337
265 * gimple-loop-interchange.cc (compute_access_stride): Handle
266 bitfield DRs properly.
268 2017-12-09 Jakub Jelinek <jakub@redhat.com>
270 PR tree-optimization/83338
271 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
272 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
273 vector of unsigned integers to vector of signed integers.
275 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
277 PR rtl-optimization/83317
278 * lra-constraints.c (process_address_1): Add insn code check.
280 2017-12-08 Michael Matz <matz@suse.de>
282 Fix PR tree-optimization/83323
283 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
284 head-controlled loops and loop BBs.
285 * common.opt (funroll-and-jam): Remove, instead ...
286 (floop-unroll-and-jam): ... reuse this option.
287 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
288 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
289 (-floop-unroll-and-jam): ... this option.
291 2017-12-08 Jakub Jelinek <jakub@redhat.com>
293 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
294 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
295 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
297 2017-12-08 Richard Biener <rguenther@suse.de>
300 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
301 handle accesses outside of zero-sized vars.
303 2017-12-08 Martin Jambor <mjambor@suse.cz>
305 PR tree-optimization/83141
306 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
307 test for MEM_REFs implicitely changing types with padding. Remove
309 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
311 2017-12-08 Jakub Jelinek <jakub@redhat.com>
313 * config/arc/arc.c (arc_attribute_table): Add exclusions to
315 * config/avr/avr.c (avr_attribute_table): Likewise.
316 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
317 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
318 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
319 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
320 of struct attribute_spec.
321 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
323 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
326 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
328 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
331 * profile-count.c (profile_count::from_gcov_type): Move from
332 profile-count.h; handle overflow.
333 * profile-count.h (profile_count::from_gcov_type): Move offline.
335 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
337 PR rtl-optimization/83304
338 * combine.c (move_deaths): If we do not know where a register died,
341 2017-12-08 Richard Biener <rguenther@suse.de>
343 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
344 Provide -fopt-info-loop feedback when we interchange in a nest.
346 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
348 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
349 for armv6 ARM CPU IDs.
351 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
353 * common/config/arm/arm-common.c: Include <algorithm>.
354 (INCLUDE_VECTOR): Define.
355 (compare_opt_names): New function.
356 (arm_rewrite_selected_arch): Only strip out extensions that can be
357 expressed through -mfpu. Sort the remaining extensions
360 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
362 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
363 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
364 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
365 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
367 2017-06-08 Tristan Gingold <gindold@adacore.com>
370 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
371 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
373 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
375 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
376 of struct attribute_spec.
378 2017-12-08 Julia Koval <julia.koval@intel.com>
380 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
381 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
382 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
383 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
384 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
385 _mm_maskz_dpwssds_epi32): Ditto.
387 2017-12-08 Richard Biener <rguenther@suse.de>
389 PR tree-optimization/81303
390 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
391 conditions try to create a comparison vector type matching
392 the data vector type.
393 (vectorizable_condition): Adjust.
394 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
395 Leave invariant conditions alone in case we can vectorize those.
397 2017-12-08 Julia Koval <julia.koval@intel.com>
399 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
400 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
401 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
402 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
403 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
405 2017-12-08 Julia Koval <julia.koval@intel.com>
407 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
408 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
409 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
410 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
411 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
412 _mm_maskz_dpbusds_epi32): New intrinsics.
414 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
416 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
418 (ld<bh>io_signed): Likewise.
419 (st<bhw>io): Likewise.
420 * config/nios2/predicates.md (ldstio_memory_operand): Allow
421 SMALL_INT12 constant integer operand.
423 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
425 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
427 (nios2_large_constant_allowed): ...this. Adjust uses.
428 (nios2_plus_symbolic_constant_p): Rename to...
429 (nios2_plus_large_constant_p): ...this. Adjust uses.
430 (nios2_legitimate_address_p): Correct CONST_INT handling.
431 (nios2_symbolic_memory_operand_p): Rename to...
432 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
433 (nios2_large_constant_p): Check for large constant integers too.
434 (nios2_split_large_constant): Handle constant integers.
435 (nios2_split_symbolic_memory_operand): Rename to...
436 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
437 (nios2_legitimize_constant_address): Handle constant integers.
438 (r0rel_constant_p): Handle small constant integers.
439 (nios2_print_operand_address): Handle r0-relative integer addresses.
440 * config/nios2/nios2-protos.h: Adjust for renamed functions.
441 * config/nios2/nios2.md: Adjust for renamed functions.
443 2017-12-07 Andrew Waterman <andrew@sifive.com>
445 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
446 (TARGET_HAVE_SRODATA_SECTION): New define.
447 (riscv_select_section): New function.
449 2017-12-08 Joseph Myers <joseph@codesourcery.com>
450 Alexander Monakov <amonakov@ispras.ru>
451 Jakub Jelinek <jakub@redhat.com>
454 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
456 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
458 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
459 a bool success value. Don't call output_operand_lossage here.
460 (aarch64_print_ldpstp_address): Return a bool success value.
461 (aarch64_print_operand_address): Call output_addr_const if
462 aarch64_print_address_internal fails.
463 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
464 'y'; call output_operand_lossage instead. Call output_operand_lossage
465 if aarch64_print_ldpstp_address fails.
467 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
469 * tree-vector-builder.h
470 (tree_vector_builder::binary_encoded_nelts): Declare.
471 * tree-vector-builder.c
472 (tree_vector_builder::binary_encoded_nelts): New function.
473 * fold-const.c (negate_expr_p): Likewise.
474 (operand_equal_p, fold_checksum_tree): Likewise.
475 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
476 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
477 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
478 (uniform_vector_p): Likewise.
479 * varasm.c (const_hash_1, compare_constant): Likewise.
480 * tree-ssa-ccp.c: Include tree-vector-builder.h.
481 (valid_lattice_transition): Operate directly on the VECTOR_CST
483 * ipa-icf.c: Include tree-vector-builder.h.
484 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
485 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
487 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
489 * tree.c (build_vector): Delete.
490 * tree.h (build_vector): Make static and move into the self-testing
493 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
495 * vector-builder.h (vector_builder::derived): New const overload.
496 (vector_builder::elt): New function.
497 * tree-vector-builder.h (tree_vector_builder::type): New function.
498 (tree_vector_builder::apply_step): Declare.
499 * tree-vector-builder.c (tree_vector_builder::apply_step): New
501 * gimple-fold.h (tree_vector_builder): Declare.
502 (gimple_build_vector): Take a tree_vector_builder instead of a
503 type and vector of elements.
504 * gimple-fold.c (gimple_build_vector): Likewise.
505 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
507 (get_initial_defs_for_reduction): Likewise.
508 (vectorizable_induction): Likewise.
510 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
512 * tree-vector-builder.h
513 (tree_vector_builder::new_binary_operation): Declare.
514 * tree-vector-builder.c
515 (tree_vector_builder::new_binary_operation): New function.
516 * fold-const.c (fold_relational_const): Use it.
517 (const_binop): Likewise. Check that both input vectors have
518 the same number of elements, thus excluding things like WIDEN_SUM.
519 Check whether it is possible to operate directly on the encodings
522 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
524 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
525 new_unary_operation, operating only on the encoded elements.
526 (const_unop): Likewise.
527 (exact_inverse): Likewise.
528 (distributes_over_addition_p): New function.
529 (const_binop): Use tree_vector_builder and new_unary_operation
530 for combinations of VECTOR_CST and INTEGER_CST. Operate only
531 on the encoded elements unless the encoding is strided and the
532 operation does not distribute over addition.
533 (fold_convert_const): Use tree_vector_builder and
534 new_unary_operation. Operate only on the encoded elements
535 for truncating integer conversions, or for non-stepped encodings.
537 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
539 * config/sparc/sparc.c: Include tree-vector-builder.h.
540 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
541 * expmed.c: Include tree-vector-builder.h.
542 (make_tree): Use tree_vector_builder instead of build_vector.
543 * fold-const.c: Include tree-vector-builder.h.
544 (const_binop): Use tree_vector_builder instead of build_vector.
545 (const_unop): Likewise.
546 (native_interpret_vector): Likewise.
547 (fold_vec_perm): Likewise.
548 (fold_ternary_loc): Likewise.
549 * gimple-fold.c: Include tree-vector-builder.h.
550 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
552 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
553 (simplify_vector_constructor): Use tree_vector_builder instead
555 * tree-vect-generic.c: Include tree-vector-builder.h.
556 (add_rshift): Use tree_vector_builder instead of build_vector.
557 (expand_vector_divmod): Likewise.
558 (optimize_vector_constructor): Likewise.
559 * tree-vect-loop.c: Include tree-vector-builder.h.
560 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
561 of build_vector. Explicitly use a stepped encoding for
563 * tree-vect-slp.c: Include tree-vector-builder.h.
564 (vect_get_constant_vectors): Use tree_vector_builder instead
566 (vect_transform_slp_perm_load): Likewise.
567 (vect_schedule_slp_instance): Likewise.
568 * tree-vect-stmts.c: Include tree-vector-builder.h.
569 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
570 (vect_gen_perm_mask_any): Likewise.
571 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
572 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
574 (build_vector_from_val): Likewise. Explicitly use a duplicate
577 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
579 * doc/generic.texi (VECTOR_CST): Describe new representation of
581 * vector-builder.h: New file.
582 * tree-vector-builder.h: Likewise.
583 * tree-vector-builder.c: Likewise.
584 * Makefile.in (OBJS): Add tree-vector-builder.o.
585 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
586 * tree-core.h (tree_base): Add a vector_cst field to the u union.
587 (tree_vector): Change the number of elements to
588 vector_cst_encoded_nelts.
589 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
590 (VECTOR_CST_ELTS): Delete.
591 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
592 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
593 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
594 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
595 (VECTOR_CST_ENCODED_ELT): Likewise.
596 (vector_cst_encoded_nelts): New function.
597 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
598 VECTOR_CST_NELTS_PER_PATTERN as arguments.
599 (vector_cst_int_elt, vector_cst_elt): Declare.
600 * tree.c: Include tree-vector-builder.h.
601 (tree_code_size): Abort if passed VECTOR_CST.
602 (tree_size): Update for new VECTOR_CST layout.
603 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
604 VECTOR_CST_NELTS_PER_PATTERN as arguments.
605 (build_vector): Use tree_vector_builder.
606 (vector_cst_int_elt, vector_cst_elt): New functions.
607 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
608 encoded elements and then create the vector in the canonical form.
609 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
610 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
611 (tree_c_tests): Call test_vector_cst_patterns.
612 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
614 (hash_tree): Likewise.
615 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
616 (streamer_write_tree_header): Likewise.
617 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
618 (streamer_alloc_tree): Likewise. Update call to make_vector.
619 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
621 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
623 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
624 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
625 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
627 2017-12-07 Bin Cheng <bin.cheng@arm.com>
628 Richard Biener <rguenther@suse.de>
630 PR tree-optimization/81303
631 * Makefile.in (gimple-loop-interchange.o): New object file.
632 * common.opt (floop-interchange): Reuse the option from graphite.
633 * doc/invoke.texi (-floop-interchange): Ditto. New document for
634 -floop-interchange and mention it for -O3.
635 * opts.c (default_options_table): Enable -floop-interchange at -O3.
636 * gimple-loop-interchange.cc: New file.
637 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
638 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
639 * passes.def (pass_linterchange): New pass.
640 * timevar.def (TV_LINTERCHANGE): New time var.
641 * tree-pass.h (make_pass_linterchange): New declaration.
642 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
643 interchange. Record IV before/after increment in new parameters.
644 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
645 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
647 (check_reduction_path): ...New function here.
648 * tree-vectorizer.h (check_reduction_path): New declaration.
650 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
653 PR rtl-optimization/80818
654 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
655 always early clobbered.
656 * lra-lives.c (process_bb_lives): Check input hard regs for early
657 clobbered non-operand hard reg.
659 2017-12-07 Jakub Jelinek <jakub@redhat.com>
662 * tree-cfg.c (verify_gimple_assign_binary): Don't require
663 types_compatible_p, just that TYPE_MODE is the same.
665 2017-12-07 Martin Sebor <msebor@redhat.com>
668 * attribs.c (empty_attribute_table): Initialize new member of
669 struct attribute_spec.
670 (decl_attributes): Add argument. Handle mutually exclusive
671 combinations of attributes.
672 (selftests::test_attribute_exclusions): New function.
673 (selftests::attribute_c_tests): Ditto.
674 * attribs.h (decl_attributes): Add default argument.
675 * selftest.h (attribute_c_tests): Declare.
676 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
677 * tree-core.h (attribute_spec::exclusions, exclude): New type and
679 * doc/extend.texi (Common Function Attributes): Update const and pure.
680 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
681 of struct attribute_spec.
682 * config/arc/arc.c (arc_attribute_table): Same.
683 * config/arm/arm.c (arm_attribute_table): Same.
684 * config/avr/avr.c ( avr_attribute_table): Same.
685 * config/bfin/bfin.c (bfin_attribute_table): Same.
686 * config/cr16/cr16.c (cr16_attribute_table): Same.
687 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
688 * config/h8300/h8300.c (h8300_attribute_table): Same.
689 * config/i386/i386.c (ix86_attribute_table): Same.
690 * config/ia64/ia64.c (ia64_attribute_table): Same.
691 * config/m32c/m32c.c (m32c_attribute_table): Same.
692 * config/m32r/m32r.c (m32r_attribute_table): Same.
693 * config/m68k/m68k.c (m68k_attribute_table): Same.
694 * config/mcore/mcore.c (mcore_attribute_table): Same.
695 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
696 * config/mips/mips.c (mips_attribute_table): Same.
697 * config/msp430/msp430.c (msp430_attribute_table): Same.
698 * config/nds32/nds32.c (nds32_attribute_table): Same.
699 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
700 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
701 * config/rl78/rl78.c (rl78__attribute_table): Same.
702 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
703 * onfig/rx/rx.c (rx_attribute_table): Same.
704 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
705 * config/sh/sh.c (sh_attribute_table): Same.
706 * config/sparc/sparc.c (sparc_attribute_table): Same.
707 * config/spu/spu.c (spu_attribute_table): Same.
708 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
709 * config/v850/v850.c (v850_attribute_table): Same.
710 * config/visium/visium.c (visium_attribute_table): Same.
712 2017-12-07 Tamar Christina <tamar.christina@arm.com>
715 * config/arm/arm.c (INCLUDE_STRING): Define.
716 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
717 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
720 2017-12-07 Michael Matz <matz@suse.de>
722 Add unroll and jam pass
724 * gimple-loop-jam.c: New file.
725 * Makefile.in (OBJS): Add gimple-loop-jam.o.
726 * common.opt (funroll-and-jam): New option.
727 * opts.c (default_options_table): Add unroll-and-jam at -O3.
728 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
729 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
730 * passes.def: Add pass_loop_jam.
731 * timevar.def (TV_LOOP_JAM): Add.
732 * tree-pass.h (make_pass_loop_jam): Declare.
733 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
734 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
735 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
736 to flow_loop_tree_node_add.
737 (duplicate_subloops, copy_loops_to): Append to sibling list.
738 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
739 * doc/invoke.texi (-funroll-and-jam): Document new option.
740 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
742 2017-12-07 Richard Biener <rguenther@suse.de>
744 PR tree-optimization/83296
745 PR tree-optimization/67769
746 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
747 flow sensitive info in an unrelated BB.
748 (value_replacement): Use reset_flow_sensitive_info.
749 (minmax_replacement): Reset flow sensitive info on the def
750 we move. Do not reset flow sensitive info in the whole BB
752 (abs_replacement): Likewise.
754 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
757 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
758 rs6000_cpu to the given -mcpu=, or to the default processor.
760 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
762 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
763 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
764 instead of rs6000_cpu_attr.
765 (rs6000_adjust_cost): Ditto.
766 (is_microcoded_insn): Ditto.
767 (rs6000_adjust_priority): Ditto.
768 (rs6000_issue_rate): Ditto.
769 (rs6000_use_sched_lookahead): Ditto.
770 (rs6000_use_sched_lookahead_guard): Ditto.
771 (rs6000_sched_reorder): Ditto.
772 (force_new_group): Ditto.
773 * config/rs6000/rs6000.md (cpu attribute): Ditto.
774 (group_ending_nop): Ditto.
776 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
778 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
779 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
780 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
781 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
783 (rs6000_reassociation_width): Ditto.
784 (rs6000_emit_epilogue): Ditto.
785 (rs6000_adjust_cost): Ditto.
786 (is_microcoded_insn): Ditto.
787 (is_cracked_insn): Ditto.
788 (rs6000_adjust_priority): Ditto.
789 (rs6000_sched_reorder): Ditto.
790 (rs6000_sched_reorder2): Ditto.
791 (insn_must_be_first_in_group): Ditto.
792 (insn_must_be_last_in_group): Ditto.
793 (rs6000_register_move_cost): Ditto.
794 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
797 2017-12-07 Julia Koval <julia.koval@intel.com>
799 * config.gcc: Add vaesintrin.h.
800 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
802 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
803 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
805 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
806 * config/i386/immintrin.h: Include vaesintrin.h.
807 * config/i386/sse.md (vaesdec_<mode>): New pattern.
808 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
809 _mm_aesdec_epi128): New intrinsics.
811 2017-12-06 David Malcolm <dmalcolm@redhat.com>
813 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
814 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
815 c-family/c-spellcheck.cc.
816 (best_macro_match::best_macro_match): Likewise.
818 (struct edit_distance_traits<cpp_hashnode *>): Move to
819 c-family/c-spellcheck.h.
820 (class best_macro_match): Likewise.
822 2017-12-06 Jakub Jelinek <jakub@redhat.com>
824 PR tree-optimization/83293
825 * gimple-ssa-strength-reduction.c (insert_initializers): Use
826 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
827 might insert into empty bb.
830 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
831 simplify for plus with :c added, and pointer_plus without that.
832 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
833 with undefined overflow and the conversion is not widening,
834 perform negation in utype and only convert to type afterwards.
835 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
836 simplify for plus with :c added, and pointer_plus without that.
837 If type is integral with undefined overflow and the conversion is
838 not widening, perform minus in utype and only convert to type
839 afterwards. Move the last pointer_diff_expr simplify into the
842 2017-12-06 Martin Sebor <msebor@redhat.com>
844 PR tree-optimization/82646
845 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
848 2017-12-06 Martin Sebor <msebor@redhat.com>
850 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
853 PR tree-optimization/83075
854 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
855 strncat/strncpy don't change length of source string.
857 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
860 2017-11-29 Martin Aberg <maberg@gaisler.com>
862 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
863 to prevent b2bst errata sequence.
864 (sqrtdf2_fix): Likewise.
866 2017-12-06 Jakub Jelinek <jakub@redhat.com>
868 PR tree-optimization/81945
869 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
870 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
871 to dest_cfun has orig_loop_num set, either remap it to the new
872 loop number if the loop got moved too, or clear it.
874 2017-12-05 Steve Ellcey <sellcey@cavium.com>
876 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
878 (thunderx2t99_nothing): New insn reservation.
879 (thunderx2t99_mrs): New insn reservation.
880 (thunderx2t99_multiple): New insn reservation.
881 (thunderx2t99_alu_basi): Add bfx to reservation.
882 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
884 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
887 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
889 2017-12-05 Bin Cheng <bin.cheng@arm.com>
891 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
892 tree-ssa-pre.c::remove_dead_inserted_code.
893 * tree-ssa-dce.h: New file.
894 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
895 (remove_dead_inserted_code): Move and rename to function
896 tree-ssa-dce.c::simple_dce_from_worklist.
897 (pass_pre::execute): Update use.
899 2017-12-05 Richard Biener <rguenther@suse.de>
901 PR tree-optimization/83277
902 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
903 to code-gen liveout vars.
905 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
907 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
908 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
909 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
912 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
914 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
916 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
917 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
920 2017-12-05 Richard Biener <rguenther@suse.de>
922 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
923 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
924 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
925 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
927 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
929 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
931 2017-12-05 Martin Liska <mliska@suse.cz>
932 Jakub Jelinek <jakub@redhat.com>
934 * doc/invoke.texi: Document the options.
935 * flag-types.h (enum sanitize_code): Add
936 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
937 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
938 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
939 * opts.c: Define new sanitizer options.
940 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
941 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
943 2017-12-05 Julia Koval <julia.koval@intel.com>
945 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
946 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
947 (ix86_handle_option): Handle -mavx512vnni.
948 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
949 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
950 * config/i386/i386-c (__AVX512VNNI__): New.
951 * config/i386/i386.c (ix86_target_string): Handle new option.
952 (ix86_valid_target_attribute_inner_p): Handle new option.
953 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
954 * config/i386/i386.opt (mavx512vnni): New option.
956 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
959 * athlon.md: Disable for generic.
960 * haswell.md: Enable for generic.
961 * i386.c (ix86_sched_init_global): Add core hooks for generic.
962 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
964 (ix86_adjust_cost): Move generic to haswell path.
966 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
968 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
969 instead of MEM_P in a couple more places. Fix formatting issues.
971 2017-12-04 Jim Wilson <jimw@sifive.com>
973 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
974 instead of GP_REG_LAST-1.
975 (riscv_adjust_libcall_cfi_prologue): Likewise.
976 (riscv_adjust_libcall_cri_epilogue): Likewise.
977 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
980 2017-12-04 Luis Machado <luis.machado@linaro.org>
982 * ipa-pure-const.c (check_decl): Add missing newline.
983 (state_from_flags): Likewise.
985 2017-12-04 Jeff Law <law@redhat.com>
987 PR tree-optimizatin/78496
988 * gimple-ssa-evrp-analyze.h
989 (evrp_range_analyzer::get_vr_values): Simplify.
990 * gimple-ssa-evrp-analyze.c: Corresponding changes.
991 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
992 and gimple-ssa-evrp-analyze.h.
993 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
994 (simplify_stmt_for_jump_threading): Copy a blob of code from
995 tree-vrp.c to use ranges to simplify statements.
996 (dom_opt_dom_walker::before_dom_children): Call
997 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
998 (dom_opt_dom_walker::after_dom_children): Similarly for
999 evrp_range_analyzer::leave.
1000 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
1003 * gimple-ssa-evrp-analyze.c
1004 (evrp_range_analyzer::extract_range_from_stmt): Always use
1005 vr_values::update_value_range so preexisting range info is
1006 medged with new range info, even if the new range is VR_VARYING.
1008 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1010 * combine.c: Adjust comment.
1011 (use_crosses_set_p): Delete.
1012 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
1013 (try_combine): Ditto.
1015 2017-12-04 Richard Biener <rguenther@suse.de>
1017 PR tree-optimization/83255
1018 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
1019 Re-add zero-iteration check.
1021 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1023 PR rtl-optimization/83245
1024 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
1025 hard registers as earlyclobber, also if not in an asm.
1027 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1031 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1034 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1035 rs6000_cpu based on cpu_index, not tune_index.
1037 2017-12-04 Richard Biener <rguenther@suse.de>
1039 PR tree-optimization/83238
1040 * graphite-scop-detection.c (scop_detection::merge_sese): Make
1041 code match comment, rejecting invalid SESE regions.
1043 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
1045 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
1046 require base operand is a REG_POINTER prior to reload on targets
1047 with non-equivalent space registers.
1049 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1051 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
1052 (update_specialized_profile): Fix updating of counts.
1053 (perhaps_add_new_callers): Likewise.
1055 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1058 * x86-tune.def: Remove obsolette FIXMEs.
1059 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
1060 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
1061 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
1063 (X86_TUNE_PAD_RETURNS): Disable for generic.
1065 2017-12-02 Jakub Jelinek <jakub@redhat.com>
1067 PR tree-optimization/83170
1068 PR tree-optimization/83241
1069 * gimple-ssa-store-merging.c
1070 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
1071 gimple_vuse (ins_stmt) in case it has changed.
1072 (imm_store_chain_info::output_merged_store): Likewise.
1074 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
1078 * tree-cfg.c (pass_warn_function_return::execute): Handle
1079 __builtin_ubsan_handle_missing_return like __builtin_unreachable
1080 with BUILTINS_LOCATION.
1084 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
1085 is BLKmode for vector field with vector raw mode, use TYPE_MODE
1086 instead of DECL_MODE.
1088 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
1089 last argument to rtx pointer.
1090 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
1091 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
1092 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
1093 depending on the chosen ISAs.
1094 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
1095 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
1096 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
1098 * config/i386/sse.md (mov<mode>_internal): Likewise.
1099 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
1101 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1103 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
1105 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1107 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
1108 parameters from prototype.
1109 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
1110 parameters. Don't print a comment.
1111 (emit_fusion_gpr_load): Adjust.
1112 (emit_fusion_load_store): Adjust.
1113 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
1114 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
1115 comment on the second line.
1117 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1120 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1121 rs6000_cpu based on cpu_index, not tune_index.
1123 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1125 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
1126 which_alternative instead of which_alternative + 1.
1127 (output_asm_insn): Print an extra tab if the template is short.
1129 2017-12-01 Jim Wilson <jimw@sifive.com>
1131 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
1133 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
1134 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
1136 * doc/tm.texi: Regenerate.
1138 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
1141 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
1142 whether we can allocate pseudos before trying to fix an address.
1143 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
1144 memory address is indexed or indirect.
1145 (floatuns_<mode>si2_hw2): Likewise.
1147 2017-12-01 Jason Merrill <jason@redhat.com>
1149 * Makefile.in (TAGS): Add c-family/*.cc.
1151 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
1153 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
1154 (call_value_insn): Likewise.
1155 (sibcall_insn): Likewise.
1156 (sibcall_value_insn): Likewise.
1157 (movsi_aarch64): Likewise.
1158 (movdi_aarch64): Likewise.
1159 (add_losym_): Likewise.
1160 (ldr_got_small_): Likewise.
1161 (ldr_got_small_sidi): Likewise.
1162 (ldr_got_small_28k_): Likewise.
1163 (ldr_got_small_28k_sidi): Likewise.
1164 * config/aarch64/aarch64.c (aarch64_print_address_internal):
1165 Move output_addr_const to symbolic case. Add error check.
1167 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1169 * config/s390/predicates.md (plus16_Q_operand): New predicate.
1170 * config/s390/s390.md: Disable MVC merging peephole if it would
1171 disable operand forwarding.
1172 (new peephole2): Split MVCs if it would turn them into up to 2
1175 2017-12-01 Richard Biener <rguenther@suse.de>
1177 PR tree-optimization/83232
1178 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
1179 detection of same access. Instead of breaking the group here
1180 do not consider the duplicate. Add comment explaining real fix.
1182 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
1184 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
1186 2017-12-01 Sudakshina Das <sudi.das@arm.com>
1188 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
1190 2017-12-01 Jakub Jelinek <jakub@redhat.com>
1192 * function.h (struct function): Remove cilk_frame_decl,
1193 is_cilk_function and calls_cilk_spawn fields.
1194 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
1196 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
1197 cilk_elemental field.
1198 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
1199 * target.def: Adjust comment.
1200 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
1201 Don't test cilk_elemental.
1203 PR tree-optimization/83233
1204 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
1205 bswap_stat name for the struct.
1208 * tree.h (SWITCH_BREAK_LABEL_P): Define.
1209 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
1210 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
1211 SWITCH_BREAK_LABEL_P set on the label.
1212 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
1213 added for default case if it was missing and not all cases covered.
1214 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
1215 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
1217 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
1218 to avoid -Wimplicit-fallthrough warning.
1219 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
1220 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
1222 PR tree-optimization/83221
1223 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
1225 (init_reassoc): Formatting fix.
1228 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
1229 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
1233 * tree-cfg.c: Include asan.h.
1234 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
1236 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1238 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
1240 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1242 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
1243 vpcmpeqd instruction.
1245 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1247 * config/i386/i386.c (standard_sse_constant_opcode): Fix
1248 registers type for 128bit mode.
1250 2017-11-30 Jakub Jelinek <jakub@redhat.com>
1252 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
1254 * gengtype-state.c (read_state_pair): Likewise.
1255 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
1256 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
1257 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
1258 * ifcvt.c (noce_try_store_flag_constants): Likewise.
1259 * tree-ssa-ccp.c (ccp_finalize): Likewise.
1260 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
1261 * builtins.c (fold_builtin_3): Likewise.
1262 * graphite-scop-detection.c
1263 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
1264 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
1267 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
1270 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
1271 parameter. Return true if ignore_decl_rtl_set_p is true and other
1272 conditions are satisfied.
1273 * asan.h (asan_protect_global): Add new parameter.
1274 * varasm.c (categorize_decl_for_section): Pass true as second parameter
1275 to asan_protect_global calls.
1277 2017-11-30 Jim Wilson <jimw@sifive.com>
1279 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
1280 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
1281 -mstrict-align, add info on default value. Delete redundant lines for
1282 -mabi. Add missing -mexplicit-relocs docs.
1284 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1286 * config/arc/arc.md (trap): New pattern.
1288 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1290 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
1291 instruction to end into a delay slot.
1292 * config/arc/arc.md (cond_delay_insn): Check if the instruction
1293 can be placed into a delay slot against reg_note.
1295 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1297 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
1298 labels number of usages.
1300 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1302 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
1304 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
1306 2017-11-30 Jakub Jelinek <jakub@redhat.com>
1309 * internal-fn.c (expand_mul_overflow): Optimize unsigned
1310 multiplication by power of 2 constant into two shifts + comparison.
1312 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
1315 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
1317 2017-11-30 Richard Biener <rguenther@suse.de>
1319 PR tree-optimization/83202
1320 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
1321 allow_peel argument and guard peeling.
1322 (canonicalize_loop_induction_variables): Likewise.
1323 (canonicalize_induction_variables): Pass false.
1324 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
1325 peeling from cunrolli.
1327 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1329 * combine.c (try_combine): Print a message to dump file whenever
1330 I0, I1, or I2 cannot be combined into I3.
1332 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1334 PR rtl-optimization/83156
1335 PR rtl-optimization/82621
1336 * combine.c (try_combine): Don't split an I2 if one of the dests is
1337 set again before I3. Allow unused dests.
1339 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1341 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
1343 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
1345 PR rtl-optimization/80818
1346 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
1347 recursively. Use insn code for clobber.
1348 (lra_set_insn_recog_data): Pass the new arg to
1349 collect_non_operand_hard_regs.
1350 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
1352 (lra_update_insn_regno_info): Pass insn to
1353 add_regs_to_insn_regno_info.
1355 2017-11-29 Jim Wilson <jimw@sifive.com>
1356 Andrew Waterman <andrew@sifive.com>
1358 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
1359 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
1360 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
1361 use SINGLE_SHIFT_COST.
1362 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
1363 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
1365 2017-11-29 Julia Koval <julia.koval@intel.com>
1367 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
1368 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
1369 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
1370 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
1371 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
1372 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
1373 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
1374 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
1375 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
1376 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
1377 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
1378 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
1379 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
1380 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
1381 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
1382 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
1383 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
1384 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
1385 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
1386 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
1387 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
1388 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
1389 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
1390 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
1391 __builtin_ia32_vpshldv_v2di_mask,
1392 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
1393 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
1394 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
1396 2017-11-29 Julia Koval <julia.koval@intel.com>
1398 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
1399 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
1400 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
1401 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
1402 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
1403 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
1404 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
1405 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
1406 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
1407 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
1408 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
1409 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
1410 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
1411 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
1412 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
1413 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
1414 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
1415 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
1416 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1417 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
1418 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
1420 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1422 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
1423 movsi_pic_gotdata_op instruction as a load for the UT699 errata
1426 2017-11-29 Martin Aberg <maberg@gaisler.com>
1428 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
1429 to prevent b2bst errata sequence.
1430 (sqrtdf2_fix): Likewise.
1432 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1434 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
1435 (div_sqrt_insn_p): New function.
1436 (sparc_do_work_around_errata): Insert NOP instructions to
1437 prevent sequences that could trigger the TN-0013 errata for
1438 certain LEON3 processors.
1439 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
1440 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
1441 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
1442 (in_branch_delay): Prevent div and sqrt in delay slot if
1444 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
1446 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1448 * config/sparc/sparc.c (atomic_insn_p): New function.
1449 (sparc_do_work_around_errata): Insert NOP instructions to
1450 prevent sequences that could trigger the TN-0010 errata for
1452 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
1453 instruction referable in atomic_insns_p.
1455 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1457 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
1458 (atomic_compare_and_swap_leon3_1): Likewise.
1461 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1463 * config/sparc/sparc.c (fpop_insn_p): New function.
1464 (sparc_do_work_around_errata): Insert NOP instructions to
1465 prevent sequences that could trigger the TN-0012 errata for
1467 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
1468 * config/sparc/sparc.md (fix_gr712rc): New attribute.
1469 (in_branch_annul_delay): Prevent floating-point instructions
1470 in delay slot of annulled integer branch.
1472 2017-11-29 Richard Biener <rguenther@suse.de>
1474 PR tree-optimization/83202
1475 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
1477 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
1478 nodes vectorized to the same stmts multiple times.
1479 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
1480 (vect_analyze_slp_instance): Adjust.
1481 (scalar_stmts_to_slp_tree_map_t): New typedef.
1482 (vect_schedule_slp_instance): Add a map recording the SLP node
1483 representing the vectorized stmts for a set of scalar stmts.
1484 Avoid code-generating redundancies.
1485 (vect_schedule_slp): Allocate map and pass it down.
1487 2017-11-29 Nathan Sidwell <nathan@acm.org>
1490 * tree.c (build_complex_type): Fix canonicalization. Only fill in
1493 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
1495 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
1496 cases for printing LDP/STP memory addresses.
1497 (aarch64_print_address_internal): Renamed from
1498 aarch64_print_operand_address, added parameter, add Pmode check.
1499 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
1500 (aarch64_print_operand_address): Indirect to
1501 aarch64_print_address_internal.
1502 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
1503 'y' operand output specifier.
1505 2017-11-29 Jakub Jelinek <jakub@redhat.com>
1508 * tree.c (build_simple_mem_ref_loc): Handle
1509 get_addr_base_and_unit_offset returning a MEM_REF.
1512 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
1515 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
1518 2017-11-28 David Malcolm <dmalcolm@redhat.com>
1520 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
1522 (layout::move_to_column): Likewise.
1524 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1527 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
1528 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
1530 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1531 Martin Jambor <mjambor@suse.cz>
1534 * tree.h (expr_type_first_operand_type_p): Declare
1535 * tree.c (expr_type_first_operand_type_p): New function.
1536 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
1537 (ipa_value_from_jfunc): Adjust declaration.
1538 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
1539 Use it as result type for arithmetics, unless it is NULL in which case
1540 be more conservative.
1541 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
1542 ipa_get_jf_pass_through_result.
1543 (propagate_vals_across_pass_through): Likewise.
1544 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
1545 is to propagate_vals_across_pass_through.
1546 (propagate_constants_across_call): Pass PARM_TYPE to
1547 propagate_scalar_across_jump_function.
1548 (find_more_scalar_values_for_callers_subset): Pass parameter type to
1549 ipa_value_from_jfunc.
1550 (cgraph_edge_brings_all_scalars_for_node): Likewise.
1551 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
1552 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
1553 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
1554 target_type, pass it to ipa_value_from_jfunc.
1555 (update_indirect_edges_after_inlining): Pass parameter type to
1556 try_make_edge_direct_simple_call.
1558 2017-11-28 Jeff Law <law@redhat.com>
1560 * gimple-ssa-evrp-analyze.c
1561 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
1562 refine ranges if scev_initialized_p returns true.
1563 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
1565 2017-11-28 Julia Koval <julia.koval@intel.com>
1567 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
1568 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
1569 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
1570 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
1571 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
1572 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
1573 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
1574 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
1575 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
1576 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
1577 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
1578 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
1579 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
1580 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
1581 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
1582 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
1583 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
1584 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
1585 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
1586 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
1587 __builtin_ia32_vpshrd_v2di_mask): New builtins.
1588 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
1590 2017-11-28 Julia Koval <julia.koval@intel.com>
1592 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
1593 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
1594 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
1595 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
1596 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
1597 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
1598 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
1599 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
1600 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
1601 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
1602 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
1603 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
1604 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
1605 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
1606 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
1607 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
1608 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
1609 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
1610 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
1611 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
1612 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
1613 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
1614 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
1615 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
1616 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
1617 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
1618 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
1619 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
1620 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
1621 __builtin_ia32_vpshld_v2di_mask): New builtins.
1622 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1623 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
1625 2017-11-28 Richard Biener <rguenther@suse.de>
1627 PR tree-optimization/80776
1628 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
1630 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
1632 (evrp_range_analyzer::record_ranges_from_incoming_edges):
1633 If the incoming edge is an effective fallthru because the other
1634 edge only reaches a __builtin_unreachable () then record ranges
1635 derived from the controlling condition in SSA info.
1636 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
1637 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
1639 2017-11-28 Olivier Hainque <hainque@adacore.com>
1641 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
1642 on mingw build hosts.
1644 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1647 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
1649 2017-11-28 Richard Biener <rguenther@suse.de>
1652 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
1653 copies generated from memcpy use a character array as reference
1656 2017-11-28 Julia Koval <julia.koval@intel.com>
1657 Sebastian Peryt <sebastian.peryt@intel.com>
1659 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
1660 c-family/c-cilkplus.o, c-family/array-notation-common.o,
1661 cilk-common.o, cilk.h, cilk-common.c): Remove.
1663 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
1664 * builtins.c (is_builtin_name): Remove cilkplus condition.
1665 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
1666 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
1667 cilk-builtins.def, cilkplus.def): Remove.
1668 * cif-code.def (CILK_SPAWN): Remove.
1669 * cilk-builtins.def: Delete.
1670 * cilk-common.c: Ditto.
1672 * cilkplus.def: Ditto.
1673 * config/darwin.h (fcilkplus): Delete.
1674 * cppbuiltin.c: Ditto.
1675 * doc/extend.texi: Remove cilkplus doc.
1676 * doc/generic.texi: Ditto.
1677 * doc/invoke.texi: Ditto.
1678 * doc/passes.texi: Ditto.
1679 * gcc.c (fcilkplus): Remove.
1680 * gengtype.c (cilk.h): Remove.
1681 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
1683 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
1685 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
1687 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
1688 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
1689 cilkplus conditions.
1690 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
1691 inline_read_section): Ditto.
1692 * ipa-inline-analysis.c (cilk.h): Remove.
1693 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
1694 * lto-wrapper.c (merge_and_complain, append_compiler_options,
1695 append_linker_options): Remove condition for fcilkplus.
1696 * lto/lto-lang.c (cilk.h): Remove.
1697 (lto_init): Remove condition for fcilkplus.
1698 * omp-expand.c (expand_cilk_for_call): Delete.
1699 (expand_omp_taskreg, expand_omp_for_static_chunk,
1700 expand_omp_for): Remove cilkplus
1702 (expand_cilk_for): Delete.
1703 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
1704 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
1705 execute_lower_omp, diagnose_sb_0): Ditto.
1706 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
1707 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
1708 * tree-nested.c: Ditto.
1709 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
1710 (dump_generic_node): Ditto.
1711 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
1712 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
1714 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
1716 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1718 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
1719 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
1720 for vector mode and !TARGET_SIMD.
1722 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1724 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
1726 * tree.h (SWITCH_LABELS): Remove.
1727 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
1728 assert SWITCH_BODY is non-NULL.
1729 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
1731 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
1733 PR tree-optimization/80788
1734 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
1735 has TREE_OVERFLOW set, call drop_tree_overflow.
1737 2017-11-28 Richard Biener <rguenther@suse.de>
1739 PR tree-optimization/83158
1740 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
1742 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
1745 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
1746 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
1748 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
1750 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
1751 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
1753 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1756 * varasm.c (compare_constant): Take the mode of the constants into
1757 account when comparing floating point constants.
1759 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
1761 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
1762 from explicit instantiation of debug_helper.
1763 * vec.h (DEFINE_DEBUG_VEC): Ditto.
1765 2017-11-27 Richard Biener <rguenther@suse.de>
1767 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
1770 2017-11-27 Richard Biener <rguenther@suse.de>
1772 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
1773 (type_hash_canon): Also clear int_cst_hash_table entry for
1775 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
1777 2017-11-27 Tamar Christina <tamar.christina@arm.com>
1779 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1780 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1781 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1783 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
1785 * hash-map.h (gt_cleare_cache): Avoid UB.
1787 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
1789 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
1790 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
1791 (decide_unroll_runtime_iterations): Likewise.
1792 (decide_unroll_stupid): Likewise.
1794 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1797 * config/i386/i386.md: Add a loop with incssp.
1799 2017-11-27 Martin Jambor <mjambor@suse.cz>
1801 PR tree-optimization/81248
1802 * tree-sra.c (splice_param_accesses): Remove size check.
1803 (decide_one_param_reduction): Fix size check.
1804 * gimple-pretty-print.c (dump_profile): Silence warning.
1805 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
1807 2017-11-27 Jakub Jelinek <jakub@redhat.com>
1810 * dbxout.c (lastlineno): New variable.
1811 (dbx_debug_hooks): Use dbxout_switch_text_section as
1812 switch_text_section debug hook.
1813 (dbxout_function_end): Switch to current_function_section
1814 rather than function_section. If crtl->has_bb_partition,
1815 output just one N_FUN, depending on in_cold_section_p.
1816 (dbxout_source_line): Remember last lineno in lastlineno.
1817 (dbxout_switch_text_section): New function.
1818 (dbxout_function_decl): Adjust dbxout_block caller.
1819 (dbx_block_with_cold_children): New function.
1820 (dbxout_block): Return true if any LBRAC/RBRAC have been
1821 emitted. Use dbx_block_with_cold_children at depth == 0
1822 in second partition. Add PARENT_BLOCKNUM argument, pass
1823 it optionally adjusted to children. Output LBRAC/RBRAC
1824 around recursive call only if the block is in the current
1825 partition, if not and anything was output, emit empty
1827 * final.c (final_scan_insn): Compute cold_function_name
1828 before calling switch_text_section debug hook. Call
1829 that hook even if dwarf2out_do_frame if not emitting
1833 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
1834 TREE_READONLY decls.
1836 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
1838 PR rtl-optimization/82488
1839 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
1841 2017-11-26 Julia Koval <julia.koval@intel.com>
1843 * config/i386/i386.c (processor_target_table): Add skylake_cost for
1845 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
1848 2017-11-26 Julia Koval <julia.koval@intel.com>
1850 * config/i386/driver-i386.c (host_detect_local_cpu):
1851 Detect skylake-avx512.
1853 2017-11-26 Julia Koval <julia.koval@intel.com>
1855 * config.gcc: Add -march=cannonlake.
1856 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
1857 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
1858 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
1859 (PTA_CANNONLAKE): New.
1860 (processor_target_table): Add cannonlake.
1861 (ix86_option_override_internal): Ditto.
1862 (fold_builtin_cpu): Ditto.
1863 (get_builtin_code_for_version): Handle cannonlake.
1864 (M_INTEL_COREI7_CANNONLAKE): New.
1865 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
1866 * doc/invoke.texi: Add -march=cannonlake.
1868 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
1870 * plugin.c (add_new_plugin): Use platform-specific library extensions.
1871 (try_init_one_plugin): Alternative implementation for MinGW.
1872 * Makefile.in (plugin_implib): New.
1873 (gengtype-lex.c): Fix broken AIX workaround.
1874 * configure: Regenerate.
1875 * doc/plugins.texi: Document support for MinGW.
1877 2017-11-25 Jakub Jelinek <jakub@redhat.com>
1879 PR rtl-optimization/81553
1880 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
1881 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
1882 is a shift where C1 has different mode than the whole shift, use C1's
1883 mode for MULT rather than the shift's mode.
1886 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
1887 builtins not enabled in the currently selected ISA.
1889 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
1891 PR tree-optimization/71026
1892 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
1893 (insert_reciprocals): Change to insert reciprocals before a division
1894 by a square and to insert the square of a reciprocal.
1895 (execute_cse_reciprocals_1): Change to consider division by a square.
1896 (register_division_in): Add importance parameter.
1898 2017-11-24 Richard Biener <rguenther@suse.de>
1900 PR tree-optimization/82402
1901 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
1902 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
1904 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
1906 * match.pd (0-ptr): New transformation.
1908 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
1911 * ipa-inline.c (inline_small_functions): Set current badnes correctly
1912 when skipping checking.
1914 2017-11-24 Richard Biener <rguenther@suse.de>
1916 PR tree-optimization/83128
1917 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
1918 (vn_reference_lookup_3): Likewise.
1920 2017-11-24 Jakub Jelinek <jakub@redhat.com>
1923 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
1924 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
1927 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
1928 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
1929 check for the latter with gimple_call_builtin_p. Do not handle
1930 BUILT_IN_STPNCPY_CHK which is not a pass through call.
1932 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
1934 * config/arm/arm_neon.h: Fix pragma GCC push_options before
1937 2017-11-23 Julia Koval <julia.koval@intel.com>
1939 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
1940 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
1941 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
1942 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
1943 _mm512_maskz_expandloadu_epi16): New intrinsics.
1944 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
1945 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
1946 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
1947 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
1948 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
1949 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
1950 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
1951 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
1952 _mm256_maskz_expandloadu_epi8): New intrinsics.
1953 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
1954 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
1955 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
1956 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
1957 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
1958 * config/i386/sse.md (VI248_VLBW): New iterator.
1959 (expand<mode>_mask, expand<mode>_maskz): New patterns.
1961 2017-11-23 Julia Koval <julia.koval@intel.com>
1963 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
1964 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
1965 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
1966 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
1967 _mm512_mask_compressstoreu_epi16): New.
1968 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
1969 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
1970 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
1971 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
1972 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
1973 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
1974 _mm256_mask_compressstoreu_epi8): New.
1975 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
1976 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
1977 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
1978 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
1979 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
1980 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
1981 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
1982 __builtin_ia32_compresshi128_mask,
1983 __builtin_ia32_compressstoreuqi512_mask,
1984 __builtin_ia32_compressstoreuhi512_mask,
1985 __builtin_ia32_compressstoreuqi256_mask,
1986 __builtin_ia32_compressstoreuqi128_mask,
1987 __builtin_ia32_compressstoreuhi256_mask,
1988 __builtin_ia32_compressstoreuhi128_mask): New builtins.
1989 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
1991 (ix86_expand_special_args_builtin): Handle new types.
1992 (s4fma_expand): Handle new builtin array.
1993 * config/i386/immintrin.h: Include new headers.
1994 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
1995 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
1997 2017-11-23 Jakub Jelinek <jakub@redhat.com>
1999 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
2000 fixes. Declare temp and g variables at the top in order to avoid
2001 {} in most of the cases.
2003 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
2005 * match.pd (ptr-0): New transformation.
2007 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2009 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
2010 (struct addr_mode_cost_table): New.
2011 (struct tune_params): Add field addr_mode_costs.
2012 * config/arm/arm.c (generic_addr_mode_costs): New.
2013 (arm_slowmul_tune): Initialise addr_mode_costs field.
2014 (arm_fastmul_tune): Likewise.
2015 (arm_strongarm_tune): Likewise.
2016 (arm_xscale_tune): Likewise.
2017 (arm_9e_tune): Likewise.
2018 (arm_marvell_pj4_tune): Likewise.
2019 (arm_v6t2_tune): Likewise.
2020 (arm_cortex_tune): Likewise.
2021 (arm_cortex_a8_tune): Likewise.
2022 (arm_cortex_a7_tune): Likewise.
2023 (arm_cortex_a15_tune): Likewise.
2024 (arm_cortex_a35_tune): Likewise.
2025 (arm_cortex_a53_tune): Likewise.
2026 (arm_cortex_a57_tune): Likewise.
2027 (arm_exynosm1_tune): Likewise.
2028 (arm_xgene1_tune): Likewise.
2029 (arm_cortex_a5_tune): Likewise.
2030 (arm_cortex_a9_tune): Likewise.
2031 (arm_cortex_a12_tune): Likewise.
2032 (arm_cortex_a73_tune): Likewise.
2033 (arm_v7m_tune): Likewise.
2034 (arm_cortex_m7_tune): Likewise.
2035 (arm_v6m_tune): Likewise.
2036 (arm_fa726te_tune): Likewise.
2037 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
2039 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2041 * config/arm/arm.c (arm_mem_costs): New function.
2042 (arm_rtx_costs_internal): Use arm_mem_costs.
2044 2017-11-23 Mark Wielaard <mark@klomp.org>
2046 * dwarf2out.c (init_sections_and_labels): Use generation to create
2047 unique ranges_section_label and ranges_base_label. Return generation.
2048 (output_rnglists): Add generation argument. Use generation to create
2049 unique ranges labels.
2050 (dwarf2out_finish): Get generation from init_sections_and_labels
2051 and pass generation to output_rnglists.
2053 2017-11-23 Mike Stump <mikestump@comcast.net>
2054 Eric Botcazou <ebotcazou@adacore.com>
2056 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
2057 * cfgloop.h (struct loop): Add unroll field.
2058 * function.h (struct function): Add has_unroll bitfield.
2059 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
2060 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
2061 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
2062 (pass_rtl_unroll_loops::gate): Likewise.
2063 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
2064 for which loop->unroll==1.
2065 (decide_unroll_constant_iterations): Use note for consistency and
2066 take loop->unroll into account. Return early if loop->unroll is set.
2067 Fix thinko in existing test.
2068 (decide_unroll_runtime_iterations): Use note for consistency and
2069 take loop->unroll into account.
2070 (decide_unroll_stupid): Likewise.
2071 * lto-streamer-in.c (input_cfg): Read loop->unroll.
2072 * lto-streamer-out.c (output_cfg): Write loop->unroll.
2073 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
2075 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
2076 (print_loop): Print loop->unroll if set.
2077 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
2078 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
2079 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
2081 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
2082 loop->unroll is set and smaller than the trip count. Otherwise bypass
2083 entirely the heuristics if loop->unroll is set. Remove dead note.
2084 Fix off-by-one bug in other note.
2085 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
2086 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
2088 (tree_unroll_loops_completely): Make static.
2089 (pass_complete_unroll::execute): Use correct type for variable.
2090 (pass_complete_unrolli::execute): Fix formatting.
2091 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
2093 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
2095 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
2096 enable when TARGET_PREFER_AVX128 is set.
2098 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2100 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
2102 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
2103 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
2105 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2107 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
2108 frequencies. Fix estimation of aggregate parameters.
2110 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2112 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
2115 2017-11-23 Tom de Vries <tom@codesourcery.com>
2117 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
2119 2017-11-23 Tom de Vries <tom@codesourcery.com>
2121 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
2124 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
2127 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
2128 sibcall_value_pcrel_fdpic): Use local variable instead of
2130 (calli_tbr_rel): Add missing operand 2.
2131 (call_valuei_tbr_rel): Add missing operand 3.
2133 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2136 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
2137 bitpos/bitsize covering the whole destination, use store_expr only if
2138 the complex mode is the same. Otherwise, use expand_normal and if
2139 it returns CONCAT, subreg each part separately instead of trying to
2140 subreg the whole result.
2142 2017-11-23 Richard Biener <rguenther@suse.de>
2144 PR tree-optimization/23094
2145 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
2146 come here from walking over backedges in the first iteration.
2147 (vn_reference_lookup_3): Skip clobbers that store the same value.
2149 2017-11-23 Richard Biener <rguenther@suse.de>
2151 PR tree-optimization/81403
2152 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
2153 a block we need a leader relative to.
2154 (phi_translate_1): For nary processing require a leader from
2155 get_representative_for given we run expression simplification
2156 using match-and-simplify. Remove previous fix.
2158 2017-11-22 Jeff Law <law@redhat.com>
2160 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
2161 Use new method allocate_value_range rather than accessing the
2162 vrp_value_range_pool data member directly.
2163 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
2164 to use extract_range_from_stmt method to avoid need for
2165 extract_range_from_assignment method.
2166 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
2167 method rather than setting values_propgated data member directly.
2168 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
2169 and values propagated data members and extract_range_from_assignment
2170 method. Reorder private data members to conform to standards.
2171 Add new methods set_lattice_propagation_complete and
2172 allocate_value_range.
2174 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
2176 PR rtl-optimization/83030
2177 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
2178 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
2179 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
2180 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
2181 CROSSING_JUMP_P flag is already set before setting it.
2182 * cfgrtl.c (fixup_partition_crossing): Likewise.
2183 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
2186 2017-11-22 Jakub Jelinek <jakub@redhat.com>
2188 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
2189 Handle the case where both arguments are using gen_const_vec_series.
2191 2017-11-22 David Malcolm <dmalcolm@redhat.com>
2194 * pretty-print.c (pp_format): Move quoting implementation to
2195 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
2196 to pass address of "quote" local.
2197 (pp_begin_quote): New function.
2198 (pp_end_quote): New function.
2199 * pretty-print.h (printer_fn): Convert penultimate param from bool
2201 (pp_begin_quote): New decl.
2202 (pp_end_quote): New decl.
2203 * tree-diagnostic.c (default_tree_printer): Convert penultimate
2204 param from bool to bool *.
2205 * tree-diagnostic.h (default_tree_printer): Likewise.
2207 2017-11-22 Jeff Law <law@redhat.com>
2209 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
2210 Initialize vr_values.
2211 (evrp_range_analyzer::try_find_new_range): Call methods attached to
2212 vr_values via vr_values class instance rather than delegators.
2213 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
2214 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
2215 (evrp_range_analyzer::push_value_range): Likewise.
2216 (evrp_range_analyzer::pop_value_range): Likewise.
2217 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
2218 most delegators. Those remaining are exposed as public interfaces.
2219 Make vr_values a pointer and private.
2220 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
2222 (evrp_range_analyzer::get_vr_value): New method.
2223 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
2224 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
2225 (class evrp_dom_walker): Attach evrp_folder class, initialize
2226 it in the ctor. Remove temporary delegators.
2227 (evrp_dom_walker::before_dom_children): Call methods in attached
2228 evrp_range_analyzer class via class instance pointer. Use
2229 free value_range_constant_singleton to remove need for
2230 op_with_constant_singleton_value delegator method. Do not
2231 create a vrp_prop class instance for every call! Narrow
2232 scope of a couple variables.
2233 (evrp_dom_walker::cleanup): Call methods in attached
2234 evrp_range_analyzer class via class instance pointer.
2235 * vr-values.h (class vr_values): Privatize many methods and
2238 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2240 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
2241 * cfgexpand.c (expand_debug_expr): Remove handling for them.
2242 * expr.c (expand_expr_real_2): Likewise.
2243 * fold-const.c (const_unop): Likewise.
2244 * optabs-tree.c (optab_for_tree_code): Likewise.
2245 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2246 * tree-inline.c (estimate_operator_cost): Likewise.
2247 * tree-pretty-print.c (dump_generic_node): Likewise.
2248 (op_code_prio): Likewise.
2249 (op_symbol_code): Likewise.
2250 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
2251 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
2252 * internal-fn.c (direct_internal_fn_optab): New function.
2253 (direct_internal_fn_array, direct_internal_fn_supported_p
2254 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
2255 * fold-const-call.c (fold_const_reduction): New function.
2256 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
2258 * tree-vect-loop.c: Include internal-fn.h.
2259 (reduction_code_for_scalar_code): Rename to...
2260 (reduction_fn_for_scalar_code): ...this and return an internal
2262 (vect_model_reduction_cost): Take an internal_fn rather than
2264 (vect_create_epilog_for_reduction): Likewise. Build calls rather
2266 (vectorizable_reduction): Use internal functions rather than tree
2267 codes for the reduction operation. Update calls to the functions
2269 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
2270 Use calls to internal functions rather than REDUC tree codes.
2271 * config/aarch64/aarch64-simd.md: Update comment accordingly.
2273 2017-11-22 Olivier Hainque <hainque@adacore.com>
2275 * config/vxworks.c (vxworks_override_options): Pick default
2276 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
2277 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
2278 DWARF_GNAT_ENCODINGS_DEFAULT.
2279 * config/vxworksae.h: Likewise.
2281 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
2283 PR tree-optimization/83104
2284 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
2287 2017-11-22 Marek Polacek <polacek@redhat.com>
2288 H.J. Lu <hongjiu.lu@intel.com>
2289 Jason Merrill <jason@redhat.com>
2294 * calls.c (initialize_argument_information): Call
2295 warn_parameter_passing_abi target hook.
2296 (store_one_arg): Use 0 for empty record size. Don't push 0 size
2297 argument onto stack.
2298 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
2299 * common.opt: Update -fabi-version description.
2300 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
2301 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
2303 (ix86_is_empty_record): New function.
2304 (ix86_warn_parameter_passing_abi): New function.
2305 (TARGET_EMPTY_RECORD_P): Redefine.
2306 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
2307 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
2308 * doc/tm.texi: Regenerated.
2309 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
2310 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
2311 * dwarf2out.c (get_ultimate_context): Move to tree.c.
2312 * explow.c (hard_function_value): Call arg_int_size_in_bytes
2313 instead of int_size_in_bytes.
2314 * expr.c (copy_blkmode_to_reg): Likewise.
2315 * function.c (aggregate_value_p): Return 0 for empty types.
2316 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
2318 (locate_and_pad_parm): Call arg size_in_bytes instead
2320 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
2321 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
2322 * target.def (empty_record_p, warn_parameter_passing_abi): New target
2324 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
2325 (std_gimplify_va_arg_expr): Skip empty records. Call
2326 arg_size_in_bytes instead size_in_bytes.
2327 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
2328 * tree-core.h (tree_type_common): Add empty_flag.
2329 (tree_decl_common): Update comments.
2330 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
2332 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
2333 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
2335 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
2336 * tree.c (default_is_empty_type): New function.
2337 (default_is_empty_record): New function.
2338 (arg_int_size_in_bytes): New function.
2339 (arg_size_in_bytes): New function.
2340 (get_ultimate_context): New function.
2341 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
2342 TRANSLATION_UNIT_WARN_EMPTY_P.
2343 (default_is_empty_record, arg_int_size_in_bytes,
2344 arg_size_in_bytes, get_ultimate_context): Declare.
2346 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2348 * config/arm/arm.c (cmse_clear_registers): New function.
2349 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
2350 code by call to cmse_clear_registers.
2351 (cmse_nonsecure_entry_clear_before_return): Likewise.
2353 2017-11-22 Tamar Christina <tamar.christina@arm.com>
2355 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
2356 (vdot_s32, vdotq_s32): New.
2357 (vdot_lane_u32, vdotq_lane_u32): New.
2358 (vdot_lane_s32, vdotq_lane_s32): New.
2361 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2364 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
2365 for unsigned values with fewer HWIs than the precision.
2366 (test_overflow): New function.
2367 (wide_int_cc_tests): Call it.
2369 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2371 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
2372 has the same mode class as Pmode.
2374 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2376 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
2377 padding_bits_to_clear_ptr.
2378 (cmse_nonsecure_entry_clear_before_return): Likewise.
2380 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2382 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
2383 auto_sbitap instead of integer bitfield to control register needing
2386 2017-11-22 Jakub Jelinek <jakub@redhat.com>
2388 PR tree-optimization/83044
2389 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
2390 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
2391 that need the upper bound. Subtract offset from
2392 get_addr_base_and_unit_offset only if positive and subtract it
2393 before division by eltsize rather than after it.
2396 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
2397 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
2398 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
2402 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
2403 expand_binop with *mul_widen_optab, make sure at least one of the
2404 operands doesn't have VOIDmode.
2407 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
2409 PR rtl-optimization/82044
2410 PR tree-optimization/82042
2411 * dse.c (record_store): Check for overflow.
2412 (check_mem_read_rtx): Properly check for overflow if width == -1, call
2413 add_wild_read instead of clear_rhs_from_active_local_stores on
2414 overflow and log it into dump_file.
2416 2017-11-22 Richard Biener <rguenther@suse.de>
2418 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
2419 fake edges to exit when looking for a place to insert.
2420 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
2422 (insert_into_preds_of_block): Commit edge insertion immediately,
2423 assert that doesn't require new BBs.
2424 (fini_pre): Release expressions.
2425 (pass_pre::execute): Shuffle things around a bit, if the fn
2426 is too large do not compute AVAIL either as this is really the
2429 2017-11-22 Richard Biener <rguenther@suse.de>
2431 PR tree-optimization/83089
2432 * tree-if-conv.c (pass_if_conversion::execute): If anything
2433 changed reset SCEV and free the number of iteration estimates.
2435 2017-11-21 Martin Sebor <msebor@redhat.com>
2437 PR tree-optimization/82945
2438 * calls.h (warn_nonstring_bound): Remove unused function.
2440 2017-11-21 Martin Sebor <msebor@redhat.com>
2442 PR tree-optimization/82945
2443 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
2444 * calls.h (maybe_warn_nonstring_arg): Declare new function.
2445 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
2447 (initialize_argument_information): Call maybe_warn_nonstring_arg.
2448 * calls.h (get_attr_nonstring_decl): Declare new function.
2449 * doc/extend.texi (attribute nonstring): Update.
2450 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
2451 get_attr_nonstring_decl and handle it.
2452 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
2453 detection of nul-termination.
2454 (strlen_to_stridx): Change to a pointer.
2455 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
2456 (pass_strlen::execute): Same.
2458 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
2460 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
2461 for the new option -mprefer-vector-width=[none|128|256|512].
2462 * config/i386/i386.c (ix86_target_string): remove old style options
2463 -mprefer-avx256 and make -mprefer-avx128 as alias.
2464 (ix86_option_override_internal): Apply defaults for the
2465 -mprefer-vector-width=[128|256] option.
2466 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
2467 Implement macros to work with -mprefer-vector-width=.
2468 * config/i386/i386.opt: Implemented option
2469 -mprefer-vector-width=[none|128|256|512].
2470 * doc/invoke.texi: Documentation for
2471 -mprefer-vector-width=[none|128|256|512].
2473 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
2475 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
2477 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
2479 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
2481 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
2483 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
2485 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
2486 * expr.c (expand_expr_real_2): Likewise.
2487 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
2488 fold_binary_loc): Likewise.
2489 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
2490 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
2491 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
2492 MINUS_EXPR transformations.
2493 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
2494 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
2495 * tree-inline.c (estimate_operator_cost): Likewise.
2496 * tree-pretty-print.c (dump_generic_node, op_code_prio,
2497 op_symbol_code): Likewise.
2498 * tree-vect-stmts.c (vectorizable_operation): Likewise.
2499 * vr-values.c (extract_range_from_binary_expr): Likewise.
2500 * varasm.c (initializer_constant_valid_p_1): Likewise.
2501 * tree.def: New tree code POINTER_DIFF_EXPR.
2503 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
2505 * config/i386/i386.md (*bswap<mode>2_movbe): Add
2506 integer suffix to movbe mnemonic.
2507 (*bswaphi2_movbe): Ditto.
2508 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
2510 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2513 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
2514 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
2515 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
2517 2017-11-21 Martin Liska <mliska@suse.cz>
2519 * tree-inline.c (expand_call_inline): Remove not needed
2522 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
2523 Jakub Jelinek <jakub@redhat.com>
2526 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
2527 Remove static keyword from f variable.
2529 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2531 PR tree-optimization/83086
2532 * gimple-ssa-store-merging.c
2533 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
2534 rather than n.base_addr.
2536 2017-11-21 Martin Liska <mliska@suse.cz>
2538 PR rtl-optimization/82044
2539 PR tree-optimization/82042
2540 * dse.c (check_mem_read_rtx): Check for overflow.
2542 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2544 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
2547 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2549 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
2550 and its corresponding call arg location note.
2552 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2553 Andrew Burgess <andrew.burgess@embecosm.com>
2555 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
2557 (arc_return_slot_offset): Likewise.
2558 (arc_eh_return_address_location): New declaration.
2559 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
2560 (MUST_SAVE_REGISTER): Add exception handler case.
2561 (MUST_SAVE_RETURN_ADDR): Likewise.
2562 (arc_frame_pointer_required): Likewise.
2563 (arc_frame_pointer_needed): New function.
2564 (arc_compute_frame_size): Changed.
2565 (arc_expand_prologue): Likewise.
2566 (arc_expand_epilogue): Likewise.
2567 (arc_initial_elimination_offset): Likewise.
2568 (arc_return_slot_offset): Delete.
2569 (arc_eh_return_address_location): New function.
2570 (arc_builtin_setjmp_frame_value): Likewise.
2571 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
2572 (EH_RETURN_STACKADJ_RTX): Define.
2573 (EH_RETURN_HANDLER_RTX): Likewise.
2574 * config/arc/arc.md (eh_return): Delete.
2576 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2578 * print-tree.h (debug_vec_tree): Remove prototype.
2579 * gdbinit.in (pvt): Remove macro.
2581 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2583 PR tree-optimization/83047
2584 * gimple-ssa-store-merging.c
2585 (imm_store_chain_info::output_merged_store): If the loads with the
2586 same vuse are in different basic blocks, for load_gsi pick a load
2587 location that is dominated by the other loads.
2590 * config/i386/i386.c (ix86_memmodel_check): Start
2591 -Winvalid-memory-model diagnostics with lowercase letter.
2594 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
2595 set in_cold_section_p to the partition containing loc_list->first.
2596 When seeing loc_list->last_before_switch node, update secname and
2597 perform range_across_switch second partition handling only after that.
2600 * run-rtl-passes.c: Include debug.h.
2601 (run_rtl_passes): Call debug_hooks->assembly_start.
2602 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
2606 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
2607 OPTAB_DIRECT in calls to expand_simple_binop.
2609 2017-11-20 David Malcolm <dmalcolm@redhat.com>
2612 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
2614 2017-11-20 David Malcolm <dmalcolm@redhat.com>
2617 * spellcheck.h (best_match::blithely_get_best_candidate): New
2620 2017-11-20 Jakub Jelinek <jakub@redhat.com>
2622 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
2623 with lower case letter.
2625 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
2627 * config/i386/i386.md (bswaphi2): New expander.
2628 (*bswaphi2_movbe): New insn pattern.
2629 (bswaphi -> rorhi pepehole2): New peephole pattern.
2631 2017-11-20 Jeff Law <law@redhat.com>
2633 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
2634 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
2635 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
2636 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
2637 gimple-ssa-evrp-analyze.h.
2639 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
2640 set BB_VISITED here.
2641 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
2643 2017-11-20 Bin Cheng <bin.cheng@arm.com>
2645 * tree-predcom.c: Add general comment on Store-Store chains.
2646 (split_data_refs_to_components): Postpone clearing eliminate_store_p
2648 (get_chain_last_ref_at): Rename into...
2649 (get_chain_last_write_at): ...this.
2650 (get_chain_last_write_before_load): New function.
2651 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
2652 CT_STORE_STORE when write reference is added.
2653 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
2654 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
2655 (initialize_root_vars_store_elim_1): Ditto.
2656 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
2657 definition is created.
2658 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
2659 chain by replacing it with dominant stored value.
2661 2017-11-20 Bin Cheng <bin.cheng@arm.com>
2663 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
2665 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
2667 * vr-values.c (extract_range_from_binary_expr): Use a full range
2670 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
2672 * config/arm/arm.md (R4_REGNUM): Define constant.
2673 (nonsecure_call_internal): Remove r4 clobber.
2674 (nonsecure_call_value_internal): Likewise.
2675 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
2676 clobber and resequence match_operands.
2677 (nonsecure_call_value_reg_thumb1_v5): Likewise.
2678 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
2679 (nonsecure_call_value_reg_thumb2): Likewise.
2681 2017-11-20 Jakub Jelinek <jakub@redhat.com>
2683 PR tree-optimization/78821
2684 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
2685 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
2686 to the address of the base rather than the base itself.
2687 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
2688 (find_bswap_or_nop_finalize): New function.
2689 (find_bswap_or_nop): Use it.
2690 (bswap_replace): Return a tree rather than bool, change first
2691 argument from gimple * to gimple_stmt_iterator, allow inserting
2692 into an empty sequence, allow ins_stmt to be NULL - then emit
2693 all stmts into gsi. Fix up MEM_REF address gimplification.
2694 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
2695 (struct store_immediate_info): Add N and INS_STMT non-static
2697 (store_immediate_info::store_immediate_info): Initialize them
2698 from newly added ctor args.
2699 (merged_store_group::apply_stores): Formatting fixes. Sort by
2701 (stmts_may_clobber_ref_p): For stores call also
2702 refs_anti_dependent_p.
2703 (gather_bswap_load_refs): New function.
2704 (imm_store_chain_info::try_coalesce_bswap): New method.
2705 (imm_store_chain_info::coalesce_immediate_stores): Use it.
2706 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
2707 (imm_store_chain_info::output_merged_store): Fail if number of
2708 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
2709 and NOP_EXPR rhs_code.
2710 (pass_store_merging::process_store): Compute n and ins_stmt, if
2711 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
2712 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
2714 (pass_store_merging::execute): Calculate dominators.
2716 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
2717 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
2718 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
2719 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
2720 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
2721 class pass_optimize_bswap, bswap_replace,
2722 pass_optimize_bswap::execute): Moved to ...
2723 * gimple-ssa-store-merging.c: ... this file.
2724 Include optabs-tree.h.
2725 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
2726 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
2727 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
2728 anonymous namespace, remove static keywords.
2729 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
2730 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
2732 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
2735 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
2737 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
2739 * vec.h (debug_helper): New function.
2740 (DEFINE_DEBUG_VEC): New macro.
2741 * hash-set.h (debug_helper): New function.
2742 (DEFINE_DEBUG_HASH_SET): New macro.
2743 * cfg.c (debug_slim (edge)): New function.
2744 Call DEFINE_DEBUG_VEC for edges.
2745 Call DEFINE_DEBUG_HASH_SET for edges.
2746 * cfghooks.c (debug_slim (basic_block)): New function.
2747 Call DEFINE_DEBUG_VEC for basic blocks.
2748 Call DEFINE_DEBUG_HASH_SET for basic blocks.
2749 * print-tree.c (debug_slim): New function to handle trees.
2750 Call DEFINE_DEBUG_VEC for trees.
2751 Call DEFINE_DEBUG_HASH_SET for trees.
2752 (debug (vec<tree, va_gc>) &): Remove.
2753 (debug (<vec<tree, va_gc>) *): Remove.
2754 * print-rtl.c (debug_slim): New function to handle const_rtx.
2755 Call DEFINE_DEBUG_VEC for rtx_def.
2756 Call DEFINE_DEBUG_VEC for rtx_insn.
2757 Call DEFINE_DEBUG_HASH_SET for rtx_def.
2758 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
2759 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
2760 (debug (vec<rtx_insn *> *ptr): Remove.
2761 (debug_insn_vector): Remove.
2762 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
2764 2017-11-20 Tom de Vries <tom@codesourcery.com>
2766 PR rtl-optimization/82020
2767 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
2768 IF_THEN_ELSE condition.
2770 2017-11-19 Jeff Law <law@redhat.com>
2772 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2773 of degenerates resulting from ignoring an edge.
2775 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2778 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
2780 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2783 * profile-count.c (profile_count::to_sreal_scale): Fix return value
2784 for uninitialied counts.
2786 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2789 * tree-inline.c (estimate_num_insns): Set to 1 at least.
2791 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2794 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
2796 2017-11-19 Tom de Vries <tom@codesourcery.com>
2798 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
2801 2017-11-19 Tom de Vries <tom@codesourcery.com>
2803 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
2806 2017-11-19 Tom de Vries <tom@codesourcery.com>
2808 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
2811 2017-11-19 Tom de Vries <tom@codesourcery.com>
2813 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
2815 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
2817 2017-11-19 Tom de Vries <tom@codesourcery.com>
2819 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
2821 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
2822 semicolon after MCORE_EXPORT_NAME call.
2824 2017-11-19 Tom de Vries <tom@codesourcery.com>
2827 * vmsdbgout.c (vmsdbgout_early_finish): New function.
2828 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
2830 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
2832 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
2833 (cgraph_edge::clone): Cleanup updating of profile.
2834 * ipa-cp.c (update_profiling_info): Likewise.
2835 * ipa-inline-transform.c (inline_transform): Likewise.
2836 * ipa-inline.c (inline_small_functions): Add missing space to dump.
2837 * ipa-split.c (execute_split_functions): Do not split when function
2839 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
2840 * profile-count.c (profile_count::dump): Add global0.
2841 (profile_count::to_cgraph_frequency): Do not ICE when entry is
2843 (profile_count::to_sreal_scale): Likewise.
2844 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
2845 (profile_count::combine_with_ipa_count): New function.
2846 * profile-count.h (profile_guessed_global0adjusted): New.
2847 (profile_count::adjusted_zero): New.
2848 (profile_count::global0adjusted): New.
2849 (profile_count::combine_with_ipa_count): New.
2850 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
2851 correct profile of return block of split functions.
2852 (copy_cfg_body): Remove unused profile_count.
2853 (copy_body): Likewise.
2854 (expand_call_inline): Update.
2855 (tree_function_versioning): Update.
2857 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
2859 * hash-set.h (hash_set::empty): New.
2860 * tree-ssa-threadbackward.h: Delete.
2861 * tree-ssa-threadbackward.c (class thread_jumps): New.
2862 Move max_threaded_paths into class.
2863 (fsm_find_thread_path): Remove arguments that are now in class.
2864 (profitable_jump_thread_path): Rename to...
2865 (thread_jumps::profitable_jump_thread_path): ...this.
2866 (convert_and_register_jump_thread_path): Rename to...
2867 (thread_jumps::convert_and_register_current_path): ...this.
2868 (check_subpath_and_update_thread_path): Rename to...
2869 (thread_jumps::check_subpath_and_update_thread_path): ...this.
2870 (register_jump_thread_path_if_profitable): Rename to...
2871 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
2872 (handle_phi): Rename to...
2873 (thread_jumps::handle_phi): ...this.
2874 (handle_assignment): Rename to...
2875 (thread_jumps::handle_assignment): ...this.
2876 (fsm_find_control_statement_thread_paths): Rename to...
2877 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
2878 (find_jump_threads_backwards): Rename to...
2879 (thread_jumps::find_jump_threads_backwards): ...this.
2880 Initialize path local data.
2881 (pass_thread_jumps::execute): Call find_jump_threads_backwards
2882 from within thread_jumps class.
2883 (pass_early_thread_jumps::execute): Same.
2885 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2887 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
2889 2017-11-17 Steve Ellcey <sellcey@cavium.com>
2891 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
2892 to canonical location.
2894 2017-11-17 Steve Ellcey <sellcey@cavium.com>
2897 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
2899 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
2901 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
2903 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
2904 rcrt1.o%s/grcrt1.o%s for -static-pie.
2906 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2908 * i386.c (ix86_multiplication_cost, ix86_division_cost,
2909 ix86_shift_rotate_cost): Break out from ...
2910 (ix86_rtx_costs): ... here.
2911 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
2914 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2916 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
2917 when entry block was promoted unlikely.
2918 (estimate_bb_frequencies): Increase frequency scale.
2919 * profile-count.h (profile_count): Export precision info.
2921 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2923 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
2924 disturbin profile of entry block.
2926 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2928 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
2931 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2933 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
2934 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
2936 2017-11-17 Jeff Law <law@redhat.com>
2938 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
2939 from evrp_dom_walker class. Various methods moved into new class.
2940 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
2941 (evrp_range_analyzer::enter): New method.
2942 (evrp_range_analyzer::leave): New method.
2943 (evrp_dom_walker): Remove delegators no longer needed by this class.
2944 Replace vr_values data member with evrp_range_analyzer
2946 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
2947 method extracted from evrp_dom_walker::before_dom_children.
2948 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
2949 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
2951 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
2952 Add private copy constructor and move assignment operators.
2953 Privatize methods and class data where trivially possible.
2954 (evrp_dom_walker::cleanup): New function, extracted from
2955 execute_early_vrp. Simplify access to class data.
2957 * vr-values.h (get_output_for_vrp): Prototype.
2958 * vr-values.c (get_output_for_vrp): New function extracted from
2959 vrp_visit_assignment_or_call and extract_range_from_stmt.
2960 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
2962 2017-11-17 Luis Machado <luis.machado@linaro.org>
2964 * config/aarch64/aarch64.c
2965 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
2966 (qdf24xx_tunings) <autoprefetcher_model>: Set to
2967 tune_params::AUTOPREFETCHER_WEAK.
2969 2017-11-17 Tamar Christina <tamar.christina@arm.com>
2972 * config/arm/arm.c (arm_valid_target_attribute_rec):
2973 Parse "arch=" and "+<ext>".
2974 (arm_valid_target_attribute_tree): Re-init global options.
2975 (arm_option_override): Make non-static.
2976 (arm_options_perform_arch_sanity_checks): Make errors fatal.
2977 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
2978 (__ARM_FEATURE_CRC32): Support undef.
2979 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
2980 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
2982 2017-11-17 David Malcolm <dmalcolm@redhat.com>
2984 * gdbinit.in (break-on-diagnostic): New command.
2986 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2988 * config/i386/i386.c (ix86_expand_epilogue): Change simple
2989 return to indirect jump for EH return if control-flow
2990 protection is enabled. Change explicit 'false' argument in
2991 pro_epilogue_adjust_stack with a value of
2993 * config/i386/i386.md (simple_return_indirect_internal):
2994 Remove SImode restriction to support 64-bit.
2996 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
2998 * combine.c (added_notes_insn): New.
2999 (try_combine): Handle added_notes_insn like added_links_insn.
3000 Rewrite return value code.
3001 (distribute_notes): Set added_notes_insn to the earliest insn we added
3004 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3006 PR rtl-optimization/82621
3007 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
3008 dest of one of those SETs is unused.
3010 2017-11-17 Richard Biener <rguenther@suse.de>
3013 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
3014 * tree-pretty-print.c (dump_generic_node): Handle
3015 annot_expr_parallel_kind.
3016 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
3017 * gimplify.c (gimple_boolify): Likewise.
3019 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3021 * config.gcc (extra_headers): Add cet.h for x86 targets.
3022 * config/i386/cet.h: New file.
3023 * doc/install.texi: Add --enable-cet/--disable-cet.
3025 2017-11-17 Richard Biener <rguenther@suse.de>
3027 PR tree-optimization/83017
3028 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
3029 (gen_parallel_loop): Properly count iterations.
3030 (parallelize_loops): Handle loop->can_be_parallel independent
3031 of flag_loop_parallelize_all. Make static profitability test match
3033 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
3034 * invoke.texi (parloops-min-per-thread): Document.
3036 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
3038 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
3039 upstreaming review comments.
3041 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3043 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
3044 with fast unaligned access.
3045 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
3047 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
3049 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
3050 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
3052 2017-11-17 Richard Biener <rguenther@suse.de>
3054 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
3055 folding of references.
3057 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
3060 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
3061 of replacing call to strncmp with corresponding call to strcmp when
3064 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
3066 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
3067 option prefer-avx256 for skylake-avx512 configuration.
3068 * config/i386/i386.c (ix86_option_override_internal): Ditto.
3069 (get_builtin_code_for_version): Ditto.
3071 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3072 Monk Chiang <sh.chiang04@gmail.com>
3074 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
3075 (FIXED_REGISTERS): Reserve more register numbers.
3076 (CALL_USED_REGISTERS): Likewise.
3077 (REG_ALLOC_ORDER): Likewise.
3078 (REG_CLASS_CONTENTS): Likewise.
3079 (REGISTER_NAMES): Likewise.
3081 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3082 Kito Cheng <kito.cheng@gmail.com>
3084 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
3086 * config/nds32/iterators.md: Add vector mode iterators and attributes.
3088 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
3090 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
3091 parameter list for vec_splats.
3093 2017-11-16 Joseph Myers <joseph@codesourcery.com>
3095 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
3097 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
3099 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
3102 * doc/install.texi: Document incompatibility of Perl >=5.6.26
3103 with the required version of automake 1.11.6.
3105 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
3107 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
3109 (DU_C2_3_power9): Correct reservation combinations.
3110 (FP_DIV_power9, VEC_DIV_power9): New.
3111 (power9-alu): Split out rotate/shift...
3112 (power9-rot): ...to here, correct dispatch resource.
3113 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
3115 (power9-fp): Correct latency.
3116 (power9-sdiv): Add div/sqrt resource.
3117 (power9-ddiv): Correct latency, add div/sqrt resource.
3118 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
3119 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
3121 (power9-qpdiv, power9-qpmul): Adjust resource usage.
3123 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
3125 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
3126 switch statement mapping KF built-ins to TF built-ins if we don't
3127 have the proper ISA 3.0 assembler support.
3129 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3131 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
3132 (lower_emutls_function_body): Do not compute it.
3134 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3136 * ipa-split.c (split_bb_info): Turn time to sreal.
3137 (split_point): Likewise.
3138 (dump_split_point): Likewise.
3139 (fine_split_points): Likewise.
3140 (execute_split_functions): Only zero split_bbs; turn time to sreals.
3142 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3144 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
3147 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3149 * predict.c (combine_predictions_for_bb): Preserve zero predicted
3151 (expensive_function_p): Remove useless assert.
3152 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
3154 2017-11-16 Martin Sebor <msebor@redhat.com>
3156 PR tree-optimization/82588
3157 PR tree-optimization/82583
3158 * tree-vrp.c (check_array_ref): Handle flexible array members,
3159 string literals, and inner indices.
3160 (search_for_addr_array): Add detail to diagnostics.
3162 2017-11-16 Nathan Sidwell <nathan@acm.org>
3166 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
3167 (SET_DECL_ASSEMBLER_NAME): Forward to
3168 overwrite_decl_assembler_name.
3169 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
3170 (overwrite_decl_assembler_name): Declare.
3171 * tree.c (overwrite_decl_assembler_name): New.
3172 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
3173 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
3174 (LANG_HOOKS_INITIALIZER): Add it.
3175 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
3176 * langhooks.c (lhd_set_decl_assembler_name): Use
3177 SET_DECL_ASSEMBLER_NAME.
3178 (lhd_overwrite_decl_assembler_name): Default implementation.
3180 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
3181 Jackson Woodruff <jackson.woodruff@arm.com>
3183 PR tree-optimization/71026
3184 * match.pd: Canonicalize constant multiplies in division.
3186 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
3188 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
3189 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
3190 Remove OPT_fomit_frame_pointer entry.
3191 * common/config/arc/arc-common.c: Likewise.
3192 * common/config/arm/arm-common.c: Likewise.
3193 * common/config/avr/avr-common.c: Likewise.
3194 * common/config/c6x/c6x-common.c: Likewise.
3195 * common/config/cr16/cr16-common.c: Likewise.
3196 * common/config/cris/cris-common.c: Likewise.
3197 * common/config/epiphany/epiphany-common.c: Likewise.
3198 * common/config/fr30/fr30-common.c: Likewise.
3199 * common/config/frv/frv-common.c: Likewise.
3200 * common/config/ia64/ia64-common.c: Likewise.
3201 * common/config/iq2000/iq2000-common.c: Likewise.
3202 * common/config/lm32/lm32-common.c: Likewise.
3203 * common/config/m32r/m32r-common.c: Likewise.
3204 * common/config/mcore/mcore-common.c: Likewise.
3205 * common/config/microblaze/microblaze-common.c: Likewise.
3206 * common/config/mips/mips-common.c: Likewise.
3207 * common/config/mmix/mmix-common.c: Likewise.
3208 * common/config/mn10300/mn10300-common.c: Likewise.
3209 * common/config/nios2/nios2-common.c: Likewise.
3210 * common/config/pa/pa-common.c: Likewise.
3211 * common/config/pdp11/pdp11-common.c: Likewise.
3212 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
3213 * common/config/riscv/riscv-common.c: Likewise.
3214 * common/config/rs6000/rs6000-common.c: Likewise.
3215 * common/config/rx/rx-common.c: Likewise.
3216 * common/config/s390/s390-common.c: Likewise.
3217 * common/config/sh/sh-common.c: Likewise.
3218 * common/config/sparc/sparc-common.c: Likewise.
3219 * common/config/tilegx/tilegx-common.c: Likewise.
3220 * common/config/tilepro/tilepro-common.c: Likewise.
3221 * common/config/v850/v850-common.c: Likewise.
3222 * common/config/visium/visium-common.c: Likewise.
3223 * common/config/xstormy16/xstormy16-common.c: Likewise.
3224 * common/config/xtensa/xtensa-common.c: Likewise.
3225 * invoke.texi (-fomit-frame-pointer): Update documentation.
3227 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3229 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
3231 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3233 * cfg.c (scale_bbs_frequencies_int,
3234 cale_bbs_frequencies_gcov_type): Remove.
3235 * cfg.h (scale_bbs_frequencies_int,
3236 cale_bbs_frequencies_gcov_type): Remove.
3238 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3240 * tree-ssa-loop-manip.c
3241 (scale_dominated_blocks_in_loop): Update to profile counts.
3242 (tree_transform_and_unroll_loop): Likewise.
3244 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3246 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
3247 scale_bbs_frequencies_int.
3249 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3251 * final.c (compute_alignments): Use counts rather than frequencies.
3253 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3255 * cfgloopanal.c: Include sreal.h
3256 (average_num_loop_insns): Use counts and sreal for accounting.
3258 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3260 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
3263 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3264 Kito Cheng <kito.cheng@gmail.com>
3266 * config/nds32/constraints.md: Provide more constraints.
3267 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
3268 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
3269 support constraints usage.
3271 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3273 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
3275 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3276 Kito Cheng <kito.cheng@gmail.com>
3278 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
3279 * config/nds32/nds32.opt: Refine the layout.
3280 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
3281 TARGET_EXT_STRING): Support new options.
3282 * config/nds32/nds32.h: Likewise.
3283 * config/nds32/nds32.md: Likewise.
3284 * config/nds32/nds32-predicates.c: Likewise.
3285 * config/nds32/constraints.md: Likewise.
3286 * common/config/nds32/nds32-common.c: Likewise.
3288 2017-11-16 Julia Koval <julia.koval@intel.com>
3291 * config/i386/gfniintrin.h: Add sse check.
3292 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
3294 2017-11-16 Julia Koval <julia.koval@intel.com>
3296 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
3297 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
3298 (ix86_handle_option): Handle -mavx512vbmi2.
3299 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
3300 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3301 * config/i386/i386-c.c (__AVX512VBMI2__): New.
3302 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
3303 (ix86_valid_target_attribute_inner_p): Ditto.
3304 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
3305 * config/i386/i386.opt (mavx512vbmi2): New option.
3306 * doc/invoke.texi: Add new option.
3308 2017-11-16 Julia Koval <julia.koval@intel.com>
3310 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
3311 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
3312 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
3313 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
3314 _mm512_gf2p8mul_epi8): New intrinsics.
3315 * config/i386/i386-builtin-types.def
3316 (V64QI_FTYPE_V64QI_V64QI): New type.
3317 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
3318 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
3319 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
3320 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
3321 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
3322 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3324 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
3326 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
3327 explicitly as a stream of bytes.
3329 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3331 * config/rs6000/altivec.h (vec_xst_be): New #define.
3332 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
3333 and externalize from *altivec_vperm_<mode>_internal.
3334 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
3336 (XL_BE_V8HI): Likewise.
3337 (XL_BE_V4SI): Likewise.
3338 (XL_BE_V4SI): Likewise.
3339 (XL_BE_V2DI): Likewise.
3340 (XL_BE_V4SF): Likewise.
3341 (XL_BE_V2DF): Likewise.
3342 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
3343 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
3344 all array entries with these keys: VSX_BUILTIN_VEC_XL,
3345 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
3346 VSX_BUILTIN_VEC_XST_BE.
3347 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
3348 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
3350 (altivec_init_builtins): Replace conditional calls to def_builtin
3351 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
3352 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
3353 with unconditional calls. Remove calls to def_builtin for
3354 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
3355 __builtin_vec_xst_be.
3356 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
3357 to define_expand, and add alternate RTL generation for P8.
3358 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
3359 vsx_ld_elemrev_v8hi.
3360 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
3361 add alternate RTL generation for P8.
3362 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
3363 vsx_ld_elemrev_v16qi.
3364 (vsx_st_elemrev_v8hi): Convert define_insn
3365 to define_expand, and add alternate RTL generation for P8.
3366 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
3367 vsx_st_elemrev_v8hi.
3368 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
3369 add alternate RTL generation for P8.
3370 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
3371 vsx_st_elemrev_v16qi.
3373 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
3376 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
3377 TARGET_AVX512ER check.
3378 (ix86_option_override_internal): Set MASK_VZEROUPPER if
3379 neither -mzeroupper nor -mno-zeroupper is used and
3380 TARGET_EMIT_VZEROUPPER is set.
3381 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
3382 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
3384 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
3386 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3387 folding of vector compares.
3388 (fold_build_vec_cmp): New helper function.
3389 (fold_compare_helper): New helper function.
3390 (builtin_function_type): Add compare builtins to the list of functions
3391 having unsigned arguments. Cosmetic updates to comment indentation.
3392 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
3393 the not+eq combination.
3395 2017-11-15 Bin Cheng <bin.cheng@arm.com>
3397 PR tree-optimization/82726
3398 PR tree-optimization/70754
3399 * tree-predcom.c (order_drefs_by_pos): New function.
3400 (combine_chains): Move code setting has_max_use_after to...
3401 (try_combine_chains): ...here. New parameter. Sort combined chains
3402 according to position information.
3403 (tree_predictive_commoning_loop): Update call to above function.
3404 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
3406 2017-11-15 Bin Cheng <bin.cheng@arm.com>
3408 PR tree-optimization/82726
3410 2017-01-23 Bin Cheng <bin.cheng@arm.com>
3412 PR tree-optimization/70754
3413 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
3414 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
3415 combined stmt before it if not NULL.
3416 (combine_chains): Process refs reversely and compute dominance point
3420 2017-02-23 Bin Cheng <bin.cheng@arm.com>
3422 PR tree-optimization/79663
3423 * tree-predcom.c (combine_chains): Process refs in reverse order
3424 only for ZERO length chains, and add explaining comment.
3426 2017-11-15 Tamar Christina <tamar.christina@arm.com>
3428 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
3429 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
3430 * doc/invoke.texi (ARM Options): Add armv8.3-a.
3432 2017-11-15 Tamar Christina <tamar.christina@arm.com>
3434 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
3436 2017-11-15 Martin Liska <mliska@suse.cz>
3438 * tree-cfg.c (pass_warn_function_return::execute):
3439 Compare warn_return_type for greater than zero.
3441 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
3445 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
3446 to return true on Xeon and not on Xeon Phi.
3447 (ix86_check_avx256_register): Changed to ...
3448 (ix86_check_avx_upper_register): ... this. Add extra check for
3449 VALID_AVX512F_REG_OR_XI_MODE.
3450 (ix86_avx_u128_mode_needed): Changed
3451 ix86_check_avx256_register to ix86_check_avx_upper_register.
3452 (ix86_check_avx256_stores): Changed to ...
3453 (ix86_check_avx_upper_stores): ... this. Changed
3454 ix86_check_avx256_register to ix86_check_avx_upper_register.
3455 (ix86_avx_u128_mode_after): Changed
3456 avx_reg256_found to avx_upper_reg_found. Changed
3457 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
3458 (ix86_avx_u128_mode_entry): Changed
3459 ix86_check_avx256_register to ix86_check_avx_upper_register.
3460 (ix86_avx_u128_mode_exit): Ditto.
3461 * config/i386/i386.h: (host_detect_local_cpu): New define.
3463 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3465 * config/arm/xgene1.md (xgene1): Split into automatons
3466 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
3467 (xgene1_f_load): Adjust reservations and/or types.
3468 (xgene1_f_store): Likewise.
3469 (xgene1_load_pair): Likewise.
3470 (xgene1_store_pair): Likewise.
3471 (xgene1_fp_load1): Likewise.
3472 (xgene1_load1): Likewise.
3473 (xgene1_store1): Likewise.
3474 (xgene1_move): Likewise.
3475 (xgene1_alu): Likewise.
3476 (xgene1_simd): Likewise.
3477 (xgene1_bfm): Likewise.
3478 (xgene1_neon_load1): Likewise.
3479 (xgene1_neon_store1): Likewise.
3480 (xgene1_neon_logic): Likewise.
3481 (xgene1_neon_st1): Likewise.
3482 (xgene1_neon_ld1r): Likewise.
3483 (xgene1_alu_cond): Added.
3484 (xgene1_shift_reg): Likwise.
3485 (xgene1_bfx): Likewise.
3486 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
3488 2017-11-15 Jakub Jelinek <jakub@redhat.com>
3491 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
3493 (can_widen_mult_without_libcall): New function.
3494 (expand_mul_overflow): If only checking unsigned mul overflow,
3495 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
3496 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
3497 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
3498 (expand_DIVMOD): Formatting fix.
3499 * expmed.h (expand_mult): Add NO_LIBCALL argument.
3500 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
3501 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
3503 PR tree-optimization/82977
3504 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
3505 constructed temporary to strlen_to_stridx.put.
3507 2017-11-15 Martin Liska <mliska@suse.cz>
3509 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
3510 * configure: Regenerate.
3512 2017-11-15 Martin Liska <mliska@suse.cz>
3515 * config/sh/sh-mem.cc: Use proper probability for
3518 2017-11-14 Jeff Law <law@redhat.com>
3520 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
3521 the red zone for stack_clash_protection_final_dynamic_probe targets
3522 when the total dynamic stack size is zero bytes.
3524 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
3525 blocks is post order.
3527 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
3529 * dumpfile.h (TDF_COMPARE_DEBUG): New.
3530 * final.c (rest_of_clean_state): Set it for the
3531 -fcompare-debug dump.
3532 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
3533 class when TDF_COMPARE_DEBUG is set.
3535 * dwarf2out.c (gen_producer_string): Discard
3538 2017-11-15 Joseph Myers <joseph@codesourcery.com>
3541 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
3542 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
3543 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
3544 (__TGMATH_REAL_2_3): Remove macros.
3545 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
3546 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
3548 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
3550 (remquo): Define using __TGMATH_REAL_3.
3552 2017-11-14 Jeff Law <law@redhat.com>
3554 * vr-values.c: New file with contents extracted from tree-vrp.c.
3555 * Makefile.in (OBJS): Add vr-values.o
3556 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
3557 (set_value_range, set_and_canonicalize_value_range): Likewise.
3558 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
3559 (value_range_constant_singleton, symbolic_range_p): Likewise.
3560 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
3561 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
3562 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
3563 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
3564 (find_case_label_range, find_case_label_index): Likewise.
3565 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
3566 (range_int_cst_singleton_p, value_inside_range): Likewise.
3567 (get_single_symbol): Likewise.
3568 (switch_update): Move structure definition here.
3569 (to_remove_edges, to_update_switch_stmts): Provide externs.
3570 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
3571 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
3572 (vrp_val_is_min, set_value_range): Likewise.
3573 (set_and_canonicalize_value_range, copy_value_range): Likewise.
3574 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
3575 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
3576 (range_is_nonnull, range_int_cst_p): Likewwise.
3577 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
3578 (get_single_symbol, operand_less_p): Likewise
3579 (compare_values_warnv, compare_values): Likewise.
3580 (value_inside_range, value_range_constant_singleton): Likewise.
3581 (zero_nonzero_bitgs_from_vr): Likewise.
3582 (extract_range_from_binary_expr_1): Likewise.
3583 (overflow_comparison_p): Likewise.
3584 (to_remove_edges, to_update_switch_stmts): Likewise.
3585 (find_case_label-index, find_case_label_range): Likewise.
3586 (switch_update, set_value_range_to_nonnegative): Remove.
3587 (set_value_range_to_truthvalue): Likewise.
3588 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
3589 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
3590 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
3591 (find_case_label_ranges, test_for_singularity): Likewise.
3592 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
3593 (x_vr_values): Move to its remaining use site.
3595 2017-11-10 Jeff Law <law@redhat.com>
3597 * vr-values.h (VR_INITIALIZER): Move #define here.
3598 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
3599 * Makefile.in (OBJS): Add tree-evrp.o
3600 * tree-vrp.h (assert_info): Move structure definition here.
3601 (set_value_range_to_varying): Prototype.
3602 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
3603 (infer_value_range, register_edge_assert_for): Likewise.
3604 (stmt_interesting_for_vrp): Likewise.
3605 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
3606 (set_value_range_to_varying): No longer static.
3607 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
3608 (infer_value_range, register_edge_assert_for): Likewise.
3610 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
3612 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
3613 generate the XXBRD instruction.
3615 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
3616 (rs6000_builtin_type_compatible): Treat _Float128 and long double
3617 as being compatible if -mabi=ieeelongdouble.
3618 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
3619 to setup float128 built-ins with hardware support.
3620 (BU_FLOAT128_HW_2): Likewise.
3621 (BU_FLOAT128_HW_3): Likewise.
3622 (BU_FLOAT128_HW_VSX_1): Likewise.
3623 (BU_FLOAT128_HW_VSX_2): Likewise.
3624 (scalar_extract_expq): Change float128 built-in functions to
3625 accommodate having both KFmode and TFmode functions. Use the
3626 KFmode variant as the default.
3627 (scalar_extract_sigq): Likewise.
3628 (scalar_test_neg_qp): Likewise.
3629 (scalar_insert_exp_q): Likewise.
3630 (scalar_insert_exp_qp): Likewise.
3631 (scalar_test_data_class_qp): Likewise.
3632 (sqrtf128_round_to_odd): Delete processing the round to odd
3633 built-in functions as special built-in functions, and define them
3634 as float128 built-ins. Use the KFmode variant as the default.
3635 (truncf128_round_to_odd): Likewise.
3636 (addf128_round_to_odd): Likewise.
3637 (subf128_round_to_odd): Likewise.
3638 (mulf128_round_to_odd): Likewise.
3639 (divf128_round_to_odd): Likewise.
3640 (fmaf128_round_to_odd): Likewise.
3641 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
3642 support for KFmode and TFmode xststdcqp calls.
3643 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
3644 point, switch the built-in handlers for the get/set float128
3645 exponent, get float128 mantissa, float128 test built-ins, and the
3646 float128 round to odd built-in functions. Eliminate creating the
3647 float128 round to odd built-in functions as special built-ins.
3648 (rs6000_init_builtins): Eliminate special creation of the float128
3649 round to odd built-in functions.
3650 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
3651 function insns to support both TFmode and KFmode variants.
3652 (xsxsigqp_<mode>): Likewise.
3653 (xsiexpqpf_<mode>): Likewise.
3654 (xsiexpqp_<mode>): Likewise.
3655 (xststdcqp_<mode>): Likewise.
3656 (xststdcnegqp_<mode>): Likewise.
3657 (xststdcqp_<mode>): Likewise.
3659 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3661 * tree-ssa-threadupdate.c (compute_path_counts): Remove
3662 unused path_in_freq_ptr parameter.
3663 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
3665 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3667 * ipa-inline.c (edge_badness): Dump sreal frequency.
3668 (compute_inlined_call_time): Match natural implementaiton ...
3669 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
3670 forgotten division by CGRAPH_FREQ_BASE.
3672 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3674 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
3675 Solaris 11. Update comment.
3676 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
3678 * config/sol2.h (STARTFILE_SPEC): Likewise.
3679 * configure: Regenerate.
3681 2017-11-14 Carl Love <cel@us.ibm.com>
3683 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
3684 le_ and be_ prefixes to swap* variables. Remove
3685 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
3688 2017-11-14 Jason Merrill <jason@redhat.com>
3690 Support GTY((cache)) on hash_map.
3691 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
3692 (ggc_cache_remove): Override it instead of ggc_mx.
3693 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
3694 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
3695 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
3696 (simple_cache_map_traits): Override maybe_mx.
3697 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
3698 (hash_map): Friend gt_cleare_cache.
3699 (gt_cleare_cache): New.
3700 * tree.h (tree_cache_traits): New hash_map traits class.
3701 (tree_cache_map): New typedef.
3703 2017-11-14 Richard Biener <rguenther@suse.de>
3705 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
3706 paramter and handling.
3707 (cleanup_control_flow_bb): Likewise.
3708 (cleanup_control_flow_pre): New helper performing a DFS walk
3709 to call cleanup_control_flow_bb in PRE order.
3710 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
3711 via cleanup_control_flow_pre.
3713 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
3715 * config/aarch64/aarch64-simd.md
3716 (aarch64_simd_bsl<mode>_internal): Remove DImode.
3717 (*aarch64_simd_bsl<mode>_alt): Likewise.
3718 (aarch64_simd_bsldi_internal): New.
3719 (aarch64_simd_bsldi_alt): Likewise.
3721 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3723 * tracer.c (better_p): Do not compare frequencies.
3724 * reg-stack.c (better_edge): Likewise.
3725 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
3728 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3730 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
3731 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
3732 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
3733 * lto-streamer-in.c (input_function): Use update_max_bb_count.
3734 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
3735 * predict.c (maybe_hot_frequency_p): Inline to ...
3736 (maybe_hot_count_p): ... here; rewrite to counts.
3737 (counts_to_freqs): Rename to ...
3738 (update_max_bb_count): ... this one.
3739 (expensive_function_p): Use counts.
3740 (estimate_bb_frequencies): Update.
3741 (rebuild_frequencies): Update.
3742 * predict.h (counts_to_freqs): Rename to ...
3743 (update_max_bb_count): ... this one.
3744 * profile.c (compute_branch_probabilities): Add debug info
3745 * tree-inline.c (expand_call_inline): Update debug info.
3746 (optimize_inline_calls): Use update_max_bb_count..
3747 (tree_function_versioning): Use update_max_bb_count..
3748 * value-prof.c (gimple_value_profile_transformations):
3749 Do not use update_max_bb_count.
3751 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3753 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
3754 always use frequencies.
3756 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3758 * bb-reorder.c: Remove frequencies from comments.
3759 (better_edge_p): Use profile counts.
3760 (find_traces): Dump profile counts.
3761 (rotate_loop): Use profile counts.
3762 (find_traces_1_round): Likewise.
3763 (connect_better_edge_p): Use counts instead of probabilities for
3765 (copy_bb_p): Drop early check for non-0 frequency.
3766 (sanitize_hot_paths): Update comments.
3768 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3770 * ipa-split.c (struct split_point): Add count.
3771 (consider_split): Do not compute incoming frequency; compute incoming
3772 count and store it to split_point.
3773 (split_function): Set count of the call to split part correctly.
3775 2017-11-13 Carl Love <cel@us.ibm.com>
3777 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
3779 2017-11-13 Tom Tromey <tom@tromey.com>
3781 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
3783 2017-11-13 Carl Love <cel@us.ibm.com>
3785 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
3786 Add support for builtins:
3787 unsigned int vec_first_{,miss}_match_{,or_eos}index,
3788 vector {un,}signed {char,int,short},
3789 vector {un,}signed {char,int,short}) arguments.
3790 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
3791 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
3792 Add BU_P9V_AV_2 expansions for the builtins.
3793 * config/rs6000/altivec.h (vec_first_match_index,
3794 vec_first_mismatch_index, vec_first_match_or_eos_index,
3795 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
3796 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
3797 new extern declaration.
3798 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
3799 * config/rs6000/vsx.md (first_match_index_<mode>,
3800 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
3801 first_mismatch_or_eos_index_<mode>): Add define expand.
3802 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
3803 * doc/extend.texi: Update the built-in documenation file for the new
3806 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
3808 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
3809 into the min/max operations for _Float<N> and _Float<N>X types.
3811 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
3814 * dwarf2out.c (do_eh_frame): New static variable.
3815 (dwarf2out_begin_prologue): Set it.
3816 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
3818 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3820 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
3822 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
3824 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
3827 * debug.h (dwarf2out_do_eh_frame): Declare.
3828 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
3829 (dwarf2out_do_frame): Use it.
3830 (dwarf2out_do_cfi_asm): Likewise.
3831 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
3832 (dwarf2out_assembly_start): Likewise.
3833 (dwarf2out_begin_prologue): Fix comment.
3834 * toplev.c (compile_file): Always call dwarf2out_frame_finish
3835 if the target needs either debug or unwind DWARF2 info.
3836 * lto-opts.c (lto_write_options): Do not save -fexceptions,
3837 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
3838 -ftrapping-math, -ftrapv and -fwrapv.
3840 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3842 * cgraph.c (cgraph_edge::sreal_frequency): New function.
3843 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
3844 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
3845 (estimate_edge_size_and_time): Likewise.
3846 (ipa_merge_fn_summary_after_inlining): Likewise.
3847 * ipa-inline.c (cgraph_freq_base_rec): Remove.
3848 (compute_uninlined_call_time): Use sreal_frequency.
3849 (compute_inlined_call_time): Likewise.
3850 (ipa_inline): Do not initialize cgraph_freq_base_rec.
3851 * profile-count.c: Include sreal.h.
3852 (profile_count::to_sreal_scale): New.
3853 * profile-count.h: Forward declare sreal.
3854 (profile_count::to_sreal_scale): Declare.
3856 2017-11-13 Nathan Sidwell <nathan@acm.org>
3858 * diagnostic.c (maybe_line_and_column): New.
3859 (diagnostic_get_location_text): Use it.
3860 (diagnostic_report_current_module): Likewise.
3861 (test_diagnostic_get_location_text): Add tests.
3863 2017-11-13 Luis Machado <luis.machado@linaro.org>
3865 * doc/md.texi (Specifying processor pipeline description): Fix
3866 incorrect latency for the div instruction example.
3868 2017-11-13 Jakub Jelinek <jakub@redhat.com>
3870 PR tree-optimization/78821
3871 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
3872 that bit_not_p is the same.
3873 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
3874 (split_group): Count precisely bit_not_p bits in each statement.
3875 (invert_op): New function.
3876 (imm_store_chain_info::output_merged_store): Use invert_op to
3877 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
3878 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
3880 2017-11-13 Martin Liska <mliska@suse.cz>
3882 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
3883 (struct source_info): Likewise.
3884 (add_branch_counts): Likewise.
3885 (add_line_counts): Likewise.
3886 (function_summary): Likewise.
3887 (output_intermediate_line): Likewise.
3888 (generate_results): Likewise.
3890 2017-11-13 Martin Liska <mliska@suse.cz>
3892 * gcov.c (struct block_info): Remove typedef for block_t.
3893 (struct line_info): Likewise.
3894 (line_info::has_block): Likewise.
3895 (EXIT_BLOCK): Likewise.
3896 (unblock): Likewise.
3897 (circuit): Likewise.
3898 (get_cycles_count): Likewise.
3899 (process_file): Likewise.
3900 (read_graph_file): Likewise.
3901 (solve_flow_graph): Likewise.
3902 (find_exception_blocks): Likewise.
3903 (add_line_counts): Likewise.
3904 (accumulate_line_info): Likewise.
3905 (output_line_details): Likewise.
3907 2017-11-13 Martin Liska <mliska@suse.cz>
3909 * gcov.c (struct arc_info): Remove typedef for arc_t.
3910 (struct line_info): Likewise.
3911 (add_branch_counts): Likewise.
3912 (output_branch_count): Likewise.
3913 (function_info::~function_info): Likewise.
3914 (circuit): Likewise.
3915 (output_intermediate_line): Likewise.
3916 (read_graph_file): Likewise.
3917 (solve_flow_graph): Likewise.
3918 (find_exception_blocks): Likewise.
3919 (add_line_counts): Likewise.
3920 (accumulate_line_info): Likewise.
3921 (output_line_details): Likewise.
3922 (output_function_details): Likewise.
3924 2017-11-13 Martin Liska <mliska@suse.cz>
3926 * gcov.c (struct function_info): Remove typedef for function_t.
3927 (struct source_info): Likewise.
3928 (source_info::get_functions_at_location): Likewise.
3929 (solve_flow_graph): Likewise.
3930 (find_exception_blocks): Likewise.
3931 (add_line_counts): Likewise.
3932 (output_intermediate_file): Likewise.
3933 (process_file): Likewise.
3934 (generate_results): Likewise.
3935 (release_structures): Likewise.
3936 (read_graph_file): Likewise.
3937 (read_count_file): Likewise.
3938 (accumulate_line_counts): Likewise.
3939 (output_lines): Likewise.
3941 2017-11-13 Martin Liska <mliska@suse.cz>
3943 * gcov.c (function_info::function_info): Remove num_counts
3944 and add vector<gcov_type>.
3945 (function_info::~function_info): Use the vector.
3946 (process_file): Likewise.
3947 (read_graph_file): Likewise.
3948 (read_count_file): Likewise.
3949 (solve_flow_graph): Likewise.
3951 2017-11-13 Martin Liska <mliska@suse.cz>
3953 * gcov.c (function_info::is_artificial): New function.
3954 (process_file): Erase all artificial early.
3955 (generate_results): Skip as all artificial are already
3958 2017-11-13 Martin Liska <mliska@suse.cz>
3960 * gcov.c (read_graph_file): Store to global vector of functions.
3961 (read_count_file): Iterate the vector.
3962 (process_file): Likewise.
3963 (generate_results): Likewise.
3964 (release_structures): Likewise.
3966 2017-11-13 Jakub Jelinek <jakub@redhat.com>
3968 PR tree-optimization/82954
3969 * gimple-ssa-store-merging.c
3970 (imm_store_chain_info::coalesce_immediate_stores): If
3971 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
3973 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
3975 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
3976 Upddate call to ENDIAN_LANE_N.
3977 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
3978 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
3979 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
3980 and use aarch64_endian_lane_rtx.
3981 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
3983 2017-11-12 Tom de Vries <tom@codesourcery.com>
3985 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
3987 2017-11-12 Tom de Vries <tom@codesourcery.com>
3989 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
3991 2017-11-12 Tom de Vries <tom@codesourcery.com>
3993 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
3995 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
3996 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
3997 * defaults.h (ASM_OUTPUT_LABELREF): Same.
3999 2017-11-11 Martin Sebor <msebor@redhat.com>
4002 * doc/extend.texi (attribute nonstring): Remove spurious argument.
4005 * prefic.c (translate_name): Replace strncpy with memcpy to
4006 avoid -Wstringop-truncation.
4008 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4010 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
4012 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4014 * predict.c (maybe_hot_frequency_p): Do not use cfun.
4016 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4018 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
4021 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
4023 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
4024 enable generating XXBRH if the value is in a vector register.
4025 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
4026 value is in a vector register.
4027 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
4028 register to register bswap64's instead of doing the GPR sequence
4029 used on previous machines.
4030 (bswapdi2_xxbrd): New insn.
4031 (bswapdi2_reg): Disallow on ISA 3.0.
4032 (register to register bswap64 splitter): Do not split the insn on
4033 ISA 3.0 systems that use XXBRD.
4035 2017-11-10 Martin Sebor <msebor@redhat.com>
4038 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
4040 (find_subframework_file): Same.
4042 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4044 * auto-profile.c (afdo_indirect_call): Drop frequency.
4045 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
4046 (cgraph_node::create_edge): Drop frequency argument.
4047 (cgraph_node::create_indirect_edge): Drop frequency argument.
4048 (cgraph_edge::make_speculative): Drop frequency arguments.
4049 (cgraph_edge::resolve_speculation): Do not update frequencies
4050 (cgraph_edge::dump_edge_flags): Do not dump frequency.
4051 (cgraph_node::dump): Check consistency in IPA mode.
4052 (cgraph_edge::maybe_hot_p): Use IPA counter.
4053 (cgraph_edge::verify_count_and_frequency): Rename to ...
4054 (cgraph_edge::verify_count): ... this one; drop frequency checking.
4055 (cgraph_node::verify_node): Update.
4056 * cgraph.h (struct cgraph_edge): Drop frequency.
4057 (cgraph_edge::frequency): New function.
4058 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
4060 (cgraph_edge::rebuild_edges): Likewise.
4061 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
4062 (duplicate_thunk_for_node): Do not pass frequency.
4063 (cgraph_node::create_clone): Scale only counts.
4064 (cgraph_node::create_virtual_clone): Do not pass frequency.
4065 (cgraph_node::create_edge_including_clones): Do not pass frequency.
4066 (cgraph_node::create_version_clone): Do not pass frequency.
4067 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
4068 (cgraph_node::expand_thunk): Do not pass frequency.
4069 (cgraph_node::create_wrapper): Do not pass frequency.
4070 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
4072 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
4073 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
4074 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
4075 (gather_caller_stats): Use frequency function.
4076 (ipcp_cloning_candidate_p): Use frequency function.
4077 (ipcp_propagate_stage): Use frequency function.
4078 (get_info_about_necessary_edges): Use frequency function.
4079 (update_profiling_info): Update only IPA profile.
4080 (update_specialized_profile): Use frequency functoin.
4081 (perhaps_add_new_callers): Update only IPA profile.
4082 * ipa-devirt.c (ipa_devirt): Use IPA profile.
4083 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
4084 (dump_ipa_call_summary): Use frequency function.
4085 (estimate_edge_size_and_time): Use frequency function.
4086 (ipa_merge_fn_summary_after_inlining): Use frequency function.
4087 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
4088 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
4089 (update_noncloned_counts): ... ths one; scale counts only.
4090 (clone_inlined_nodes): Do not scale frequency.
4091 (inline_call): Do not pass frequency.
4092 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
4093 (compute_inlined_call_time): Use IPA profile.
4094 (want_inline_small_function_p): Use IPA profile.
4095 (want_inline_self_recursive_call_p): Use IPA profile.
4096 (edge_badness): Use IPA profile.
4097 (lookup_recursive_calls): Use IPA profile.
4098 (recursive_inlining): Do not pass frequency.
4099 (resolve_noninline_speculation): Do not update frequency.
4100 (inline_small_functions): Collect max of IPA profile.
4101 (dump_overall_stats): Dump IPA porfile.
4102 (dump_inline_stats): Dump IPA porfile.
4103 (ipa_inline): Collect IPA stats.
4104 * ipa-inline.h (clone_inlined_nodes): Update prototype.
4105 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
4106 (ipa_propagate_frequency): Use frequency function.
4107 (ipa_profile): Cleanup.
4108 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
4109 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
4110 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
4111 (input_node): Do not stream frequency.
4112 (input_edge): Do not stream frequency.
4113 (merge_profile_summaries): Scale only IPA profiles.
4114 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
4115 * predict.c (drop_profile): Do not recompute frequency.
4116 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
4117 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
4118 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
4119 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
4120 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
4121 * tree-inline.c (copy_bb): Do not scale frequency.
4122 (expand_call_inline): Do not scale frequency.
4123 (tree_function_versioning): Do not scale frequency.
4124 * ubsan.c (ubsan_create_edge): Do not pass frequency.
4126 2017-11-10 Julia Koval <julia.koval@intel.com>
4128 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
4129 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
4130 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
4131 (_mm256_maskz_gf2p8affine_epi64_epi8)
4132 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
4133 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
4134 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
4135 (__builtin_ia32_vgf2p8affineqb_v32qi)
4136 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
4137 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
4139 2017-11-10 Tamar Christina <tamar.christina@arm.com>
4143 (arm_option_override): Refactor.
4144 (arm_option_reconfigure_globals): New.
4145 (arm_options_perform_arch_sanity_checks): New.
4146 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
4148 (arm_options_perform_arch_sanity_checks): Likewise
4150 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
4152 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
4153 (power9-qpmul): New.
4154 * rs6000/rs6000.md ("type" attr): Add qmul.
4155 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
4156 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
4157 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
4159 2017-11-10 Martin Sebor <msebor@redhat.com>
4162 * builtins.c (compute_objsize): Handle arrays that
4163 compute_builtin_object_size likes to fail for. Make extern.
4164 * builtins.h (compute_objsize): Declare.
4165 (check_strncpy_sizes): New function.
4166 (expand_builtin_strncpy): Call check_strncpy_sizes.
4167 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
4168 -Wstringop-truncation.
4169 (gimple_fold_builtin_strncat): Same.
4170 * gimple.c (gimple_build_call_from_tree): Set call location.
4171 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
4172 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
4173 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
4174 (handle_builtin_strlen): Use strlen_to_stridx.
4175 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
4177 Use strlen_to_stridx.
4178 (pass_strlen::execute): Release strlen_to_stridx.
4179 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
4180 (-Wstringop-truncation): Document new option.
4182 2017-11-10 Martin Liska <mliska@suse.cz>
4184 PR gcov-profile/82702
4185 * gcov.c (main): Handle intermediate files in a different
4187 (get_gcov_intermediate_filename): New function.
4188 (output_gcov_file): Remove support of intermediate files.
4189 (generate_results): Allocate intermediate file.
4190 (release_structures): Clean-up properly fn_end.
4191 (output_intermediate_file): Start iterating with line 1.
4193 2017-11-10 Jakub Jelinek <jakub@redhat.com>
4195 PR tree-optimization/82929
4196 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
4197 ops_swapped_p non-static data member.
4198 (store_immediate_info::store_immediate_info): Clear it.
4199 (imm_store_chain_info::coalesce_immediate_stores): If swapping
4200 ops set ops_swapped_p.
4201 (count_multiple_uses): Handle ops_swapped_p.
4203 2017-11-10 Martin Liska <mliska@suse.cz>
4205 * coverage.c (coverage_init): Stream information about
4206 support of has_unexecuted_blocks.
4207 * doc/gcov.texi: Document that.
4208 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
4209 * gcov.c (read_graph_file): Likewise.
4210 (output_line_beginning): Fix a small issue with
4213 2017-11-10 Bin Cheng <bin.cheng@arm.com>
4215 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
4216 reference of trivial component.
4218 2017-11-10 Jakub Jelinek <jakub@redhat.com>
4221 * gimple-ssa-store-merging.c
4222 (pass_store_merging::terminate_all_aliasing_chains): For
4223 gimple_store_p stmts also call refs_output_dependent_p.
4225 PR rtl-optimization/82913
4226 * compare-elim.c (try_merge_compare): Punt if def_insn is not
4229 2017-11-09 Jeff Law <law@redhat.com>
4231 * vr-values.h: New file with vr_values class.
4232 * tree-vrp.c: Include vr-values.h
4233 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
4234 data objects into the vr_values class.
4235 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
4236 (get_value_range): Make it a member function within vr_values class.
4237 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
4238 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
4239 (op_with_constant_singleton_value_range): Likewise.
4240 (extract_range_for_var_from_comparison_expr): Likewise.
4241 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
4242 (extract_range_from_binary_expr): Likewise.
4243 (extract_range_from_unary_expr): Likewise.
4244 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
4245 (check_for_binary_op_overflow, extract_range_basic): Likewise.
4246 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
4247 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
4248 (compare_name_with_value, compare_names): Likewise.
4249 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
4250 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
4251 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
4252 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
4253 (extract_range_from_phi_node): Likewise.
4254 (simplify_truth_ops_using_ranges): Likewise.
4255 (simplify_div_or_mod_using_ranges): Likewise.
4256 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
4257 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
4258 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
4259 (simplify_float_conversion_using_ranges): Likewise.
4260 (simplify_internal_call_using_ranges): Likewise.
4261 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
4262 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
4263 poitner via x_vr_values for calls into gimple folder.
4264 (vrp_initialize_lattice): Make this the vr_values ctor.
4265 (vrp_free_lattice): Make this the vr_values dtor.
4266 (set_vr_value): New function.
4267 (class vrp_prop): Add vr_values data member. Add various member
4268 functions as well as member functions that delegate to vr_values.
4269 (check_array_ref): Make a member function within vrp_prop class.
4270 (search_for_addr_array, vrp_initialize): Likewise.
4271 (vrp_finalize): Likewise. Revamp to avoid direct access to
4272 vr_value, values_propagated, etc.
4273 (check_array_bounds): Extract vrp_prop class instance pointer from
4274 walk info structure. Use it to call member functions.
4275 (check_all_array_refs): Make a member function within vrp_prop class.
4276 Smuggle class instance pointer via walk info structure.
4277 (x_vr_values): New local static.
4278 (vrp_valueize): Use x_vr_values to get class instance.
4279 (vr_valueize_1): Likewise.
4280 (class vrp_folder): Add vr_values data member. Add various member
4281 functions as well as member functions that delegate to vr_values.
4282 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
4283 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
4284 class instance from vr_values. Use it to call member functions.
4285 (vrp_dom_walker): Add vr_values data member.
4286 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
4287 instance via x_vr_values.
4288 (identify_jump_threads): Accept vr_values as argument. Store
4289 it into the walker structure.
4290 (evrp_dom_walker): Add vr_values class data member. Add various
4292 (evrp_dom_walker::try_find_new_range): Use vr_values data
4293 member to access the memory allocator.
4294 (evrp_dom_walker::before_dom_children): Store vr_values class
4295 instance into the vrp_folder class.
4296 (evrp_dom_walker::push_value_range): Rework to avoid direct
4297 access to num_vr_values and vr_value.
4298 (evrp_dom_walker::pop_value_range): Likewise.
4299 (execute_early_vrp): Remove call to vrp_initialize_lattice.
4300 Use vr_values to get to dump_all_value_ranges member function.
4301 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
4302 and simplify_cond_using_ranges_2 via vrp_prop class instance.
4303 Pass vr_values class instance down to identify_jump_threads.
4304 Remove call to vrp_free_lattice.
4305 (debug_all_value_ranges): Remove.
4307 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
4308 (vrp_folder): Likewise.
4310 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
4311 Get it from the existing bitmap instead.
4312 (vrp_intersect_ranges_1): Likewise.
4314 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4316 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
4318 (store_immediate_info::store_immediate_info): Add bitnotp argument,
4319 set bit_not_p to it.
4320 (imm_store_chain_info::coalesce_immediate_stores): Break group
4321 if bit_not_p is different.
4322 (count_multiple_uses, split_group,
4323 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
4324 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
4325 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
4326 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
4327 to store_immediate_info ctor.
4329 2017-11-09 Jim Wilson <jimw@sifive.com>
4331 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
4332 (scan_prog_file): Likewise.
4334 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
4336 * bb-reorder.c (max_entry_frequency): Remove.
4337 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
4338 connect_traces, push_to_next_round_p): Remove prototypes.
4339 (find_traces_1_round): Use counts only.
4340 (push_to_next_round_p): Likewise.
4341 (find_traces): Likewise.
4342 (rotate_loop): Likewise.
4343 (find_traces_1_round): Likewise.
4344 (connect_traces): Likewise.
4345 (edge_order): Likewise.
4347 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
4349 * config/arm/arm.c (output_return_instruction): Add comments to
4350 indicate requirement for cmse_nonsecure_entry return to account
4351 for the size of clearing instruction output here.
4352 (thumb_exit): Likewise.
4353 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
4354 return in hardfloat mode.
4356 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
4358 * config/rs6000/rs6000.c (machine_function): Add a bool,
4359 "toc_is_wrapped_separately".
4360 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
4361 if it wasn't explicitly set or unset, we are optimizing for speed, and
4362 doing separate shrink-wrapping.
4363 (rs6000_get_separate_components): Enable the TOC component if
4364 saving the TOC register in the prologue.
4365 (rs6000_components_for_bb): Handle the TOC component.
4366 (rs6000_emit_prologue_components): Store the TOC register where needed.
4367 (rs6000_set_handled_components): Mark TOC as handled, if handled.
4368 (rs6000_emit_prologue): Don't save the TOC if that is already done.
4370 2017-11-09 Martin Jambor <mjambor@suse.cz>
4372 * ipa-param-manipulation.c: New file.
4373 * ipa-param-manipulation.h: Likewise.
4374 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
4375 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
4376 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
4377 (ipa_parm_adjustment): Likewise.
4378 (ipa_parm_adjustment_vec): Likewise.
4379 (ipa_get_vector_of_formal_parms): Moved declaration to
4380 ipa-param-manipulation.h.
4381 (ipa_get_vector_of_formal_parm_types): Likewise.
4382 (ipa_modify_formal_parameters): Likewise.
4383 (ipa_modify_call_arguments): Likewise.
4384 (ipa_combine_adjustments): Likewise.
4385 (ipa_dump_param_adjustments): Likewise.
4386 (ipa_modify_expr): Likewise.
4387 (ipa_get_adjustment_candidate): Likewise.
4388 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
4389 ipa-param-manipulation.c.
4390 (ipa_get_vector_of_formal_parm_types): Likewise.
4391 (ipa_modify_formal_parameters): Likewise.
4392 (ipa_modify_call_arguments): Likewise.
4393 (ipa_modify_expr): Likewise.
4394 (get_ssa_base_param): Likewise.
4395 (ipa_get_adjustment_candidate): Likewise.
4396 (index_in_adjustments_multiple_times_p): Likewise.
4397 (ipa_combine_adjustments): Likewise.
4398 (ipa_dump_param_adjustments): Likewise.
4399 * tree-sra.c: Also include ipa-param-manipulation.h
4400 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
4403 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4404 Alan Hayward <alan.hayward@arm.com>
4405 David Sherwood <david.sherwood@arm.com>
4407 * doc/sourcebuild.texi (vect_masked_store): Document.
4409 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4410 Alan Hayward <alan.hayward@arm.com>
4411 David Sherwood <david.sherwood@arm.com>
4413 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
4415 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4416 Alan Hayward <alan.hayward@arm.com>
4417 David Sherwood <david.sherwood@arm.com>
4419 * doc/sourcebuild.texi (vect_variable_length): Document.
4421 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4422 Alan Hayward <alan.hayward@arm.com>
4423 David Sherwood <david.sherwood@arm.com>
4425 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
4427 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4428 Alan Hayward <alan.hayward@arm.com>
4429 David Sherwood <david.sherwood@arm.com>
4431 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
4433 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4434 Alan Hayward <alan.hayward@arm.com>
4435 David Sherwood <david.sherwood@arm.com>
4437 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
4438 previously undocumented selectors.
4439 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
4441 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4443 * doc/rtl.texi (const_vector): Say that elements can be
4444 const_wide_ints too.
4445 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
4446 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
4447 (gen_vec_duplicate): Use it instead of CONSTANT_P.
4448 * optabs.c (expand_vector_broadcast): Likewise.
4450 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4451 Alan Hayward <alan.hayward@arm.com>
4452 David Sherwood <david.sherwood@arm.com>
4454 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
4455 scaled index even if the unscaled address was invalid.
4456 Don't increase the complexity of using a scale in that case.
4458 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4459 Alan Hayward <alan.hayward@arm.com>
4460 David Sherwood <david.sherwood@arm.com>
4462 * doc/rtl.texi: Rewrite the subreg rules so that they partition
4463 the inner register into REGMODE_NATURAL_SIZE bytes rather than
4464 UNITS_PER_WORD bytes.
4465 * emit-rtl.c (validate_subreg): Divide subregs into blocks
4466 based on REGMODE_NATURAL_SIZE of the inner mode.
4467 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
4468 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
4469 * expmed.c (lowpart_bit_field_p): Divide the value up into
4470 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
4471 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
4472 whether something is likely to occupy more than one register.
4474 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
4477 * ipa-inline-transform.c (update_noncloned_frequencies): Use
4478 profile_count::adjust_for_ipa_scaling.
4479 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
4480 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
4482 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
4484 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4486 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
4487 (split_group): Add total_orig and total_new arguments, estimate the
4488 number of statements related to the store group without store merging
4489 and with store merging.
4490 (imm_store_chain_info::output_merged_store): Adjust split_group
4491 callers, punt if estimated number of statements with store merging
4492 is not smaller than estimated number of statements without it.
4494 (handled_load): Remove has_single_use checks.
4495 (pass_store_merging::process_store): Likewise.
4497 2017-11-09 Richard Biener <rguenther@suse.de>
4499 PR tree-optimization/82902
4500 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
4502 2017-11-09 Martin Liska <mliska@suse.cz>
4505 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
4508 2017-11-09 Martin Liska <mliska@suse.cz>
4510 PR tree-optimization/82669
4511 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
4513 2017-11-09 Martin Liska <mliska@suse.cz>
4515 PR gcov-profile/48463
4516 * coverage.c (coverage_begin_function): Output also end locus
4517 of a function and information whether the function is
4519 * gcov-dump.c (tag_function): Parse and print the information.
4520 * gcov.c (INCLUDE_MAP): Add include.
4521 (INCLUDE_SET): Likewise.
4522 (struct line_info): Move earlier in the source file because
4523 of vector<line_info> in function_info structure.
4524 (line_info::line_info): Likewise.
4525 (line_info::has_block): Likewise.
4526 (struct source_info): Add new member index.
4527 (source_info::get_functions_at_location): New function.
4528 (function_info::group_line_p): New function.
4529 (output_intermediate_line): New function.
4530 (output_intermediate_file): Use the mentioned function.
4531 (struct function_start): New.
4532 (struct function_start_pair_hash): Likewise.
4533 (process_file): Add code that identifies group functions.
4534 Assign lines either to global or function scope.
4535 (generate_results): Skip artificial functions.
4536 (find_source): Assign index for each source file.
4537 (read_graph_file): Read new flag artificial and end_line.
4538 (add_line_counts): Assign it either to global of function scope.
4539 (accumulate_line_counts): Isolate core of the function to
4540 accumulate_line_info and call it for both function and global
4542 (accumulate_line_info): New function.
4543 (output_line_beginning): Fix GNU coding style.
4544 (print_source_line): New function.
4545 (output_line_details): Likewise.
4546 (output_function_details): Likewise.
4547 (output_lines): Iterate both source (global) scope and function
4549 (struct function_line_start_cmp): New class.
4550 * doc/gcov.texi: Reflect changes in documentation.
4552 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4555 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
4556 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
4557 and similarly for not instead of neg.
4559 2017-11-08 Andi Kleen <ak@linux.intel.com>
4561 * config/i386/i386.opt: Add -mforce-indirect-call.
4562 * config/i386/predicates.md: Check for flag_force_indirect_call.
4563 * doc/invoke.texi: Document -mforce-indirect-call
4565 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
4567 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
4569 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
4570 config/riscv/riscv.c (predict.h): New include.
4571 (riscv_slow_unaligned_access_p): No longer static.
4572 (riscv_block_move_straight): Add require.
4573 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
4575 2017-11-08 Jakub Jelinek <jakub@redhat.com>
4578 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
4579 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
4580 nonimmediate_operand predicate for operand 1 instead of
4583 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4585 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
4587 * config/aarch64/constraints.md (Uml): New constraint.
4588 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
4591 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4593 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
4594 of two vec_duplicates into a vec_concat.
4596 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4598 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
4599 Simplify vec_merge of vec_duplicate and vec_concat.
4600 * config/aarch64/constraints.md (Utq): New constraint.
4601 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
4604 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4606 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
4607 Simplify vec_merge of vec_duplicate and const_vector.
4608 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
4610 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
4611 mode iterator. Update predicate on operand 1 to
4612 handle non-const_vec constants. Delete constraints.
4613 (*aarch64_combinez_be<mode>): Likewise for operand 2.
4615 2017-11-08 Jakub Jelinek <jakub@redhat.com>
4617 PR tree-optimization/78821
4618 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
4620 (store_operand_info::store_operand_info): Initialize it to false.
4621 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
4622 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
4623 store in the group, and if chain_info is non-NULL, to ignore altogether
4625 (compatible_load_p): Fail if bit_not_p does not match.
4626 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
4627 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
4628 (pass_store_merging::process_store): Adjust
4629 terminate_all_aliasing_chains calls to pass NULL in all current spots,
4630 call terminate_all_aliasing_chains newly when adding a store into
4631 a chain with non-NULL chain_info.
4633 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
4635 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
4637 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
4639 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4641 (aarch64_layout_frame): Initialise emit_frame_chain.
4642 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
4643 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
4645 2017-11-08 Martin Liska <mliska@suse.cz>
4647 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
4648 of gimple_call_internal_p.
4650 2017-11-07 Tom de Vries <tom@codesourcery.com>
4652 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
4654 2017-11-07 Tom de Vries <tom@codesourcery.com>
4656 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
4659 2017-11-08 Martin Liska <mliska@suse.cz>
4662 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
4664 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
4666 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
4668 2017-11-07 Jakub Jelinek <jakub@redhat.com>
4671 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
4672 (*cmp<mode>_ccz_1): New insn with $k alternative.
4675 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
4676 RTX_COMM_COMPARE as commutative as well.
4677 (ix86_binary_operator_ok): Formatting fix.
4678 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
4679 *<code><mode>3<mask_name><round_saeonly_name>,
4680 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
4681 *<s>mul<mode>3_highpart<mask_name>,
4682 *vec_widen_umult_even_v16si<mask_name>,
4683 *vec_widen_umult_even_v8si<mask_name>,
4684 *vec_widen_umult_even_v4si<mask_name>,
4685 *vec_widen_smult_even_v16si<mask_name>,
4686 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
4687 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
4688 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
4689 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
4690 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
4691 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
4692 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
4693 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
4694 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
4695 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
4696 ix86_binary_operator_ok. Formatting fixes.
4697 (*<plusminus_insn><mode>3<mask_name><round_name>,
4698 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
4701 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
4703 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
4704 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
4705 eq and ne if TARGET_ISEL.
4706 (cmp): New code_iterator.
4707 (UNS, UNSU_, UNSIK): New code_attrs.
4708 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
4709 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
4711 ("eq<mode>3"): ... this.
4712 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
4714 ("ne<mode>3"): ... this.
4716 2017-11-07 Julia Koval <julia.koval@intel.com>
4719 * common/config/i386/i386-common.c
4720 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
4721 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
4722 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
4723 * config/i386/i386.opt: Ditto.
4724 * config/i386/i386.c (ix86_target_string): Ditto.
4725 (ix86_option_override_internal): Ditto.
4726 (ix86_init_mpx_builtins): Move MPX to args2.
4727 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
4728 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
4729 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
4730 __builtin_ia32_vgf2p8affineinvqb_v32qi,
4731 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
4732 __builtin_ia32_vgf2p8affineinvqb_v16qi,
4733 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
4735 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
4738 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
4739 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
4740 (zero-extendsidi peephole2): Remove peephole.
4742 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
4745 * stor-layout.c: Include attribs.h.
4746 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
4747 explicit lookup of "aligned" attribute.
4749 2017-11-07 Andrew Waterman <andrew@sifive.com>
4751 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
4752 (riscv_expand_block_move): Likewise.
4753 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
4755 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
4756 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
4757 * config/riscv/riscv.c (riscv_block_move_straight): New function.
4758 (riscv_adjust_block_mem): Likewise.
4759 (riscv_block_move_loop): Likewise.
4760 (riscv_expand_block_move): Likewise.
4761 * config/riscv/riscv.md (movmemsi): New pattern.
4763 2017-11-07 Michael Clark <michaeljclark@mac.com>
4765 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
4766 (MUSL_DYNAMIC_LINKER): Likewise.
4768 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
4770 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
4771 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
4774 2017-11-07 Richard Biener <rguenther@suse.de>
4776 * match.pd: Fix build.
4778 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
4779 Jackson Woodruff <jackson.woodruff@arm.com>
4781 PR tree-optimization/71026
4782 * match.pd: Canonicalize negate in division.
4784 2017-11-07 Sudakshina Das <sudi.das@arm.com>
4787 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
4789 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
4791 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
4792 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
4794 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
4796 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
4797 non-scalar integral types.
4798 * match.pd (negate_expr_p): Handle MINUS_EXPR.
4799 (-(A-B), -(~A)): New transformations.
4801 2017-11-07 Tom de Vries <tom@codesourcery.com>
4803 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
4804 semicolon after "do {} while (0)".
4805 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4806 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4807 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4808 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4809 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4811 2017-11-07 Tom de Vries <tom@codesourcery.com>
4813 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
4814 after "do {} while (0)".
4815 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4816 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4817 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4818 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4819 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4821 2017-11-07 Tom de Vries <tom@codesourcery.com>
4824 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
4826 (arm_rtx_costs_internal): Add missing semicolon after
4827 HANDLE_NARROW_SHIFT_ARITH call.
4829 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
4831 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
4832 disable isel if it was not set explicitly.
4834 2017-11-06 James Bowman <james.bowman@ftdichip.com>
4836 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
4837 (add_type_attribute) likewise.
4839 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
4841 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
4842 of struct ix86_frame.
4843 (ix86_initial_elimination_offset): Likewise.
4844 (ix86_expand_split_stack_prologue): Likewise.
4846 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
4848 * tree-vrp.h (enum value_range_type): Update stale comment.
4850 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4851 Alan Hayward <alan.hayward@arm.com>
4852 David Sherwood <david.sherwood@arm.com>
4854 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
4855 (aarch64_expand_vec_perm_const): Take the number of units too.
4856 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
4857 (aarch64_expand_vec_perm_const): Likewise.
4858 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
4859 (vec_perm<mode>): Update accordingly.
4861 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4862 Alan Hayward <alan.hayward@arm.com>
4863 David Sherwood <david.sherwood@arm.com>
4865 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
4866 Take the number of units too.
4867 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
4868 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
4869 but check for a vector mode before rather than after the call.
4870 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
4871 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
4872 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
4873 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
4874 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
4875 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
4876 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
4877 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
4878 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
4879 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
4880 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
4881 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
4882 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
4883 (aarch64_sqdmull2_n<mode>): Update accordingly.
4885 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4886 Alan Hayward <alan.hayward@arm.com>
4887 David Sherwood <david.sherwood@arm.com>
4889 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
4890 the number of units too.
4891 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
4892 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
4893 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
4894 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
4895 (vec_store_lanesxi<mode>): Update accordingly.
4897 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4898 Alan Hayward <alan.hayward@arm.com>
4899 David Sherwood <david.sherwood@arm.com>
4901 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
4902 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
4903 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
4904 of units rather than the mode.
4905 * config/aarch64/iterators.md (nunits): New mode attribute.
4906 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
4907 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
4908 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
4909 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
4910 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
4911 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
4912 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
4913 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
4914 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
4915 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
4916 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4917 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
4918 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
4919 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
4920 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
4921 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
4922 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
4923 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
4924 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
4925 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
4926 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
4927 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
4928 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
4929 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4930 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
4931 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
4932 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4933 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
4934 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
4935 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
4936 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
4937 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
4938 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
4939 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
4940 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
4941 (aarch64_simd_vec_setv2di): Likewise.
4943 2017-11-06 Carl Love <cel@us.ibm.com>
4945 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
4947 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
4948 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
4950 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
4952 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
4953 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
4954 Add power 8 macro expansions.
4955 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
4956 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
4957 power 8 instructions. (VSX_XXBR): Add iterator.
4959 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
4961 * config/arm/arm.md (predicable_short_it): Change default to "no",
4962 improve documentation, remove uses that are identical to the default.
4963 (enabled_for_depr_it): Rename to enabled_for_short_it.
4964 * gcc/config/arm/arm-fixed.md (predicable_short_it):
4965 Remove default uses.
4966 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
4967 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
4968 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
4969 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
4971 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
4974 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
4975 float128 helper macros, which are no longer used after deleting
4976 the old 'q' built-in functions, and moving the round to odd
4977 built-in functions to being special built-in functions.
4978 (BU_FLOAT128_2): Likewise.
4979 (BU_FLOAT128_1_HW): Likewise.
4980 (BU_FLOAT128_2_HW): Likewise.
4981 (BU_FLOAT128_3_HW): Likewise.
4982 (FABSQ): Delete old 'q' built-in functions.
4983 (COPYSIGNQ): Likewise.
4984 (SQRTF128_ODD): Move round to odd built-in functions to be
4985 special built-in functions, so that we can handle
4986 -mabi=ieeelongdouble.
4987 (TRUNCF128_ODD): Likewise.
4988 (ADDF128_ODD): Likewise.
4989 (SUBF128_ODD): Likewise.
4990 (MULF128_ODD): Likewise.
4991 (DIVF128_ODD): Likewise.
4992 (FMAF128_ODD): Likewise.
4993 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
4994 built-in names to 'f128'.
4995 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
4996 old 'q' built-in functions, as the machine independent code for
4997 'f128' built-in functions handles this.
4998 (rs6000_expand_builtin): Add expansion for float128 round to odd
4999 functions, keying off on -mabi=ieeelongdouble of whether to use
5000 the KFmode or TFmode variant.
5001 (rs6000_init_builtins): Initialize the _Float128 round to odd
5003 * doc/extend.texi (PowerPC Built-in Functions): Document the old
5004 _Float128 'q' built-in functions are now mapped into the new
5005 'f128' built-in functions.
5007 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
5009 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
5010 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
5012 2017-11-06 David Malcolm <dmalcolm@redhat.com>
5015 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
5016 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
5017 * toplev.c: Include "ipa-fnsummary.h".
5018 (toplev::finalize): Call ipa_fnsummary_c_finalize.
5020 2017-11-06 Jakub Jelinek <jakub@redhat.com>
5022 PR tree-optimization/82838
5023 * gimple-ssa-store-merging.c
5024 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
5025 on a separate gimple_seq which is then appended to seq.
5027 2017-11-06 Jeff Law <law@redhat.com>
5030 * config/i386/i386.c (PROBE_INTERVAL): Remove.
5031 (get_probe_interval): New functions.
5032 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
5033 (ix86_adjust_stack_and_probe): Likewise.
5034 (output_adjust_stack_and_probe): Likewise.
5035 (ix86_emit_probe_stack_range): Likewise.
5036 (ix86_expand_prologue): Likewise.
5038 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5040 PR tree-optimization/82816
5041 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
5042 if the modes of the two types are the same.
5043 (convert_plusminus_to_widen): Likewise.
5045 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5047 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
5049 (usadv16qi): New define_expand.
5050 (usadv8hi): New define_expand.
5052 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
5055 * ipa-inline-transform.c (update_noncloned_frequencies): Always
5057 (inline_transform): Likewise.
5058 * predict.c (counts_to_freqs): Remove useless conditional.
5059 * profile-count.h (profile_count::apply_scale): Move sanity check.
5060 * tree-inline.c (copy_bb): Always scale.
5061 (copy_cfg_body): Likewise.
5063 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
5066 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
5068 (*cmp_ite0): Add enabled_for_depr_it attribute.
5069 (*cmp_ite1): Likewise.
5071 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5073 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
5076 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5078 * tree-vrp.c (vrp_int_const_binop): Return true on success and
5079 return the value by pointer.
5080 (extract_range_from_multiplicative_op_1): Update accordingly.
5081 Return as soon as an operation fails.
5083 2017-11-05 Tom de Vries <tom@codesourcery.com>
5086 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
5087 (DEF_SANITIZER_BUILTIN): ... here.
5088 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
5089 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
5091 2017-11-05 Tom de Vries <tom@codesourcery.com>
5094 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
5096 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
5097 ASM_OUTPUT_BEFORE_CASE_LABEL call.
5098 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
5100 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5101 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5102 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5104 2017-11-05 Tom de Vries <tom@codesourcery.com>
5107 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
5110 2017-11-04 Michael Clark <michaeljclark@mac.com>
5112 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
5113 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
5115 (*addsi3_extended): Likewise.
5116 (*addsi3_extended2): Likewise.
5117 (<optab>si3): Likewise.
5118 (<optab>di3): Likewise.
5119 (<optab><mode>3): Likewise.
5120 (<*optabe>si3_internal): Likewise.
5121 (zero_extendqi<SUPERQI:mode>2): Likewise.
5122 (*add<mode>hi3): Likewise.
5123 (*xor<mode>hi3): Likewise.
5124 (<optab>di3): Likewise.
5125 (*<optab>si3_extend): Likewise.
5126 (*sge<u>_<X:mode><GPR:mode>): Likewise.
5127 (*slt<u>_<X:mode><GPR:mode>): Likewise.
5128 (*sle<u>_<X:mode><GPR:mode>): Likewise.
5130 2017-11-04 Andrew Waterman <andrew@sifive.com>
5132 * config/riscv/riscv.c (riscv_option_override): Conditionally set
5133 TARGET_STRICT_ALIGN based upon -mtune argument.
5135 2017-11-04 Andrew Waterman <andrew@sifive.com>
5137 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
5139 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
5141 * config/i386/i386.c (choose_basereg): Use optional scratch
5142 register and add assertion.
5143 (x86_emit_outlined_ms2sysv_save): Use scratch register when
5144 needed, and don't allocate stack.
5145 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
5146 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
5147 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
5149 2017-11-03 Jeff Law <law@redhat.com>
5151 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
5152 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
5153 to probe at the start of a noreturn function.
5155 2017-11-03 Jakub Jelinek <jakub@redhat.com>
5157 PR tree-optimization/78821
5158 * gimple-ssa-store-merging.c: Update the file comment.
5159 (MAX_STORE_ALIAS_CHECKS): Define.
5160 (struct store_operand_info): New type.
5161 (store_operand_info::store_operand_info): New constructor.
5162 (struct store_immediate_info): Add rhs_code and ops data members.
5163 (store_immediate_info::store_immediate_info): Add rhscode, op0r
5164 and op1r arguments to the ctor, initialize corresponding data members.
5165 (struct merged_store_group): Add load_align_base and load_align
5167 (merged_store_group::merged_store_group): Initialize them.
5168 (merged_store_group::do_merge): Update them.
5169 (merged_store_group::apply_stores): Pick the constant for
5170 encode_tree_to_bitpos from one of the two operands, or skip
5171 encode_tree_to_bitpos if neither operand is a constant.
5172 (class pass_store_merging): Add process_store method decl. Remove
5173 bool argument from terminate_all_aliasing_chains method decl.
5174 (pass_store_merging::terminate_all_aliasing_chains): Remove
5175 var_offset_p argument and corresponding handling.
5176 (stmts_may_clobber_ref_p): New function.
5177 (compatible_load_p): New function.
5178 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
5179 if there is overlap and rhs_code is not INTEGER_CST. For
5180 non-overlapping stores terminate group if rhs is not mergeable.
5181 (get_alias_type_for_stmts): Change first argument from
5182 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
5183 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
5184 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
5186 (get_location_for_stmts): Change first argument from
5187 auto_vec<gimple *> & to vec<gimple *> &.
5188 (struct split_store): Remove orig_stmts data member, add orig_stores.
5189 (split_store::split_store): Create orig_stores rather than orig_stmts.
5190 (find_constituent_stmts): Renamed to ...
5191 (find_constituent_stores): ... this. Change second argument from
5192 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
5193 to info structures rather than the statements.
5194 (split_group): Rename ALLOW_UNALIGNED argument to
5195 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
5196 it. Adjust find_constituent_stores caller.
5197 (imm_store_chain_info::output_merged_store): Handle rhs_code other
5198 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
5199 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
5200 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
5201 (mem_valid_for_store_merging): New function.
5202 (handled_load): New function.
5203 (pass_store_merging::process_store): New method.
5204 (pass_store_merging::execute): Use process_store method. Adjust
5205 terminate_all_aliasing_chains caller.
5207 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5209 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
5210 Return true for more constants, symbols and label references.
5211 (aarch64_valid_floating_const): Remove unused function.
5213 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5216 * config/aarch64/aarch64.c (aarch64_layout_frame):
5217 Undo forcing of LR at bottom of frame.
5219 2017-11-03 Jeff Law <law@redhat.com>
5222 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
5223 for int_registers_saved.
5225 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
5226 extracted from tree-ssa-dom.c.
5227 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
5228 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
5229 (record_equivalences_from_incoming_edge): Add additional argument
5230 to single_pred_edge_ignoring_loop_edges call.
5231 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
5232 (uncprop_dom_walker::before_dom_children): Add additional argument
5233 to single_pred_edge_ignoring_loop_edges call.
5234 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
5235 single_pred_edge_ignoring_loop_edges rather than open coding.
5236 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
5238 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
5240 * match.pd (-(-A)): Rewrite.
5242 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5244 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
5245 (rs6000_emit_int_cmove): New declaration.
5246 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
5247 (rs6000_emit_sISEL): Delete.
5248 (rs6000_emit_int_cmove): Make non-static.
5249 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
5250 instead of rs6000_emit_sISEL.
5252 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
5254 * asan.c (create_cond_insert_point): Maintain profile.
5255 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
5257 * basic-block.h (struct basic_block_def): Remove frequency.
5258 (EDGE_FREQUENCY): Use to_frequency
5259 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
5261 (find_traces): Update to use to_frequency.
5262 (find_traces_1_round): Likewise; use only IPA counts.
5263 (bb_to_key): Likewise.
5264 (connect_traces): Use IPA counts only.
5265 (copy_bb_p): Update to use to_frequency.
5266 (fix_up_crossing_landing_pad): Likewise.
5267 (sanitize_hot_paths): Likewise.
5268 * bt-load.c (basic_block_freq): Likewise.
5269 * cfg.c (init_flow): Set count_max to uninitialized.
5270 (check_bb_profile): Remove frequencies; check counts.
5271 (dump_bb_info): Do not dump frequencies.
5272 (update_bb_profile_for_threading): Update counts only.
5273 (scale_bbs_frequencies_int): Likewise.
5274 (MAX_SAFE_MULTIPLIER): Remove.
5275 (scale_bbs_frequencies_gcov_type): Update counts only.
5276 (scale_bbs_frequencies_profile_count): Update counts only.
5277 (scale_bbs_frequencies): Update counts only.
5278 * cfg.h (struct control_flow_graph): Add count-max.
5279 (update_bb_profile_for_threading): Update prototype.
5280 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
5281 (find_many_sub_basic_blocks): Likewise.
5282 * cfgcleanup.c (try_forward_edges): Likewise.
5283 (try_crossjump_to_edge): Likewise.
5284 * cfgexpand.c (expand_gimple_cond): Likewise.
5285 (expand_gimple_tailcall): Likewise.
5286 (construct_init_block): Likewise.
5287 (construct_exit_block): Likewise.
5288 * cfghooks.c (verify_flow_info): Check consistency of counts.
5289 (dump_bb_for_graph): Do not dump frequencies.
5290 (split_block_1): Do not update frequencies.
5291 (split_edge): Do not update frequencies.
5292 (make_forwarder_block): Do not update frequencies.
5293 (duplicate_block): Do not update frequencies.
5294 (account_profile_record): Do not update frequencies.
5295 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
5296 for global heuristics.
5297 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
5298 (expected_loop_iterations_unbounded): Use counts only.
5299 * cfgloopmanip.c (scale_loop_profile): Simplify.
5300 (create_empty_loop_on_edge): Simplify
5302 (duplicate_loop_to_header_edge): Simplify
5303 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
5304 (update_br_prob_note): Take care of removing note when profile
5306 (relink_block_chain): Do not dump frequency.
5307 (rtl_account_profile_record): Use to_frequency.
5308 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
5309 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
5310 (cgraph_update_edges_for_call_stmt_node): Likewise.
5311 (cgraph_edge::verify_count_and_frequency): Update.
5312 (cgraph_node::verify_node): Temporarily disable frequency verification.
5313 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
5314 to_cgraph_frequency.
5315 (cgraph_edge::rebuild_edges): Convert to ipa counts.
5316 * cgraphunit.c (init_lowered_empty_function): Do not initialize
5318 (cgraph_node::expand_thunk): Update profile.
5319 * except.c (dw2_build_landing_pads): Do not update frequency.
5320 * final.c (compute_alignments): Use to_frequency.
5321 (dump_basic_block_info): Do not dump frequency.
5322 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
5323 (dump_gimple_bb_header): Do not dump frequency.
5324 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
5326 * gimple-streamer-in.c (input_bb): Do not stream frequency.
5327 * gimple-streamer-out.c (output_bb): Do not stream frequency.
5328 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
5329 (init_before_recovery): Do not update frequency.
5330 (sched_create_recovery_edges): Do not update frequency.
5331 * hsa-gen.c (convert_switch_statements): Do not update frequency.
5332 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
5333 (ipa_cp_c_finalize): Set max_count to uninitialized.
5334 * ipa-fnsummary.c (get_minimal_bb): Use counts.
5335 (param_change_prob): Use counts.
5336 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
5338 * ipa-split.c (consider_split): Use to_frequency.
5339 (split_function): Use to_frequency.
5340 * ira-build.c (loop_compare_func): Likewise.
5341 (mark_loops_for_removal): Likewise.
5342 (mark_all_loops_for_removal): Likewise.
5343 * loop-doloop.c (doloop_modify): Do not update frequency.
5344 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
5346 * lto-streamer-in.c (input_function): Update count_max.
5347 * omp-expand.c (expand_omp_taskreg): Update count_max.
5348 * omp-simd-clone.c (simd_clone_adjust): Update profile.
5349 * predict.c (maybe_hot_frequency_p): Use to_frequency.
5350 (maybe_hot_count_p): Use ipa counts only.
5351 (maybe_hot_bb_p): Simplify.
5352 (maybe_hot_edge_p): Simplify.
5353 (probably_never_executed): Do not take frequency argument.
5354 (probably_never_executed_bb_p): Do not pass frequency.
5355 (probably_never_executed_edge_p): Likewise.
5356 (combine_predictions_for_bb): Check that profile is nonzero.
5357 (propagate_freq): Do not set frequency.
5358 (drop_profile): Simplify.
5359 (counts_to_freqs): Simplify.
5360 (expensive_function_p): Use to_frequency.
5361 (propagate_unlikely_bbs_forward): Simplify.
5362 (determine_unlikely_bbs): Simplify.
5363 (estimate_bb_frequencies): Add hack to silence graphite issues.
5364 (compute_function_frequency): Use ipa counts.
5365 (pass_profile::execute): Update.
5366 (rebuild_frequencies): Use counts only.
5367 (force_edge_cold): Use counts only.
5368 * profile-count.c (profile_count::dump): Dump new count types.
5369 (profile_count::differs_from_p): Check compatiblity.
5370 (profile_count::to_frequency): New function.
5371 (profile_count::to_cgraph_frequency): New function.
5372 * profile-count.h (struct function): Declare.
5373 (enum profile_quality): Add profile_guessed_local and
5374 profile_guessed_global0.
5375 (class profile_proability): Decrease number of bits to 29;
5376 update from_reg_br_prob_note and to_reg_br_prob_note.
5377 (class profile_count: Update comment; decrease number of bits
5378 to 61. Check compatibility.
5379 (profile_count::compatible_p): New private member function.
5380 (profile_count::ipa_p): New member function.
5381 (profile_count::operator<): Handle global zero correctly.
5382 (profile_count::operator>): Handle global zero correctly.
5383 (profile_count::operator<=): Handle global zero correctly.
5384 (profile_count::operator>=): Handle global zero correctly.
5385 (profile_count::nonzero_p): New member function.
5386 (profile_count::force_nonzero): New member function.
5387 (profile_count::max): New member function.
5388 (profile_count::apply_scale): Handle IPA scalling.
5389 (profile_count::guessed_local): New member function.
5390 (profile_count::global0): New member function.
5391 (profile_count::ipa): New member function.
5392 (profile_count::to_frequency): Declare.
5393 (profile_count::to_cgraph_frequency): Declare.
5394 * profile.c (OVERLAP_BASE): Delete.
5395 (compute_frequency_overlap): Delete.
5396 (compute_branch_probabilities): Do not use compute_frequency_overlap.
5397 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
5398 * sched-ebb.c (rank): Use counts only.
5399 * shrink-wrap.c (handle_simple_exit): Use counts only.
5400 (try_shrink_wrapping): Use counts only.
5401 (place_prologue_for_one_component): Use counts only.
5402 * tracer.c (find_best_predecessor): Use to_frequency.
5403 (find_trace): Use to_frequency.
5404 (tail_duplicate): Use to_frequency.
5405 * trans-mem.c (expand_transaction): Do not update frequency.
5406 * tree-call-cdce.c: Do not update frequency.
5407 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
5408 (gimple_merge_blocks): Likewise.
5409 (gimple_split_edge): Likewise.
5410 (gimple_duplicate_sese_region): Likewise.
5411 (gimple_duplicate_sese_tail): Likewise.
5412 (move_sese_region_to_fn): Likewise.
5413 (gimple_account_profile_record): Likewise.
5414 (insert_cond_bb): Likewise.
5415 * tree-complex.c (expand_complex_div_wide): Likewise.
5416 * tree-eh.c (lower_resx): Update profile.
5417 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
5419 (initialize_cfun): Do not initialize frequencies
5420 (freqs_to_counts): Delete.
5421 (copy_cfg_body): Ignore count parameter.
5422 (copy_body): Update.
5423 (expand_call_inline): Update count_max.
5424 (optimize_inline_calls): Update count_max.
5425 (tree_function_versioning): Update count_max.
5426 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
5427 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
5429 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
5430 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
5431 (try_peel_loop): Likewise.
5432 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
5434 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
5435 (tree_transform_and_unroll_loop): Do not use frequencies
5436 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
5437 Use reliable prediction only.
5438 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
5439 * tree-ssa-sink.c (select_best_block): Use to_frequency.
5440 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
5441 probability scaling.
5442 * tree-ssa-threadupdate.c (create_block_for_threading): Do
5443 not update frequency
5444 (any_remaining_duplicated_blocks): Likewise.
5445 (update_profile): Likewise.
5446 (estimated_freqs_path): Delete.
5447 (freqs_to_counts_path): Delete.
5448 (clear_counts_path): Delete.
5449 (ssa_fix_duplicate_block_edges): Likewise.
5450 (duplicate_thread_path): Likewise.
5451 * tree-switch-conversion.c (gen_inbound_check): Use counts.
5452 * tree-tailcall.c (decrease_profile): Do not update frequency.
5453 (eliminate_tail_call): Likewise.
5454 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
5455 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
5456 (optimize_mask_stores): Likewise.
5457 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
5458 * ubsan.c (ubsan_expand_null_ifn): Update profile.
5459 (ubsan_expand_ptr_ifn): Update profile.
5460 * value-prof.c (gimple_ic): Simplify.
5461 * value-prof.h (gimple_ic): Update prototype.
5462 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
5463 * ipa-inline.c (compute_uninlined_call_time): Be sure that
5465 (want_inline_self_recursive_call_p): Likewise.
5466 (resolve_noninline_speculation): Only cummulate defined counts.
5467 (inline_small_functions): Use nonzero_p.
5468 (ipa_inline): Do not access freed node.
5470 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5472 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5473 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
5475 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
5477 * config/riscv/riscv.c (riscv_legitimize_move): Handle
5478 non-legitimate address.
5480 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5482 * config/rs6000/rs6000.md (*lt0_disi): Delete.
5483 (*lt0_<mode>di, *lt0_<mode>si): New.
5485 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5487 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
5488 TARGET_PAIRED_FLOAT.
5490 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
5491 Jim Wilson <jim.wilson@linaro.org>
5493 * config/aarch64/aarch64-cores.def (saphira): New CPU.
5494 * config/aarch64/aarch64-tune.md: Regenerated.
5495 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
5496 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
5498 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
5500 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
5501 (arc_expand_prologue): Restore blink for millicode.
5502 * config/arc/linux.h (LINK_EH_SPEC): Defined.
5504 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
5507 * config/i386/i386.c (ix86_vector_duplicate_value): Use
5508 gen_vec_duplicate after forcing the scalar into a register.
5510 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
5512 * combine (try_combine): Print the insns input to try_combine to the
5515 2017-11-02 Steve Ellcey <sellcey@cavium.com>
5518 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
5519 Remove second argument from aarch64_process_target_attr call.
5520 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
5522 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
5524 (aarch64_handle_attr_arch): Remove second argument.
5525 (aarch64_handle_attr_cpu): Ditto.
5526 (aarch64_handle_attr_tune): Ditto.
5527 (aarch64_handle_attr_isa_flags): Ditto.
5528 (aarch64_process_one_target_attr): Ditto.
5529 (aarch64_process_target_attr): Ditto.
5530 (aarch64_option_valid_attribute_p): Remove second argument.
5531 on aarch64_process_target_attr call.
5533 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5535 * diagnostic.c: Include "selftest-diagnostic.h".
5536 (selftest::assert_location_text): New function.
5537 (selftest::test_diagnostic_get_location_text): New function.
5538 (selftest::diagnostic_c_tests): Call it.
5540 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5542 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
5543 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
5544 (class selftest::test_diagnostic_context): Move to...
5545 * selftest-diagnostic.c: New file.
5546 * selftest-diagnostic.h: New file.
5548 2017-11-02 James Bowman <james.bowman@ftdichip.com>
5550 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
5551 offset range for FT32B.
5552 * config/ft32/ft32.h: option "mcompress" enables relaxation.
5553 * config/ft32/ft32.md: Add TARGET_NOPM.
5554 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
5555 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
5557 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
5559 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
5561 2017-11-02 Jeff Law <law@redhat.com>
5563 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
5564 virtual keyword on FINAL OVERRIDE members.
5566 * tree-ssa-propagate.h (ssa_propagation_engine): Group
5567 virtuals together. Add virtual destructor.
5568 (substitute_and_fold_engine): Similarly.
5570 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
5572 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
5574 2017-11-02 Richard Biener <rguenther@suse.de>
5576 PR tree-optimization/82795
5577 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
5579 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5581 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
5583 (gcc_GAS_CHECK_FEATURE): Remove.
5584 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
5585 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
5586 * configure: Regenerate.
5588 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
5590 * config/arc/arc.c (hwloop_optimize): Account for empty
5593 2017-11-02 Richard Biener <rguenther@suse.de>
5596 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
5597 Truncate ARRAY_REF index and element size.
5599 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
5601 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
5603 2017-11-01 Jeff Law <law@redhat.com>
5605 * tree-ssa-ccp.c (ccp_folder): New class derived from
5606 substitute_and_fold_engine.
5607 (ccp_folder::get_value): New member function.
5608 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
5609 (ccp_fold_stmt): Remove prototype.
5610 (ccp_finalize): Call substitute_and_fold from the ccp_class.
5611 * tree-ssa-copy.c (copy_folder): New class derived from
5612 substitute_and_fold_engine.
5613 (copy_folder::get_value): Renamed from get_value.
5614 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
5615 * tree-vrp.c (vrp_folder): New class derived from
5616 substitute_and_fold_engine.
5617 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
5618 (vrp_folder::get_value): New member function.
5619 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
5620 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
5621 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
5622 provide a class interface to folder/substitute routines.
5623 (ssa_prop_fold_stmt_fn): Remove typedef.
5624 (ssa_prop_get_value_fn): Likewise.
5625 (subsitute_and_fold): Remove prototype.
5626 (replace_uses_in): Likewise.
5627 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
5628 Renamed from replace_uses_in. Call the virtual member function
5629 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
5630 (substitute_and_fold_dom_walker): Remove initialization of
5631 data member entries for calbacks. Add substitute_and_fold_engine
5632 member and initialize it.
5633 (substitute_and_fold_dom_walker::before_dom_children0: Use the
5634 member functions for get_value, replace_phi_args_in c
5635 replace_uses_in, and fold_stmt calls.
5636 (substitute_and_fold_engine::substitute_and_fold): Renamed from
5637 substitute_and_fold. Remove assert. Update ctor call.
5639 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
5640 (ssa_prop_visit_phi_fn): Likewise.
5641 (class ssa_propagation_engine): New class to provide an interface
5643 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
5645 (ssa_prop_visit_phi): Likewise.
5646 (ssa_propagation_engine::simulate_stmt): Moved into class.
5647 Call visit_phi/visit_stmt from the class rather than via
5648 file scoped static variables.
5649 (ssa_propagation_engine::simulate_block): Moved into class.
5650 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
5651 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
5652 set file scoped statics for the visit_stmt/visit_phi callbacks.
5653 * tree-complex.c (complex_propagate): New class derived from
5654 ssa_propagation_engine.
5655 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
5656 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
5657 (tree_lower_complex): Call ssa_propagate via the complex_propagate
5659 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
5660 ssa_propagation_engine.
5661 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
5662 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
5663 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
5664 * tree-ssa-copy.c (copy_prop): New class derived from
5665 ssa_propagation_engine.
5666 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
5667 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
5668 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
5669 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
5670 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
5671 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
5672 (execute_vrp): Call ssa_propagate from the vrp_prop class.
5674 2017-11-01 Jakub Jelinek <jakub@redhat.com>
5676 PR rtl-optimization/82778
5677 PR rtl-optimization/82597
5678 * compare-elim.c (struct comparison): Add in_a_setter field.
5679 (find_comparison_dom_walker::before_dom_children): Remove killed
5680 bitmap and df_simulate_find_defs call, instead walk the defs.
5681 Compute last_setter and initialize in_a_setter. Merge definitions
5682 with first initialization for a few variables.
5683 (try_validate_parallel): Use insn_invalid_p instead of
5684 recog_memoized. Return insn rather than just the pattern.
5685 (try_merge_compare): Fix up comment. Don't uselessly test if
5686 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
5688 (execute_compare_elim_after_reload): Remove df_chain_add_problem
5691 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5692 Alan Hayward <alan.hayward@arm.com>
5693 David Sherwood <david.sherwood@arm.com>
5695 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
5696 aarch64_hard_regno_nregs to get the number of registers
5699 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5700 Alan Hayward <alan.hayward@arm.com>
5701 David Sherwood <david.sherwood@arm.com>
5703 * config/aarch64/constraints.md (Upl): Rename to...
5705 * config/aarch64/aarch64.md
5706 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
5709 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5710 Alan Hayward <alan.hayward@arm.com>
5711 David Sherwood <david.sherwood@arm.com>
5713 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
5714 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
5717 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5718 Alan Hayward <alan.hayward@arm.com>
5719 David Sherwood <david.sherwood@arm.com>
5721 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
5722 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
5723 (aarch64_evpc_dup): Generate rtl direcly, rather than using
5725 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
5726 of a single element.
5727 * config/aarch64/iterators.md: Add a comment above the permute
5728 unspecs to say that they are generated directly by
5729 aarch64_expand_vec_perm_const.
5730 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
5732 2017-11-01 Nathan Sidwell <nathan@acm.org>
5734 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
5736 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
5738 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
5739 and medany code models, and describe what they do.
5741 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5743 Revert accidental duplicate:
5745 * combine.c (can_change_dest_mode): Reject changes in
5746 REGMODE_NATURAL_SIZE.
5748 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
5750 PR rtl-optimization/64682
5751 PR rtl-optimization/69567
5752 PR rtl-optimization/69737
5753 PR rtl-optimization/82683
5754 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
5755 register mentioned in the note, drop the note, unless it came from I3,
5756 in which case it should go to I3 again.
5758 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5760 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
5761 and return false if not.
5762 (clear_bytes_written_by, live_bytes_read): Update accordingly.
5764 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5766 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
5767 range is known to be empty.
5769 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5770 Alan Hayward <alan.hayward@arm.com>
5771 David Sherwood <david.sherwood@arm.com>
5773 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
5774 and CONST_VECTOR_NUNITS instead of computing the number of units from
5775 the byte sizes of the vector and element.
5776 (simplify_binary_operation_1): Likewise.
5777 (simplify_const_binary_operation): Likewise.
5778 (simplify_ternary_operation): Likewise.
5780 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5781 Alan Hayward <alan.hayward@arm.com>
5782 David Sherwood <david.sherwood@arm.com>
5784 * var-tracking.c (INT_MEM_OFFSET): Replace with...
5785 (int_mem_offset): ...this new function.
5786 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
5787 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
5788 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
5791 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5792 Alan Hayward <alan.hayward@arm.com>
5793 David Sherwood <david.sherwood@arm.com>
5795 * lower-subreg.c (interesting_mode_p): New function.
5796 (compute_costs, find_decomposable_subregs, decompose_register)
5797 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
5798 (resolve_clobber, dump_choices): Use it.
5800 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5801 Alan Hayward <alan.hayward@arm.com>
5802 David Sherwood <david.sherwood@arm.com>
5804 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
5806 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5807 Alan Hayward <alan.hayward@arm.com>
5808 David Sherwood <david.sherwood@arm.com>
5810 * alias.c (find_base_value, find_base_term): Only process integer
5811 truncations. Check the precision rather than the size.
5813 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5814 Alan Hayward <alan.hayward@arm.com>
5815 David Sherwood <david.sherwood@arm.com>
5817 * machmode.h (is_narrower_int_mode): New function
5818 * optabs.c (expand_float, expand_fix): Use it.
5819 * dwarf2out.c (rotate_loc_descriptor): Likewise.
5821 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5822 Alan Hayward <alan.hayward@arm.com>
5823 David Sherwood <david.sherwood@arm.com>
5825 * rtl.h (narrower_subreg_mode): New function.
5826 * ira-color.c (update_costs_from_allocno): Use it.
5828 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5829 Alan Hayward <alan.hayward@arm.com>
5830 David Sherwood <david.sherwood@arm.com>
5832 * optabs-query.h (convert_optab_p): New function, split out from...
5833 (convert_optab_handler): ...here.
5834 (widening_optab_handler): Delete.
5835 (find_widening_optab_handler): Remove permit_non_widening parameter.
5836 (find_widening_optab_handler_and_mode): Likewise. Provide an
5837 override that operates on mode class wrappers.
5838 * optabs-query.c (widening_optab_handler): Delete.
5839 (find_widening_optab_handler_and_mode): Remove permit_non_widening
5840 parameter. Assert that the two modes are the same class and that
5841 the "from" mode is narrower than the "to" mode. Use
5842 convert_optab_handler instead of widening_optab_handler.
5843 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
5844 instead of widening_optab_handler.
5845 * expr.c (expand_expr_real_2): Update calls to
5846 find_widening_optab_handler.
5847 * optabs.c (expand_widen_pattern_expr): Likewise.
5848 (expand_binop_directly): Take the insn_code as a parameter.
5849 (expand_binop): Only call find_widening_optab_handler for
5850 conversion optabs; use optab_handler otherwise. Update calls
5851 to find_widening_optab_handler and expand_binop_directly.
5852 Use convert_optab_handler instead of widening_optab_handler.
5853 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
5854 find_widening_optab_handler and use scalar_mode rather than
5856 (convert_plusminus_to_widen): Likewise.
5858 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5859 Alan Hayward <alan.hayward@arm.com>
5860 David Sherwood <david.sherwood@arm.com>
5862 * machmode.h (fixed_size_mode): New class.
5863 * rtl.h (get_pool_mode): Return fixed_size_mode.
5864 * gengtype.c (main): Add fixed_size_mode.
5865 * target.def (get_raw_result_mode): Return a fixed_size_mode.
5866 (get_raw_arg_mode): Likewise.
5867 * doc/tm.texi: Regenerate.
5868 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
5869 * targhooks.c (default_get_reg_raw_mode): Likewise.
5870 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
5871 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
5872 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
5873 (msp430_get_raw_result_mode): Likewise.
5874 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
5875 * dbxout.c (dbxout_parms): Require fixed-size modes.
5876 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
5877 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
5878 * omp-low.c (lower_oacc_reductions): Likewise.
5879 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
5880 (simplify_subreg): Update accordingly.
5881 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
5882 (force_const_mem): Update accordingly. Return NULL_RTX for modes
5883 that aren't fixed-size.
5884 (get_pool_mode): Return a fixed_size_mode.
5885 (output_constant_pool_2): Take a fixed_size_mode.
5887 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5888 Alan Hayward <alan.hayward@arm.com>
5889 David Sherwood <david.sherwood@arm.com>
5891 * doc/rtl.texi (vec_series): Document.
5892 (const): Say that the operand can be a vec_series.
5893 * rtl.def (VEC_SERIES): New rtx code.
5894 * rtl.h (const_vec_series_p_1): Declare.
5895 (const_vec_series_p): New function.
5896 * emit-rtl.h (gen_const_vec_series): Declare.
5897 (gen_vec_series): Likewise.
5898 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
5899 (gen_vec_series): Likewise.
5900 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
5901 * simplify-rtx.c (simplify_unary_operation): Handle negations
5903 (simplify_binary_operation_series): New function.
5904 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
5905 (test_vector_ops_series): New function.
5906 (test_vector_ops): Call it.
5907 * config/powerpcspe/altivec.md (altivec_lvsl): Use
5908 gen_const_vec_series.
5909 (altivec_lvsr): Likewise.
5910 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
5912 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5913 Alan Hayward <alan.hayward@arm.com>
5914 David Sherwood <david.sherwood@arm.com>
5916 * doc/rtl.texi (const): Update description of address constants.
5917 Say that vector constants are allowed too.
5918 * common.md (E, F): Use CONSTANT_P instead of checking for
5920 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
5921 checking for CONST_VECTOR.
5922 * expmed.c (make_tree): Use build_vector_from_val for a CONST
5924 * expr.c (expand_expr_real_2): Check for vector modes instead
5925 of checking for CONST_VECTOR.
5926 * rtl.h (const_vec_p): New function.
5927 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
5928 (unwrap_const_vec_duplicate): Handle them here too.
5930 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5931 David Malcolm <dmalcolm@redhat.com>
5932 Alan Hayward <alan.hayward@arm.com>
5933 David Sherwood <david.sherwood@arm.com>
5935 * rtl.h (vec_duplicate_p): New function.
5936 * selftest-rtl.c (assert_rtx_eq_at): New function.
5937 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
5938 (assert_rtx_eq_at): Declare.
5939 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
5940 * selftest-run-tests.c (selftest::run_tests): Call it.
5941 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
5942 (simplify_unary_operation_1): Recursively handle vector duplicates.
5943 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
5945 (simplify_subreg): Handle subregs of vector duplicates.
5946 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
5947 (selftest::simplify_rtx_c_tests): New functions.
5949 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5950 Alan Hayward <alan.hayward@arm.com>
5951 David Sherwood <david.sherwood@arm.com>
5953 * emit-rtl.h (gen_const_vec_duplicate): Declare.
5954 (gen_vec_duplicate): Likewise.
5955 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
5957 (gen_const_vector): ...here.
5958 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
5959 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
5960 whose elements are all equal.
5961 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
5962 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
5963 (simplify_relational_operation): Likewise.
5964 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
5966 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
5967 (aarch64_expand_vector_init): Likewise.
5968 * config/arm/arm.c (neon_vdup_constant): Likewise.
5969 (neon_expand_vector_init): Likewise.
5970 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
5971 (arm_block_set_unaligned_vect): Likewise.
5972 (arm_block_set_aligned_vect): Likewise.
5973 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
5974 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
5975 (expand_vec_perm_even_odd_pack): Likewise.
5976 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
5977 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
5978 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
5979 gen_const_vec_duplicate.
5980 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
5981 * config/mips/mips.c (mips_gen_const_int_vector): Use
5982 gen_const_vec_duplicate.
5983 (mips_expand_vector_init): Use CONST0_RTX.
5984 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
5985 (define_split): Use gen_const_vec_duplicate.
5986 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
5987 (define_split): Use gen_const_vec_duplicate.
5988 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
5989 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
5990 * config/spu/spu.c (spu_const): Likewise.
5992 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5993 Alan Hayward <alan.hayward@arm.com>
5994 David Sherwood <david.sherwood@arm.com>
5996 * combine.c (can_change_dest_mode): Reject changes in
5997 REGMODE_NATURAL_SIZE.
5999 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
6001 * configure.ac (--enable-libssp): New.
6002 (gcc_cv_libc_provides_ssp): Check for explicit setting before
6003 trying to determine target-specific default. Adjust indentation.
6004 * configure: Regenerated.
6005 * doc/install.texi (Configuration): Expand --disable-libssp
6008 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
6010 config/i386/i386.c (ix86_expand_epilogue): Correct stack
6013 2017-10-31 Martin Jambor <mjambor@suse.cz>
6016 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
6018 2017-10-31 David Malcolm <dmalcolm@redhat.com>
6020 * auto-profile.c (autofdo_source_profile::read): Use
6021 UNKNOWN_LOCATION rather than 0.
6022 * diagnostic-core.h (warning_at_rich_loc): Rename to...
6023 (warning_at): ...this overload.
6024 (warning_at_rich_loc_n): Rename to...
6025 (warning_n): ...this overload.
6026 (error_at_rich_loc): Rename to...
6027 (error_at): ...this overload.
6028 (pedwarn_at_rich_loc): Rename to...
6029 (pedwarn): ...this overload.
6030 (permerror_at_rich_loc): Rename to...
6031 (permerror): ...this overload.
6032 (inform_at_rich_loc): Rename to...
6033 (inform): ...this overload.
6034 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
6035 (diagnostic_n_impl_richloc): Rename to...
6036 (diagnostic_n_impl): ...this rich_location *-based decl.
6037 (inform_at_rich_loc): Rename to...
6038 (inform): ...this, and add an assertion.
6039 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
6040 (warning_at_rich_loc): Rename to...
6041 (warning_at): ...this, and add an assertion.
6042 (warning_at_rich_loc_n): Rename to...
6043 (warning_n): ...this, and add an assertion.
6044 (warning_n): Update location_t-based implementation for removal of
6045 location_t-based diagnostic_n_impl.
6046 (pedwarn_at_rich_loc): Rename to...
6047 (pedwarn): ...this, and add an assertion.
6048 (permerror_at_rich_loc): Rename to...
6049 (permerror): ...this, and add an assertion.
6050 (error_n): Update for removal of location_t-based diagnostic_n_impl.
6051 (error_at_rich_loc): Rename to...
6052 (error_at): ...this, and add an assertion.
6053 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
6054 (driver::do_spec_on_infiles): Likewise.
6055 * substring-locations.c (format_warning_va): Update for renaming
6056 of inform_at_rich_loc.
6058 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
6060 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
6061 _Float<N>X built-in functions so that the variant without the
6062 "__builtin_" prefix is only enabled for the GNU C and Objective C
6063 languages when they are in non-strict ANSI/ISO mode.
6064 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
6065 * target.def (floatn_builtin_p): Add a target hook to control
6066 whether _Float<N> and _Float<N>X built-in functions without the
6067 "__builtin_" prefix are enabled, and return true for C and
6068 Objective C in the default hook. Include langhooks.h in
6070 * targhooks.h (default_floatn_builtin_p): Likewise.
6071 * targhooks.c (default_floatn_builtin_p): Likewise.
6072 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
6073 floatn_builtin_p target hook.
6074 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
6076 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
6077 Eric Botcazou <ebotcazou@adacore.com>
6079 PR rtl-optimization/81803
6080 * lra-constraints.c (curr_insn_transform): Also reload the whole
6081 register for a strict subreg no wider than a word if this is for
6082 a WORD_REGISTER_OPERATIONS target.
6084 2017-10-31 Jason Merrill <jason@redhat.com>
6086 * gdbinit.in: Skip over inlines from timevar.h.
6088 2017-10-31 Martin Liska <mliska@suse.cz>
6090 * doc/gcov.texi: Document new option.
6091 * gcov.c (print_usage): Likewise print it.
6092 (process_args): Support the argument.
6093 (format_count): New function.
6094 (format_gcov): Use the function.
6096 2017-10-31 Martin Liska <mliska@suse.cz>
6098 * gcov.c (struct name_map): do not use typedef.
6099 Define operator== and operator<.
6100 (name_search): Remove.
6101 (name_sort): Remove.
6102 (main): Do not allocate names.
6103 (process_file): Add vertical space.
6104 (generate_results): Use std::find.
6105 (release_structures): Do not release memory.
6106 (find_source): Use std::find.
6108 2017-10-31 Martin Liska <mliska@suse.cz>
6110 * gcov.c (struct line_info): Remove it's typedef.
6111 (line_info::line_info): Add proper ctor.
6112 (line_info::has_block): Do not use a typedef.
6113 (struct source_info): Do not use typedef.
6114 (circuit): Likewise.
6115 (get_cycles_count): Likewise.
6116 (output_intermediate_file): Iterate via vector iterator.
6117 (add_line_counts): Use std::vector methods.
6118 (accumulate_line_counts): Likewise.
6119 (output_lines): Likewise.
6121 2017-10-31 Martin Liska <mliska@suse.cz>
6123 * gcov.c (struct source_info): Remove typedef.
6124 (source_info::source_info): Add proper ctor.
6125 (accumulate_line_counts): Use struct, not it's typedef.
6126 (output_gcov_file): Likewise.
6127 (output_lines): Likewise.
6128 (main): Do not allocate an array.
6129 (output_intermediate_file): Use size of vector container.
6130 (process_file): Resize the vector.
6131 (generate_results): Do not preallocate, use newly added vector
6133 (release_structures): Do not release sources.
6134 (find_source): Use vector methods.
6135 (add_line_counts): Do not use typedef.
6137 2017-10-31 Martin Liska <mliska@suse.cz>
6139 * doc/gcov.texi: Document that.
6140 * gcov.c (add_line_counts): Mark lines with a non-executed
6142 (output_line_beginning): Handle such lines.
6143 (output_lines): Pass new argument.
6144 (output_intermediate_file): Print it in intermediate format.
6146 2017-10-31 Martin Liska <mliska@suse.cz>
6148 * color-macros.h: New file.
6149 * diagnostic-color.c: Factor out color related to macros to
6151 * doc/gcov.texi: Document -k option.
6152 * gcov.c (INCLUDE_STRING): Include string.h.
6153 (print_usage): Add -k option.
6154 (process_args): Parse it.
6155 (pad_count_string): New function.
6156 (output_line_beginning): Likewise.
6157 (DEFAULT_LINE_START): New macro.
6158 (output_lines): Support color output.
6160 2017-10-31 Martin Liska <mliska@suse.cz>
6162 PR gcov-profile/82633
6163 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
6164 their interaction with GCOV infrastructure.
6165 * configure.ac: Add -fkeep-{inline,static}-functions to
6167 * configure: Regenerate.
6169 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
6172 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
6174 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
6177 * config/rs6000/rs6000.md (allocate_stack): Force update interval
6178 into a register if it does not fit into an immediate offset field.
6180 2017-10-31 Olivier Hainque <hainque@adacore.com>
6182 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
6184 2017-10-31 Julia Koval <julia.koval@intel.com>
6186 * config.gcc: Add gfniintrin.h.
6187 * config/i386/gfniintrin.h: New.
6188 * config/i386/i386-builtin-types.def
6189 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
6190 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
6191 __builtin_ia32_vgf2p8affineinvqb_v32qi,
6192 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
6193 __builtin_ia32_vgf2p8affineinvqb_v16qi,
6194 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
6195 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
6196 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
6197 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
6198 V64QI_FTYPE_V64QI_V64QI_INT): New types.
6199 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
6200 * config/i386/immintrin.h: Include gfniintrin.h.
6201 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
6203 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
6205 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
6207 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
6209 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
6210 (arm_ashldi3_1bit): Remove pattern.
6211 (ashrdi3): Remove shift by 1 expansion.
6212 (arm_ashrdi3_1bit): Remove pattern.
6213 (lshrdi3): Remove shift by 1 expansion.
6214 (arm_lshrdi3_1bit): Remove pattern.
6215 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
6216 cost of ashldi3 by 1.
6217 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
6218 (<shift>di3_neon): Likewise.
6220 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6222 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
6223 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
6224 and (*aarch64_simd_mov<VQ:mode>).
6225 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
6226 pattern alternative.
6227 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
6228 attributes to match pattern alternative.
6230 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
6232 * config.gcc (powerpc*-*-*): Add emmintrin.h.
6233 * config/rs6000/emmintrin.h: New file.
6234 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
6236 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
6238 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
6239 * (movdi_vfp_cortexa8): Remove pattern.
6241 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6243 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
6245 (Specific, alpha*-dec-osf5.1): Remove.
6246 (Specific, mips-sgi-irix5): Remove.
6247 (Specific, mips-sgi-irix6): Remove.
6249 2017-10-30 Jakub Jelinek <jakub@redhat.com>
6252 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
6253 arguments to clear_bit_region_be.
6255 2017-10-30 Jim Wilson <wilson@tuliptree.org>
6257 * gimplify.c: Include memmodel.h.
6259 2017-10-30 Martin Jambor <mjambor@suse.cz>
6261 * omp-grid.c (grid_attempt_target_gridification): Also insert a
6262 condition whether loop should be executed at all.
6264 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
6266 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6267 gimple folding of vec_madd() intrinsics.
6268 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
6269 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
6270 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
6272 2017-10-30 Richard Biener <rguenther@suse.de>
6274 PR tree-optimization/82762
6276 2017-10-23 Richard Biener <rguenther@suse.de>
6278 PR tree-optimization/82129
6280 2017-08-01 Richard Biener <rguenther@suse.de>
6282 PR tree-optimization/81181
6283 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6284 (compute_antic): ... end of iteration here.
6286 2017-10-30 Joseph Myers <joseph@codesourcery.com>
6288 * doc/invoke.texi (C Dialect Options): Document -std=c17,
6289 -std=iso9899:2017 and -std=gnu17.
6290 * doc/standards.texi (C Language): Document C17 support.
6291 * doc/cpp.texi (Overview): Mention -std=c17.
6292 (Standard Predefined Macros): Document C11 and C17 values of
6293 __STDC_VERSION__. Do not refer to C99 support as incomplete.
6294 * doc/extend.texi (Inline): Do not list individual options for
6295 standards newer than C99.
6296 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
6298 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
6301 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
6303 * asan.c (asan_finish_file): Align asan globals array by shadow
6306 2017-10-30 Jakub Jelinek <jakub@redhat.com>
6309 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
6310 (struct store_immediate_info): Add bitregion_start and bitregion_end
6312 (store_immediate_info::store_immediate_info): Add brs and bre
6313 arguments and initialize bitregion_{start,end} from those.
6314 (struct merged_store_group): Add bitregion_start, bitregion_end,
6315 align_base and mask fields. Drop unnecessary struct keyword from
6316 struct store_immediate_info. Add do_merge method.
6317 (clear_bit_region_be): Use memset instead of loop storing zeros.
6318 (merged_store_group::do_merge): New method.
6319 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
6320 stores as long as the surrounding bitregions have no gaps.
6321 (merged_store_group::merge_overlapping): Use do_merge.
6322 (merged_store_group::apply_stores): Test that bitregion_{start,end}
6323 is byte aligned, rather than requiring that start and width are
6324 byte aligned. Drop unnecessary struct keyword from
6325 struct store_immediate_info. Allocate and populate also mask array.
6326 Make start of the arrays relative to bitregion_start rather than
6327 start and size them according to bitregion_{end,start} difference.
6328 (struct imm_store_chain_info): Drop unnecessary struct keyword from
6329 struct store_immediate_info.
6330 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
6331 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
6332 struct keyword from struct store_immediate_info.
6333 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
6334 between stores as long as the surrounding bitregions have no gaps.
6336 (struct split_store): Add orig non-static data member.
6337 (split_store::split_store): Initialize orig to false.
6338 (find_constituent_stmts): Return store_immediate_info *, non-NULL
6339 if there is exactly a single original stmt. Change stmts argument
6340 to pointer from reference, if NULL, don't push anything to it. Add
6341 first argument, use it to optimize skipping over orig stmts that
6342 are known to be before bitpos already. Simplify.
6343 (split_group): Return unsigned int count how many stores are or
6344 would be needed rather than a bool. Add allow_unaligned argument.
6345 Change split_stores argument from reference to pointer, if NULL,
6346 only do a dry run computing how many stores would be produced.
6347 Rewritten algorithm to use both alignment and misalign if
6348 !allow_unaligned and handle bitfield stores with gaps.
6349 (imm_store_chain_info::output_merged_store): Set start_byte_pos
6350 from bitregion_start instead of start. Compute allow_unaligned
6351 here, if true, do 2 split_group dry runs to compute which one
6352 produces fewer stores and prefer aligned if equal. Punt if
6353 new count is bigger or equal than original before emitting any
6354 statements, rather than during that. Remove no longer needed
6355 new_ssa_names tracking. Replace num_stmts with
6356 split_stores.length (). Use 32-bit stack allocated entries
6357 in split_stores auto_vec. Try to reuse original store lhs/rhs1
6358 if possible. Handle bitfields with gaps.
6359 (pass_store_merging::execute): Ignore bitsize == 0 stores.
6360 Compute bitregion_{start,end} for the stores and construct
6361 store_immediate_info with that. Formatting fixes.
6363 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
6366 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
6367 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
6369 2017-10-29 Jim Wilson <wilson@tuliptree.org>
6371 * gimplify.c: Include tm_p.h.
6373 * common.opt (gcoff): Re-add as ignored option.
6374 (gcoff1, gcoff2, gcoff3): Likewise.
6376 * Makefile.in (OBJS): Delete sdbout.o.
6377 (GTFILES): Delete $(srcdir)/sdbout.c.
6378 * debug.h: Delete sdb_debug_hooks.
6379 * final.c: Delete sdbout.h include.
6380 (final_scan_insn): Delete SDB_DEBUG check.
6381 (rest_of_clean_state): Likewise.
6382 * output.h: Delete sdb_begin_function_line.
6385 * toplev.c: Delete sdbout.h include.
6386 (process_options): Delete SDB_DEBUG check.
6387 * tree-core.h (tree_type_common): Delete pointer field of
6389 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
6390 TYPE_SYMTAB_POINTER.
6391 * tree.h (TYPE_SYMTAB_POINTER): Delete.
6392 (TYPE_SYMTAB_IS_POINTER): Delete.
6393 (TYPE_SYMTAB_IS_DIE): Renumber.
6394 * xcoffout.c: Refer to former sdbout.c file.
6395 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
6397 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
6398 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
6399 (Debugging Options): Delete -gcoff.
6400 (-gstabs): Delete SDB reference.
6402 (-gcoff@var{level}): Delete.
6403 * doc/passes.texi (Debugging information output): Delete SDB and
6404 sdbout.c references.
6405 * doc/tm.texi: Regenerate.
6406 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
6407 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
6409 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
6410 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
6411 (SDB_DEBUGGING_INFO): Delete.
6412 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
6413 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
6414 * target.def (output_source_filename): Delete COFF reference.
6416 * common.opt (gcoff): Delete.
6417 (gxcoff+): Update Negative chain.
6418 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
6420 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
6421 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
6422 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
6424 (expand_function_start): Change sdb reference to past tense.
6425 (expand_function_end): Change sdb reference to past tense.
6426 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
6427 * opts.c (debug_type_names): Delete coff entry.
6428 (common_handle_option): Delete OPT_gcoff case.
6429 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
6431 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
6432 * config/cris/cris.h: Delete SDB reference in comment.
6433 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
6434 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
6435 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
6436 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
6438 (ix86_expand_prologue): Likewise.
6439 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
6440 * config/ia64/ia64.h: Likewise.
6441 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
6442 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
6444 * config/mmix/mmix.h: Likewise.
6445 * config/nds32/nds32.c: Likewise.
6446 * config/stormy/storym16.h: Likewise.
6447 * config/visium/visium.h: Likewise.
6448 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
6450 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
6452 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
6453 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
6454 FRAME_POINTER_REGNUM point at high end of local var area.
6456 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
6458 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
6459 Move comment around. Do not reset best_edge for a copiable
6460 destination if the copy would cause a partition change.
6461 (better_edge_p): Remove redundant check.
6463 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6465 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
6467 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
6469 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
6470 for math functions that have _Float<N> and _Float<N>X variants.
6471 (mathfn_built_in_2): Add support for math functions that have
6472 _Float<N> and _Float<N>X variants.
6473 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
6474 (expand_builtin_mathfn_ternary): Add support for fma with
6475 _Float<N> and _Float<N>X variants.
6476 (expand_builtin): Likewise.
6477 (fold_builtin_3): Likewise.
6478 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
6479 create math function _Float<N> and _Float<N>X variants as external
6481 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
6482 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
6483 the __builtin_ prefix and if not strict ansi, without the prefix.
6484 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
6485 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
6486 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
6487 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
6488 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
6489 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
6490 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
6491 function signatures for fma _Float<N> and _Float<N>X variants.
6492 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
6493 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
6494 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
6495 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
6496 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
6497 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
6498 * gencfn-macros.c (print_case_cfn): Add support for math functions
6499 that have _Float<N> and _Float<N>X variants.
6500 (print_define_operator_list): Likewise.
6501 (fltfn_suffixes): Likewise.
6503 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
6504 for math functions that have _Float<N> and _Float<N>X variants.
6505 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
6506 and _Float<N>X variants.
6507 (COPYSIGN): Likewise.
6510 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
6511 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
6513 (integer_valued_read_call_p): Likewise.
6514 * fold-const-call.c (fold_const_call_ss): Likewise.
6515 (fold_const_call_sss): Add support for copysign, fmin, and fmax
6516 _Float<N> and _Float<N>X variants.
6517 (fold_const_call_ssss): Add support for fma _Float<N> and
6518 _Float<N>X variants.
6519 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
6520 support for copysign and fma _Float<N> and _Float<N>X variants.
6521 (backprop::process_builtin_call_use): Likewise.
6522 * tree-call-cdce.c (can_test_argument_range); Add support for
6523 sqrt _Float<N> and _Float<N>X variants.
6524 (edom_only_function): Likewise.
6525 (get_no_error_domain): Likewise.
6526 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
6527 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
6528 copysign _Float<N> and _Float<N>X variants.
6529 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
6530 handled by machine independent code.
6531 (FMAF128): Likewise.
6532 * doc/cpp.texi (Common Predefined Macros): Document defining
6533 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
6534 fma _Float<N> and _Float<N>X variants.
6536 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6539 * config/i386/i386-modes.def (CCFPU): Remove definition.
6540 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
6541 (ix86_cc_modes_compatible): Ditto.
6542 (ix86_expand_carry_flag_compare): Ditto.
6543 (ix86_expand_int_movcc): Ditto.
6544 (ix86_expand_int_addcc): Ditto.
6545 (ix86_reverse_condition): Ditto.
6546 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
6547 Return true/false for unordered/ordered fp comparisons.
6548 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
6549 (ix86_prepare_fp_compare_args): Update for rename.
6550 (ix86_expand_fp_compare): Update for rename. Generate unordered
6551 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
6552 (ix86_expand_sse_compare_and_jump): Ditto.
6553 * config/i386/predicates.md (fcmov_comparison_operator):
6554 Remove CCFPU mode handling.
6555 (ix86_comparison_operator): Ditto.
6556 (ix86_carry_flag_operator): Ditto.
6557 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
6558 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
6559 (*cmpu<mode>_cc_i387): Ditto.
6560 (FPCMP): Remove mode iterator.
6561 (unord): Remove mode attribute.
6562 (unord_subst): New define_subst transformation
6563 (unord): New define_subst attribute.
6565 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
6566 (*cmpi<unord>xf_i387): Ditto.
6567 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
6568 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
6569 using unord_subst transformation.
6570 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
6571 (round_saeonly): Also handle CCFP mode.
6572 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
6573 Remove UNSPEC_SAHF unspec handling.
6575 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6577 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
6579 2017-10-27 Jeff Law <law@redhat.com>
6581 * gimple-ssa-sprintf.c: Include domwalk.h.
6582 (class sprintf_dom_walker): New class, derived from dom_walker.
6583 (sprintf_dom_walker::before_dom_children): New function.
6584 (struct call_info): Moved into sprintf_dom_walker class
6585 (compute_formath_length, handle_gimple_call): Likewise.
6586 (sprintf_length::execute): Call the dominator walker rather
6587 than walking the statements.
6589 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
6590 gimple statement locations.
6591 (check_array_bounds): Corresponding changes. Get the statement's
6592 location directly from wi->stmt.
6594 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
6597 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
6599 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6601 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
6602 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
6604 2017-10-27 Jakub Jelinek <jakub@redhat.com>
6607 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
6608 * config/i386/i386.c (maybe_get_pool_constant): Removed.
6609 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
6610 maybe_get_pool_constant.
6611 * config/i386/predicates.md (zero_extended_scalar_load_operand):
6614 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6616 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
6617 2.26 caveat. Update gas and gld versions.
6618 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
6621 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6623 * cgraph.h (set_malloc_flag): Declare.
6624 * cgraph.c (set_malloc_flag_1): New function.
6625 (set_malloc_flag): Likewise.
6626 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
6627 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
6629 (read_ipa_call_summary): Add support for reading is_return_callee.
6630 (write_ipa_call_summary): Stream is_return_callee.
6631 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
6632 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
6633 ipa-prop.h, ipa-fnsummary.h.
6634 (pure_const_names): Change to static.
6635 (malloc_state_e): Define.
6636 (malloc_state_names): Define.
6637 (funct_state_d): Add field malloc_state.
6638 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
6639 (check_retval_uses): New function.
6640 (malloc_candidate_p): Likewise.
6641 (analyze_function): Add support for malloc attribute.
6642 (pure_const_write_summary): Stream malloc_state.
6643 (pure_const_read_summary): Add support for reading malloc_state.
6644 (dump_malloc_lattice): New function.
6645 (propagate_malloc): New function.
6646 (warn_function_malloc): New function.
6647 (ipa_pure_const::execute): Call propagate_malloc and
6648 ipa_free_fn_summary.
6649 (pass_local_pure_const::execute): Add support for malloc attribute.
6650 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
6651 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
6653 2017-10-27 Martin Liska <mliska@suse.cz>
6655 PR gcov-profile/82457
6656 * doc/invoke.texi: Document that one needs a non-strict ISO mode
6657 for fork-like functions to be properly instrumented.
6659 2017-10-27 Richard Biener <rguenther@suse.de>
6662 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
6663 info when we redirected EH.
6665 2017-10-26 Michael Collison <michael.collison@arm.com>
6667 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
6669 (<optab>_trunchf<GPI:mode>2: New pattern.
6670 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
6671 * config/aarch64/iterators.md (wv): New mode attribute.
6672 (vf, VF): New mode attributes.
6673 (vgp, VGP): New mode attributes.
6674 (s): Update attribute with SImode and DImode prefixes.
6676 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
6678 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
6679 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
6680 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
6681 (nios2_option_overide): Initialize it. Don't allow R0-relative
6682 addressing with PIC.
6683 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
6684 (nios2_symbolic_constant_p): Likewise.
6685 (nios2_legitimate_address_p): Likewise.
6686 (nios2_r0rel_section_name_p): New.
6687 (nios2_symbol_ref_in_r0rel_data_p): New.
6688 (nios2_emit_move_sequence): Handle r0rel_constant_p.
6689 (r0rel_constant_p): New.
6690 (nios2_print_operand_address): Handle r0rel_constant_p.
6691 (nios2_cdx_narrow_form_p): Likewise.
6692 * config/nios2/nios2.opt (mr0rel-sec=): New option.
6693 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
6694 (Nios II Options): Document -mr0rel-sec.
6696 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
6698 * config/nios2/nios2.c: Include xregex.h.
6699 (nios2_gprel_sec_regex): New.
6700 (nios2_option_overide): Initialize it. Don't allow GP-relative
6701 addressing with PIC.
6702 (nios2_small_section_name_p): Check for regex match.
6703 * config/nios2/nios2.opt (mgprel-sec=): New option.
6704 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
6705 (Nios II Options): Document -mgprel-sec.
6707 2017-10-26 Jim Wilson <wilson@tuliptree.org>
6709 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
6711 2017-10-26 Tom de Vries <tom@codesourcery.com>
6713 PR tree-optimization/82707
6714 * gimple.c (gimple_copy): Fix unsharing of
6715 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
6717 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
6719 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
6720 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
6721 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
6722 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
6723 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
6724 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
6725 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
6726 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
6727 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
6728 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
6729 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
6730 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
6731 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
6732 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
6733 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
6734 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
6735 _mm512_mask_cmpunord_ps_mask): New intrinsics.
6737 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6739 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
6741 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
6742 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
6743 warning to rs6000.c. Remove the Undocumented flag, since it has
6745 (-mabi=ibmlongdouble): Likewise.
6746 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
6747 already set, set the default format for long double.
6748 (rs6000_debug_reg_global): Print whether long double is IBM or
6750 (rs6000_option_override_internal): Rework setting long double
6751 format. Only warn if the user is changing the long double default
6752 and they did not use -Wno-psabi.
6753 * doc/invoke.texi (PowerPC options): Update the documentation for
6754 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
6756 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6757 Alan Hayward <alan.hayward@arm.com>
6758 David Sherwood <david.sherwood@arm.com>
6760 * rtl.h (wider_subreg_mode): New function.
6761 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
6762 rather than an unsigned int *.
6763 * ira-color.c (regno_max_ref_width): Replace with...
6764 (regno_max_ref_mode): ...this new variable.
6765 (coalesced_pseudo_reg_slot_compare): Update accordingly.
6766 Use wider_subreg_mode.
6767 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
6768 rather than an unsigned int *.
6769 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
6770 (process_alt_operands): Likewise.
6771 (invariant_p): Likewise.
6772 * lra-spills.c (assign_mem_slot): Likewise.
6773 (add_pseudo_to_slot): Likewise.
6774 * lra.c (collect_non_operand_hard_regs): Likewise.
6775 (add_regs_to_insn_regno_info): Likewise.
6776 * reload1.c (regno_max_ref_width): Replace with...
6777 (regno_max_ref_mode): ...this new variable.
6778 (reload): Update accordingly. Update call to
6779 ira_sort_regnos_for_alter_reg.
6780 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
6781 (init_eliminable_invariants): Update to use regno_max_ref_mode.
6782 (scan_paradoxical_subregs): Likewise.
6784 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6786 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
6787 (aarch64_frame): Add emit_frame_chain boolean.
6788 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6789 Move eh_return case to aarch64_layout_frame.
6790 (aarch64_layout_frame): Initialize emit_frame_chain.
6791 (aarch64_expand_prologue): Use emit_frame_chain.
6793 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6795 * config/aarch64/aarch64.c (aarch64_layout_frame):
6796 Ensure LR is always stored at the bottom of the callee-saves.
6797 Remove rarely used frame layout which saves callee-saves at top of
6798 frame, so the store of LR can be used as a valid probe in all cases.
6800 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6802 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
6803 Improve unaligned TImode/TFmode base/offset split.
6805 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6806 Alan Hayward <alan.hayward@arm.com>
6807 David Sherwood <david.sherwood@arm.com>
6809 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
6810 * combine.c (find_single_use_1): Likewise.
6811 (expand_field_assignment): Likewise.
6812 (move_deaths): Likewise.
6813 * lra-constraints.c (simplify_operand_subreg): Likewise.
6814 (curr_insn_transform): Likewise.
6815 * lra.c (collect_non_operand_hard_regs): Likewise.
6816 (add_regs_to_insn_regno_info): Likewise.
6817 * rtlanal.c (reg_referenced_p): Likewise.
6818 (covers_regno_no_parallel_p): Likewise.
6820 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6822 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
6823 Don't print any bits outside the precision of the value.
6824 * wide-int.cc (test_printing): Add some new tests.
6826 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6828 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
6829 supports -xbrace_comment option.
6830 * configure: Regenerate.
6831 * config.in: Regenerate.
6832 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
6833 (ASM_CPU_SPEC): Use it.
6835 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6837 * target.def (static_rtx_alignment): New hook.
6838 * targhooks.h (default_static_rtx_alignment): Declare.
6839 * targhooks.c (default_static_rtx_alignment): New function.
6840 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
6841 * doc/tm.texi: Regenerate.
6842 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
6843 instead of targetm.constant_alignment. Remove call to
6845 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6846 (cris_preferred_mininum_alignment): New function, split out from...
6847 (cris_constant_alignment): ...here.
6848 (cris_static_rtx_alignment): New function.
6849 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
6851 (ix86_constant_alignment): ...here.
6852 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6853 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6854 (mmix_static_rtx_alignment): New function.
6855 * config/spu/spu.c (spu_static_rtx_alignment): New function.
6856 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6858 2017-10-26 Tamar Christina <tamar.christina@arm.com>
6861 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
6862 Add flag_trapping_math and flag_fp_int_builtin_inexact.
6864 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
6866 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
6867 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
6868 mark as a sign-extending load.
6869 (local_pic_load_u): Define.
6871 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
6874 * fold-const.c (operand_equal_for_comparison_p): Also return true
6875 if ARG0 is a simple variant of ARG1 with narrower precision.
6876 (fold_ternary_loc): Always pass unstripped operands to the predicate.
6878 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
6880 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
6882 * i386.h (processor_costs): Add gather_static, gather_per_elt,
6883 scatter_static, scatter_per_elt.
6884 * x86-tune-costs.h: Add new cost entries.
6886 2017-10-25 Richard Biener <rguenther@suse.de>
6888 * tree-ssa-sccvn.h (vn_eliminate): Declare.
6889 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
6890 class pass_fre): Move to ...
6891 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
6892 class pass_fre): ... here and adjust for statistics.
6894 2017-10-25 Jakub Jelinek <jakub@redhat.com>
6897 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
6898 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
6899 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
6900 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
6903 2017-10-25 Richard Biener <rguenther@suse.de>
6905 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
6906 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
6907 eliminate_push_avail, eliminate_insert): Move inside...
6908 (class eliminate_dom_walker): ... this class in preparation
6910 (fini_eliminate): Remove by merging with ...
6911 (eliminate): ... this function. Adjust for class changes.
6912 (pass_pre::execute): Remove fini_eliminate call.
6913 (pass_fre::execute): Likewise.
6915 2017-10-24 Jakub Jelinek <jakub@redhat.com>
6918 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
6919 (VPERMI2, VPERMI2I): New mode iterators.
6920 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
6921 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
6923 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
6924 mode iterator. Remove 3 old define_insn patterns.
6925 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
6926 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
6927 VPERMI2 mode iterator, remove the other two expanders.
6928 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
6929 to use VPERMI2 mode iterator, add another alternative for vpermi2*
6930 instructions, remove the other two patterns.
6931 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
6932 mode iterator, remove the other two patterns.
6933 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
6934 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
6935 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
6936 and adjust argument order accordingly.
6937 (ix86_expand_vec_perm): Adjust caller.
6938 (expand_vec_perm_1): Likewise.
6939 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
6940 (expand_vec_perm_vpermt2_vpshub2): ... this.
6941 (ix86_expand_vec_perm_const_1): Adjust caller.
6942 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
6945 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
6946 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
6947 (vec_shl_<mode>): Remove unused expander.
6948 (avx512bw_<shift_insn><mode>3): New define_insn.
6949 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
6950 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
6952 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
6955 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
6958 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
6960 PR rtl-optimization/82396
6961 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
6962 (autopref_multipass_init): Simplify initialization.
6963 (autopref_rank_data): Simplify sort order.
6964 * gcc/sched-int.h (autopref_multipass_data_): Remove
6965 multi_mem_insn_p, min_offset and max_offset.
6967 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
6970 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6971 Check special value of flag_omit_frame_pointer.
6972 (aarch64_can_eliminate): Likewise.
6973 (aarch64_override_options_after_change_1): Simplify handling of
6974 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
6976 2017-10-24 Richard Biener <rguenther@suse.de>
6978 PR tree-optimization/82697
6979 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
6980 zero for conditional load and unconditional store.
6982 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
6984 * doc/install.texi: Document bootstrap-cet.
6986 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
6989 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
6990 ENDBR instruction at function entrance if function is only
6993 2017-10-24 Jakub Jelinek <jakub@redhat.com>
6996 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
6997 patterns to better describe from which operation the CF is computed.
6998 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
6999 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
7000 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
7001 is 0, use _0 suffixed expanders instead of emitting a comparison
7004 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
7006 * config/i386/i386.md(*movsf_internal, *movdf_internal):
7007 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
7009 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
7012 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
7013 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
7014 * loop-iv.c (iv_get_reaching_def): Likewise.
7015 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
7016 variable is promoted and the partition contains undefined values.
7018 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7020 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
7022 (nios2_address_cost): Define.
7023 (nios2_legitimize_address): Recognize (exp + constant) directly.
7024 (TARGET_ADDRESS_COST): Define.
7026 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7028 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
7029 (nios2_symbolic_memory_operand_p): Declare.
7030 (nios2_split_large_constant): Declare.
7031 (nios2_split_symbolic_memory_operand): Declare.
7032 * config/nios2/nios2.c: Adjust includes.
7033 (nios2_symbolic_constant_allowed): New.
7034 (nios2_symbolic_constant_p): New.
7035 (nios2_plus_symbolic_constant_p): New.
7036 (nios2_valid_addr_expr_p): Recognize addresses involving
7038 (nios2_legitimate_address_p): Likewise, also LO_SUM.
7039 (nios2_symbolic_memory_operand_p): New.
7040 (nios2_large_constant_p): New.
7041 (nios2_split_large_constant): New.
7042 (nios2_split_plus_large_constant): New.
7043 (nios2_split_symbolic_memory_operand): New.
7044 (nios2_legitimize_address): Code refactoring. Handle addresses
7045 involving symbolic constants.
7046 (nios2_emit_move_sequence): Likewise.
7047 (nios2_print_operand): Improve error output.
7048 (nios2_print_operand_address): Handle LO_SUM.
7049 (nios2_cdx_narrow_form_p): Likewise.
7050 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
7051 operands involving symbolic constants.
7052 (movhi_internal, movsi_internal): Likewise.
7053 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
7054 (extendhisi2, extendqi<mode>2): Likewise.
7056 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7058 * tree-pass.h (PROP_rtl_split_insns): Define.
7059 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
7061 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7063 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
7065 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7068 * target.def (const_not_ok_for_debug_p): Default to
7069 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
7070 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
7071 * targhooks.c (default_const_not_ok_for_debug_p): New function.
7072 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
7073 which targetm.const_not_ok_for_debug_p returned true.
7074 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
7076 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
7078 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
7079 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
7080 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
7082 (ix86_const_not_ok_for_debug_p): New function.
7083 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
7084 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
7086 2017-10-23 David Malcolm <dmalcolm@redhat.com>
7089 * system.h: Conditionally include "unique-ptr.h" if
7090 INCLUDE_UNIQUE_PTR is defined.
7091 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
7092 of defining INCLUDE_UNIQUE_PTR before including "system.h".
7094 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
7096 * config/rl78/rl78.md: New define_expand "subdi3".
7098 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
7101 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
7102 DF_REF_INSN if DF_REF_INSN_INFO is false.
7104 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
7106 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
7107 xmm_move instead of sse_move.
7108 (sse_store_index): New function.
7109 (ix86_register_move_cost): Be more sensible about mismatch stall;
7110 model AVX moves correctly; make difference between sse->integer and
7112 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
7113 moves; make difference between SSE and AVX.
7114 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
7115 and zmm_move. Increase size of sse load and store tables;
7116 add unaligned load and store tables; add ssemmx_to_integer.
7117 * x86-tune-costs.h: Update all entries according to real
7118 move latencies from Agner Fog's manual and chip documentation.
7120 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7123 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
7124 * config/i386/constraints.md (Wf): New constraint.
7125 * config/i386/i386.md (UNSPEC_SBB): New unspec.
7126 (cmp<dwi>_doubleword): Removed.
7127 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
7128 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
7129 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
7130 expand with cmp<dwi>_doubleword. For LTU and GEU use
7131 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
7133 * common.opt (gcolumn-info): Enable by default.
7134 * doc/invoke.texi (gcolumn-info): Document new default.
7136 2017-10-23 Richard Biener <rguenther@suse.de>
7138 PR tree-optimization/82672
7139 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
7140 Fold the stmt if we propagated into it.
7142 2017-10-23 Richard Biener <rguenther@suse.de>
7144 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
7145 (bitmap_remove_expr_from_set): ... this. All callers call this
7146 for non-constant values.
7147 (bitmap_set_subtract): Rename to...
7148 (bitmap_set_subtract_expressions): ... this. Adjust and
7150 (bitmap_set_contains_value): Remove superfluous check.
7151 (bitmap_set_replace_value): Inline into single caller ...
7152 (bitmap_value_replace_in_set): ... here and simplify.
7153 (dependent_clean): Merge into ...
7154 (clean): ... this using an overload. Adjust.
7155 (prune_clobbered_mems): Adjust.
7156 (compute_antic_aux): Likewise.
7157 (compute_partial_antic_aux): Likewise.
7159 2017-10-23 Richard Biener <rguenther@suse.de>
7161 PR tree-optimization/82129
7163 2017-08-01 Richard Biener <rguenther@suse.de>
7165 PR tree-optimization/81181
7166 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7167 (compute_antic): ... end of iteration here.
7169 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7171 * target.def (starting_frame_offset): New hook.
7172 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
7173 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
7174 * doc/tm.texi.in: Regenerate.
7175 * hooks.h (hook_hwi_void_0): Declare.
7176 * hooks.c (hook_hwi_void_0): New function.
7177 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
7178 STARTING_FRAME_OFFSET.
7179 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
7180 * reload1.c (reload): Likewise.
7181 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
7182 instead of STARTING_FRAME_OFFSET.
7183 * function.c (try_fit_stack_local): Likewise.
7184 (assign_stack_local_1): Likewise
7185 (instantiate_virtual_regs): Likewise.
7186 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
7187 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
7188 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
7189 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
7190 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
7191 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
7192 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
7193 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
7194 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
7195 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
7196 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
7197 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
7198 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
7199 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
7200 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
7201 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
7202 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
7203 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
7204 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
7205 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
7206 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
7207 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
7208 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
7209 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
7210 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
7211 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
7212 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
7213 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
7214 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
7215 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
7216 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
7217 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
7218 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
7219 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
7220 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
7221 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
7222 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
7223 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
7224 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
7225 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
7226 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
7227 * config/avr/avr.c (avr_starting_frame_offset): Make static and
7228 return a HOST_WIDE_INT.
7229 (avr_builtin_setjmp_frame_value): Use it instead of
7230 STARTING_FRAME_OFFSET.
7231 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7232 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
7233 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
7235 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7236 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
7237 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
7238 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7239 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
7240 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
7241 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7242 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
7243 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
7244 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7245 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
7246 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
7248 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7249 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
7250 * config/mips/mips.c (mips_compute_frame_info): Refer to
7251 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
7252 (mips_starting_frame_offset): New function.
7253 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7254 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
7255 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
7256 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
7257 and return a HOST_WIDE_INT.
7258 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7259 (mmix_initial_elimination_offset): Refer to
7260 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
7261 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
7262 * config/pa/pa.c (pa_starting_frame_offset): New function.
7263 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
7264 (pa_expand_prologue): Likewise.
7265 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7266 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
7267 !FRAME_GROWS_DOWNWARD handling to...
7268 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7269 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
7270 !FRAME_GROWS_DOWNWARD handling to...
7271 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7272 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
7273 !FRAME_GROWS_DOWNWARD handling to...
7274 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7275 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
7277 (rs6000_starting_frame_offset): New function.
7278 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
7279 !FRAME_GROWS_DOWNWARD handling to...
7280 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7281 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
7282 !FRAME_GROWS_DOWNWARD handling to...
7283 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7284 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
7285 !FRAME_GROWS_DOWNWARD handling to...
7286 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7287 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
7288 (rs6000_starting_frame_offset): New function.
7289 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
7290 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
7291 * config/vax/vax.c (vax_starting_frame_offset): New function.
7292 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
7293 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7294 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
7295 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
7296 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7297 * system.h (STARTING_FRAME_OFFSET): Poison.
7299 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7301 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
7302 SCALAR_TYPE_MODE instead of TYPE_MODE.
7304 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7305 Alan Hayward <alan.hayward@arm.com>
7306 David Sherwood <david.sherwood@arm.com>
7308 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
7310 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7311 Alan Hayward <alan.hayward@arm.com>
7312 David Sherwood <david.sherwood@arm.com>
7314 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
7316 2017-10-23 Richard Biener <rguenther@suse.de>
7318 PR tree-optimization/82129
7319 * tree-ssa-pre.c (bitmap_set_and): Remove.
7320 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
7321 canonicalizing expressions in the set to those with lowest
7322 ID rather than taking that from the first edge.
7324 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7326 * combine.c (rtx_equal_for_field_assignment_p): Use
7327 byte_lowpart_offset.
7329 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7330 Alan Hayward <alan.hayward@arm.com>
7331 David Sherwood <david.sherwood@arm.com>
7333 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
7334 to a SUBREG_PROMOTED_VAR.
7336 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7337 Alan Hayward <alan.hayward@arm.com>
7338 David Sherwood <david.sherwood@arm.com>
7340 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
7341 (expand_debug_source_expr): Likewise.
7342 * combine.c (combine_simplify_rtx): Likewise.
7343 * cse.c (fold_rtx): Likewise.
7344 * optabs.c (expand_float): Likewise.
7345 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7346 (simplify_binary_operation_1): Likewise.
7348 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7349 Alan Hayward <alan.hayward@arm.com>
7350 David Sherwood <david.sherwood@arm.com>
7352 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
7353 (record_promoted_value): Likewise.
7354 * expr.c (expand_expr_real_2): Likewise.
7355 * ree.c (update_reg_equal_equiv_notes): Likewise.
7356 (combine_set_extension): Likewise.
7357 * rtlanal.c (low_bitmask_len): Likewise.
7358 * simplify-rtx.c (neg_const_int): Likewise.
7359 (simplify_binary_operation_1): Likewise.
7361 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7362 Alan Hayward <alan.hayward@arm.com>
7363 David Sherwood <david.sherwood@arm.com>
7365 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
7366 * regcprop.c (maybe_mode_change): Likewise.
7367 * reload1.c (alter_reg): Likewise.
7369 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7371 * inchash.h (inchash::hash::add_wide_int): New function.
7372 * lto-streamer-out.c (hash_tree): Use it.
7374 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7376 * inchash.h (inchash::hash::add_wide_int): Rename to...
7377 (inchash::hash::add_hwi): ...this.
7378 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
7379 (polymorphic_call_target_hasher::hash): Likewise.
7380 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
7381 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
7382 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
7383 * lto-streamer-out.c (hash_tree): Likewise.
7384 * optc-save-gen.awk: Likewise.
7385 * tree.c (add_expr): Likewise.
7387 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
7390 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
7391 for ordered inequality comparisons even with TARGET_IEEE_FP.
7393 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
7396 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
7397 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
7398 Expand with cmp<dwi>_doubleword.
7400 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7402 * extend.texi: Add x86 specific to 'nocf_check' attribute.
7403 List CET intrinsics.
7404 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
7405 specific to -fcf-protection option.
7407 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7409 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
7410 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
7411 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
7412 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
7413 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
7414 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
7415 (extra_objs): Add cet.o for Linux/x86 targets.
7416 (tmake_file): Add i386/t-cet for Linux/x86 targets.
7417 * config/i386/cet.c: New file.
7418 * config/i386/cetintrin.h: Likewise.
7419 * config/i386/t-cet: Likewise.
7420 * config/i386/cpuid.h (bit_SHSTK): New.
7421 (bit_IBT): Likewise.
7422 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
7423 pass IBT and SHSTK bits.
7424 * config/i386/i386-builtin-types.def
7425 (VOID_FTYPE_UNSIGNED_PVOID): New.
7426 (VOID_FTYPE_UINT64_PVOID): Likewise.
7427 * config/i386/i386-builtin.def: Add CET intrinsics.
7428 * config/i386/i386-c.c (ix86_target_macros_internal): Add
7429 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
7430 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
7431 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
7433 * config/i386/i386.c (rest_of_insert_endbranch): New.
7434 (pass_data_insert_endbranch): Likewise.
7435 (pass_insert_endbranch): Likewise.
7436 (make_pass_insert_endbranch): Likewise.
7437 (ix86_notrack_prefixed_insn_p): Likewise.
7438 (ix86_target_string): Add -mibt, -mshstk flags.
7439 (ix86_option_override_internal): Add flag_cf_protection
7441 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
7442 (ix86_print_operand): Add 'notrack' prefix output.
7443 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
7444 (ix86_expand_builtin): Expand CET intrinsics.
7445 (x86_output_mi_thunk): Add 'endbranch' instruction.
7446 * config/i386/i386.h (TARGET_IBT): New.
7447 (TARGET_IBT_P): Likewise.
7448 (TARGET_SHSTK): Likewise.
7449 (TARGET_SHSTK_P): Likewise.
7450 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
7451 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
7452 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
7453 (builtin_setjmp_setup): New pattern.
7454 (builtin_longjmp): Likewise.
7455 (rdssp<mode>): Likewise.
7456 (incssp<mode>): Likewise.
7457 (saveprevssp): Likewise.
7458 (rstorssp): Likewise.
7459 (wrss<mode>): Likewise.
7460 (wruss<mode>): Likewise.
7461 (setssbsy): Likewise.
7462 (clrssbsy): Likewise.
7463 (nop_endbr): Likewise.
7464 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
7466 * config/i386/immintrin.h: Include <cetintrin.h>.
7467 * config/i386/linux-common.h
7468 (file_end_indicate_exec_stack_and_cet): New prototype.
7469 (TARGET_ASM_FILE_END): New.
7471 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7473 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
7474 latencies instead of having separate table; make difference between
7475 integer and float costs.
7476 * i386.h (processor_costs): Remove scalar_stmt_cost,
7477 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
7478 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
7480 * x86-tune-costs.h: Remove entries which has been removed in
7481 procesor_costs from all tables; make cond_taken_branch_cost
7482 and cond_not_taken_branch_cost COST_N_INSNS based.
7484 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7486 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
7488 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7490 * config/i386/i386.md (isa): Remove fma_avx512f.
7491 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
7492 <avx512>_fmadd_<mode>_mask3<round_name>,
7493 <avx512>_fmsub_<mode>_mask<round_name>,
7494 <avx512>_fmsub_<mode>_mask3<round_name>,
7495 <avx512>_fnmadd_<mode>_mask<round_name>,
7496 <avx512>_fnmadd_<mode>_mask3<round_name>,
7497 <avx512>_fnmsub_<mode>_mask<round_name>,
7498 <avx512>_fnmsub_<mode>_mask3<round_name>,
7499 <avx512>_fmaddsub_<mode>_mask<round_name>,
7500 <avx512>_fmaddsub_<mode>_mask3<round_name>,
7501 <avx512>_fmsubadd_<mode>_mask<round_name>,
7502 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
7503 (*vec_widen_umult_even_v16si<mask_name>,
7504 *vec_widen_smult_even_v16si<mask_name>): Likewise.
7505 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
7507 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7509 * extend.texi: Add 'nocf_check' documentation.
7510 * gimple.texi: Add second parameter to
7511 gimple_build_call_from_tree.
7512 * invoke.texi: Add -fcf-protection documentation.
7513 * rtl.texi: Add REG_CALL_NOTRACK documenation.
7515 2017-10-20 Richard Biener <rguenther@suse.de>
7517 PR tree-optimization/82473
7518 * tree-vect-loop.c (vectorizable_reduction): Properly get at
7519 the largest input type.
7521 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7523 * c-attribs.c (handle_nocf_check_attribute): New function.
7524 (c_common_attribute_table): Add 'nocf_check' handling.
7525 * gimple-parser.c: Add second argument NULL to
7526 gimple_build_call_from_tree.
7527 * attrib.c (comp_type_attributes): Check nocf_check attribute.
7528 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
7530 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
7531 * common.opt: Add fcf-protection flag.
7532 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
7533 * flag-types.h: Add enum cf_protection_level.
7534 * gimple.c (gimple_build_call_from_tree): Add second parameter.
7535 Add 'nocf_check' attribute propagation to gimple call.
7536 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
7537 (gimple_build_call_from_tree): Update prototype.
7538 (gimple_call_nocf_check_p): New function.
7539 (gimple_call_set_nocf_check): Likewise.
7540 * gimplify.c: Add second argument to gimple_build_call_from_tree.
7541 * ipa-icf.c: Add nocf_check attribute in statement hash.
7542 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
7543 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
7544 * toplev.c (process_options): Add flag_cf_protection handling.
7546 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7548 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
7550 2017-10-20 Richard Biener <rguenther@suse.de>
7552 PR tree-optimization/82603
7553 * tree-if-conv.c (predicate_mem_writes): Make sure to only
7554 remove false predicated stores.
7556 2017-10-20 Richard Biener <rguenther@suse.de>
7558 * graphite-isl-ast-to-gimple.c
7559 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
7560 Remove return value and simplify, dump copied stmt after lhs
7562 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
7563 Reduce dump verbosity.
7564 (gsi_insert_earliest): Likewise.
7565 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
7566 * graphite.c (print_global_statistics): Adjust dumping.
7567 (print_graphite_scop_statistics): Likewise.
7568 (print_graphite_statistics): Do not dump loops here.
7569 (graphite_transform_loops): But here.
7571 2017-10-20 Nicolas Roche <roche@adacore.com>
7573 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
7574 * configure: Regenerate.
7576 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7579 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
7580 functions when optimizing replace GIMPLE_RETURN stmts with
7581 calls to __builtin_unreachable ().
7584 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
7585 for -fsanitize=thread link of executables.
7586 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
7587 link of executables.
7590 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
7592 (<shift_insn><mode>3<mask_name>): Change the last of the 3
7593 define_insns for logical vector shifts to use VI248_AVX512BW
7594 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
7595 condition, useless isa and prefix attributes. Change the first
7596 2 of these define_insns to ...
7597 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
7598 define_insn for avx512vl.
7599 (<shift_insn><mode>3): ... and this, new define_insn without
7600 masking for non-avx512vl.
7603 * config/i386/sse.md (*andnot<mode>3,
7604 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
7605 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
7606 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
7607 not applied use empty suffix even for TARGET_AVX512VL.
7608 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
7609 is applied, supply evex,evex or evex,evex,evex instead of just
7612 2017-10-20 Julia Koval <julia.koval@intel.com>
7614 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
7615 (OPTION_MASK_ISA_GFNI_UNSET): New.
7616 (ix86_handle_option): Handle OPT_mgfni.
7617 * config/i386/cpuid.h (bit_GFNI): New.
7618 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
7619 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
7620 * config/i386/i386.c (ix86_target_string): Add -mgfni.
7621 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
7622 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
7623 * config/i386/i386.opt: Add mgfni.
7625 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
7627 * config/msp430/msp430.c (msp430_option_override): Disable
7628 -fdelete-null-pointer-checks.
7629 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
7631 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7633 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
7634 of x87 and SSE instructions.
7636 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7638 * asan.c (create_cond_insert_point): Do not update edge count.
7639 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
7640 (afdo_propagate_circuit): Likewise.
7641 (afdo_calculate_branch_prob): Likewise.
7642 (afdo_annotate_cfg): Likewise.
7643 * basic-block.h (struct edge_def): Remove count.
7644 (edge_def::count): New accessor.
7645 * bb-reorder.c (rotate_loop): Update.
7646 (find_traces_1_round): Update.
7647 (connect_traces): Update.
7648 (sanitize_hot_paths): Update.
7649 * cfg.c (unchecked_make_edge): Update.
7650 (make_single_succ_edge): Update.
7651 (check_bb_profile): Update.
7652 (dump_edge_info): Update.
7653 (update_bb_profile_for_threading): Update.
7654 (scale_bbs_frequencies_int): Update.
7655 (scale_bbs_frequencies_gcov_type): Update.
7656 (scale_bbs_frequencies_profile_count): Update.
7657 (scale_bbs_frequencies): Update.
7658 * cfganal.c (connect_infinite_loops_to_exit): Update.
7659 * cfgbuild.c (compute_outgoing_frequencies): Update.
7660 (find_many_sub_basic_blocks): Update.
7661 * cfgcleanup.c (try_forward_edges): Update.
7662 (try_crossjump_to_edge): Update
7663 * cfgexpand.c (expand_gimple_cond): Update
7664 (expand_gimple_tailcall): Update
7665 (construct_exit_block): Update
7666 * cfghooks.c (verify_flow_info): Update
7667 (redirect_edge_succ_nodup): Update
7668 (split_edge): Update
7669 (make_forwarder_block): Update
7670 (duplicate_block): Update
7671 (account_profile_record): Update
7672 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
7673 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
7674 * cfgloopmanip.c (scale_loop_profile): Update.
7676 (lv_adjust_loop_entry_edge): Update.
7677 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
7678 (force_nonfallthru_and_redirect): Update.
7679 (purge_dead_edges): Update.
7680 (rtl_flow_call_edges_add): Update.
7681 * cgraphunit.c (init_lowered_empty_function): Update.
7682 (cgraph_node::expand_thunk): Update.
7683 * gimple-pretty-print.c (dump_probability): Update.
7684 (dump_edge_probability): Update.
7685 * gimple-ssa-isolate-paths.c (isolate_path): Update.
7686 * haifa-sched.c (sched_create_recovery_edges): Update.
7687 * hsa-gen.c (convert_switch_statements): Update.
7688 * ifcvt.c (dead_or_predicable): Update.
7689 * ipa-inline-transform.c (inline_transform): Update.
7690 * ipa-split.c (split_function): Update.
7691 * ipa-utils.c (ipa_merge_profiles): Update.
7692 * loop-doloop.c (add_test): Update.
7693 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
7694 * lto-streamer-in.c (input_cfg): Update.
7695 (input_function): Update.
7696 * lto-streamer-out.c (output_cfg): Update.
7697 * modulo-sched.c (sms_schedule): Update.
7698 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
7699 * predict.c (maybe_hot_edge_p): Update.
7700 (unlikely_executed_edge_p): Update.
7701 (probably_never_executed_edge_p): Update.
7702 (dump_prediction): Update.
7703 (drop_profile): Update.
7704 (propagate_unlikely_bbs_forward): Update.
7705 (determine_unlikely_bbs): Update.
7706 (force_edge_cold): Update.
7707 * profile.c (compute_branch_probabilities): Update.
7708 * reg-stack.c (better_edge): Update.
7709 * shrink-wrap.c (handle_simple_exit): Update.
7710 * tracer.c (better_p): Update.
7711 * trans-mem.c (expand_transaction): Update.
7712 (split_bb_make_tm_edge): Update.
7713 * tree-call-cdce.c: Update.
7714 * tree-cfg.c (gimple_find_sub_bbs): Update.
7715 (gimple_split_edge): Update.
7716 (gimple_duplicate_sese_region): Update.
7717 (gimple_duplicate_sese_tail): Update.
7718 (gimple_flow_call_edges_add): Update.
7719 (insert_cond_bb): Update.
7720 (execute_fixup_cfg): Update.
7721 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
7722 * tree-complex.c (expand_complex_div_wide): Update.
7723 * tree-eh.c (lower_resx): Update.
7724 (unsplit_eh): Update.
7725 (cleanup_empty_eh_move_lp): Update.
7726 * tree-inline.c (copy_edges_for_bb): Update.
7727 (freqs_to_counts): Update.
7728 (copy_cfg_body): Update.
7729 * tree-ssa-dce.c (remove_dead_stmt): Update.
7730 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
7731 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
7732 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
7733 (unloop_loops): Update.
7734 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
7735 * tree-ssa-loop-split.c (connect_loops): Update.
7736 (split_loop): Update.
7737 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
7738 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
7739 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
7740 * tree-ssa-reassoc.c (branch_fixup): Update.
7741 * tree-ssa-tail-merge.c (replace_block_by): Update.
7742 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
7743 (compute_path_counts): Update.
7744 (update_profile): Update.
7745 (recompute_probabilities): Update.
7746 (update_joiner_offpath_counts): Update.
7747 (estimated_freqs_path): Update.
7748 (freqs_to_counts_path): Update.
7749 (clear_counts_path): Update.
7750 (ssa_fix_duplicate_block_edges): Update.
7751 (duplicate_thread_path): Update.
7752 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
7753 (case_bit_test_cmp): Update.
7754 (collect_switch_conv_info): Update.
7755 (gen_inbound_check): Update.
7756 (do_jump_if_equal): Update.
7757 (emit_cmp_and_jump_insns): Update.
7758 * tree-tailcall.c (decrease_profile): Update.
7759 (eliminate_tail_call): Update.
7760 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
7761 (vect_do_peeling): Update.
7762 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
7763 * ubsan.c (ubsan_expand_null_ifn): Update.
7764 (ubsan_expand_ptr_ifn): Update.
7765 * value-prof.c (gimple_divmod_fixed_value): Update.
7766 (gimple_mod_pow2): Update.
7767 (gimple_mod_subtract): Update.
7768 (gimple_ic): Update.
7769 (gimple_stringop_fixed_value): Update.
7771 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
7774 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
7776 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
7778 PR rtl-optimization/82395
7779 * ira-color.c (allocno_priority_compare_func): Fix comparison step
7780 based on non_spilled_static_chain_regno_p.
7782 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
7784 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
7785 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
7786 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
7788 2017-10-19 Martin Sebor <msebor@redhat.com>
7790 PR tree-optimization/82596
7791 * tree.c (array_at_struct_end_p): Handle STRING_CST.
7793 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
7795 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
7796 (get_mem_refs_of_builtin_call): Likewise.
7797 * builtins.c (expand_builtin_apply): Adjust call to
7798 allocate_dynamic_stack_space.
7799 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
7800 the third argument to allocate_dynamic_stack_space, otherwise -1.
7801 (expand_builtin): Deal with all alloca variants.
7802 (is_inexpensive_builtin): Likewise.
7803 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
7804 * calls.c (special_function_p): Deal with all alloca variants.
7805 (initialize_argument_information): Adjust call to
7806 allocate_dynamic_stack_space.
7807 (expand_call): Likewise.
7808 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
7809 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
7810 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
7811 use it for the stack usage computation.
7812 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
7813 * function.c (gimplify_parameters): Call build_alloca_call_expr.
7814 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
7815 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
7816 (in_loop_p): Remove first argument and useless check.
7817 (pass_walloca::execute): Remove useless test and adjust call to above.
7818 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
7819 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
7820 (gimplify_call_expr): Deal with all alloca variants.
7821 * hsa-gen.c (gen_hsa_alloca): Likewise.
7822 (gen_hsa_insns_for_call): Likewise.
7823 * ipa-pure-const.c (special_builtin_state): Likewise.
7824 * tree-chkp.c (chkp_build_returned_bound): Likewise.
7825 * tree-object-size.c (alloc_object_size): Likewise.
7826 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
7827 (call_may_clobber_ref_p_1): Likewise.
7828 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
7829 (ccp_fold_stmt): Likewise.
7830 (optimize_stack_restore): Likewise.
7831 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
7832 (mark_all_reaching_defs_necessary_1): Likewise.
7833 (propagate_necessity): Likewise.
7834 (eliminate_unnecessary_stmts): Likewise.
7835 * tree.c (build_common_builtin_nodes): Build
7836 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
7837 (build_alloca_call_expr): New function.
7838 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
7839 (CASE_BUILT_IN_ALLOCA): Likewise.
7840 (build_alloca_call_expr): Declare.
7841 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
7843 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
7846 * dwarf2out.c (new_die_raw): New static inline function.
7847 (new_die): Use it to create the DIE.
7848 (add_AT_external_die_ref): Likewise.
7849 (clone_die): Likewise.
7850 (clone_as_declaration): Likewise.
7851 (dwarf2out_vms_debug_main_pointer): Likewise.
7852 (base_type_die): Likewise. Remove early return for corner cases.
7853 Do not call add_pubtype on the DIE here.
7854 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
7855 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
7856 typedefs for base types with DW_AT_endianity. Make sure a DIE with
7857 native order exists for base types, attach the DIE manually and call
7858 add_pubtype on it. Do not equate a reverse order DIE to the type.
7860 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
7862 * config/arm/arm.c (align_ok_ldrd_strd): New function.
7863 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
7865 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
7867 2017-10-19 Jakub Jelinek <jakub@redhat.com>
7869 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
7870 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
7871 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
7872 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
7873 * opts.c (sanitizer_opts): Add builtin.
7874 * ubsan.c (instrument_builtin): New function.
7875 (pass_ubsan::execute): Call it.
7876 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
7877 * doc/invoke.texi: Document -fsanitize=builtin.
7879 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
7880 builtins, store max (log2 (align), 0) into uchar field instead of
7881 align into uptr field.
7882 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
7883 store uchar 0 field instead of uptr 0 field.
7884 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
7885 instead of passing one address of struct with 2 locations pass
7886 two addresses of structs with 1 location each.
7887 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
7888 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
7889 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
7890 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
7891 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
7892 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
7893 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
7894 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
7896 2017-10-19 Martin Liska <mliska@suse.cz>
7899 * file-find.c (remove_prefix): Remove.
7900 * file-find.h (remove_prefix): Likewise.
7901 * gcc-ar.c: Remove smartness of lookup.
7903 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
7905 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
7906 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
7907 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
7909 2017-10-19 Jakub Jelinek <jakub@redhat.com>
7912 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
7913 (setcc + and to xor + setcc): New peephole2.
7915 2017-10-19 Tom de Vries <tom@codesourcery.com>
7917 * doc/sourcebuild.texi (Test Directives, Variants of
7918 dg-require-support): Add dg-require-stack-size.
7920 2017-10-19 Martin Liska <mliska@suse.cz>
7923 * gimplify.c (gimplify_decl_expr): Do not instrument variables
7924 that have a large alignment.
7925 (gimplify_target_expr): Likewise.
7927 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
7929 PR rtl-optimization/82602
7930 * ira.c (rtx_moveable_p): Return false for volatile asm.
7932 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
7935 * config/i386/i386-modes.def (CCGZ): New CC mode.
7936 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
7937 * config/i386/predicates.md (ix86_comparison_operator):
7939 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
7940 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
7941 with double-word subtraction.
7942 (put_condition_code): Handle CCGZmode.
7944 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
7946 * wide-int.cc (debug (const wide_int &)): New.
7947 (debug (const wide_int *)): New.
7948 (debug (const widest_int &)): New.
7949 (debug (const widest_int *)): New.
7951 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
7954 * lra-constraints.c (curr_insn_transform): Use non-input operand
7955 instead of output one for matched reload.
7957 2017-10-18 Bin Cheng <bin.cheng@arm.com>
7959 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
7960 (tree-ssa-loop-ivopts.h): New header file.
7961 (struct builtin_info): New fields.
7962 (classify_builtin_1): Compute and record base and offset parts for
7963 memset builtin partition by calling strip_offset.
7964 (offset_cmp, fuse_memset_builtins): New functions.
7965 (finalize_partitions): Fuse adjacent memset partitions by calling
7967 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
7968 Expose the interface.
7969 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
7971 2017-10-18 Bin Cheng <bin.cheng@arm.com>
7973 PR tree-optimization/82574
7974 * tree-loop-distribution.c (find_single_drs): New parameter. Check
7975 that data reference must be executed exactly once per iteration
7976 against the outermost loop in nest.
7977 (classify_partition): Update call to above function.
7979 2017-10-18 Richard Biener <rguenther@suse.de>
7981 PR tree-optimization/82591
7982 * graphite.c (graphite_transform_loops): Move code gen message
7984 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7985 Here. Handle scop_to_isl_ast failing.
7986 (scop_to_isl_ast): Limit the number of ISL operations.
7988 2017-10-18 Richard Biener <rguenther@suse.de>
7990 * graphite-isl-ast-to-gimple.c
7991 (translate_isl_ast_to_gimple::set_rename): Simplify.
7992 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
7993 (graphite_copy_stmts_from_block): ... here.
7994 (copy_bb_and_scalar_dependences): Simplify.
7995 (add_parameters_to_ivs_params): Canonicalize.
7996 (generate_entry_out_of_ssa_copies): Simplify.
7997 * graphite-sese-to-poly.c (extract_affine_name): Simplify
7998 by passing in ISL dimension.
7999 (parameter_index_in_region_1): Rename to ...
8000 (parameter_index_in_region): ... this.
8001 (extract_affine): Adjust assert, pass down parameter index.
8002 (add_param_constraints): Use range-info when available.
8003 (build_scop_context): Adjust.
8004 * sese.c (new_sese_info): Adjust.
8005 (free_sese_info): Likewise.
8006 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
8007 Remove unused typedefs.
8008 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
8010 2017-10-18 Martin Liska <mliska@suse.cz>
8012 * combine.c (simplify_compare_const): Add gcc_fallthrough.
8014 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8016 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
8017 (s390_sched_init): Do not reset s390_sched_state if we entered the
8018 current basic block via a fallthru edge and all others are unlikely.
8020 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8022 * config/s390/s390.c (NUM_SIDES): New variable.
8023 (LONGRUNNING_THRESHOLD): New variable.
8024 (LATENCY_FACTOR): New variable.
8025 (s390_sched_score): Decrease score for long-running instructions on
8027 (s390_sched_variable_issue): Perform bookkeeping for long-running
8030 2017-10-18 Richard Biener <rguenther@suse.de>
8032 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
8033 Simplify with removal of the parameter rename map.
8034 (set_rename): Likewise.
8035 (should_copy_to_new_region): Likewise.
8036 (graphite_copy_stmts_from_block): Likewise.
8037 (copy_bb_and_scalar_dependences): Remove initialization of
8038 unused copied_bb_map.
8040 (copy_internal_parameters): Likewise.
8041 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
8042 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
8043 Use INTEGRAL_TYPE_P.
8044 (parameter_index_in_region_1): Rename to ...
8045 (assign_parameter_index_in_region): ... this. Assert we have
8046 a parameter we handle.
8047 (scan_tree_for_params): Adjust.
8048 * sese.h (parameter_rename_map_t): Remove.
8049 (struct sese_info_t): Remove unused parameter_rename_map and
8050 copied_bb_map members.
8051 * sese.c (new_sese_info): Adjust.
8052 (free_sese_info): Likewise.
8054 2017-10-18 Martin Liska <mliska@suse.cz>
8057 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
8058 on an abnormal edge.
8060 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
8062 * doc/invoke.texi (ffunction-sections and fdata-sections):
8065 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8067 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
8068 the use statement can throw internally.
8070 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8072 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
8073 any RTX present on the RHS of a SET.
8074 * compare-elim.c (try_eliminate_compare): Restore comment.
8076 2017-10-17 Jakub Jelinek <jakub@redhat.com>
8078 * langhooks.h (struct lang_hooks): Document that tree_size langhook
8079 may be also called on tcc_type nodes.
8080 * langhooks.c (lhd_tree_size): Likewise.
8082 2017-10-17 David Malcolm <dmalcolm@redhat.com>
8084 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
8085 format_warning_at_substring.
8086 (maybe_warn): Convert source_range * param to a location_t. Pass
8087 UNKNOWN_LOCATION rather than NULL to fmtwarn.
8088 (format_directive): Remove code to extract source_ranges and
8089 source_range * in favor of just a location_t.
8090 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
8092 * substring-locations.c (format_warning_va): Convert
8093 source_range * param to a location_t.
8094 (format_warning_at_substring): Likewise.
8095 * substring-locations.h (format_warning_va): Likewise.
8096 (format_warning_at_substring): Likewise.
8098 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
8100 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
8102 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
8103 and scatter/gather ops.
8105 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
8106 vec_gather_load and vec_scatter_store.
8107 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
8108 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
8109 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
8110 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
8111 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
8112 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
8114 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
8116 * reg-stack.c (compare_for_stack_reg): Add bool argument.
8117 Detect FTST instruction and handle its register pops. Only pop
8118 second operand if can_pop_second_op is true.
8119 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
8120 set can_pop_second_op to false in the compare_for_stack_reg call.
8122 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
8123 output_fp_compare for stack register operands.
8124 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
8125 instructions here. Do not emit stack register pops here. Assert
8126 that FCOMPP pops next to top stack register. Rewrite function.
8128 2017-10-17 Nathan Sidwell <nathan@acm.org>
8131 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
8132 use DECL_ASSEMBLER_NAME_RAW.
8135 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
8138 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
8139 Wilco Dijkstra <wilco.dijkstra@arm.com>
8141 * builtins.c (expand_builtin_update_setjmp_buf): Add a
8142 converstion to Pmode from the buf_addr.
8144 2017-10-17 Richard Biener <rguenther@suse.de>
8146 * graphite-dependences.c (scop_get_reads_and_writes): Change
8147 output parameters to references.
8149 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
8151 PR 71026/tree-optimization
8152 * fold-const.c (distribute_real_division): Removed.
8153 (fold_binary_loc): Remove calls to distribute_real_divison.
8155 2017-10-17 Richard Biener <rguenther@suse.de>
8157 * graphite-scop-detection.c
8158 (scop_detection::stmt_has_simple_data_refs_p): Always use
8159 the full nest as region.
8160 (try_generate_gimple_bb): Likewise.
8161 * sese.c (scalar_evolution_in_region): Simplify now that
8162 SCEV can handle instantiation in regions.
8163 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
8164 in the non-loop part of a function if requested.
8166 2017-10-17 Richard Biener <rguenther@suse.de>
8168 PR tree-optimization/82563
8169 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
8171 (graphite_regenerate_ast_isl): Call it.
8172 * graphite-scop-detection.c (build_scops): Remove entry edge split.
8174 2017-10-17 Jakub Jelinek <jakub@redhat.com>
8176 PR tree-optimization/82549
8177 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
8178 Formatting fixes. Instead of calling make_bit_field_ref with negative
8181 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
8183 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
8184 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
8185 _mm_maskz_reduce_ss): New.
8186 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
8187 __builtin_ia32_reducess_mask): Ditto..
8188 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
8189 * config/i386/sse.md (reduces<mode>): Renamed to ...
8190 (reduces<mode><mask_scalar_name>): ... this.
8191 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
8193 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
8194 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
8196 2017-10-16 David Malcolm <dmalcolm@redhat.com>
8198 * Makefile.in (OBJS): Add unique-ptr-tests.o.
8199 * selftest-run-tests.c (selftest::run_tests): Call
8200 selftest::unique_ptr_tests_cc_tests.
8201 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
8202 * unique-ptr-tests.cc: New file.
8204 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
8207 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8209 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8210 (make_hard_regno_born, make_hard_regno_dead): Update
8211 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
8213 2017-10-16 Jeff Law <law@redhat.com>
8215 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
8217 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
8219 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
8221 2017-10-16 Olivier Hainque <hainque@adacore.com>
8223 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
8224 with_cpu if we were configured for an e500v2 target cpu name.
8226 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
8228 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
8229 * doc/invoke.texi: Document +nodsp as a valid extension for
8232 2017-10-16 Martin Liska <mliska@suse.cz>
8234 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
8235 (test_set_range): Likewise.
8236 (test_range_functions): Rename to ...
8237 (test_bit_in_range): ... this.
8238 (sbitmap_c_tests): Add new test.
8240 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8242 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
8244 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
8245 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
8247 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8249 * config/aarch64/aarch64-builtins.c
8250 (aarch64_types_quadopu_lane_qualifiers): New.
8251 (TYPES_QUADOPU_LANE): New.
8252 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
8253 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
8254 (aarch64_<sur>dot_laneq<vsi2qi>): New.
8255 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
8256 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
8257 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
8258 (Vdottype, DOTPROD): New.
8259 (sur): Add SDOT and UDOT.
8261 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8263 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
8264 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
8265 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
8267 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
8268 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
8269 Enable TARGET_DOTPROD.
8270 (cortex-a75.cortex-a55): Likewise.
8271 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
8273 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8275 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
8276 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
8278 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
8280 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
8281 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
8282 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
8283 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
8284 * config/arm/types.md (neon_dot, neon_dot_q): New.
8285 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
8287 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8289 * config/arm/arm.h (TARGET_DOTPROD): New.
8290 * config/arm/arm.c (arm_arch_dotprod): New.
8291 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
8292 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
8293 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
8294 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
8295 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
8296 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
8297 * doc/invoke.texi (armv8.2-a): Document dotprod
8299 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
8301 * i386.c (ix86_vec_cost): New function.
8302 (ix86_rtx_costs): Handle vector operations better.
8303 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
8304 * x86-tune-costs.h: Add new costs to all tables.
8306 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
8308 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
8310 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
8311 divsd, sqrtss and sqrtsd
8312 * x86-tune-costs.h: Add new entries to all costs.
8313 (znver1_cost): Fix to match real instruction latencies.
8315 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8316 Michael Collison <michael.collison@arm.com>
8318 * compare-elim.c: Include emit-rtl.h.
8319 (can_merge_compare_into_arith): New function.
8320 (try_validate_parallel): Likewise.
8321 (try_merge_compare): Likewise.
8322 (try_eliminate_compare): Call the above when no previous clobber
8324 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
8327 2017-10-14 Jakub Jelinek <jakub@redhat.com>
8331 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
8332 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
8336 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
8337 to be any operand_equal_p operands. For & (B - 1) require
8338 B to be power of 2. Recognize
8339 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
8341 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8344 * optabs.c (expand_memory_blockage): Fix call of
8345 targetm.have_memory_blockage.
8347 2017-10-14 Jakub Jelinek <jakub@redhat.com>
8350 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
8351 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
8352 objects to extra_objs instead of overwriting it.
8354 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8356 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
8357 Use any_fp_register_operand as operand[3] predicate. Simplify
8358 equality test for operands[2] and operands[4] memory location.
8359 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
8360 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
8361 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
8362 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
8363 any_fp_register_operand as operand[1] predicate. Simplify
8364 equality test for operands[0] and operands[3] memory location.
8365 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
8366 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
8367 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
8369 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8371 * target-insns.def: Add memory_blockage.
8372 * optabs.c (expand_memory_blockage): New function.
8373 (expand_asm_memory_barrier): Rename ...
8374 (expand_asm_memory_blockage): ... to this.
8375 (expand_mem_thread_fence): Call expand_memory_blockage
8376 instead of expand_asm_memory_barrier.
8377 (expand_mem_singnal_fence): Ditto.
8378 (expand_atomic_load): Ditto.
8379 (expand_atomic_store): Ditto.
8380 * doc/md.texi (Standard Pattern Names For Generation):
8381 Document memory_blockage instruction pattern.
8383 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
8385 * config/rl78/rl78.c (rl78_emit_libcall): New function.
8386 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
8387 * config/rl78/rl78.md: New define_expand "adddi3".
8389 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
8391 * cfghooks.c (verify_flow_info): Disable check that all probabilities
8394 2017-10-13 Jeff Law <law@redhat.com>
8396 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
8398 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8401 * internal-fn.c (expand_mul_overflow): If both operands have
8402 the same highpart of -1 or 0 and the topmost bit of lowpart
8403 is different, overflow is if res <= 0 rather than res < 0.
8405 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
8407 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
8408 TARGET_P9_VECTOR code for unaligned_load case.
8410 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
8412 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
8414 2017-10-13 Nathan Sidwell <nathan@acm.org>
8416 * tree-core.h (tree_contains_struct): Make bool.
8417 * tree.c (tree_contains_struct): Likewise.
8418 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
8419 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
8420 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
8421 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
8422 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
8424 2017-10-13 Richard Biener <rguenther@suse.de>
8426 * graphite-isl-ast-to-gimple.c
8427 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
8428 parameters and dominance check.
8429 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
8430 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
8431 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
8432 Do not update SSA form here or do intermediate IL verification.
8433 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
8434 (graphite_initialize): Remove check on the number of loops in
8435 the function and inline into graphite_transform_loops.
8436 (graphite_finalize): Inline into graphite_transform_loops.
8437 (graphite_transform_loops): Perform SSA update and IL verification
8439 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
8441 2017-10-13 Richard Biener <rguenther@suse.de>
8443 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
8444 graphite_expression_type_precision): Avoid global constructor
8446 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
8447 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
8448 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
8449 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
8450 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
8451 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
8453 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
8456 * config/i386/i386.h (ix86_red_zone_size): New.
8457 * config/i386/i386.md (push peephole2s): Replace
8458 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
8460 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8461 Alan Hayward <alan.hayward@arm.com>
8462 David Sherwood <david.sherwood@arm.com>
8464 * combine.c (can_change_dest_mode): Reject changes in
8465 REGMODE_NATURAL_SIZE.
8467 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8468 Alan Hayward <alan.hayward@arm.com>
8469 David Sherwood <david.sherwood@arm.com>
8471 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
8472 (expand_debug_source_expr): Likewise.
8473 * combine.c (combine_simplify_rtx): Likewise.
8474 * cse.c (fold_rtx): Likewise.
8475 * fwprop.c (canonicalize_address): Likewise.
8476 * targhooks.c (default_shift_truncation_mask): Likewise.
8478 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8479 Alan Hayward <alan.hayward@arm.com>
8480 David Sherwood <david.sherwood@arm.com>
8482 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
8483 (widened_mode): Likewise.
8484 (expand_unop): Likewise.
8485 * ree.c (transform_ifelse): Likewise.
8486 (merge_def_and_ext): Likewise.
8487 (combine_reaching_defs): Likewise.
8488 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8490 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8491 Alan Hayward <alan.hayward@arm.com>
8492 David Sherwood <david.sherwood@arm.com>
8494 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
8495 * combine.c (gen_lowpart_for_combine): Likewise.
8496 * dwarf2out.c (rtl_for_decl_location): Likewise.
8497 * final.c (alter_subreg): Likewise.
8498 * rtlhooks.c (gen_lowpart_general): Likewise.
8499 (gen_lowpart_if_possible): Likewise.
8501 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8502 Alan Hayward <alan.hayward@arm.com>
8503 David Sherwood <david.sherwood@arm.com>
8505 * calls.c (expand_call): Use subreg_lowpart_offset.
8506 * cse.c (cse_insn): Likewise.
8507 * regcprop.c (copy_value): Likewise.
8508 (copyprop_hardreg_forward_1): Likewise.
8510 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8513 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
8514 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
8515 =Q constraints to +Q and into insn condition add check
8516 that operands[0] and operands[1] are equal.
8517 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
8518 =Q constraints to +Q and into insn condition add check
8519 that operands[0] is equal to either operands[1] or operands[2].
8522 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
8523 instead of handling MINUS_EXPR twice (once for each argument),
8524 canonicalize operand order and handle just once, use rtype where
8525 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
8528 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
8529 any values of __C while still being pattern recognizable as a simple
8532 2017-10-13 Richard Biener <rguenther@suse.de>
8534 PR tree-optimization/82451
8536 2017-10-02 Richard Biener <rguenther@suse.de>
8538 PR tree-optimization/82355
8539 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
8540 a mapping for the enclosing loop but avoid generating one for
8542 (copy_bb_and_scalar_dependences): Remove premature codegen
8543 error on PHIs in blocks duplicated into multiple places.
8544 * graphite-scop-detection.c
8545 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
8546 in the region use it as loop and nest to analyze the DR in.
8547 (try_generate_gimple_bb): Likewise.
8548 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
8549 (add_loop_constraints): For blocks in a loop not in the region
8550 create a dimension with a single iteration.
8551 * sese.h (gbb_loop_at_index): Remove assert.
8553 * cfgloop.c (loop_preheader_edge): For the loop tree root
8554 return the single successor of the entry block.
8555 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8556 Reset the SCEV hashtable and niters.
8557 * graphite-scop-detection.c
8558 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
8559 assert that we only have POLYNOMIAL_CHREC that vary in loops
8560 contained in the region.
8561 (scop_detection::graphite_can_represent_expr): Adjust.
8562 (scop_detection::stmt_has_simple_data_refs_p): For loops
8563 not in the region set loop to NULL. The nest is now the
8564 entry edge to the region.
8565 (try_generate_gimple_bb): Likewise.
8566 * sese.c (scalar_evolution_in_region): Adjust for
8567 instantiate_scev change.
8568 * tree-data-ref.h (graphite_find_data_references_in_stmt):
8569 Make nest parameter the edge into the region.
8570 (create_data_ref): Likewise.
8571 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
8572 entry edge into a region and adjust instantiate_scev calls.
8573 (create_data_ref): Likewise.
8574 (graphite_find_data_references_in_stmt): Likewise.
8575 (find_data_references_in_stmt): Pass the loop preheader edge
8576 from the nest argument.
8577 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
8578 parameter the edge into the region.
8579 (instantiate_parameters): Use the loop preheader edge as entry.
8580 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
8582 (get_instantiated_value_entry): Make instantiate_below parameter
8583 the edge into the region.
8584 (instantiate_scev_name): Likewise. Adjust dominance checks,
8585 when we cannot use loop-based instantiation instantiate by
8586 walking use-def chains.
8587 (instantiate_scev_poly): Adjust.
8588 (instantiate_scev_binary): Likewise.
8589 (instantiate_scev_convert): Likewise.
8590 (instantiate_scev_not): Likewise.
8591 (instantiate_array_ref): Remove.
8592 (instantiate_scev_3): Likewise.
8593 (instantiate_scev_2): Likewise.
8594 (instantiate_scev_1): Likewise.
8595 (instantiate_scev_r): Do not blindly handle N-operand trees.
8596 Do not instantiate array-refs. Handle all constants and invariants.
8597 (instantiate_scev): Make instantiate_below parameter
8598 the edge into the region.
8599 (resolve_mixers): Use the loop preheader edge for the region
8600 parameter to instantiate_scev_r.
8601 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
8603 2017-10-13 Richard Biener <rguenther@suse.de>
8605 PR tree-optimization/82525
8606 * graphite-isl-ast-to-gimple.c
8607 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
8609 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
8610 Fail code generation when we cannot represent the isl integer.
8611 (binary_op_to_tree): Elide modulo operations that are no-ops
8612 in the type we code generate. Remove now superfluous code
8615 2017-10-13 Richard Biener <rguenther@suse.de>
8617 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
8618 (scop_detection::harmful_loop_in_region): Remove premature
8619 IV type restriction.
8620 (scop_detection::graphite_can_represent_scev): We can handle
8621 pointer IVs just fine.
8623 2017-10-13 Alan Modra <amodra@gmail.com>
8625 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
8626 "Clobbers and Scratch Registers". Add paragraph on
8627 alternative to clobbers for scratch registers and OpenBLAS
8630 2017-10-13 Alan Modra <amodra@gmail.com>
8632 * doc/extend.texi (Clobbers): Correct vax example. Delete old
8633 example of a memory input for a string of known length. Move
8634 commentary out of table. Add a number of new examples
8635 covering array memory inputs.
8637 2017-10-12 Martin Liska <mliska@suse.cz>
8639 PR tree-optimization/82493
8640 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
8641 (test_range_functions): New function.
8642 (sbitmap_c_tests): Likewise.
8643 * selftest-run-tests.c (selftest::run_tests): Run new tests.
8644 * selftest.h (sbitmap_c_tests): New function.
8646 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
8648 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
8650 * config/rs6000/amo.h: Fix spacing issue.
8652 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8655 * config/i386/i386.md (*ashl<mode>3_mask_1,
8656 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
8657 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
8660 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
8662 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
8663 (profile_probability): Set max_probability
8664 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
8666 * profile-count.c (profile_probability::differs_from_p): Do not
8667 rely on max_probaiblity == 10000
8669 2017-10-12 Jeff Law <law@redhat.com>
8671 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
8674 2017-10-12 Martin Sebor <msebor@redhat.com>
8678 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
8679 (handle_alias_pairs): Call it.
8680 * common.opt (-Wattribute-alias): New option.
8681 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
8682 * doc/invoke.texi (-Wattribute-alias): Document.
8684 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
8687 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
8689 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8691 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8692 (make_hard_regno_born, make_hard_regno_dead): Update
8693 bb_killed_pseudos and bb_gen_pseudos.
8695 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
8697 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
8699 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
8701 * config/alpha/alpha.c (alpha_split_conditional_move):
8702 Use std::swap instead of manually swapping.
8703 (alpha_stdarg_optimize_hook): Ditto.
8704 (alpha_canonicalize_comparison): Ditto.
8706 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8708 * tree-loop-distribution.c (struct builtin_info): New struct.
8709 (struct partition): Refactor fields into struct builtin_info.
8710 (partition_free): Free struct builtin_info.
8711 (build_size_arg_loc, build_addr_arg_loc): Delete.
8712 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
8713 information from struct builtin_info.
8714 (find_single_drs): New function refactored from classify_partition.
8715 Also moved builtin validity checks to this function.
8716 (compute_access_range, alloc_builtin): New functions.
8717 (classify_builtin_st, classify_builtin_ldst): New functions.
8718 (classify_partition): Refactor code into functions find_single_drs,
8719 classify_builtin_st and classify_builtin_ldst.
8720 (distribute_loop): Don't do runtime alias check when distributing
8722 (find_seed_stmts_for_distribution): New function.
8723 (pass_loop_distribution::execute): Refactor code finding seed
8724 stmts into above function. Support distribution for the innermost
8725 two-level loop nest. Adjust dump information.
8727 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8729 * tree-loop-distribution.c: Adjust the general comment.
8730 (NUM_PARTITION_THRESHOLD): New macro.
8731 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
8732 (classify_partition): Skip builtin pattern of loop nest's inner loop.
8733 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
8734 in call to build_partition_graph.
8735 (finalize_partitions): New parameter. Make loop distribution more
8736 conservative by fusing more partitions.
8737 (distribute_loop): Don't do runtime alias check in case of loop nest
8739 (find_seed_stmts_for_distribution): New function.
8740 (prepare_perfect_loop_nest): New function.
8741 (pass_loop_distribution::execute): Refactor code finding seed stmts
8742 and loop nest into above functions. Support loop nest distribution.
8743 Adjust dump information accordingly.
8745 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8747 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
8748 and set PTYPE_SEQUENTIAL for merged partition.
8750 2017-10-12 Richard Biener <rguenther@suse.de>
8752 PR tree-optimization/69728
8754 2017-09-19 Richard Biener <rguenther@suse.de>
8756 PR tree-optimization/69728
8757 * graphite-sese-to-poly.c (schedule_error): New global.
8758 (add_loop_schedule): Handle empty domain by failing the
8760 (build_original_schedule): Handle schedule_error.
8762 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
8763 domain by returning an unchanged schedule.
8765 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8767 * genrecog.c (validate_pattern): For VEC_SELECT verify that
8768 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
8770 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
8772 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
8775 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8778 * expr.c (store_field): Don't optimize away bitsize == 0 store
8779 from CALL_EXPR with addressable return type.
8781 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
8783 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
8784 * config/rs6000/rs6000.md (sel): Delete mode attribute.
8785 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
8786 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
8787 TARGET_ISEL instead of TARGET_ISEL<sel>.
8789 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
8791 * config/rs6000/rs6000.c
8792 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
8794 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
8796 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
8798 (reg_or_cint_operand): Fix comment.
8799 (reg_or_zero_operand): New predicate.
8800 * config/rs6000/rs6000-protos.h (output_isel): Delete.
8801 * config/rs6000/rs6000.c (output_isel): Delete.
8802 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
8803 instead of reg_or_cint_operand. Output instruction directly (not via
8805 (isel_unsigned_<mode>): Ditto.
8806 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
8807 gpc_reg_operand. Add an instruction alternative for this. Output
8808 instruction directly.
8809 (*isel_reversed_unsigned_<mode>): Ditto.
8811 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
8813 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
8814 (TARGET_CANONICALIZE_COMPARISON): Define.
8816 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
8819 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
8820 Check whether the dest is REG before adding REG_EQUIV note.
8822 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
8825 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8827 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8828 (make_hard_regno_born, make_hard_regno_dead): Update
8829 bb_killed_pseudos and bb_gen_pseudos.
8831 2017-10-11 Nathan Sidwell <nathan@acm.org>
8833 * incpath.h (enum incpath_kind): Name enum, prefix values.
8834 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
8835 * incpath.c (heads, tails): Use INC_MAX.
8836 (add_env_var_paths, add_standard_paths): Use incpath_kind.
8837 (merge_include_chains, split_quote_chain,
8838 register_include_chains): Update incpath_kind names.
8839 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
8840 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
8842 (add_framework_path, darwin_register_objc_includes): Likewise.
8843 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
8845 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
8847 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
8848 Do not use float_operator operator predicate.
8849 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
8850 * config/i386/predicates.md (float_operator): Remove predicate.
8852 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
8854 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
8855 (*jccxf_i387): Ditto.
8856 (*jcc<mode>_i387): Ditto.
8857 (*jccu<mode>_i387): Ditto.
8858 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
8859 (*jcc_*_i387 splitters): Remove.
8860 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
8861 * config/i386/i386.c (ix86_split_fp_branch): Remove.
8862 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
8865 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
8867 * profile-count.h (slow_safe_scale_64bit): New function.
8868 (safe_scale_64bit): New inline.
8869 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
8870 * profile-count.c: Include wide-int.h
8871 (slow_safe_scale_64bit): New.
8873 2017-10-11 Nathan Sidwell <nathan@acm.org>
8875 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
8876 HAS_DECL_ASSEMBLER_NAME_P.
8877 * gimple-expr.c (gimple_decl_printable_name: Check
8878 HAS_DECL_ASSEMBLER_NAME_P too.
8879 * ipa-utils.h (type_in_anonymous_namespace_p): Check
8880 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
8881 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
8882 * passes.c (rest_of_decl_compilation): Check
8883 HAS_DECL_ASSEMBLER_NAME_P too.
8884 * recog.c (verify_changes): Likewise.
8885 * tree-pretty-print.c (dump_decl_name): Likewise.
8886 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
8888 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
8889 (SET_DECL_ASSEMBLER_NAME): Use it.
8890 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
8891 (COPY_DECL_ASSEMBLER_NAME): Likewise.
8892 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
8894 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
8896 * config.gcc (i386, x86_64): Add extra objects.
8897 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
8898 (ix86_min_insn_size): Declare.
8899 (ix86_issue_rate): Declare.
8900 (ix86_adjust_cost): Declare.
8901 (ia32_multipass_dfa_lookahead): Declare.
8902 (ix86_macro_fusion_p): Declare.
8903 (ix86_macro_fusion_pair_p): Declare.
8904 (ix86_bd_has_dispatch): Declare.
8905 (ix86_bd_do_dispatch): Declare.
8906 (ix86_core2i7_init_hooks): Declare.
8907 (ix86_atom_sched_reorder): Declare.
8908 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
8909 (COSTS_N_BYTES): Move to x86-tune-costs.h.
8910 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
8911 (rip_relative_addr_p): Rename to ...
8912 (ix86_rip_relative_addr_p): ... this one; export.
8913 (memory_address_length): Update.
8914 (ix86_issue_rate): Move to x86-tune-sched.c.
8915 (ix86_flags_dependent): Move to x86-tune-sched.c.
8916 (ix86_agi_dependent): Move to x86-tune-sched.c.
8917 (exact_dependency_1): Move to x86-tune-sched.c.
8918 (exact_store_load_dependency): Move to x86-tune-sched.c.
8919 (ix86_adjust_cost): Move to x86-tune-sched.c.
8920 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
8921 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
8922 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
8923 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
8924 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
8925 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
8926 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
8927 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
8928 (min_insn_size): Rename to ...
8929 (ix86_min_insn_size): ... this one; export.
8930 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
8931 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
8932 (core2i7_first_cycle_multipass_backtrack): Move to
8933 x86-tune-sched-core.c.
8934 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
8935 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
8936 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
8937 (ix86_avoid_jump_mispredicts): Update.
8938 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
8939 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
8940 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
8941 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
8942 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
8943 (MAX_INSN): Move to ix86-tune-sched-bd.c.
8944 (MAX_IMM): Move to ix86-tune-sched-bd.c.
8945 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
8946 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
8947 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
8948 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
8949 (MAX_STORE): Move to ix86-tune-sched-bd.c.
8950 (BIG): Move to ix86-tune-sched-bd.c.
8951 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
8952 (enum insn_path): Move to ix86-tune-sched-bd.c.
8953 (get_mem_group): Move to ix86-tune-sched-bd.c.
8954 (is_cmp): Move to ix86-tune-sched-bd.c.
8955 (dispatch_violation): Move to ix86-tune-sched-bd.c.
8956 (is_branch): Move to ix86-tune-sched-bd.c.
8957 (is_prefetch): Move to ix86-tune-sched-bd.c.
8958 (init_window): Move to ix86-tune-sched-bd.c.
8959 (allocate_window): Move to ix86-tune-sched-bd.c.
8960 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
8961 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
8962 (process_end_window): Move to ix86-tune-sched-bd.c.
8963 (allocate_next_window): Move to ix86-tune-sched-bd.c.
8964 (find_constant): Move to ix86-tune-sched-bd.c.
8965 (get_num_immediates): Move to ix86-tune-sched-bd.c.
8966 (has_immediate): Move to ix86-tune-sched-bd.c.
8967 (get_insn_path): Move to ix86-tune-sched-bd.c.
8968 (get_insn_group): Move to ix86-tune-sched-bd.c.
8969 (count_num_restricted): Move to ix86-tune-sched-bd.c.
8970 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
8971 (add_insn_window): Move to ix86-tune-sched-bd.c.
8972 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
8973 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
8974 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
8975 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
8976 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
8977 (do_dispatch): Move to ix86-tune-sched-bd.c.
8978 (has_dispatch): Move to ix86-tune-sched-bd.c.
8979 * config/i386/t-i386: Add new object files.
8980 * config/i386/x86-tune-costs.h: New file.
8981 * config/i386/x86-tune-sched-atom.c: New file.
8982 * config/i386/x86-tune-sched-bd.c: New file.
8983 * config/i386/x86-tune-sched-core.c: New file.
8984 * config/i386/x86-tune-sched.c: New file.
8986 2017-10-11 Liu Hao <lh_mouse@126.com>
8988 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
8989 the generic version below instead.
8990 (should_colorize): Recognize Windows consoles as terminals
8992 * pretty-print.c [__MINGW32__] (write_all): New function.
8993 [__MINGW32__] (find_esc_head): Likewise.
8994 [__MINGW32__] (find_esc_terminator): Likewise.
8995 [__MINGW32__] (eat_esc_sequence): Likewise.
8996 [__MINGW32__] (mingw_ansi_fputs): New function that handles
8998 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
9001 2017-10-11 Richard Biener <rguenther@suse.de>
9003 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
9004 Properly call analyze_scalar_evolution with the loop of the stmt.
9006 2017-10-11 Richard Biener <rguenther@suse.de>
9008 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
9009 * tree-core.h (tree_base): Add chrec_var union member.
9010 * tree.h (CHREC_VAR): Remove.
9011 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
9012 * tree-chrec.h (build_polynomial_chrec): Adjust.
9013 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
9014 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
9016 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
9018 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
9019 * match.pd: ... here.
9020 ((T) X == (T) Y): Relax condition.
9022 2017-10-11 Bin Cheng <bin.cheng@arm.com>
9024 PR tree-optimization/82472
9025 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
9027 (break_alias_scc_partitions): Update postorder number.
9029 2017-10-11 Martin Liska <mliska@suse.cz>
9032 * opts.c (parse_no_sanitize_attribute): Do not use error_value
9034 * opts.h (parse_no_sanitize_attribute): Remove last argument.
9036 2017-10-11 Martin Liska <mliska@suse.cz>
9038 * print-rtl.c (print_insn): Move declaration of idbuf
9039 to same scope as name.
9041 2017-10-11 Martin Liska <mliska@suse.cz>
9046 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9049 2017-10-11 Martin Liska <mliska@suse.cz>
9052 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9055 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
9057 * config.gcc (default_gnu_indirect_function): Default to yes for
9058 arm*-*-linux* with glibc.
9060 2017-10-11 Richard Biener <rguenther@suse.de>
9062 * tree-scalar-evolution.c (get_scalar_evolution): Handle
9063 default-defs and types we do not want to analyze.
9064 (interpret_loop_phi): Replace unreachable code with an assert.
9065 (compute_scalar_evolution_in_loop): Remove and inline ...
9066 (analyze_scalar_evolution_1): ... here, replacing condition with
9067 what makes the intent clearer. Remove handling of cases
9068 get_scalar_evolution now handles.
9070 2017-10-10 Jim Wilson <wilson@tuliptree.org>
9072 PR rtl-optimization/81434
9073 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
9074 comment for main loop. In sched_group_found if, also add checks for
9075 pass and min_cost_group.
9077 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
9079 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
9080 (rs6000_insn_cost): New function.
9081 * config/rs6000/rs6000.md (cost): New attribute.
9083 2017-10-10 Jakub Jelinek <jakub@redhat.com>
9084 H.J. Lu <hongjiu.lu@intel.com>
9088 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
9089 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
9090 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
9091 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
9092 to OPTION_MASK_ISA_AVX512VL - builtins that have both
9093 OPTION_MASK_ISA_MMX and some other bit set require both
9094 mmx and the ISAs without the mmx bit.
9095 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
9096 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
9097 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
9098 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
9099 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
9100 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
9101 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
9102 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
9103 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
9104 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
9105 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
9106 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
9107 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
9108 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
9109 Add OPTION_MASK_ISA_MMX.
9111 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
9113 * config.gcc (armv7*-*-freebsd*): New target.
9114 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
9116 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
9118 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
9119 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
9122 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
9124 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
9125 a template parameter.
9126 (WIDE_INT_REF_FOR): Update accordingly.
9127 * tree.h (wi::int_traits <const_tree>): Delete.
9128 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
9129 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
9130 (wi::tree_to_wide_ref): New typedef.
9131 (wi::to_wide): New function.
9132 * calls.c (get_size_range): Use wi::to_wide when operating on
9134 * cgraph.c (cgraph_node::create_thunk): Likewise.
9135 * config/i386/i386.c (ix86_data_alignment): Likewise.
9136 (ix86_local_alignment): Likewise.
9137 * dbxout.c (stabstr_O): Likewise.
9138 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
9139 * expr.c (const_vector_from_tree): Likewise.
9140 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
9141 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
9142 (fold_negate_expr_1, int_const_binop_1, const_binop)
9143 (fold_convert_const_int_from_real, optimize_bit_field_compare)
9144 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
9145 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
9146 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
9147 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
9148 (fold_not_const, round_up_loc): Likewise.
9149 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
9150 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
9151 (alloca_call_type): Likewise.
9152 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
9153 * godump.c (go_output_typedef): Likewise.
9154 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
9155 * internal-fn.c (get_min_precision): Likewise.
9156 * ipa-cp.c (ipcp_store_vr_results): Likewise.
9157 * ipa-polymorphic-call.c
9158 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
9159 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
9160 (ipa_modify_call_arguments): Likewise.
9161 * match.pd: Likewise.
9162 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
9163 * print-tree.c (print_node_brief, print_node): Likewise.
9164 * stmt.c (expand_case): Likewise.
9165 * stor-layout.c (layout_type): Likewise.
9166 * tree-affine.c (tree_to_aff_combination): Likewise.
9167 * tree-cfg.c (group_case_labels_stmt): Likewise.
9168 * tree-data-ref.c (dr_analyze_indices): Likewise.
9169 (prune_runtime_alias_test_list): Likewise.
9170 * tree-dump.c (dequeue_and_dump): Likewise.
9171 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
9172 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
9173 * tree-pretty-print.c (dump_generic_node): Likewise.
9174 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
9175 (simple_iv_with_niters): Likewise.
9176 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
9177 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
9178 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
9179 * tree-ssa-loop-niter.c (split_to_var_and_offset)
9180 (refine_value_range_using_guard, number_of_iterations_ne_max)
9181 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
9182 (get_cst_init_from_scev, record_nonwrapping_iv)
9183 (scev_var_range_cant_overflow): Likewise.
9184 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
9185 * tree-ssa-pre.c (compute_avail): Likewise.
9186 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
9187 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
9188 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
9189 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
9190 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
9191 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
9192 (dump_case_nodes, try_switch_expansion): Likewise.
9193 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
9194 (vect_do_peeling): Likewise.
9195 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
9196 * tree-vect-stmts.c (vectorizable_load): Likewise.
9197 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
9198 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
9199 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
9200 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
9201 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
9202 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
9203 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
9204 (evrp_dom_walker::before_dom_children): Likewise.
9205 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
9206 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
9207 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
9208 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
9209 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
9210 (get_range_pos_neg): Likewise.
9211 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
9212 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
9213 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
9214 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
9215 * config/avr/avr.c (avr_fold_builtin): Likewise.
9216 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
9217 * config/msp430/msp430.c (msp430_attr): Likewise.
9218 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
9219 * config/powerpcspe/powerpcspe-c.c
9220 (altivec_resolve_overloaded_builtin): Likewise.
9221 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
9222 (rs6000_expand_ternop_builtin): Likewise.
9223 * config/rs6000/rs6000-c.c
9224 (altivec_resolve_overloaded_builtin): Likewise.
9225 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
9226 (rs6000_expand_ternop_builtin): Likewise.
9227 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
9229 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9231 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
9232 when copying loop nest with only one inner loop.
9234 2017-10-10 Richard Biener <rguenther@suse.de>
9236 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
9237 blocks if SCEV is active.
9238 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
9240 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
9241 (scev_initialize): Assert we are not yet initialized.
9243 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9245 * tree-loop-distribution.c (generate_loops_for_partition): Remove
9246 inner loop's exit stmt by making it always exit the loop, otherwise
9247 we would generate an infinite empty loop.
9249 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9251 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
9252 renaming variables in new preheader if it's deleted.
9254 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9256 * tree-loop-distribution.c (struct partition): Remove unused field
9257 loops of the structure.
9258 (partition_alloc, partition_free): Ditto.
9259 (build_rdg_partition_for_vertex): Ditto.
9261 2017-10-09 Jeff Law <law@redhat.com>
9263 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
9264 return type to match prototype and documentation.
9266 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9268 * config/rs6000/rs6000.c (processor_costs): Move to ...
9269 * config/rs6000/rs6000.h: ... here.
9270 (rs6000_cost): Declare.
9272 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
9274 * except.c (setjmp_fn): New global variable.
9275 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
9276 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
9277 if DONT_USE_BUILTIN_SETJMP is defined.
9279 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9281 * target.def (insn_cost): New hook.
9282 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
9283 * doc/tm.texi: Regenerate.
9284 * rtlanal.c (insn_cost): Use the new hook.
9286 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9288 * combine.c (combine_validate_cost): Compute the new insn_cost,
9289 not just pattern_cost.
9290 (try_combine): Adjust comment.
9292 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9294 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
9296 * combine.c (uid_insn_cost): Adjust comment.
9297 (combine_validate_cost): Adjust comment. Use pattern_cost instead
9299 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
9300 * dse.c (find_shift_sequence): Ditto.
9301 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
9302 (bb_valid_for_noce_process_p): Use pattern_cost.
9303 * rtl.h (insn_rtx_cost): Delete.
9304 (pattern_cost): New prototype.
9305 (insn_cost): New prototype.
9306 * rtlanal.c (insn_rtx_cost): Rename to...
9307 (pattern_cost): ... this.
9310 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
9312 * config/i386/i386.md (*jcc_2): Remove insn pattern.
9313 (*jcc<mode>_0_r_i387): Ditto.
9314 (*jccxf_r_i387): Ditto.
9315 (*jcc<mode>_r_i387): Ditto.
9316 (*jccu<mode>_r_i387): Ditto.
9317 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
9318 (*jcc): Rename from *jcc_1.
9320 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9322 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
9323 deferred rescans after the lvx/stvx recombination pre-pass.
9325 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
9327 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
9328 memory operation instruction support.
9329 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
9330 (rs6000-ibm-aix[789]*): Likewise.
9331 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
9332 Document new functions.
9334 2017-10-09 Richard Biener <rguenther@suse.de>
9336 PR tree-optimization/82397
9337 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
9338 equality only for semantically equal trees.
9340 2017-10-09 Richard Biener <rguenther@suse.de>
9342 PR tree-optimization/82449
9343 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
9344 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
9345 allow constant addresses.
9346 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
9349 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9351 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
9354 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9357 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
9360 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9363 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
9365 2017-10-09 Jakub Jelinek <jakub@redhat.com>
9368 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
9369 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
9371 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
9373 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
9374 (WI_BINARY_PREDICATE_RESULT): Likewise.
9375 (wi::binary_traits::operator_result): New type.
9376 (wi::binary_traits::predicate_result): Likewise.
9377 (generic_wide_int::operator~, unary generic_wide_int::operator-)
9378 (generic_wide_int::operator==, generic_wide_int::operator!=)
9379 (generic_wide_int::operator&, generic_wide_int::and_not)
9380 (generic_wide_int::operator|, generic_wide_int::or_not)
9381 (generic_wide_int::operator^, generic_wide_int::operator+
9382 (binary generic_wide_int::operator-, generic_wide_int::operator*):
9384 (operator~, unary operator-, operator==, operator!=, operator&)
9385 (operator|, operator^, operator+, binary operator-, operator*): New
9387 * expr.c (get_inner_reference): Use wi::bit_and_not.
9388 * fold-const.c (fold_binary_loc): Likewise.
9389 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
9390 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
9391 (bit_value_binop): Likewise.
9392 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
9393 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
9394 (extract_range_from_binary_expr_1): Likewise.
9395 (masked_increment): Likewise.
9396 (simplify_bit_ops_using_ranges): Likewise.
9398 2017-10-09 Martin Jambor <mjambor@suse.cz>
9401 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
9402 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
9403 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
9405 (hsa_fixup_mov_insn_type): New function.
9406 (hsa_op_with_type::get_in_type): Use it.
9407 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
9408 immediates in an assert.
9409 (hsa_op_with_type::extend_int_to_32bit): New method.
9410 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
9411 types. Convert to dest type if necessary.
9412 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
9413 (reg_for_gimple_ssa): Pass false as min32int to
9414 hsa_type_for_scalar_tree_type.
9415 (gen_hsa_addr): Fixup type when creating addresable temporary.
9416 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
9417 (gen_hsa_unary_operation): Extend operands and convert to dest type if
9418 necessary. Call hsa_fixup_mov_insn_type.
9419 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
9420 extend operands and convert to dest type if necessary.
9421 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
9422 to dest type if necessary.
9423 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
9424 if conversion nt necessary and size matches.
9425 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
9426 to dest type if necessary.
9427 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
9428 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
9430 (gen_hsa_clrsb): Likewise.
9431 (gen_hsa_ffs): Likewise.
9432 (gen_hsa_divmod): Extend operands and convert to dest type if
9434 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
9436 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
9438 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
9439 Remove empty default arguments. Use a brace block as output
9441 (conditional return): Ditto.
9443 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
9444 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
9446 (group_ending_nop): Ditto.
9447 (doloop_end): Ditto.
9448 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
9449 (splitters for those): Ditto.
9451 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
9453 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
9454 a conditional jump (and the compare for it) so that pc_rtx is the
9456 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
9457 for the deleted and renamed ctr<mode>_internal[234] patterns.
9458 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
9459 Delete second conditional return pattern.
9460 (ctr<mode>_internal2): Delete this second bdnz pattern.
9461 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
9462 (ctr<mode>_internal4): Delete this second bdz pattern.
9464 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
9466 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
9467 (always_initialized_rtx_for_ssa_name_p): New predicate.
9468 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
9469 (finish_out_of_ssa): Free new field of SA.
9470 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
9471 * tree-ssa-coalesce.c: Include tree-ssa.h.
9472 (get_parm_default_def_partitions): Remove extern keyword.
9473 (get_undefined_value_partitions): New function.
9474 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
9475 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
9477 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
9479 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
9481 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
9483 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9485 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
9486 for targets that preffer 128bit.
9488 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9490 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
9492 2017-10-08 Olivier Hainque <hainque@adacore.com>
9494 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
9495 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
9497 (thumb_set_return_address): Likewise.
9499 2017-10-08 Olivier Hainque <hainque@adacore.com>
9501 * common/config/arm/arm-common.c (arm_except_unwind_info):
9502 Handle DWARF2_UNWIND_INFO.
9504 2017-10-07 Michael Collison <michael.collison@arm.com>
9506 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
9509 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
9511 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
9512 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
9513 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
9514 defined, force the creation of a new block for a dispatch label.
9516 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
9518 * invoke.texi (Wsuggest-attribute=cold): Document.
9519 * common.opt (Wsuggest-attribute=cold): New
9520 * ipa-pure-const.c (warn_function_cold): New function.
9521 * predict.c (compute_function_frequency): Use it.
9522 * predict.h (warn_function_cold): Declare.
9524 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
9526 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
9529 2017-10-06 Martin Liska <mliska@suse.cz>
9531 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
9532 keyword for member functions.
9533 (struct sanopt_tree_couple): New struct.
9534 (struct sanopt_tree_couple_hash): New function.
9535 (struct sanopt_ctx): Add new hash_map.
9536 (has_dominating_ubsan_ptr_check): New function.
9537 (record_ubsan_ptr_check_stmt): Likewise.
9538 (maybe_optimize_ubsan_ptr_ifn): Likewise.
9539 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
9540 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
9542 2017-10-06 Sudakshina Das <sudi.das@arm.com>
9545 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
9546 aarch64_simd_valid_immediate on CONST_VECTORs.
9547 (aarch64_reg_or_bic_imm): Likewise.
9549 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
9551 PR rtl-optimization/82396
9552 * haifa-sched.c (ready_sort_real): Disable qsort checking.
9554 2017-10-06 Sebastian Pop <sebpop@gmail.com>
9556 * graphite-dependences.c (scop_get_reads): Move code to...
9557 (scop_get_must_writes): Move code to...
9558 (scop_get_may_writes): Move code to...
9559 (scop_get_reads_and_writes): ... here.
9560 (scop_get_dependences): Call scop_get_reads_and_writes.
9562 2017-10-06 Jakub Jelinek <jakub@redhat.com>
9564 PR tree-optimization/82434
9565 * fold-const.h (can_native_encode_type_p,
9566 can_native_encode_string_p): Remove.
9567 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
9568 don't encode anything, just return what would be otherwise returned.
9569 (native_encode_fixed, native_encode_complex, native_encode_vector):
9571 (native_encode_string): Likewise. Inline by hand
9572 can_native_encode_string_p.
9573 (can_native_encode_type_p): Remove.
9574 (can_native_encode_string_p): Remove.
9575 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
9576 STRING_CSTs using can_native_encode_string_p, test all
9577 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
9578 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
9579 argument from native_encode_expr.
9580 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
9581 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
9584 2017-10-06 Richard Biener <rguenther@suse.de>
9586 PR tree-optimization/82397
9587 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
9588 operand_equal_p but rely on data_ref_compare_tree for detecting
9590 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
9591 to match up with dr_group_sort_cmp.
9593 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9596 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
9598 * config/s390/s390-builtin-types.def: Regenerate.
9600 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9603 * config/s390/s390-builtin-types.def: Regenerate.
9604 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
9605 Change flag from B_VXE to B_VX.
9606 (s390_vec_min_dbl): Remove B_VXE flag.
9608 2017-10-06 Richard Biener <rguenther@suse.de>
9610 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
9611 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
9612 translate_isl_ast_to_gimple::is_valid_rename,
9613 translate_isl_ast_to_gimple::get_rename,
9614 translate_isl_ast_to_gimple::get_def_bb_for_const,
9615 translate_isl_ast_to_gimple::get_new_name,
9616 translate_isl_ast_to_gimple::collect_all_ssa_names,
9617 translate_isl_ast_to_gimple::copy_loop_phi_args,
9618 translate_isl_ast_to_gimple::collect_all_ssa_names,
9619 translate_isl_ast_to_gimple::copy_loop_phi_args,
9620 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
9621 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
9622 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
9623 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
9624 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
9625 translate_isl_ast_to_gimple::copy_cond_phi_args,
9626 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
9627 translate_isl_ast_to_gimple::edge_for_new_close_phis,
9628 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
9629 translate_isl_ast_to_gimple::rename_uses,
9630 translate_isl_ast_to_gimple::rename_all_uses): Remove.
9631 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
9632 (set_rename_for_each_def): Likewise.
9633 (graphite_copy_stmts_from_block): Handle debug stmt resetting
9634 here. Handle rewriting SCEV analyzable uses here.
9635 (copy_bb_and_scalar_dependences): Generate code for PHI
9637 (graphite_regenerate_ast_isl): Adjust.
9638 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
9639 (add_write, add_read): New functions.
9640 (build_cross_bb_scalars_def): Use it and simplify.
9641 (build_cross_bb_scalars_use): Likewise.
9642 (graphite_find_cross_bb_scalar_vars): Inline into...
9643 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
9644 simulating out-of-SSA. Compute liveout and add dependencies.
9645 (build_scops): Force an empty entry block.
9646 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
9648 (sese_build_liveouts): Declare.
9649 (sese_trivially_empty_bb_p): Likewise.
9650 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
9651 compute liveout and debug_liveout.
9652 (sese_bad_liveouts_use): Remove.
9653 (sese_reset_debug_liveouts_bb): Likewise.
9654 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
9655 (sese_build_liveouts): Build liveout and debug_liveout and store
9657 (new_sese_info): Adjust.
9658 (free_sese_info): Likewise.
9659 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
9660 do not build liveout here.
9661 (move_sese_in_condition): Adjust region entry.
9662 (scev_analyzable_p): Match up with chrec_apply requirements.
9663 (sese_trivially_empty_bb_p): New.
9664 * tree-into-ssa.c (get_reaching_def): Properly support generating
9665 default-defs for incremental rewrite of anonymous names.
9667 2017-10-06 Richard Biener <rguenther@suse.de>
9669 * graphite-sese-to-poly.c (extract_affine): For casts increasing
9670 precision do not perform modulo reduction.
9672 2017-10-06 Richard Biener <rguenther@suse.de>
9674 PR tree-optimization/82436
9675 * tree-vect-slp.c (vect_supported_load_permutation_p): More
9676 conservatively choose the vectorization factor when checking
9677 whether we can perform the required load permutation.
9678 (vect_transform_slp_perm_load): Assert when we may not fail.
9680 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
9682 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
9683 message for incompatible -msdata=* and -mcall-* options.
9685 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9687 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
9688 rate for post-reload scheduling.
9690 2017-10-05 Tamar Christina <tamar.christina@arm.com>
9692 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
9694 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9696 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
9697 to improve monte carlo in scimark.
9699 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9701 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
9702 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
9703 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
9704 pentium4_cost, nocona_cost): Set reassociation width to 1.
9705 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
9706 width to 2 for fp operations and 1 otherwise.
9707 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
9709 (atom_cost): Set reassociation width to 2.
9710 (slm_cost, generic_cost): Set fp reassociation width
9711 to 2 and 1 otherwise.
9712 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
9713 (core_cost): Set fp reassociation width to 4 and vector to 2.
9714 (ix86_reassociation_width): Rewrite using cost table; special case
9715 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
9716 and TARGET_AVX128_OPTIMAL.
9717 * config/i386/i386.h (processor_costs): Add
9718 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
9719 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
9720 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
9721 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
9722 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
9723 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
9725 2017-10-05 Nathan Sidwell <nathan@acm.org>
9727 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
9729 2017-10-05 Tamar Christina <tamar.christina@arm.com>
9731 * config/arm/arm.c (arm_test_fpu_data): New.
9732 (arm_run_selftests): Call arm_test_fpu_data.
9734 2017-10-04 Nathan Sidwell <nathan@acm.org>
9736 * toplev.c (toplev::main): Remove excess parens on pretty_printer
9738 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
9740 2017-10-04 Sudakshina Das <sudi.das@arm.com>
9742 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
9743 check type for aarch64_simd_valid_immediate.
9744 (aarch64_output_simd_mov_immediate): Update prototype.
9745 (aarch64_simd_valid_immediate): Update prototype.
9746 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
9747 support for ORR-immediate.
9748 (and<mode>3): modified pattern to add support for BIC-immediate.
9749 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
9750 now checks for valid immediate for BIC and ORR based on new enum
9752 (aarch64_output_simd_mov_immediate): Function now used to output
9753 BIC/ORR imm as well based on new enum argument.
9754 * config/aarch64/constraints.md (Do): New vector immediate constraint.
9756 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
9757 (aarch64_reg_or_bic_imm): Likewise.
9759 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9761 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
9762 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
9764 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
9768 PR rtl-optimization/82396
9769 * haifa-sched.c (autopref_multipass_init): Simplify
9771 (autopref_rank_data): Simplify sort order.
9772 * sched-int.h (autopref_multipass_data_): Remove
9773 multi_mem_insn_p, min_offset and max_offset.
9775 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9777 * doc/sourcebuild.texi: Document vect_peeling_profitable.
9779 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9781 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
9784 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9786 * doc/sourcebuild.texi: Document vect_long_mult.
9788 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
9790 PR tree-optimization/82413
9791 * fold-const.c (build_range_check): Use widest_int when comparing
9792 the maximum ETYPE value with HIGH.
9794 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
9796 PR rtl-optimization/82396
9797 * haifa-sched.c (autopref_multipass_init): Simplify
9799 (autopref_rank_data): Simplify sort order.
9800 * sched-int.h (autopref_multipass_data_): Remove
9801 multi_mem_insn_p, min_offset and max_offset.
9803 2017-10-04 Jakub Jelinek <jakub@redhat.com>
9805 PR tree-optimization/82381
9806 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
9807 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
9810 PR tree-optimization/82374
9811 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
9812 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
9813 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
9814 current_function_decl to the new decl.
9816 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
9818 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
9819 helper macro for IEEE float128 hardware built-in functions.
9820 (SQRTF128_ODD): Add built-in functions with the round-to-odd
9822 (TRUNCF128_ODD): Likewise.
9823 (ADDF128_ODD): Likewise.
9824 (SUBF128_ODD): Likewise.
9825 (MULF128_ODD): Likewise.
9826 (DIVF128_ODD): Likewise.
9827 (FMAF128_ODD): Likewise.
9828 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
9829 UNSPEC_TRUNC_ROUND_TO_ODD.
9830 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
9831 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
9832 floating point round to odd instructions.
9833 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
9834 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
9835 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
9836 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
9837 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
9838 (trunc<mode>sf2_hw): Change the truncate with round to odd
9839 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
9840 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
9841 to odd hardware instructions.
9842 (sub<mode>3_odd): Likewise.
9843 (mul<mode>3_odd): Likewise.
9844 (div<mode>3_odd): Likewise.
9845 (sqrt<mode>2_odd): Likewise.
9846 (fma<mode>4_odd): Likewise.
9847 (fms<mode>4_odd): Likewise.
9848 (nfma<mode>4_odd): Likewise.
9849 (nfms<mode>4_odd): Likewise.
9850 (trunc<mode>df2_odd): Change the truncate with round to odd
9851 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
9853 * doc/extend.texi (PowerPC built-in functions): Update documentation
9854 for existing IEEE float128-bit built-in functions. Add built-in
9855 functions that generate the IEEE 128-bit floating point round to
9858 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
9860 PR rtl-optimization/77729
9861 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
9862 to (X&(C1&~C2))|C2 transformations.
9864 2017-10-03 Martin Jambor <mjambor@suse.cz>
9866 PR tree-optimization/82363
9867 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
9868 mismatch, mark lacc written regardless of racc.
9870 2017-10-03 Jakub Jelinek <jakub@redhat.com>
9872 PR tree-optimization/82381
9873 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
9874 stmt_to_insert nor wheather SSA_NAMEs are default defs.
9875 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
9876 fallthrough into reassoc_stmt_dominates_stmt_p.
9879 * combine.c (combine_instructions): Don't combine in unreachable
9882 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
9885 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
9886 function to not use the have_cpu variable. Do not set cpu_index,
9887 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
9889 (rs6000_valid_attribute_p): Remove duplicate initializations of
9890 old_optimize and func_optimize.
9891 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
9892 (rs6000_activate_target_options): Make global.
9893 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
9896 2017-10-02 Jakub Jelinek <jakub@redhat.com>
9898 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
9899 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
9900 Set *poffset to 0 and *psize and *pmax_size to -1 if
9901 *poffset + *psize overflows in HOST_WIDE_INT.
9903 PR tree-optimization/82387
9904 PR tree-optimization/82388
9905 PR tree-optimization/82389
9906 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
9907 instead of live_bytes non-NULL.
9909 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
9912 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
9915 2017-10-02 Richard Biener <rguenther@suse.de>
9917 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
9918 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
9919 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
9921 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
9923 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
9924 requested precision matches the type's.
9925 * calls.c (alloc_max_size): Calculate the new candidate size as
9926 a widest_int and use wi::to_widest when comparing it with the
9927 current candidate size.
9928 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
9929 zero rather than integer_zero_node.
9930 * match.pd: Check for a no-op conversion before using wi::add
9931 rather than after. Use tree_to_uhwi when summing small shift
9932 counts into an unsigned int.
9934 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
9935 Alan Hayward <alan.hayward@arm.com>
9936 David Sherwood <david.sherwood@arm.com>
9939 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
9940 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
9941 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
9942 POINTER_AND_FP_REGS.
9944 2017-10-02 Richard Biener <rguenther@suse.de>
9946 PR tree-optimization/82355
9947 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
9948 a mapping for the enclosing loop but avoid generating one for
9950 (copy_bb_and_scalar_dependences): Remove premature codegen
9951 error on PHIs in blocks duplicated into multiple places.
9952 * graphite-scop-detection.c
9953 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
9954 in the region use it as loop and nest to analyze the DR in.
9955 (try_generate_gimple_bb): Likewise.
9956 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
9957 (add_loop_constraints): For blocks in a loop not in the region
9958 create a dimension with a single iteration.
9959 * sese.h (gbb_loop_at_index): Remove assert.
9961 2017-10-01 Kevin Buettner <kevinb@redhat.com>
9963 * omp-expand.c (adjust_context_scope): New function.
9964 (expand_parallel_call): Call adjust_context_scope.
9966 2017-10-01 Jeff Law <law@redhat.com>
9968 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
9969 dom_opt_dom_walker class with direct access to private members.
9970 Add comments. Call test_for_singularity.
9971 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
9972 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
9973 m_dummy_cond anymore.
9974 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
9976 (pass_dominator:execute): Build the dummy_cond here and pass it
9977 to the dom_opt_dom_walker ctor.
9978 (test_for_singularity): New function.
9980 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
9981 Maya Rashish <coypu@sdf.org>
9983 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
9984 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
9985 (alpha*-*-netbsd*) Use nbsd_tm_file.
9986 (arm*-*-netbsdelf*) Likewise.
9987 (i[34567]86-*-netbsdelf*) Likewise.
9988 (x86_64-*-netbsd*) Likewise.
9989 (mips*-*-netbsd*) Likewise.
9990 (powerpc-*-netbsd*) Likewise.
9991 (sh*-*-netbsd*) Likewise.
9992 (sparc-*-netbsdelf*) Likewise.
9993 (sparc64-*-netbsd*) Likewise.
9994 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
9996 (vax-*-netbsdelf*) Likewise.
9997 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
9998 (UINT_FAST8_TYPE) Likewise.
9999 (INT_FAST16_TYPE) Check CHAR_FAST16.
10000 (UINT_FAST16_TYPE) Likewise.
10002 2017-09-30 Jakub Jelinek <jakub@redhat.com>
10005 * config/i386/i386.md
10006 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
10007 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
10008 *divmodsi4_zext_2): New define_insn_and_split.
10009 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
10010 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
10011 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
10012 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
10013 New define_insn_and_split.
10014 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
10015 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
10016 operands[1] having DImode when mode is SImode.
10018 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
10019 always SImode for DIV and MOD in REG_EQUAL notes.
10021 2017-09-29 Yury Gribov <tetra2005@gmail.com>
10023 PR middle-end/82319
10024 * match.pd: Fix handling of NaNs in pattern.
10026 2017-09-29 Jeff Law <law@redhat.com>
10028 * sbitmap.c (bitmap_bit_in_range_p): New function.
10029 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
10030 * tree-ssa-dse.c (live_bytes_read): New function.
10031 (dse_classify_store): Ignore reads of dead bytes.
10033 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
10034 typos and whitespace errors.
10035 * config/i386/predicates.md (address_no_seg_operand): Likewise.
10036 * config/s390/s390.c (s390_emit_prologue): Likewise.
10038 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10041 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
10042 with a symbol for LRA.
10044 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10046 PR rtl-optimization/82338
10047 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
10049 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
10051 * genmodes.c (calc_wider_mode): Suppress qsort macro.
10052 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
10053 (qsort_chk): Declare.
10054 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
10055 (qsort_chk): New function.
10057 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10059 PR tree-optimization/82337
10060 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
10061 phi definition if the PHI result appears in an abnormal PHI.
10062 (find_basis_for_base_expr): Don't record a basis if the LHS of the
10063 basis appears in an abnormal PHI.
10065 2017-09-29 Richard Biener <rguenther@suse.de>
10067 * graphite-isl-ast-to-gimple.c
10068 (translate_isl_ast_to_gimple::set_codegen_error): New function.
10069 (binary_op_to_tree): Use it.
10070 (get_rename_from_scev): Likewise.
10071 (copy_loop_phi_nodes): Likewise.
10072 (copy_bb_and_scalar_dependences): Likewise.
10073 (translate_pending_phi_nodes): Likewise.
10075 2017-09-29 Jakub Jelinek <jakub@redhat.com>
10078 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
10079 for movabsq $(i32 << shift), r64.
10081 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
10083 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
10084 index when encoding %esp as %rsp to avoid 0x67 prefix.
10086 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
10088 * config/i386/i386.md (*movsf_internal, *movdf_internal):
10089 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
10091 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
10093 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
10094 Extensions with more than 16 double VFP registers.
10095 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
10096 to_clear_mask and all code related to it. Replace the remaining
10097 entry by a sbitmap and adapt code accordingly.
10099 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
10101 * brig-builtins.def: Change pure attributes to const.
10103 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10105 * config.gcc (default_gnu_indirect_function): Default to yes for
10106 sparc*-*-linux* with glibc.
10108 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10110 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
10111 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
10112 when creating .init_array and .fini_array sections with priority
10115 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
10118 * config/aarch64/aarch64.c
10119 (aarch64_builtin_support_vector_misalignment): Always return false
10120 when misalignment is unknown.
10122 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
10124 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
10125 this function to return false if the definition used by the swap
10126 instruction is artificial, or if the memory address from which the
10127 constant value is loaded is not represented by a base address held
10128 in a register or if the base address register is a frame or stack
10129 pointer. Additionally, return false if the base address of the
10130 loaded constant is a SYMBOL_REF but is not considered to be a
10132 (replace_swapped_load_constant): New function.
10133 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
10134 loaded constant vector with a load of a swapped constant vector.
10136 2017-09-27 Carl Love <cel@us.ibm.com>
10138 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
10139 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
10140 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
10143 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
10145 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
10146 first, always call autopref_rank_data otherwise.
10148 2017-09-27 Richard Biener <rguenther@suse.de>
10150 * graphite-scop-detection.c (find_scop_parameters): Move
10151 loop bound handling ...
10152 (gather_bbs::before_dom_children): ... here, avoiding the need
10153 to build scop_info->loop_nest.
10154 (record_loop_in_sese): Remove.
10155 * sese.h (sese_info_t::loop_nest): Remove.
10156 * sese.c (new_sese_info): Do not allocate loop_nest.
10157 (free_sese_info): Do not free loop_nest.
10159 2017-09-27 Jakub Jelinek <jakub@redhat.com>
10162 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
10163 lhs from calls if the lhs has addressable type.
10165 2017-09-27 Richard Biener <rguenther@suse.de>
10167 * graphite.h (scop::max_alias_set): New member.
10168 * graphite-scop-detection.c: Remove references to non-existing
10169 --param in comments.
10170 (build_alias_sets): Record the maximum alias set used for drs.
10171 (build_scops): Support zero as unlimited for
10172 --param graphite-max-arrays-per-scop.
10173 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
10174 and inline into ...
10175 (build_poly_sr_1): ... here. Compute alias set based on the
10176 maximum alias set used for drs rather than
10177 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
10179 2017-09-27 Richard Biener <rguenther@suse.de>
10181 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
10182 --param loop-block-tile-size=0 to disable tiling.
10184 2017-09-27 Richard Biener <rguenther@suse.de>
10186 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
10187 (graphite-max-nb-scop-params): Document special value zero.
10188 * domwalk.h (dom_walker::STOP): New symbolical constant.
10189 (dom_walker::dom_walker): Add optional parameter for bb to
10191 (dom_walker::~dom_walker): Declare.
10192 (dom_walker::before_dom_children): Document STOP return value.
10193 (dom_walker::m_user_bb_to_rpo): New member.
10194 (dom_walker::m_bb_to_rpo): Likewise.
10195 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
10196 mapping here if not provided by the user.
10197 (dom_walker::~dom_walker): Free bb to RPO mapping if not
10198 provided by the user.
10199 (dom_walker::STOP): Define.
10200 (dom_walker::walk): Do not compute bb to RPO mapping here.
10201 Support STOP return value from before_dom_children to stop
10203 * graphite-optimize-isl.c (optimize_isl): If the schedule
10204 is the same still generate code if -fgraphite-identity
10205 or -floop-parallelize-all are given.
10206 * graphite-scop-detection.c: Include cfganal.h.
10207 (gather_bbs::gather_bbs): Get and pass through bb to RPO
10209 (gather_bbs::before_dom_children): Return STOP for BBs
10211 (build_scops): Compute bb to RPO mapping and pass it to
10212 the domwalk. Treat --param graphite-max-nb-scop-params=0
10213 as not limiting the number of params.
10214 * graphite.c (graphite_initialize): Remove limit on the
10215 number of basic-blocks in a function.
10216 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
10217 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
10218 default value of 10.
10220 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10222 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
10223 Adjust code to eliminate needing to do the shift right 32-bits
10224 operation after XSCVDPSPN.
10226 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10228 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
10229 ((X / Y) != 0 -> X >= Y): Likewise.
10231 2017-09-26 Carl Love <cel@us.ibm.com>
10233 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
10234 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
10235 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
10236 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
10237 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
10238 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
10239 definitions and overloading.
10240 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
10241 statement for P9V_BUILTIN_XST_LEN_R.
10242 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
10243 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
10244 define_expand and define_insn for the instructions and builtins.
10245 * doc/extend.texi: Update the built-in documentation file for the new
10246 built-in functions.
10247 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
10248 define_insn for the instructions
10250 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
10253 * gcc/config/netbsd-protos.h: New file.
10254 * gcc/config/netbsd.c: New file.
10255 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
10256 * gcc/config/t-netbsd: New file.
10257 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
10258 (tmake_file) Add t-netbsd.
10259 (extra_objs) Add netbsd.o.
10261 2017-09-26 Janus Weil <janus@gcc.gnu.org>
10265 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
10267 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10269 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
10270 sign extension from a vector register to a GPR by doing a 32-bit
10271 direct move and then an EXTSW.
10272 (extendsi<mode>2 splitter): Likewise.
10273 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
10274 right or vector extract after doing XSCVDPSPN. Use
10275 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
10277 (movdi_from_sf_zero_ext): Likewise.
10278 (reload_gpr_from_vsxsf): Likewise.
10279 (p8_mfvsrd_4_disf): Delete, no longer used.
10280 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
10281 then needing to move the SFmode to a GPR to use the XSCVDPSP
10282 instruction instead of FRSP and XSCVDPSPN.
10283 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
10284 it is adjacent to the other XSCVSPDP insns.
10285 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
10286 SFmode to be in traditional Altivec registers.
10287 (vsx_xscvdpspn): Eliminate useless alternative constraint.
10288 (vsx_xscvspdpn): Likewise.
10289 (vsx_xscvspdpn_scalar): Likewise.
10291 2017-09-26 Martin Jambor <mjambor@suse.cz>
10293 * tree-sra.c (compare_access_positions): Put integral types first,
10294 stabilize sorting of integral types, remove conditions putting
10295 non-full-precision integers last.
10296 (sort_and_splice_var_accesses): Disable scalarization if a
10297 non-integert would be represented by a non-full-precision integer.
10299 2017-09-26 Joseph Myers <joseph@codesourcery.com>
10301 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
10302 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
10303 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
10304 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
10305 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
10306 conditionals inside the function instead of around it. Call
10307 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
10308 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
10310 2017-09-26 Richard Biener <rguenther@suse.de>
10312 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
10314 (scop_detection::build_scop_breadth): ... this. Removed.
10315 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
10316 (scop_detection::harmful_stmt_in_bb): Likewise.
10317 (scop_detection::graphite_can_represent_stmt): Likewise.
10318 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
10319 (scop_detection::can_represent_loop): Remove recursion, fold in ...
10320 (scop_detection::can_represent_loop_1): ... this. Removed.
10321 (scop_detection::harmful_loop_in_region): Simplify after inlining
10322 the above and remove more quadraticness.
10323 (build_scops): Adjust.
10324 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
10327 2017-09-26 Jakub Jelinek <jakub@redhat.com>
10330 * config/i386/i386.c (ix86_print_operand_address_as): Only test
10331 REGNO (base) == SP_REG if base is a REG.
10333 PR middle-end/35691
10334 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
10335 if it is different SSA_NAME.
10336 (optimize_range_tests_cmp_bitwise): New function.
10337 (optimize_range_tests): Call it.
10339 2017-09-26 Richard Biener <rguenther@suse.de>
10341 PR tree-optimization/82321
10342 * graphite.c (canonicalize_loop_closed_ssa): Properly check
10343 for the def being inside the loop.
10345 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10347 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
10349 * config/s390/s390-builtins.def: Fix constraint on op4.
10351 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10353 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
10354 independent expanders.
10355 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
10356 ("vec_ordered", "vec_unordered"): New expanders.
10358 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10360 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
10363 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10365 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
10366 vec_unpacks_lo_v16qi.
10367 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
10369 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10371 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
10372 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
10373 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
10375 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10377 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
10379 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
10381 ("*vec_slb<mode>"): New insn pattern.
10382 ("vec_shr_<mode>"): New expander.
10383 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
10384 and force the shift count operand to V16QImode.
10385 ("vec_srb<mode>"): Set shift count mode to V16QI.
10387 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10389 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
10390 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
10391 ("vec_widen_smult_hi_<mode>"): New expander definitions.
10393 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
10396 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
10398 2017-09-26 Richard Biener <rguenther@suse.de>
10400 PR tree-optimization/82320
10401 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
10404 2017-09-25 Jeff Law <law@redhat.com>
10406 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
10407 prototype for new argument.
10408 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
10409 mostly extracted from rs6000_emit_allocate_stack.
10410 (rs6000_emit_probe_stack_range_stack_clash): New function.
10411 (rs6000_emit_allocate_stack): Call
10412 rs6000_emit_probe_stack_range_stack_clash as needed.
10413 (rs6000_emit_probe_stack_range): Add additional argument
10414 to call to gen_probe_stack_range{si,di}.
10415 (output_probe_stack_range): New.
10416 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
10417 (output_probe_stack_range_stack_clash): New.
10418 (rs6000_emit_prologue): Emit notes into dump file as requested.
10419 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
10420 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
10421 Add additional operand and pass it to output_probe_stack_range.
10423 2017-09-25 Bin Cheng <bin.cheng@arm.com>
10425 PR tree-optimization/82163
10426 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
10427 (checking_verify_loop_closed_ssa): New parameter.
10428 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
10429 (check_loop_closed_ssa_stmt): Delete.
10430 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
10431 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
10432 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
10435 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
10437 * brig-builtins.def: Treat HSAIL barrier builtins as
10438 setjmp/longjump style functions.
10440 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
10442 * target.def (constant_alignment): New hook.
10443 * defaults.h (CONSTANT_ALIGNMENT): Delete.
10444 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
10445 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
10446 * doc/tm.texi: Regenerate.
10447 * targhooks.h (default_constant_alignment): Declare.
10448 (constant_alignment_word_strings): Likewise.
10449 * targhooks.c (default_constant_alignment): New function.
10450 (constant_alignment_word_strings): Likewise.
10451 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
10452 instead of CONSTANT_ALIGNMENT.
10453 * varasm.c (align_variable, get_variable_align, build_constant_desc)
10454 (force_const_mem): Likewise.
10455 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
10456 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
10457 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
10458 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10459 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
10461 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
10462 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10463 constant_alignment_word_strings.
10464 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
10465 (CONSTANT_ALIGNMENT): Likewise.
10466 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10467 (arm_constant_alignment): New function.
10468 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
10469 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10470 constant_alignment_word_strings.
10471 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
10472 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10473 constant_alignment_word_strings.
10474 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
10475 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10476 (cris_constant_alignment): New function.
10477 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
10478 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10479 (epiphany_constant_alignment): New function.
10480 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
10481 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10482 constant_alignment_word_strings.
10483 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
10484 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10485 constant_alignment_word_strings.
10486 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
10487 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10488 constant_alignment_word_strings.
10489 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
10490 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
10491 * config/i386/i386.c (ix86_constant_alignment): Make static.
10492 Use the same interface as the target hook.
10493 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10494 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
10495 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10496 constant_alignment_word_strings.
10497 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
10498 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
10499 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10500 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
10501 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10502 constant_alignment_word_strings.
10503 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
10504 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10505 constant_alignment_word_strings.
10506 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
10507 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10508 constant_alignment_word_strings.
10509 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
10510 * config/microblaze/microblaze.c (microblaze_constant_alignment):
10512 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10513 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
10514 * config/mips/mips.c (mips_constant_alignment): New function.
10515 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10516 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
10517 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
10518 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10519 (mmix_constant_alignment): Make static. Use the same interface
10520 as the target hook.
10521 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
10522 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10523 constant_alignment_word_strings.
10524 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
10525 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10526 constant_alignment_word_strings.
10527 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
10528 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10529 constant_alignment_word_strings.
10530 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
10531 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10532 (rs6000_constant_alignment): New function.
10533 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
10534 * config/riscv/riscv.c (riscv_constant_alignment): New function.
10535 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10536 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
10537 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10538 (rs6000_constant_alignment): New function.
10539 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
10540 * config/s390/s390.c (s390_constant_alignment): New function.
10541 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10542 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
10543 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10544 constant_alignment_word_strings.
10545 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
10546 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10547 (sparc_constant_alignment): New function.
10548 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
10549 * config/spu/spu.c (spu_constant_alignment): New function.
10550 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10551 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
10552 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10553 constant_alignment_word_strings.
10554 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
10555 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10556 constant_alignment_word_strings.
10557 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
10558 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10559 constant_alignment_word_strings.
10560 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
10561 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10562 (visium_constant_alignment): New function.
10563 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
10564 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10565 (xtensa_constant_alignment): New function.
10566 * system.h (CONSTANT_ALIGNMENT): Poison.
10568 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
10570 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10571 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
10572 (rs6000_builtin_valid_without_lhs): New helper function.
10573 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10574 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
10576 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
10578 * target.h (vec_perm_indices): Use unsigned short rather than
10580 (auto_vec_perm_indices): Likewise.
10581 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
10582 Use unsigned int rather than unsigned char.
10583 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
10585 2017-09-25 Richard Biener <rguenther@suse.de>
10587 * cfgloop.h (sort_sibling_loops): Declare.
10588 * cfgloop.c (sort_sibling_loops_cmp): New helper.
10589 (sort_sibling_loops): New function sorting the sibling loop list
10591 * graphite.c (graphite_transform_loops): Sort sibling loops.
10593 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
10595 * target.def (vec_perm_const_ok): Change sel parameter to
10597 * optabs-query.c (can_vec_perm_p): Update accordingly.
10598 * doc/tm.texi: Regenerate.
10599 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
10600 auto_vec_perm_indices and remove separate nelt field.
10601 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
10602 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
10603 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
10604 (aarch64_expand_vec_perm_const): Update accordingly.
10605 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
10606 to vec_perm_indices.
10607 * config/arm/arm.c (expand_vec_perm_d): Change perm to
10608 auto_vec_perm_indices and remove separate nelt field.
10609 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
10610 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
10611 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
10613 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
10614 to vec_perm_indices.
10615 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
10616 sel to vec_perm_indices.
10617 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
10618 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
10619 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
10621 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
10624 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
10627 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
10628 on the FUNCTION_DECL function context if it has a DIE that is a
10631 2017-09-25 Richard Biener <rguenther@suse.de>
10633 PR tree-optimization/82285
10634 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
10637 2017-09-25 Tom de Vries <tom@codesourcery.com>
10641 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
10644 2017-09-25 Richard Biener <rguenther@suse.de>
10646 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
10647 ISL errors other than isl_error_quota happen. Dump if the
10648 schedule is the same.
10649 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
10650 errors instead of aborting inside ISL.
10652 2017-09-25 Iain Sandoe <iain@codesourcery.com>
10655 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
10656 of libgcc_eh for m64.
10657 * config/i386/darwin64.h: Likewise.
10659 2017-09-25 Richard Biener <rguenther@suse.de>
10661 PR middle-end/82144
10662 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
10663 attribute for incomplete types nor twice for complete ones.
10665 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
10668 * config/i386/i386.c (ix86_print_operand_address_as): Encode
10669 %esp as %rsp to avoid 0x67 prefix if there is no index or base
10672 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
10675 * config/i386/i386.opt (mprefer-avx256): Use
10676 ix86_target_flags variable.
10677 * config/i386/i386.c (ix86_target_string): Move
10678 -mprefer-avx256 to flag2_opts.
10680 2017-09-22 Jakub Jelinek <jakub@redhat.com>
10682 PR middle-end/35691
10683 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
10684 and x != -1 | y != -1 into (x & y) != -1.
10686 2017-09-22 Steve Ellcey <sellcey@cavium.com>
10688 * config.gcc: Add new case statement to set
10689 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
10690 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
10691 s390x-*-linux* case statements. Added aarch64 to the list of
10692 supported architectures.
10694 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10696 PR tree-optimization/82289
10697 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
10698 STMT_VINFO_RELEVANT_P.
10700 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10701 Alan Hayward <alan.hayward@arm.com>
10702 David Sherwood <david.sherwood@arm.com>
10704 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
10705 for VR_RANGE only; don't allow VR_ANTI_RANGE.
10706 (extract_range_from_binary_expr_1): Don't call
10707 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
10709 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10710 Alan Hayward <alan.hayward@arm.com>
10711 David Sherwood <david.sherwood@arm.com>
10713 * target.def (preferred_vector_alignment): New hook.
10714 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
10716 * doc/tm.texi: Regenerate.
10717 * targhooks.h (default_preferred_vector_alignment): Declare.
10718 * targhooks.c (default_preferred_vector_alignment): New function.
10719 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
10721 (DR_TARGET_ALIGNMENT): New macro.
10722 (aligned_access_p): Update commentary.
10723 (vect_known_alignment_in_bytes): New function.
10724 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
10726 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
10727 Calculate the misalignment based on the target alignment rather than
10729 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
10730 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
10731 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
10732 the target alignment, rather than masking the element misalignment
10733 with the number of elements in a vector. Also use the target
10734 alignment when calculating the maximum number of peels.
10735 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
10736 instead of TYPE_ALIGN_UNIT.
10737 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
10738 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
10739 (vect_create_addr_base_for_vector_ref): Update call accordingly.
10740 (vect_create_data_ref_ptr): Likewise.
10741 (vect_setup_realignment): Realign by ANDing with
10742 -DR_TARGET_MISALIGNMENT.
10743 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
10744 the number of peels based on DR_TARGET_ALIGNMENT.
10745 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
10746 with the guaranteed alignment boundary when deciding whether
10748 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
10749 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
10750 (ensure_base_align): Remove stmt_info parameter. Get the
10751 target base alignment from DR_TARGET_ALIGNMENT.
10752 (vectorizable_store): Update call accordingly. Interpret
10753 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
10755 (vectorizable_load): Likewise.
10757 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10758 Alan Hayward <alan.hayward@arm.com>
10759 David Sherwood <david.sherwood@arm.com>
10761 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
10762 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
10763 (vect_enhance_data_refs_alignment): Likewise.
10765 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
10767 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
10768 error. Only quit immediately if parsing is complete.
10769 (BEGIN): Initialize fatal_err and parse_done.
10770 (begin fpu, end fpu): Check number of arguments.
10771 (begin arch, end arch): Likewise.
10772 (begin cpu, end cpu): Likewise.
10773 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
10774 (optalias): Likewise.
10776 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
10778 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
10779 * config/arm/arm-isa.h: Delete. Move definitions to ...
10780 * arm-cpus.in: ... here. Use new feature and fgroup values.
10781 * config/arm/arm.c (arm_option_override): Use lower case for feature
10783 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
10784 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
10785 * config/arm/parsecpu.awk (END): Add new command 'isa'.
10787 (print_isa_bits_for): New function.
10788 (gen_isa): New function.
10789 (gen_comm_data): Use print_isa_bits_for.
10790 (define feature): New keyword.
10791 (define fgroup): New keyword.
10792 * config/arm/t-arm (TM_H): Remove.
10793 (GTM_H): Add arm-isa.h.
10794 (arm-isa.h): Add rule to generate file.
10795 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
10796 case for feature bit names.
10798 2017-09-22 Richard Biener <rguenther@suse.de>
10800 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
10802 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
10803 print of no dependency loops ...
10804 * graphite.c (graphite_transform_loops): ... here.
10805 (canonicalize_loop_closed_ssa_form): Work from inner to outer
10807 (same_close_phi_node, remove_duplicate_close_phi,
10808 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
10809 (canonicalize_loop_closed_ssa): ... here and simplify.
10810 * graphite-optimize-isl.c: Include tree-vectorizer.h.
10811 (optimize_isl): Use dump_printf_loc to tell when we stopped
10812 optimizing because of an ISL timeout.
10814 2017-09-22 Richard Biener <rguenther@suse.de>
10816 PR tree-optimization/82291
10817 * tree-if-conv.c (predicate_mem_writes): Make sure to
10818 remove writes in blocks predicated with false.
10820 2017-09-22 Richard Biener <rguenther@suse.de>
10822 * sese.c: Include cfganal.h.
10823 (if_region_set_false_region): Remove.
10824 (create_if_region_on_edge): Likewise.
10825 (move_sese_in_condition): Re-implement without destroying
10828 2017-09-22 Richard Biener <rguenther@suse.de>
10830 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
10831 Verify both BBs contain loop PHI nodes before dispatching to
10832 copy_loop_phi_args.
10833 (graphite_regenerate_ast_isl): Do not recompute dominators,
10834 do not verify three times. Restructure for clarity.
10835 * graphite-scop-detection.c (same_close_phi_node,
10836 remove_duplicate_close_phi, make_close_phi_nodes_unique,
10837 defined_in_loop_p, canonicalize_loop_closed_ssa,
10838 canonicalize_loop_closed_ssa_form): Simplify, remove excess
10839 checking and SSA rewrite, move to ...
10840 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
10841 (graphite_initialize): Do not pass in ctx, do not reset the
10842 SCEV cache, compute only dominators.
10843 (graphite_transform_loops): Allocate ISL ctx after
10844 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
10845 Maintain post-dominators only around build_scops.
10846 * sese.c (if_region_set_false_region): Make static. Free
10847 and recompute dominators.
10848 (move_sese_in_condition): Assert we don't get called with
10849 post-dominators computed.
10850 * sese.h (if_region_set_false_region): Remove.
10852 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
10854 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
10855 mode attribute for TARGET_AVX512VL.
10857 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
10859 * config/i386/i386.opt (mprefer-avx256): New option.
10860 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
10862 (ix86_preferred_simd_mode): Return 256-bit AVX modes
10863 for TARGET_PREFER_AVX256.
10864 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
10866 2017-09-21 Jeff Law <law@redhat.com>
10868 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
10869 Fix dump output if the only stack space is for pushed registers.
10871 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10873 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
10876 2017-09-21 Martin Sebor <msebor@redhat.com>
10879 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
10880 code (in C++) or code that triggers warnings.
10882 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
10884 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
10886 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
10888 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
10889 * sched-rgn.c: Ditto.
10890 * sel-sched-ir.c: Ditto.
10892 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
10894 * toplev.h (set_random_seed): Adjust return type.
10895 * toplev.c (init_local_tick): Move eager initialization of random_seed
10896 to get_random_seed. Adjust comment.
10897 (init_random_seed): Inline to get_random_seed, delete.
10898 (get_random_seed): Initialize random_seed lazily.
10899 (set_random_seed): Do not return previous value.
10900 (print_switch_value): Do not call get_random_seed.
10902 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
10904 * cgraph.c (delete_function_version): New, broken out from...
10905 (cgraph_node::delete_function_version): ...here. Rename to
10906 cgraph_node::delete_function_version_by_decl. Update all uses.
10907 (cgraph_node::remove): Call delete_function_version.
10909 2017-09-21 Jakub Jelinek <jakub@redhat.com>
10912 * tree-inline.c (expand_call_inline): Emit clobber stmts for
10913 VAR_DECLs to which addressable non-volatile parameters are mapped
10914 and for id->retvar after the return value assignment. Clear
10915 id->retval and id->retbnd after inlining.
10917 2017-09-21 Richard Biener <rguenther@suse.de>
10919 PR tree-optimization/82276
10920 PR tree-optimization/82244
10921 * tree-vrp.c (build_assert_expr_for): Set
10922 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
10924 (remove_range_assertions): Revert earlier change.
10926 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
10929 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
10931 2017-09-21 Richard Biener <rguenther@suse.de>
10933 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10934 Restore valid IL after code generation errors.
10935 * graphite.c (graphite_transform_loops): Diagnose code
10936 generation issues as MSG_MISSED_OPTIMIZATION and continue
10937 with processing SCOPs.
10939 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10940 Alan Hayward <alan.hayward@arm.com>
10941 David Sherwood <david.sherwood@arm.com>
10943 * calls.c (compute_argument_addresses): Use simplify_gen_binary
10944 rather than choosing between plus_constant and gen_rtx_<CODE>.
10945 * expr.c (emit_push_insn): Likewise.
10946 (expand_expr_real_2): Likewise.
10948 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10949 Alan Hayward <alan.hayward@arm.com>
10950 David Sherwood <david.sherwood@arm.com>
10952 * loop-unroll.c (split_iv): Call copy_rtx on the step.
10954 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10955 Alan Hayward <alan.hayward@arm.com>
10956 David Sherwood <david.sherwood@arm.com>
10958 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
10959 calling tree_to_uhwi.
10961 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10962 Alan Hayward <alan.hayward@arm.com>
10963 David Sherwood <david.sherwood@arm.com>
10965 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
10966 INTEGER_CST rather than a negative test for ADDR_EXPR.
10968 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10969 Alan Hayward <alan.hayward@arm.com>
10970 David Sherwood <david.sherwood@arm.com>
10972 * tree-vrp.c (extract_range_from_binary_expr_1): Check
10973 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
10975 2017-09-21 Richard Biener <rguenther@suse.de>
10977 PR tree-optimization/71351
10978 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
10979 graphite_create_new_loop_guard): Remove, fold remaining parts
10981 (translate_isl_ast_node_for): ... here and simplify.
10983 2017-09-21 Jakub Jelinek <jakub@redhat.com>
10986 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
10987 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
10988 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
10989 alternative always use QI mode, for -Os imov (=R,R) alternative
10990 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
10993 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10994 Jeff Law <law@redhat.com>
10996 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
10997 (allocate_stack_space): New function, partially extracted from
10998 s390_emit_prologue.
10999 (s390_emit_prologue): Track offset to most recent stack probe.
11000 Code to allocate space moved into allocate_stack_space.
11001 Dump actions when no stack is allocated.
11002 (s390_prologue_plus_offset): New function.
11003 (s390_emit_stack_probe): Likewise.
11005 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
11007 * common.opt (Wa, Wl, Wp, g, gz=): Add
11009 (gno-column-info): Remove.
11010 (gcolumn-info): Drop RejectNegative.
11011 (gno-): New prefix.
11012 (gno-record-gcc-switches): Remove.
11013 (grecord-gcc-switches): Drop RejectNegative.
11014 (gno-split-dwarf): Remove.
11015 (gsplit-dwarf): Drop RejectNegative.
11016 (gno-strict-dwarf): Remove.
11017 (gstrict-dwarf): Drop RejectNegative.
11018 * config/darwin.opt (gfull, gused): Add RejectNegative.
11019 * dwarf2out.c (gen_producer_string): Drop
11020 gno-record-gcc-switches handler.
11021 * optc-gen.awk: Add g to prefixes with negative forms.
11022 * opts-common.c (remapping_prefix_p): New.
11023 (find_opt): Check it.
11024 (generate_canonical_option): Test g prefix.
11025 (option_map): Add -gno- mapping.
11026 (add_misspelling_candidates): Check remapping_prefix_p.
11028 2017-09-20 Jeff Law <law@redhat.com>
11030 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
11031 thinko in stack clash protection support.
11033 * explow.c (compute_stack_clash_protection_loop_data): Use
11034 CONST_INT_P instead of explicit test. Verify object is a
11035 CONST_INT_P before looking at INTVAL.
11036 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
11037 instead of explicit test.
11039 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
11042 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
11043 address instead of to r1 and r11.
11045 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
11047 * config.gcc: Support "knm".
11048 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
11049 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
11051 * config/i386/i386.c (m_KNM): Define.
11052 (processor_target_table): Add "knm".
11054 (ix86_option_override_internal): Add "knm".
11055 (ix86_issue_rate): Add PROCESSOR_KNM.
11056 (ix86_adjust_cost): Ditto.
11057 (ia32_multipass_dfa_lookahead): Ditto.
11058 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
11059 (fold_builtin_cpu): Add M_INTEL_KNM.
11060 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
11061 (processor_type): Add PROCESSOR_KNM.
11062 * config/i386/x86-tune.def: Add m_KNM.
11063 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
11065 2017-09-20 Richard Biener <rguenther@suse.de>
11067 PR tree-optimization/80213
11068 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
11069 are allowed in empty BBs as well.
11070 (canonicalize_loop_closed_ssa): Also look for other complex
11072 (scop_detection::get_sese): Include the loop-closed PHI block
11074 (scop_detection::merge_sese): Remove code adding extra blocks.
11075 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
11076 (build_scops): Assert the final returned scop is invalid.
11078 2017-09-20 Richard Biener <rguenther@suse.de>
11080 PR tree-optimization/82264
11081 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
11083 (vn_phi_lookup): Likewise.
11084 (vn_phi_insert): Likewise.
11086 2017-09-20 Jakub Jelinek <jakub@redhat.com>
11088 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
11089 that fits into uhwi or shwi, add DW_AT_const_value regardless
11090 of early_dwarf without going through RTL, using add_AT_unsigned
11093 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
11094 (DEBUG_LTO_ABBREV_SECTION): Likewise.
11095 (DEBUG_LTO_MACINFO_SECTION): Likewise.
11096 (DEBUG_MACRO_SECTION): Likewise.
11097 (DEBUG_LTO_MACRO_SECTION): Likewise.
11098 (DEBUG_STR_DWO_SECTION): Likewise.
11099 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
11100 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
11101 (DEBUG_LTO_DWO_LINE_SECTION): Define.
11102 (DEBUG_LTO_LINE_STR_SECTION): Define.
11103 (init_sections_and_labels): Initialize debug_line_str_section
11104 variable. Initialize debug_loc_section for -gdwarf-5 to
11105 DEBUG_LOCLISTS_SECTION. Formatting fixes.
11107 2017-09-20 Richard Biener <rguenther@suse.de>
11109 * graphite-sese-to-poly.c (extract_affine): Properly handle
11110 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
11112 2017-09-20 Richard Biener <rguenther@suse.de>
11114 PR tree-optimization/81373
11115 * graphite-scop-detection.c (build_cross_bb_scalars_def):
11116 Force SESE live-out defs to be handled even if they are
11119 2017-09-19 Jeff Law <law@redhat.com>
11121 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
11122 nothing for stack adjustments with REG_STACK_CHECK.
11123 * sched-deps.c (parse_add_or_inc): Reject insns with
11124 REG_STACK_CHECK from dependency breaking.
11125 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
11126 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
11127 * reg-notes.def (STACK_CHECK): New note.
11129 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
11130 (ix86_expand_prologue): Dump stack clash info as needed.
11131 Call ix86_adjust_stack_and_probe_stack_clash as needed.
11133 * function.c (dump_stack_clash_frame_info): New function.
11134 * function.h (dump_stack_clash_frame_info): Prototype.
11135 (enum stack_clash_probes): New enum.
11137 * config/alpha/alpha.c (alpha_expand_prologue): Also check
11138 flag_stack_clash_protection.
11139 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
11140 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
11141 (arm_frame_pointer_required): Likewise.
11142 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
11143 (ia64_expand_prologue): Likewise.
11144 * config/mips/mips.c (mips_expand_prologue): Likewise.
11145 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
11146 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
11147 (sparc_flat_expand_prologue): Likewise.
11148 * config/spu/spu.c (spu_expand_prologue): Likewise.
11150 * explow.c: Include "params.h".
11151 (anti_adjust_stack_and_probe_stack_clash): New function.
11152 (get_stack_check_protect): Likewise.
11153 (compute_stack_clash_protection_loop_data): Likewise.
11154 (emit_stack_clash_protection_loop_start): Likewise.
11155 (emit_stack_clash_protection_loop_end): Likewise.
11156 (allocate_dynamic_stack_space): Use get_stack_check_protect.
11157 Use anti_adjust_stack_and_probe_stack_clash.
11158 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
11159 (emit_stack_clash_protection_loop_start): Likewise.
11160 (emit_stack_clash_protection_loop_end): Likewise.
11161 * rtl.h (get_stack_check_protect): Prototype.
11162 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
11163 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
11164 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
11166 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
11168 * doc/tm.texi: Rebuilt.
11169 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
11170 get_stack_check_protect.
11171 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
11172 * config/arm/arm.c (arm_expand_prologue): Likewise.
11173 (arm_frame_pointer_required): Likewise.
11174 * config/i386/i386.c (ix86_expand_prologue): Likewise.
11175 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
11176 * config/mips/mips.c (mips_expand_prologue): Likewise.
11177 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
11178 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
11179 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
11180 (sparc_flat_expand_prologue): Likewise.
11182 * common.opt (-fstack-clash-protection): New option.
11183 * flag-types.h (enum stack_check_type): Note difference between
11184 -fstack-check= and -fstack-clash-protection.
11185 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
11186 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
11187 * toplev.c (process_options): Issue warnings/errors for cases
11188 not handled with -fstack-clash-protection.
11189 * doc/invoke.texi (-fstack-clash-protection): Document new option.
11190 (-fstack-check): Note additional problem with -fstack-check=generic.
11191 Note that -fstack-check is primarily for Ada and refer users
11192 to -fstack-clash-protection for stack-clash-protection.
11193 Document new params for stack clash protection.
11195 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
11197 * config/i386/i386.c (ix86_split_long_move): Do not handle
11198 address used for LEA in a special way.
11200 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
11202 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
11204 2017-09-19 Martin Sebor <msebor@redhat.com>
11207 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
11208 of incompatible types.
11210 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
11212 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11213 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
11214 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11215 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
11217 2017-09-19 Richard Biener <rguenther@suse.de>
11219 PR tree-optimization/82244
11220 * tree-vrp.c (remove_range_assertions): Do not propagate
11221 a constant to abnormals but replace the assert with a copy.
11223 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
11225 PR rtl-optimization/57878
11226 PR rtl-optimization/68988
11227 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
11228 avoidance test involving non_reload_pseudos. Move frequency test
11229 below the general fragmentation avoidance test.
11231 2017-09-19 Richard Biener <rguenther@suse.de>
11233 PR tree-optimization/69728
11234 * graphite-sese-to-poly.c (schedule_error): New global.
11235 (add_loop_schedule): Handle empty domain by failing the
11237 (build_original_schedule): Handle schedule_error.
11239 2017-09-19 Richard Biener <rguenther@suse.de>
11241 * graphite-scop-detection.c (scop_detection::can_represent_loop):
11242 Do not iterate to sibling loops but only to siblings of inner
11245 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
11248 * config/m68k/m68k.md (moveq feeding equality comparison): Check
11249 that the registers are different.
11251 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
11253 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
11254 to processor_model and "amdfam17h" to arch_names_table.
11255 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
11257 2017-09-18 Jakub Jelinek <jakub@redhat.com>
11260 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
11262 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11263 Alan Hayward <alan.hayward@arm.com>
11264 David Sherwood <david.sherwood@arm.com>
11266 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
11268 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
11270 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
11271 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
11272 vect_schedule_slp_instance.
11273 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
11274 Update call to vect_slp_analyze_node_operations. Simplify return
11276 (vect_slp_analyze_bb_1): Update call accordingly.
11277 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
11278 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
11279 (vect_schedule_slp): Update call accordingly.
11281 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11282 Alan Hayward <alan.hayward@arm.com>
11283 David Sherwood <david.sherwood@arm.com>
11285 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
11286 with types that aren't in fact scalar.
11288 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11290 * tree-vect-slp.c (vect_record_max_nunits): New function,
11292 (vect_build_slp_tree_1): ...here.
11293 (vect_build_slp_tree_2): Call it for phis too.
11295 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11297 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
11298 to vect_get_vec_def_for_operand when getting the mask operand.
11300 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11301 Alan Hayward <alan.hayward@arm.com>
11302 David Sherwood <david.sherwood@arm.com>
11304 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
11307 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11308 Alan Hayward <alan.hayward@arm.com>
11309 David Sherwood <david.sherwood@arm.com>
11311 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
11312 calculation for vector booleans.
11314 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11315 Alan Hayward <alan.hayward@arm.com>
11316 David Sherwood <david.sherwood@arm.com>
11318 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
11320 (vect_transform_stmt): ...here.
11321 (vect_analyze_stmt): Use it instead of calling
11322 vectorizable_live_operation directly.
11324 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
11326 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
11327 non-SIMT targets in acc vector loops.
11329 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
11331 * configure.ac: Add arc and check if assembler supports gdwarf2.
11332 * configure: Regenerate.
11334 2017-09-18 Richard Biener <rguenther@suse.de>
11336 PR tree-optimization/82220
11337 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
11338 epilogue niters from the min_profitable_iters compute.
11340 2017-09-18 Jakub Jelinek <jakub@redhat.com>
11343 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
11344 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
11345 (ix86_init_pic_reg): Revert 2017-09-01 changes.
11347 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
11350 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
11351 switching to a new text section.
11353 2017-09-18 Richard Biener <rguenther@suse.de>
11355 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
11357 (build_alias_set): Reject aliases with no access function.
11359 2017-09-18 Richard Biener <rguenther@suse.de>
11361 PR tree-optimization/79622
11362 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
11364 (build_cross_bb_scalars_use): Likewise.
11366 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
11368 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
11370 2017-09-18 Alan Modra <amodra@gmail.com>
11373 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
11374 stack_pointer_rtx for count 0. Update comments. Break up
11375 large rtl expression.
11377 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
11379 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
11380 Increase to 20 bytes.
11381 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
11382 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
11383 or avx version of the stub.
11385 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
11388 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
11389 compute the minimum stack alignment. Also update preferred stack
11390 boundary for leaf functions.
11392 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
11394 PR tree-optimization/82228
11395 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
11398 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
11400 * common/config/nds32/nds32-common.c
11401 (nds32_option_optimization_table): Refine formatting.
11402 (nds32_option_optimization_table): Use -fsched-pressure and
11403 -fomit-frame-pointer for specific optimization level.
11405 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
11407 * config/nds32/nds32.c: Refine formatting and comments.
11408 * config/nds32/nds32.h: Likewise.
11409 * config/nds32/nds32.md: Likewise.
11410 * config/nds32/nds32-cost.c: Likewise.
11411 * config/nds32/nds32-isr.c: Likewise.
11412 * config/nds32/nds32-md-auxiliary.c: Likewise.
11413 * config/nds32/nds32-multiple.md: Likewise.
11414 * config/nds32/nds32-predicates.c: Likewise.
11416 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
11417 Jakub Jelinek <jakub@redhat.com>
11419 Add support for -std=c++2a.
11420 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
11422 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
11424 2017-09-15 Steve Ellcey <sellcey@cavium.com>
11427 * doc/extend.texi (Common Function Attributes): Add
11428 references to ARM, AArch64, and S/390 specific attributes.
11429 (Function Specific Option Pragmas): Add AArch64 and S/390
11430 to list of back ends that support the target pragma.
11432 2017-09-15 Nathan Sidwell <nathan@acm.org>
11434 * doc/standards.texi: Fix C++17 description. Update URLs for
11437 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
11439 * common.opt (Wcast-align=strict): New warning option.
11440 * doc/invoke.texi: Document -Wcast-align=strict.
11442 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
11444 * cgraph.h (cgraph_thunk_info): Add comments.
11445 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
11446 assert for VIRTUAL_* arguments stricter.
11448 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
11450 PR tree-optimization/71026
11451 * match.pd: Move RDIV patterns from fold-const.c
11452 * fold-const.c (distribute_real_division): Removed.
11453 (fold_binary_loc): Remove calls to distribute_real_divison.
11455 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11457 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
11458 c++1z and gnu++1z as deprecated. Change other references to
11459 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
11460 Change -Wc++1z-compat to -Wc++17-compat.
11461 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
11462 * dwarf2out.c (highest_c_language): Handle C++17.
11463 (gen_compile_unit_die): Likewise.
11465 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11467 PR rtl-optimization/82192
11468 * combine.c (make_extraction): Don't look through non-paradoxical
11469 SUBREGs or TRUNCATE if pos + len is or might be bigger than
11472 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11473 Alan Hayward <alan.hayward@arm.com>
11474 David Sherwood <david.sherwood@arm.com>
11476 * target.def (function_arg_offset): New hook.
11477 * targhooks.h (default_function_arg_offset): Declare.
11478 * targhooks.c (default_function_arg_offset): New function.
11479 * function.c (locate_and_pad_parm): Use
11480 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
11481 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
11482 (TARGET_FUNCTION_ARG_OFFSET): ...this.
11483 * doc/tm.texi: Regenerate.
11484 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
11485 * config/spu/spu.c (spu_function_arg_offset): New function.
11486 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
11487 * system.h (FUNCTION_ARG_OFFSET): Poison.
11489 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11490 Alan Hayard <alan.hayward@arm.com>
11491 David Sherwood <david.sherwood@arm.com>
11493 * target.def (truly_noop_truncation): New hook.
11494 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
11495 than TRULY_NOOP_TRUNCATION.
11496 * hooks.h (hook_bool_uint_uint_true): Declare.
11497 * hooks.c (hook_bool_uint_uint_true): New function.
11498 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
11499 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
11500 * doc/tm.texi: Regenerate.
11501 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
11502 rather than TRULY_NOOP_TRUNCATION in comments.
11503 (simplify_comparison): Likewise.
11504 (record_truncated_value): Likewise.
11505 * expmed.c (extract_bit_field_1): Likewise.
11506 (extract_split_bit_field): Likewise.
11507 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
11508 instead of TRULY_NOOP_TRUNCATION.
11509 * function.c (assign_parm_setup_block): Likewise.
11510 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
11511 * rtlhooks.c: Include target.h.
11512 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
11513 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
11514 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
11515 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
11516 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
11517 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
11518 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
11519 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
11520 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
11521 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
11522 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
11523 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
11524 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
11525 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
11526 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
11527 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
11528 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
11529 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
11530 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
11531 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
11532 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
11533 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
11534 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
11535 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
11536 * config/mips/mips.c (mips_truly_noop_truncation): New function.
11537 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11538 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11539 rather than TRULY_NOOP_TRUNCATION in comments.
11540 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
11541 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
11542 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
11543 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
11544 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
11545 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
11546 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
11547 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
11548 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
11549 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
11550 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
11551 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11552 rather than TRULY_NOOP_TRUNCATION in comments.
11553 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
11554 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
11555 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
11556 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
11557 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
11558 TRULY_NOOP_TRUNCATION condition.
11559 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
11560 (TRULY_NOOP_TRUNCATION): Delete.
11561 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
11562 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
11563 * config/spu/spu.c (spu_truly_noop_truncation): New function.
11564 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11565 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
11566 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
11567 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
11568 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11569 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11570 rather than TRULY_NOOP_TRUNCATION in comments.
11571 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
11572 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
11573 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
11574 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
11575 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
11576 * system.h (TRULY_NOOP_TRUNCATION): Poison.
11578 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
11581 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
11582 (*cmp_ior): Likewise.
11583 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
11584 (*ior_scc_scc_cmp): Likewise.
11585 (*and_scc_scc): Likewise.
11586 (*and_scc_scc_cmp): Likewise.
11588 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11589 Alan Hayard <alan.hayward@arm.com>
11590 David Sherwood <david.sherwood@arm.com>
11592 * target.def (can_change_mode_class): New hook.
11593 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
11594 (hard_regno_nregs): Likewise.
11595 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
11596 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
11597 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
11598 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
11599 (LOAD_EXTEND_OP): Update accordingly.
11600 * doc/tm.texi: Regenerate.
11601 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
11602 CANNOT_CHANGE_MODE_CLASS.
11603 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
11604 (REG_CAN_CHANGE_MODE_P): ...this new macro.
11605 * combine.c (simplify_set): Update accordingly.
11606 * emit-rtl.c (validate_subreg): Likewise.
11607 * recog.c (general_operand): Likewise.
11608 * regcprop.c (mode_change_ok): Likewise.
11609 * reload1.c (choose_reload_regs): Likewise.
11610 (inherit_piecemeal_p): Likewise.
11611 * rtlanal.c (simplify_subreg_regno): Likewise.
11612 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
11613 instead of CANNOT_CHANGE_MODE_CLASS.
11614 (reload_cse_simplify_operands): Likewise.
11615 * reload.c (push_reload): Use targetm.can_change_mode_class
11616 instead of CANNOT_CHANGE_MODE_CLASS.
11617 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
11618 REG_CANNOT_CHANGE_MODE_P.
11619 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11620 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
11621 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11622 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11623 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11624 (arm_can_change_mode_class): New function.
11625 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
11626 than CANNOT_CHANGE_MODE_CLASS in comments.
11627 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11628 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
11629 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
11630 (ix86_can_change_mode_class): ...this new function, inverting the
11631 sense of the return value.
11632 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11633 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11634 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11635 (ia64_can_change_mode_class): New function.
11636 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11637 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
11638 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
11639 (m32c_can_change_mode_class): ...this new function, inverting the
11640 sense of the return value.
11641 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11642 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11643 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
11644 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
11645 (mips_can_change_mode_class): ...this new function, inverting the
11646 sense of the return value.
11647 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11648 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11649 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11650 (msp430_can_change_mode_class): New function.
11651 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11652 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
11653 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11654 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11655 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11656 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
11657 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11658 (pa_cannot_change_mode_class): Replace with...
11659 (pa_can_change_mode_class): ...this new function, inverting the
11660 sense of the return value.
11661 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
11662 than CANNOT_CHANGE_MODE_CLASS in comments.
11663 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11664 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
11665 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11666 (pdp11_cannot_change_mode_class): Replace with...
11667 (pdp11_can_change_mode_class): ...this new function, inverting the
11668 sense of the return value.
11669 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11670 * config/powerpcspe/powerpcspe-protos.h
11671 (rs6000_cannot_change_mode_class_ptr): Delete.
11672 * config/powerpcspe/powerpcspe.c
11673 (rs6000_cannot_change_mode_class_ptr): Delete.
11674 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11675 (rs6000_option_override_internal): Assign to
11676 targetm.can_change_mode_class instead of
11677 rs6000_cannot_change_mode_class_ptr.
11678 (rs6000_cannot_change_mode_class): Replace with...
11679 (rs6000_can_change_mode_class): ...this new function, inverting the
11680 sense of the return value.
11681 (rs6000_debug_cannot_change_mode_class): Replace with...
11682 (rs6000_debug_can_change_mode_class): ...this new function.
11683 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11684 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
11685 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11686 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11687 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
11689 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
11690 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11691 (rs6000_option_override_internal): Assign to
11692 targetm.can_change_mode_class instead of
11693 rs6000_cannot_change_mode_class_ptr.
11694 (rs6000_cannot_change_mode_class): Replace with...
11695 (rs6000_can_change_mode_class): ...this new function, inverting the
11696 sense of the return value.
11697 (rs6000_debug_cannot_change_mode_class): Replace with...
11698 (rs6000_debug_can_change_mode_class): ...this new function.
11699 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11700 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
11701 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
11702 (s390_can_change_mode_class): ...this new function, inverting the
11703 sense of the return value.
11704 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11705 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11706 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
11707 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11708 (sh_cannot_change_mode_class): Replace with...
11709 (sh_can_change_mode_class): ...this new function, inverting the
11710 sense of the return value.
11711 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11712 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11713 (sparc_can_change_mode_class): New function.
11714 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11715 * config/spu/spu.c (spu_can_change_mode_class): New function.
11716 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11717 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11718 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11719 (visium_can_change_mode_class): New function.
11720 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
11722 2017-09-15 Richard Biener <rguenther@suse.de>
11724 PR tree-optimization/82217
11725 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
11726 but not undefined case.
11728 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11731 * postreload.c (reload_cse_simplify_operands): Skip
11732 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
11734 2017-09-15 Richard Biener <rguenther@suse.de>
11736 PR tree-optimization/68823
11737 * graphite-scop-detection.c (build_alias_set): If we have a
11738 possible dependence check whether we can handle them by just
11739 looking at the DRs DR_ACCESS_FNs.
11740 (build_scops): If build_alias_set fails, fail the SCOP.
11742 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
11744 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
11745 to support float128 built-in functions that require the ISA 3.0
11747 (BU_FLOAT128_3_HW): Likewise.
11748 (SQRTF128): Add support for the IEEE 128-bit square root and fma
11749 built-in functions.
11750 (FMAF128): Likewise.
11752 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
11753 support for built-in functions that need the ISA 3.0 IEEE 128-bit
11754 floating point instructions.
11755 (rs6000_invalid_builtin): Likewise.
11756 (rs6000_builtin_mask_names): Likewise.
11757 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
11758 (RS6000_BTM_FLOAT128_HW): Likewise.
11759 (RS6000_BTM_COMMON): Likewise.
11760 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
11762 * doc/extend.texi (RS/6000 built-in functions): Document the
11763 IEEE 128-bit floating point square root and fused multiply-add
11764 built-in functions.
11766 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
11768 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
11769 reg (r2) isn't in the set of registers defined in the prologue.
11771 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11772 Alan Hayward <alan.hayward@arm.com>
11773 David Sherwood <david.sherwood@arm.com>
11775 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
11776 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
11777 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
11778 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
11779 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
11781 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
11782 max_vectorization_factor.
11783 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
11785 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11786 Alan Hayward <alan.hayward@arm.com>
11787 David Sherwood <david.sherwood@arm.com>
11789 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
11790 (vect_worthwhile_without_simd_p): Declare.
11791 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
11792 (vectorizable_reduction): Use it.
11793 * tree-vect-stmts.c (vectorizable_shift): Likewise.
11794 (vectorizable_operation): Likewise.
11796 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11797 Alan Hayward <alan.hayward@arm.com>
11798 David Sherwood <david.sherwood@arm.com>
11800 * tree-vectorizer.h (vect_get_num_copies): New function.
11801 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
11802 * tree-vect-loop.c (vectorizable_reduction): Likewise.
11803 (vectorizable_induction): Likewise.
11804 (vectorizable_live_operation): Likewise.
11805 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11806 (vectorizable_bswap): Likewise.
11807 (vectorizable_call): Likewise.
11808 (vectorizable_conversion): Likewise.
11809 (vectorizable_assignment): Likewise.
11810 (vectorizable_shift): Likewise.
11811 (vectorizable_operation): Likewise.
11812 (vectorizable_store): Likewise.
11813 (vectorizable_load): Likewise.
11814 (vectorizable_condition): Likewise.
11815 (vectorizable_comparison): Likewise.
11816 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
11818 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11819 Alan Hayward <alan.hayward@arm.com>
11820 David Sherwood <david.sherwood@arm.com>
11822 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
11823 of vect_init_vector.
11825 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11826 Alan Hayward <alan.hayward@arm.com>
11827 David Sherwood <david.sherwood@arm.com>
11829 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
11830 an inline wrapper that provides a location.
11831 (gimple_build_vector): Likewise.
11832 * gimple-fold.c (gimple_build_vector_from_val): New function.
11833 (gimple_build_vector): Likewise.
11834 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
11835 functions to build the initial value. Always return a gimple value.
11836 (get_initial_defs_for_reduction): Likewise. Only compute
11838 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
11839 vect_init_vector on the results from get_initial_def(s)_for_reduction.
11840 (vectorizable_induction): Use gimple_build_vector rather than
11843 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11844 Alan Hayward <alan.hayward@arm.com>
11845 David Sherwood <david.sherwood@arm.com>
11847 * target.h (vec_perm_indices): New typedef.
11848 (auto_vec_perm_indices): Likewise.
11849 * optabs-query.h: Include target.h
11850 (can_vec_perm_p): Take a vec_perm_indices *.
11851 * optabs-query.c (can_vec_perm_p): Likewise.
11852 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
11853 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
11854 * tree-vect-generic.c (lower_vec_perm): Likewise.
11855 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
11856 (vect_grouped_load_supported): Likewise.
11857 (vect_shift_permute_load_chain): Likewise.
11858 (vect_permute_store_chain): Use auto_vec_perm_indices.
11859 (vect_permute_load_chain): Likewise.
11860 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
11861 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
11862 Update uses of can_vec_perm_p.
11863 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
11864 mode with a number of elements. Take a vec_perm_indices *.
11865 (vect_create_epilog_for_reduction): Update accordingly.
11866 Use auto_vec_perm_indices.
11867 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
11868 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
11869 (vect_transform_slp_perm_load): Likewise.
11870 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
11871 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
11872 (vect_gen_perm_mask_checked): Likewise.
11873 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
11874 (vect_gen_perm_mask_checked): Likewise.
11875 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
11876 (vectorizable_store): Likewise.
11877 (vectorizable_load): Likewise.
11878 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
11879 (vectorizable_bswap): Likewise.
11881 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11882 Alan Hayward <alan.hayward@arm.com>
11883 David Sherwood <david.sherwood@arm.com>
11885 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
11886 * tree.c (build_vector): Likewise.
11887 (build_vector_from_ctor): Update accordingly.
11888 (build_vector_from_val): Likewise.
11889 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
11890 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
11891 * tree-vect-generic.c (add_rshift): Likewise.
11892 (expand_vector_divmod): Likewise.
11893 (optimize_vector_constructor): Likewise.
11894 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
11895 (vect_transform_slp_perm_load): Likewise.
11896 (vect_schedule_slp_instance): Likewise.
11897 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
11898 (vectorizable_call): Likewise.
11899 (vect_gen_perm_mask_any): Likewise. Add elements in order.
11900 * expmed.c (make_tree): Likewise.
11901 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
11902 a vector passed to build_vector.
11903 (fold_convert_const): Likewise.
11904 (exact_inverse): Likewise.
11905 (fold_ternary_loc): Likewise.
11906 (fold_relational_const): Likewise.
11907 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
11908 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
11909 (const_unop): Likewise. Store the reduction accumulator in a
11910 variable rather than an array.
11911 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
11912 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
11913 the new vector, rather than constructing it after the input arrays.
11914 (native_interpret_vector): Use auto_vec<tree> when building
11915 a vector passed to build_vector. Add elements in order.
11916 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
11917 auto_vec<tree> when building a vector passed to build_vector.
11918 (vect_create_epilog_for_reduction): Likewise.
11919 (vectorizable_induction): Likewise.
11920 (get_initial_def_for_reduction): Likewise. Fix indentation of
11922 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
11924 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
11925 passed to build_vector.
11927 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11928 Alan Hayward <alan.hayward@arm.com>
11929 David Sherwood <david.sherwood@arm.com>
11931 * tree-core.h (tree_base::u): Add an "nelts" field.
11932 (tree_vector): Use VECTOR_CST_NELTS as the length.
11933 * tree.c (tree_size): Likewise.
11934 (make_vector): Initialize VECTOR_CST_NELTS.
11935 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
11936 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
11937 TYPE_VECTOR_SUBPARTS.
11938 * expr.c (const_vector_mask_from_tree): Consistently use "units"
11939 as the number of units, setting it from VECTOR_CST_NELTS.
11940 (const_vector_from_tree): Likewise.
11941 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
11942 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
11943 (fold_negate_expr_1): Likewise.
11944 (fold_convert_const): Likewise.
11945 (const_binop): Likewise. Differentiate the number of output and
11947 (const_unop): Likewise.
11948 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
11949 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
11950 in cases that did the opposite.
11952 2017-09-14 Richard Biener <rguenther@suse.de>
11954 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
11957 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
11959 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
11961 2017-09-14 Jakub Jelinek <jakub@redhat.com>
11964 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
11965 if and where to split a bb, except for splitting before debug insn
11966 sequences followed by non-label real insn. Delete debug insns
11967 in between basic blocks.
11969 * combine.c (make_compound_operation_int): Formatting fixes.
11971 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
11972 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11973 * config/netbsd.h (LINK_EH_SPEC): Likewise.
11974 * config/sol2.h (LINK_EH_SPEC): Likewise.
11975 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11976 * config/s390/linux.h (LINK_SPEC): Likewise.
11977 * config/freebsd.h (LINK_EH_SPEC): Likewise.
11978 * config/openbsd.h (LINK_EH_SPEC): Likewise.
11979 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11980 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
11981 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
11982 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11983 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
11984 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
11986 2017-09-13 Jakub Jelinek <jakub@redhat.com>
11988 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
11990 (ENDFILE_LINUX_SPEC): Likewise.
11991 (LINK_EH_SPEC): Likewise.
11992 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
11993 (LINK_OS_LINUX_SPEC32): Likewise.
11994 (LINK_OS_LINUX_SPEC64): Likewise.
11995 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
11996 (LINK_OS_LINUX_SPEC): Likewise.
11998 2017-09-13 Martin Liska <mliska@suse.cz>
12000 PR middle-end/82154
12001 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
12002 CASE_HIGH is NULL_TREE.
12004 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12005 Alan Hayward <alan.hayward@arm.com>
12006 David Sherwood <david.sherwood@arm.com>
12008 * target.def (secondary_memory_needed): New hook.
12009 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
12010 instead of SECONDARY_MEMORY_NEEDED.
12011 (secondary_memory_needed_mode): Likewise.
12012 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
12013 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
12014 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
12015 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
12016 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
12017 * doc/tm.texi: Regenerate.
12018 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
12019 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
12020 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12021 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
12022 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
12023 * config/i386/i386.c (inline_secondary_memory_needed): Put the
12024 mode argument first and change the reg_class arguments to reg_class_t.
12025 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
12026 Make static. Update the call to inline_secondary_memory_needed.
12027 (ix86_register_move_cost): Update the call to
12028 inline_secondary_memory_needed.
12029 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12030 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
12032 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
12033 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12035 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
12036 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
12037 * config/mips/mips.c (mips_secondary_memory_needed): Make static
12038 and match hook interface. Add comment from mips.h.
12039 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12040 * config/mmix/mmix.md (truncdfsf2): Refer to
12041 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12043 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
12044 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
12045 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12046 (pa_secondary_memory_needed): New function.
12047 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
12048 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
12049 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12050 (pdp11_secondary_memory_needed): Make static and match hook interface.
12051 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
12052 * config/powerpcspe/powerpcspe-protos.h
12053 (rs6000_secondary_memory_needed_ptr): Delete.
12054 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
12056 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12057 (rs6000_option_override_internal): Assign to
12058 targetm.secondary_memory_needed rather than
12059 rs6000_secondary_memory_needed_ptr.
12060 (rs6000_secondary_memory_needed): Match hook interface.
12061 (rs6000_debug_secondary_memory_needed): Likewise.
12062 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
12063 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
12064 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
12065 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12066 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
12067 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
12069 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
12070 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12071 (rs6000_option_override_internal): Assign to
12072 targetm.secondary_memory_needed rather than
12073 rs6000_secondary_memory_needed_ptr.
12074 (rs6000_secondary_memory_needed): Match hook interface.
12075 (rs6000_debug_secondary_memory_needed): Likewise.
12076 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
12077 * config/s390/s390.c (s390_secondary_memory_needed): New function.
12078 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12079 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
12080 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12081 (sparc_secondary_memory_needed): New function.
12082 * lra-constraints.c (check_and_process_move): Refer to
12083 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12085 (curr_insn_transform): Likewise.
12086 (process_alt_operands): Use targetm.secondary_memory_needed
12087 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12088 (check_secondary_memory_needed_p): Likewise.
12089 (choose_split_class): Likewise.
12090 * reload.c: Unconditionally include code that was previously
12091 conditional on SECONDARY_MEMORY_NEEDED.
12092 (push_secondary_reload): Use targetm.secondary_memory_needed
12093 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12094 (push_reload): Likewise.
12095 * reload1.c: Unconditionally include code that was previously
12096 conditional on SECONDARY_MEMORY_NEEDED.
12097 (choose_reload_regs): Use targetm.secondary_memory_needed
12098 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12099 (gen_reload): Likewise.
12100 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
12102 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12103 Alan Hayward <alan.hayward@arm.com>
12104 David Sherwood <david.sherwood@arm.com>
12106 * target.def (secondary_memory_needed_mode): New hook:
12107 * targhooks.c (default_secondary_memory_needed_mode): Declare.
12108 * targhooks.h (default_secondary_memory_needed_mode): New function.
12109 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
12110 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
12111 * doc/tm.texi: Regenerate.
12112 * lra-constraints.c (check_and_process_move): Use
12113 targetm.secondary_memory_needed_mode instead of
12114 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
12115 (curr_insn_transform): Likewise.
12116 * reload.c (get_secondary_mem): Likewise.
12117 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12118 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
12120 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12121 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12122 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
12123 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12124 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
12126 * config/powerpcspe/powerpcspe-protos.h
12127 (rs6000_secondary_memory_needed_mode): Delete.
12128 * config/powerpcspe/powerpcspe.c
12129 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12130 (rs6000_secondary_memory_needed_mode): Make static.
12131 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12132 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
12134 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
12136 (rs6000_secondary_memory_needed_mode): Make static.
12137 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12138 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
12139 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12140 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12141 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
12143 (sparc_secondary_memory_needed_mode): New function.
12144 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
12146 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
12148 * config/aarch64/constraints.md (Umq): New constraint.
12149 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
12151 (mov<mode>): Update condition.
12153 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12155 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
12156 when bitposition is the same.
12158 2017-09-13 Richard Biener <rguenther@suse.de>
12160 * dwarf2out.c (output_die_symbol): Remove.
12161 (output_die): Do not output a DIEs symbol.
12163 2017-09-13 Richard Biener <rguenther@suse.de>
12165 PR middle-end/82128
12166 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
12167 default-def to avoid breaking iterator update with the weird
12168 interaction with cgraph_update_edges_for_call_stmt_node.
12170 2017-09-13 Richard Biener <rguenther@suse.de>
12172 * tree-cfg.c (verify_gimple_assign_binary): Add verification
12173 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
12174 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
12175 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
12177 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
12179 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
12180 Disable pc relative literal load irrespective of
12181 TARGET_FIX_ERR_A53_84341 for default.
12183 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
12185 * config/sparc/sparc.c (output_return): Output the source location of
12186 the insn in the delay slot, if any.
12187 (output_sibcall): Likewise.
12189 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
12192 * common.opt (-static-pie): New alias.
12193 (shared): Negate static-pie.
12194 (-no-pie): Update help text.
12196 (static-pie): New option.
12197 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
12198 -static-pie support.
12199 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
12200 (LINK_EH_SPEC): Likewise.
12201 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12202 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
12203 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
12204 * gcc.c (LINK_COMMAND_SPEC): Likewise.
12205 (init_gcc_specs): Likewise.
12206 (init_spec): Likewise.
12207 (display_help): Update help message for -pie.
12208 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
12211 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
12213 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
12214 (movdi_aarch64): Likewise.
12215 (movti_aarch64): Likewise.
12217 2017-09-12 Simon Wright <simon@pushface.org>
12220 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
12221 calculation of the minor version, always output as 0.
12223 2017-09-12 Jakub Jelinek <jakub@redhat.com>
12226 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
12227 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
12228 on it early, rather than manual conversion late. For
12229 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
12230 instead of performing manual conversion.
12232 2017-09-12 Carl Love <cel@us.ibm.com>
12234 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
12235 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
12237 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12238 VMULOSW): Add definitions.
12239 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12240 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12241 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12242 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12243 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12245 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
12247 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
12249 (movti_aarch64): Likewise.
12250 (movdf_aarch64): Likewise.
12251 (movtf_aarch64): Likewise.
12252 (load_pairdi): Likewise.
12253 (store_pairdi): Likewise.
12254 (load_pairdf): Likewise.
12255 (store_pairdf): Likewise.
12256 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
12257 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
12258 (ldr_got_small_<mode>): Likewise.
12259 (ldr_got_small_28k_<mode>): Likewise.
12260 (ldr_got_tiny): Likewise.
12261 * config/aarch64/iterators.md (ldst_sz): New.
12262 (ldpstp_sz): Likewise.
12263 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
12265 (thunderx_load): Split load_8 to load_16.
12266 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
12268 (thunderx2t99_storepair_basic): Split store_8 to store_16.
12269 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
12270 (xgene1_store_pair): Split store_8 to store_16.
12271 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
12272 (falkor_st_0_st_sd): Split store_8 to store_16.
12274 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
12276 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
12277 and store1/2/3/4 to store_4/8/12/16.
12278 * config/aarch64/aarch64.md: Update for rename.
12279 * config/arm/arm.md: Likewise.: Likewise.
12280 * config/arm/arm.c: Likewise.
12281 * config/arm/thumb1.md: Likewise.
12282 * config/arm/thumb2.md: Likewise.
12283 * config/arm/vfp.md: Likewise.
12284 * config/arm/arm-generic.md: Likewise.
12285 * config/arm/arm1020e.md: Likewise.
12286 * config/arm/arm1026ejs.md: Likewise.
12287 * config/arm/arm1136jfs.md: Likewise.
12288 * config/arm/arm926ejs.md: Likewise.
12289 * config/arm/cortex-a15.md: Likewise.
12290 * config/arm/cortex-a17.md: Likewise.
12291 * config/arm/cortex-a5.md: Likewise.
12292 * config/arm/cortex-a53.md: Likewise.
12293 * config/arm/cortex-a57.md: Likewise.
12294 * config/arm/cortex-a7.md: Likewise.
12295 * config/arm/cortex-a8.md: Likewise.
12296 * config/arm/cortex-a9.md: Likewise.
12297 * config/arm/cortex-m4.md: Likewise.
12298 * config/arm/cortex-m7.md: Likewise.
12299 * config/arm/cortex-r4.md: Likewise.
12300 * config/arm/exynos-m1.md: Likewise.
12301 * config/arm/fa526.md: Likewise.
12302 * config/arm/fa606te.md: Likewise.
12303 * config/arm/fa626te.md: Likewise.
12304 * config/arm/fa726te.md: Likewise.
12305 * config/arm/fmp626.md: Likewise.
12306 * config/arm/iwmmxt.md: Likewise.
12307 * config/arm/ldmstm.md: Likewise.
12308 * config/arm/marvell-pj4.md: Likewise.
12309 * config/arm/xgene1.md: Likewise.
12310 * config/aarch64/thunderx.md: Likewise.
12311 * config/aarch64/thunderx2t99.md: Likewise.
12312 * config/aarch64/falkor.md: Likewise.
12314 2017-09-12 Martin Liska <mliska@suse.cz>
12316 * attribs.c (private_lookup_attribute): New function.
12317 * attribs.h (private_lookup_attribute): Declared here.
12318 (lookup_attribute): Called from this place.
12320 2017-09-12 Richard Biener <rguenther@suse.de>
12322 PR tree-optimization/82157
12323 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
12324 stmts with side-effects.
12326 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12327 Alan Hayward <alan.hayward@arm.com>
12328 David Sherwood <david.sherwood@arm.com>
12330 * target.def (hard_regno_nregs): New hook.
12331 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
12332 * targhooks.h (default_hard_regno_nregs): Declare.
12333 * targhooks.c (default_hard_regno_nregs): New function.
12334 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
12335 (TARGET_HARD_REGNO_NREGS): ...this hook.
12336 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
12337 (CLASS_MAX_NREGS): Likewise.
12338 * doc/tm.texi: Regenerate.
12339 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
12340 instead of HARD_REGNO_NREGS.
12341 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
12342 HARD_REGNO_NREGS in the comment.
12343 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
12344 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
12345 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
12346 Return an unsigned int.
12347 (TARGET_HARD_REGNO_NREGS): Redefine.
12348 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
12349 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
12350 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
12351 (arc_hard_regno_nregs): New function.
12352 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
12353 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
12354 (arm_hard_regno_nregs): New function.
12355 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
12356 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
12357 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
12358 (TARGET_HARD_REGNO_NREGS): Redefine.
12359 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
12360 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
12361 (HARD_REGNO_NREGS): Delete.
12362 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
12363 (cr16_hard_regno_nregs): New function.
12364 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
12365 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
12366 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
12367 (cris_hard_regno_nregs): New function.
12368 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
12369 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
12370 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
12371 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
12372 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
12373 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
12374 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
12375 (frv_hard_regno_nregs): Make static. Take and return an
12377 (frv_class_max_nregs): Remove outdated copy of documentation.
12378 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
12379 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
12380 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
12381 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
12382 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
12383 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
12384 (TARGET_HARD_REGNO_NREGS): Redefine.
12385 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
12386 (CLASS_MAX_NREGS): Update comment.
12387 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
12388 (ia64_hard_regno_nregs): New function.
12389 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
12390 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
12391 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
12392 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
12393 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
12395 (m32c_hard_regno_nregs): Likewise. Make static.
12396 (TARGET_HARD_REGNO_NREGS): Redefine.
12397 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
12398 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
12399 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
12400 (m68k_hard_regno_nregs): New function.
12401 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
12402 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
12403 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
12404 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
12405 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
12406 Take and return an unsigned int.
12407 (TARGET_HARD_REGNO_NREGS): Redefine.
12408 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
12409 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
12410 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
12411 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
12412 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
12413 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
12414 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
12415 (msp430_hard_regno_nregs): Make static. Take and return an
12417 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
12418 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
12419 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
12420 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
12421 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
12422 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
12423 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
12424 (TARGET_HARD_REGNO_NREGS): Redefine.
12425 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
12426 (PA_HARD_REGNO_NREGS): ...this.
12427 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
12428 (PA_HARD_REGNO_NREGS): ...this.
12429 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
12430 (pa_hard_regno_nregs): New function.
12431 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
12432 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
12433 (pdp11_hard_regno_nregs): New function.
12434 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
12435 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
12436 (rs6000_hard_regno_nregs_hook): New function.
12437 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
12438 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
12439 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
12440 Take and return an unsigned int. Move earlier in file.
12441 (TARGET_HARD_REGNO_NREGS): Redefine.
12442 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
12443 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
12444 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
12445 (rl78_hard_regno_nregs): Make static. Take and return an
12447 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
12448 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
12449 (rs6000_hard_regno_nregs_hook): New function.
12450 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
12451 * config/rx/rx.c (rx_hard_regno_nregs): New function.
12452 (TARGET_HARD_REGNO_NREGS): Redefine.
12453 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
12454 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
12455 instead of HARD_REGNO_NREGS.
12456 (s390_hard_regno_nregs): New function.
12457 (s390_hard_regno_mode_ok): Add comment from s390.h.
12458 (TARGET_HARD_REGNO_NREGS): Redefine.
12459 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
12460 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
12461 (sh_hard_regno_nregs): New function.
12462 (sh_pass_in_reg_p): Use it.
12463 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
12464 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
12465 (sparc_hard_regno_nregs): New function.
12466 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
12467 * config/spu/spu.c (spu_hard_regno_nregs): New function.
12468 (spu_function_arg_advance): Use it, supplying a valid register number.
12469 (TARGET_HARD_REGNO_NREGS): Redefine.
12470 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
12471 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
12472 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
12473 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
12474 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
12475 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
12476 (CLASS_MAX_NREGS): Remove copy of old documentation.
12477 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
12478 (visium_hard_regno_nregs): New function.
12479 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
12480 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
12481 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
12482 xtensa_hard_regno_nregs): New function.
12483 * system.h (HARD_REGNO_NREGS): Poison.
12485 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12487 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
12488 hard_regno_nregs instead of HARD_REGNO_NREGS.
12489 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
12490 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
12491 (c6x_expand_epilogue): Likewise.
12492 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
12493 (frv_read_iacc_argument): Likewise.
12494 * config/sh/sh.c: Include regs.h.
12495 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12496 (regs_used): Likewise.
12497 (output_stack_adjust): Likewise.
12498 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
12499 * expmed.c: Include regs.h.
12500 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12501 * ree.c: Include regs.h.
12502 (combine_reaching_defs): Use hard_regno_nregs instead of
12504 (add_removable_extension): Likewise.
12506 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12508 * regs.h (hard_regno_nregs): Turn into a function.
12509 (end_hard_regno): Update accordingly.
12510 * caller-save.c (setup_save_areas): Likewise.
12511 (save_call_clobbered_regs): Likewise.
12512 (replace_reg_with_saved_mem): Likewise.
12513 (insert_restore): Likewise.
12514 (insert_save): Likewise.
12515 * combine.c (can_change_dest_mode): Likewise.
12516 (move_deaths): Likewise.
12517 (distribute_notes): Likewise.
12518 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
12519 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
12520 (rs6000_split_multireg_move): Likewise.
12521 (rs6000_register_move_cost): Likewise.
12522 (rs6000_memory_move_cost): Likewise.
12523 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
12524 (rs6000_split_multireg_move): Likewise.
12525 (rs6000_register_move_cost): Likewise.
12526 (rs6000_memory_move_cost): Likewise.
12527 * cselib.c (cselib_reset_table): Likewise.
12528 (cselib_lookup_1): Likewise.
12529 * emit-rtl.c (set_mode_and_regno): Likewise.
12530 * function.c (aggregate_value_p): Likewise.
12531 * ira-color.c (setup_profitable_hard_regs): Likewise.
12532 (check_hard_reg_p): Likewise.
12533 (calculate_saved_nregs): Likewise.
12534 (assign_hard_reg): Likewise.
12535 (improve_allocation): Likewise.
12536 (calculate_spill_cost): Likewise.
12537 * ira-emit.c (modify_move_list): Likewise.
12538 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
12539 (ira_hard_reg_in_set_p): Likewise.
12540 * ira.c (setup_reg_mode_hard_regset): Likewise.
12541 (clarify_prohibited_class_mode_regs): Likewise.
12542 (check_allocation): Likewise.
12543 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12544 (lra_setup_reg_renumber): Likewise.
12545 (setup_try_hard_regno_pseudos): Likewise.
12546 (spill_for): Likewise.
12547 (assign_hard_regno): Likewise.
12548 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
12549 * lra-constraints.c (in_class_p): Likewise.
12550 (lra_constraint_offset): Likewise.
12551 (simplify_operand_subreg): Likewise.
12552 (lra_constraints): Likewise.
12553 (split_reg): Likewise.
12554 (split_if_necessary): Likewise.
12555 (invariant_p): Likewise.
12556 (inherit_in_ebb): Likewise.
12557 * lra-lives.c (process_bb_lives): Likewise.
12558 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
12559 (get_hard_regs): Likewise.
12560 (do_remat): Likewise.
12561 * lra-spills.c (assign_spill_hard_regs): Likewise.
12562 * mode-switching.c (create_pre_exit): Likewise.
12563 * postreload.c (reload_combine_recognize_pattern): Likewise.
12564 * recog.c (peep2_find_free_register): Likewise.
12565 * regcprop.c (kill_value_regno): Likewise.
12566 (set_value_regno): Likewise.
12567 (copy_value): Likewise.
12568 (maybe_mode_change): Likewise.
12569 (find_oldest_value_reg): Likewise.
12570 (copyprop_hardreg_forward_1): Likewise.
12571 * regrename.c (check_new_reg_p): Likewise.
12572 (regrename_do_replace): Likewise.
12573 * reload.c (push_reload): Likewise.
12574 (combine_reloads): Likewise.
12575 (find_dummy_reload): Likewise.
12576 (operands_match_p): Likewise.
12577 (find_reloads): Likewise.
12578 (find_equiv_reg): Likewise.
12579 (reload_adjust_reg_for_mode): Likewise.
12580 * reload1.c (count_pseudo): Likewise.
12581 (count_spilled_pseudo): Likewise.
12582 (find_reg): Likewise.
12583 (clear_reload_reg_in_use): Likewise.
12584 (free_for_value_p): Likewise.
12585 (allocate_reload_reg): Likewise.
12586 (choose_reload_regs): Likewise.
12587 (reload_adjust_reg_for_temp): Likewise.
12588 (emit_reload_insns): Likewise.
12589 (delete_output_reload): Likewise.
12590 * rtlanal.c (subreg_get_info): Likewise.
12591 * sched-deps.c (sched_analyze_reg): Likewise.
12592 * sel-sched.c (init_regs_for_mode): Likewise.
12593 (mark_unavailable_hard_regs): Likewise.
12594 (choose_best_reg_1): Likewise.
12595 (verify_target_availability): Likewise.
12596 * valtrack.c (dead_debug_insert_temp): Likewise.
12597 * var-tracking.c (track_loc_p): Likewise.
12598 (emit_note_insn_var_location): Likewise.
12599 * varasm.c (make_decl_rtl): Likewise.
12600 * reginfo.c (choose_hard_reg_mode): Likewise.
12601 (init_reg_modes_target): Refer directly to
12602 this_target_regs->x_hard_regno_nregs.
12604 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12606 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
12607 instead of hard_regno_nregs.
12609 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12611 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
12612 end_hard_regno instead of hard_regno_nregs.
12613 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
12614 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
12615 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
12616 * ira-color.c (improve_allocation): Likewise.
12617 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12618 * lra-lives.c (mark_regno_live): Likewise.
12619 (mark_regno_dead): Likewise.
12620 * lra-remat.c (operand_to_remat): Likewise.
12621 * lra.c (collect_non_operand_hard_regs): Likewise.
12622 * postreload.c (reload_combine_note_store): Likewise.
12623 (move2add_valid_value_p): Likewise.
12624 * reload.c (regno_clobbered_p): Likewise.
12626 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12628 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
12630 * config/v850/v850.c (v850_reorg): Likewise.
12631 * reload.c (refers_to_regno_for_reload_p): Likewise.
12632 (find_equiv_reg): Likewise.
12633 * reload1.c (reload_reg_reaches_end_p): Likewise.
12635 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12637 * caller-save.c (add_used_regs): Use REG_NREGS instead of
12639 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
12640 * config/arm/arm.c (output_move_neon): Likewise.
12641 (arm_attr_length_move_neon): Likewise.
12642 (neon_split_vcombine): Likewise.
12643 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
12644 (c6x_mark_reg_written): Likewise.
12645 (c6x_dwarf_register_span): Likewise.
12646 * config/i386/i386.c (ix86_save_reg): Likewise.
12647 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
12648 (rws_access_reg): Likewise.
12649 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
12650 * mode-switching.c (create_pre_exit): Likewise.
12651 * ree.c (combine_reaching_defs): Likewise.
12652 (add_removable_extension): Likewise.
12653 * regcprop.c (find_oldest_value_reg): Likewise.
12654 (copyprop_hardreg_forward_1): Likewise.
12655 * reload.c (reload_inner_reg_of_subreg): Likewise.
12656 (push_reload): Likewise.
12657 (combine_reloads): Likewise.
12658 (find_dummy_reload): Likewise.
12659 (reload_adjust_reg_for_mode): Likewise.
12660 * reload1.c (find_reload_regs): Likewise.
12661 (forget_old_reloads_1): Likewise.
12662 (reload_reg_free_for_value_p): Likewise.
12663 (reload_adjust_reg_for_temp): Likewise.
12664 (emit_reload_insns): Likewise.
12665 (delete_output_reload): Likewise.
12666 * sel-sched.c (choose_best_reg_1): Likewise.
12667 (choose_best_pseudo_reg): Likewise.
12669 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12670 Alan Hayward <alan.hayward@arm.com>
12671 David Sherwood <david.sherwood@arm.com>
12673 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
12674 * target.def (slow_unaligned_access): New hook.
12675 * targhooks.h (default_slow_unaligned_access): Declare.
12676 * targhooks.c (default_slow_unaligned_access): New function.
12677 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
12678 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
12679 * doc/tm.texi: Regenerate.
12680 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
12681 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
12682 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
12684 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
12685 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
12687 (rs6000_slow_unaligned_access): New function.
12688 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
12689 (expand_block_compare): Likewise.
12690 (expand_strn_compare): Likewise.
12691 (rs6000_rtx_costs): Likewise.
12692 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
12693 (riscv_slow_unaligned_access): Likewise.
12694 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
12695 (riscv_slow_unaligned_access_p): ...this and make static.
12696 (riscv_option_override): Update accordingly.
12697 (riscv_slow_unaligned_access): New function.
12698 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
12699 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
12700 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
12701 (rs6000_slow_unaligned_access): New function.
12702 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
12703 (rs6000_rtx_costs): Likewise.
12704 * config/rs6000/rs6000-string.c (expand_block_compare)
12705 (expand_strn_compare): Use targetm.slow_unaligned_access instead
12706 of SLOW_UNALIGNED_ACCESS.
12707 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
12708 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
12709 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
12710 of SLOW_UNALIGNED_ACCESS.
12711 * expmed.c (simple_mem_bitfield_p): Likewise.
12712 * expr.c (alignment_for_piecewise_move): Likewise.
12713 (emit_group_load_1): Likewise.
12714 (emit_group_store): Likewise.
12715 (copy_blkmode_from_reg): Likewise.
12716 (emit_push_insn): Likewise.
12717 (expand_assignment): Likewise.
12718 (store_field): Likewise.
12719 (expand_expr_real_1): Likewise.
12720 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
12721 * lra-constraints.c (simplify_operand_subreg): Likewise.
12722 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12723 * gimple-ssa-store-merging.c: Likewise in block comment at start
12725 * tree-ssa-strlen.c: Include target.h.
12726 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
12727 of SLOW_UNALIGNED_ACCESS.
12728 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
12730 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12732 PR rtl-optimization/82185
12733 * expmed.c (emit_store_flag_int): Only test tem if it has been
12736 2017-09-12 Richard Biener <rguenther@suse.de>
12738 PR middle-end/82149
12739 * match.pd ((FTYPE) N CMP CST): Fix typo.
12741 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
12743 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
12745 (mips_near_type_p): Add 'short_call' attribute as a synonym
12747 * doc/extend.texi (short_call): Document new function attribute.
12749 2017-09-12 Jakub Jelinek <jakub@redhat.com>
12752 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
12753 assertion check that in the condition.
12754 (get_atomic_generic_size): Likewise. Before testing if parameter
12755 has pointer type, if it has array type, call for C++
12756 default_conversion to perform array-to-pointer conversion.
12758 2017-09-12 Richard Biener <rguenther@suse.de>
12760 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
12761 for operations we cannot scalarize.
12763 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
12765 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
12766 vectors heap vectors. Clean up comments.
12767 Make visited_bbs a reference.
12768 (profitable_jump_thread_path): Make GC
12769 vectors heap vectors. Clean up comments.
12771 (convert_and_register_jump_thread_path): Make GC vectors heap
12773 (check_subpath_and_update_thread_path): Same. Clean up comments.
12774 Make visited_bbs a reference.
12775 (handle_phi): Abstract common code to to
12776 register_jump_thread_path_if_profitable.
12777 Rename VAR_BB to DEF_BB.
12779 Make GC vectors heap vectors.
12780 Make visited_bbs a reference.
12781 (handle_assignment): Same.
12782 (register_jump_thread_path_if_profitable): New.
12783 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
12785 Make GC vectors heap vectors. Clean up comments.
12786 Make visited_bbs a reference.
12787 (find_jump_threads_backwards): Make visited_bbs live in the stack.
12788 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
12791 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
12794 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
12795 words of E_DImode object are reachable by xtensa_uimm8x4 access.
12797 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
12799 Revert r251800 and r251799.
12801 2017-09-11 Martin Jambor <mjambor@suse.cz>
12804 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
12805 arguments in advance.
12806 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
12807 use it to find predecessor edges.
12808 (naive_outof_ssa): Collect vector of predecessors.
12810 2017-09-08 Jason Merrill <jason@redhat.com>
12812 PR c++/70029 - ICE with ref-qualifier and -flto
12813 * langhooks.h (struct lang_hooks_for_types): Add
12814 copy_lang_qualifiers.
12815 * attribs.c (build_type_attribute_qual_variant): Use it.
12816 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
12818 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
12819 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
12821 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
12824 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
12825 (*mulsi3_sp64): New instruction.
12826 (mulsi3): New expander.
12828 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
12830 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
12832 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12834 * sancov.c: Include memmodel.h.
12836 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
12839 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
12842 2017-09-07 Carl Love <cel@us.ibm.com>
12844 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
12845 the sldi instruction.
12847 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
12849 * sancov.c: Include tm_p.h.
12851 2017-09-07 Jakub Jelinek <jakub@redhat.com>
12854 * output.h (switch_to_other_text_partition): New declaration.
12855 * varasm.c (switch_to_other_text_partition): New function.
12856 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
12857 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
12858 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
12859 to the other text partition before emitting LCL label and switch back
12860 after emitting the word after it.
12862 2017-09-07 Richard Biener <rguenther@suse.de>
12864 * passes.def (pass_split_crit_edges): Remove instance before PRE.
12865 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
12866 critical edges here, after loop init.
12867 (pass_data_pre): Remove PROP_no_crit_edges flags.
12868 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
12869 for valueization of call args to avoid leaking VN_TOP.
12870 (visit_use): Assert we do not visit default defs.
12871 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
12872 Use error_mark_node to more easily detect leaking VN_TOP.
12873 All default-defs are varying, not VN_TOP. Mark them visited.
12874 (run_scc_vn): Make code match comment.
12876 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
12878 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
12879 OPTION_MASK_FLOAT128_KEYWORD.
12880 (POWERPC_MASKS): Likewise.
12881 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
12882 support for the -mfloat128-type option, and make -mfloat128
12883 default on PowerPC Linux systems. Define or undefine
12884 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
12885 Define __float128 to be __ieee128 if IEEE 128-bit support is
12886 enabled, or undefine it.
12887 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
12888 Delete defining __FLOAT128_TYPE__.
12889 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
12890 -mfloat128-type option and make -mfloat128 default on PowerPC
12892 (TARGET_FLOAT128_TYPE): Likewise.
12893 (-mfloat128-type): Likewise.
12894 * config/rs6000/rs6000.c (rs6000_option_override_internal):
12895 Delete the -mfloat128-type option and make -mfloat128 default on
12896 PowerPC Linux systems. Always use __ieee128 to be the keyword for
12897 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
12898 128-bit floating point is enabled. Change tests from using
12899 -mfloat128-type to -mfloat128.
12900 (rs6000_mangle_type): Use the correct mangling for the __float128
12901 type even if normal long double is restricted to 64-bits.
12902 (floatn_mode): Enable the _Float128 type by default on VSX Linux
12904 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
12905 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
12906 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
12908 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
12909 documentation for -mfloat128.
12911 2017-09-06 Olivier Hainque <hainque@adacore.com>
12913 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
12915 2017-09-06 Wish Wu <wishwu007@gmail.com>
12916 Jakub Jelinek <jakub@redhat.com>
12918 * asan.c (initialize_sanitizer_builtins): Add
12919 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
12920 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
12921 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
12922 BT_FN_VOID_UINT64_PTR variables.
12923 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
12924 (BT_FN_VOID_UINT16_UINT16): Likewise.
12925 (BT_FN_VOID_UINT32_UINT32): Likewise.
12926 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
12927 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
12928 (BT_FN_VOID_UINT64_PTR): Likewise.
12929 * common.opt (flag_sanitize_coverage): New variable.
12930 (fsanitize-coverage=trace-pc): Remove.
12931 (fsanitize-coverage=): Add.
12932 * flag-types.h (enum sanitize_coverage_code): New enum.
12933 * fold-const.c (fold_range_test): Disable non-short-circuit
12934 optimization if flag_sanitize_coverage.
12935 (fold_truth_andor): Likewise.
12936 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
12937 * opts.c (COVERAGE_SANITIZER_OPT): Define.
12938 (coverage_sanitizer_opts): New array.
12939 (get_closest_sanitizer_option): Add OPTS argument, handle also
12940 OPT_fsanitize_coverage_.
12941 (parse_sanitizer_options): Adjusted to also handle
12942 OPT_fsanitize_coverage_.
12943 (common_handle_option): Add OPT_fsanitize_coverage_.
12944 * sancov.c (instrument_comparison, instrument_switch): New function.
12945 (sancov_pass): Add trace-cmp support.
12946 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
12947 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
12948 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
12949 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
12950 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
12951 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
12952 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
12953 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
12954 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
12955 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
12957 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
12959 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
12960 error. Only quit immediately if parsing is complete.
12961 (BEGIN): Initialize fatal_err and parse_done.
12962 (begin fpu, end fpu): Check number of arguments.
12963 (begin arch, end arch): Likewise.
12964 (begin cpu, end cpu): Likewise.
12965 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
12966 (optalias): Likewise.
12968 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
12970 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
12971 * config/arm/arm-isa.h: Delete. Move definitions to ...
12972 * arm-cpus.in: ... here. Use new feature and fgroup values.
12973 * config/arm/arm.c (arm_option_override): Use lower case for feature
12975 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
12976 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
12977 * config/arm/parsecpu.awk (END): Add new command 'isa'.
12979 (print_isa_bits_for): New function.
12980 (gen_isa): New function.
12981 (gen_comm_data): Use print_isa_bits_for.
12982 (define feature): New keyword.
12983 (define fgroup): New keyword.
12984 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
12985 (arm-isa.h): Add rule to generate file.
12986 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
12987 case for feature bit names.
12989 2017-09-06 Richard Biener <rguenther@suse.de>
12991 * tree-ssa-pre.c (NECESSARY): Remove.
12992 (create_expression_by_pieces): Do not touch pass-local flags.
12993 (insert_into_preds_of_block): Likewise.
12994 (do_pre_regular_insertion): Likewise.
12995 (eliminate_insert): Likewise.
12996 (eliminate_dom_walker::before_dom_children): Likewise.
12997 (fini_eliminate): Do not look at inserted_exprs.
12998 (mark_operand_necessary): Remove.
12999 (remove_dead_inserted_code): Replace with simple work-list
13000 algorithm based on inserted_exprs and SSA uses.
13001 (pass_pre::execute): Re-order fini_eliminate and
13002 remove_dead_inserted_code.
13004 2017-09-06 Olivier Hainque <hainque@adacore.com>
13006 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
13007 for VxWorks 7. Adjust surrounding comments.
13009 2017-09-06 Richard Biener <rguenther@suse.de>
13011 * gimple-ssa-strength-reduction.c
13012 (find_candidates_dom_walker::before_dom_children): Also allow
13015 2017-09-06 Richard Biener <rguenther@suse.de>
13017 PR tree-optimization/82108
13018 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
13019 for gap in the non-permutation SLP case.
13021 2017-09-06 Martin Jambor <mjambor@suse.cz>
13023 PR tree-optimization/82078
13024 * tree-sra.c (sort_and_splice_var_accesses): Move call to
13025 add_access_to_work_queue...
13026 (build_accesses_from_assign): ...here.
13027 (propagate_all_subaccesses): Make sure racc is the group
13028 representative, if there is one.
13030 2017-09-06 Jakub Jelinek <jakub@redhat.com>
13032 PR middle-end/82095
13033 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
13036 2017-09-06 Richard Biener <rguenther@suse.de>
13038 * gimple-ssa-strength-reduction.c
13039 (find_candidates_dom_walker::before_doom_children): Use a
13040 type and not a mode check.
13042 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
13045 * config/arm/predicates.md (arm_general_adddi_operand): Create new
13047 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
13049 2017-09-05 Jeff Law <law@redhat.com>
13051 PR tree-optimization/64910
13052 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
13053 cases where we have 3 or more operands.
13055 2017-09-05 Jakub Jelinek <jakub@redhat.com>
13057 PR middle-end/81768
13058 * omp-low.c (lower_omp_for): Recompute tree invariant if
13059 gimple_omp_for_initial/final is ADDR_EXPR.
13061 PR middle-end/81768
13062 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
13063 into gimple val before gimplification fo the COND_EXPR.
13065 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
13067 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
13068 REGION_COPY argument.
13069 (thread_through_all_blocks): Remove unused argument to
13070 duplicate_thread_path.
13072 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13073 Alan Hayward <alan.hayward@arm.com>
13074 David Sherwood <david.sherwood@arm.com>
13076 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
13077 Take a scalar_mode rather than a machine_mode.
13078 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13079 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
13080 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13081 (aarch64_gen_adjusted_ldpstp): Likewise.
13082 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
13084 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13085 Alan Hayward <alan.hayward@arm.com>
13086 David Sherwood <david.sherwood@arm.com>
13088 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
13089 Take a scalar_int_mode instead of a machine_mode.
13090 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13091 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13092 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13093 (aarch64_simd_attr_length_rglist): Delete.
13094 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
13095 a scalar_int_mode instead of a machine_mode.
13096 (aarch64_add_offset): Likewise.
13097 (aarch64_internal_mov_immediate): Likewise
13098 (aarch64_add_constant_internal): Likewise.
13099 (aarch64_add_constant): Likewise.
13100 (aarch64_movw_imm): Likewise.
13101 (aarch64_rtx_arith_op_extract_p): Likewise.
13102 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13103 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13104 Remove assert that the mode isn't a vector.
13105 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13106 (aarch64_expand_mov_immediate): Update calls after above changes.
13107 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
13108 (aarch64_and_bitmask_imm): Check for scalar integer modes.
13109 (aarch64_move_imm): Likewise.
13110 (aarch64_can_const_movi_rtx_p): Likewise.
13111 (aarch64_strip_extend): Likewise.
13112 (aarch64_extr_rtx_p): Likewise.
13113 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
13114 a CONST_INT when the mode parameter is VOIDmode.
13115 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
13117 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13119 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
13120 * stor-layout.c (bitwise_mode_for_mode): Likewise.
13121 (bitwise_type_for_mode): Update accordingly.
13123 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13125 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
13126 * stor-layout.c (mode_for_size_tree): Likewise.
13127 (mode_for_array): Update accordingly.
13128 (layout_decl): Likewise.
13129 (compute_record_mode): Likewise. Only set the mode once.
13131 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13133 * target.def (get_mask_mode): Change return type to opt_mode.
13135 * doc/tm.texi: Regenerate.
13136 * targhooks.h (default_get_mask_mode): Return an opt_mode.
13137 * targhooks.c (default_get_mask_mode): Likewise.
13138 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
13139 * optabs-query.c (can_vec_mask_load_store_p): Update use of
13140 targetm.get_mask_mode.
13141 * tree.c (build_truth_vector_type): Likewise.
13143 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13145 * machmode.h (mode_for_vector): Return an opt_mode.
13146 * stor-layout.c (mode_for_vector): Likewise.
13147 (mode_for_int_vector): Update accordingly.
13148 (layout_type): Likewise.
13149 * config/i386/i386.c (emit_memmov): Likewise.
13150 (ix86_expand_set_or_movmem): Likewise.
13151 (ix86_expand_vector_init): Likewise.
13152 (ix86_get_mask_mode): Likewise.
13153 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
13155 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
13156 * expmed.c (extract_bit_field_1): Likewise.
13157 * expr.c (expand_expr_real_2): Likewise.
13158 * optabs-query.c (can_vec_perm_p): Likewise.
13159 (can_vec_mask_load_store_p): Likewise.
13160 * optabs.c (expand_vec_perm): Likewise.
13161 * targhooks.c (default_get_mask_mode): Likewise.
13162 * tree-vect-stmts.c (vectorizable_store): Likewise.
13163 (vectorizable_load): Likewise.
13164 (get_vectype_for_scalar_type_and_size): Likewise.
13166 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13168 * machmode.h (mode_for_int_vector): New function.
13169 * stor-layout.c (mode_for_int_vector): Likewise.
13170 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
13171 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
13172 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
13173 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
13174 (s390_expand_vcond): Likewise.
13176 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13178 * machmode.h (opt_machine_mode): New type.
13179 (opt_mode<T>): Allow construction from anything that can be
13181 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
13182 (mode_for_size): Return an opt_machine_mode.
13183 * stor-layout.c (mode_for_size): Likewise.
13184 (mode_for_size_tree): Update call accordingly.
13185 (bitwise_mode_for_mode): Likewise.
13186 (make_fract_type): Likewise.
13187 (make_accum_type): Likewise.
13188 * caller-save.c (replace_reg_with_saved_mem): Update call
13190 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13191 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13192 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13193 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13194 * expmed.c (extract_bit_field_1): Likewise.
13195 * reload.c (get_secondary_mem): Likewise.
13196 * varasm.c (assemble_integer): Likewise.
13197 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
13200 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13202 * machmode.h (decimal_float_mode_for_size): New function.
13203 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
13204 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
13205 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
13206 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
13207 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
13208 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
13210 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13212 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
13213 (get_builtin_sync_mode): Likewise.
13214 (expand_ifn_atomic_compare_exchange): Likewise.
13215 (expand_builtin_atomic_clear): Likewise.
13216 (expand_builtin_atomic_test_and_set): Likewise.
13217 (fold_builtin_atomic_always_lock_free): Likewise.
13218 * calls.c (compute_argument_addresses): Likewise.
13219 (emit_library_call_value_1): Likewise.
13220 (store_one_arg): Likewise.
13221 * combine.c (combine_instructions): Likewise.
13222 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
13223 * config/arm/arm.c (arm_function_value): Likewise.
13224 (aapcs_allocate_return_reg): Likewise.
13225 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
13226 * config/i386/i386.c (construct_container): Likewise.
13227 (ix86_gimplify_va_arg): Likewise.
13228 (ix86_expand_sse_cmp): Likewise.
13229 (emit_memmov): Likewise.
13230 (emit_memset): Likewise.
13231 (expand_small_movmem_or_setmem): Likewise.
13232 (ix86_expand_pextr): Likewise.
13233 (ix86_expand_pinsr): Likewise.
13234 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
13235 * config/microblaze/microblaze.c (microblaze_block_move_straight):
13237 * config/mips/mips.c (mips_function_value_1) Likewise.
13238 (mips_block_move_straight): Likewise.
13239 (mips_expand_ins_as_unaligned_store): Likewise.
13240 * config/powerpcspe/powerpcspe.c
13241 (rs6000_darwin64_record_arg_advance_flush): Likewise.
13242 (rs6000_darwin64_record_arg_flush): Likewise.
13243 * config/rs6000/rs6000.c
13244 (rs6000_darwin64_record_arg_advance_flush): Likewise.
13245 (rs6000_darwin64_record_arg_flush): Likewise.
13246 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
13247 (sparc_function_value_1): Likewise.
13248 * config/spu/spu.c (adjust_operand): Likewise.
13249 (spu_emit_branch_or_set): Likewise.
13250 (arith_immediate_p): Likewise.
13251 * emit-rtl.c (gen_lowpart_common): Likewise.
13252 * expr.c (expand_expr_real_1): Likewise.
13253 * function.c (assign_parm_setup_block): Likewise.
13254 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
13255 * reload1.c (alter_reg): Likewise.
13256 * stor-layout.c (mode_for_vector): Likewise.
13257 (layout_type): Likewise.
13259 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13261 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
13262 (spu_convert_move): Likewise.
13263 * lower-subreg.c (resolve_simple_move): Likewise.
13265 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13268 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
13269 define_insn to a define_expand.
13270 (altivec_vsum2sws_direct): New define_insn.
13271 (altivec_vsumsws): Convert from a define_insn to a define_expand.
13273 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
13275 * config/arm/arm.c (arm_option_params_internal): Improve setting of
13278 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
13283 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
13284 realign stack if stack alignment needed is less than incoming
13287 2017-09-05 Marek Polacek <polacek@redhat.com>
13290 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
13293 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
13295 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
13297 2017-09-05 Richard Biener <rguenther@suse.de>
13299 PR tree-optimization/82084
13300 * fold-const.c (can_native_encode_string_p): Handle wide characters.
13302 2017-09-05 Richard Biener <rguenther@suse.de>
13304 PR tree-optimization/82102
13305 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
13307 2017-09-05 Martin Liska <mliska@suse.cz>
13309 PR tree-optimization/82032
13310 * tree-cfg.c (generate_range_test): New function.
13311 * tree-cfg.h (generate_range_test): Declared here.
13312 * tree-cfgcleanup.c (convert_single_case_switch): New function.
13313 (cleanup_control_expr_graph): Use it.
13314 * tree-switch-conversion.c (try_switch_expansion): Remove
13316 (emit_case_nodes): Use generate_range_test.
13318 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
13321 * config/i386/i386.md (*<btsc><mode>_mask): Add
13322 TARGET_USE_BT to insn constraint.
13323 (*btr<mode>_mask): Ditto.
13325 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
13327 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
13328 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
13330 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
13333 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
13334 TARGET_NEON and TARGET_IWMMXT.
13335 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
13336 TARGET_NEON and TARGET_IWMMXT.
13337 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
13339 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
13341 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
13342 (ix86_rewrite_tls_address): Ditto.
13343 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
13344 (ix86_rewrite_tls_address_1): Ditto.
13345 (ix86_rewrite_tls_address): Ditto.
13346 * config/i386/predicates.md (tls_address_pattern): New predicate.
13347 * config/i386/i386.md (TLS address splitter): New splitter.
13349 2017-09-04 Richard Biener <rguenther@suse.de>
13351 PR tree-optimization/82084
13352 * fold-const.h (can_native_encode_string_p): Declare.
13353 * fold-const.c (can_native_encode_string_p): Factor out from ...
13354 (native_encode_string): ... here.
13355 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
13356 vectorizing stores from constants we later cannot handle.
13358 2017-09-04 Marek Polacek <polacek@redhat.com>
13361 * doc/invoke.texi: Update -Wtautological-compare documentation.
13363 2017-09-04 Jeff Law <law@redhat.com>
13365 PR tree-optimization/64910
13366 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
13367 swap the first and last operand if the last is a constant.
13369 2017-09-04 Marek Polacek <polacek@redhat.com>
13372 * convert.c (do_narrow): When sanitizing signed integer overflows,
13373 bail out for signed types.
13374 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
13376 2017-09-04 Richard Biener <rguenther@suse.de>
13378 PR tree-optimization/82060
13379 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13380 Move devirtualization after stmt folding and before EH/AB/noreturn
13381 cleanup to get the stmt refs canonicalized. Use a bool instead
13382 of gimple_modified_p since that doesn't work for NOPs. Schedule
13383 NOPs generated by folding for removal.
13385 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13386 Alan Hayward <alan.hayward@arm.com>
13387 David Sherwood <david.sherwood@arm.com>
13389 * coretypes.h (pad_direction): New enum.
13390 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
13391 (FUNCTION_ARG_PADDING): Likewise.
13392 * target.def (function_arg_padding): New hook.
13393 * targhooks.h (default_function_arg_padding): Declare.
13394 * targhooks.c (default_function_arg_padding): New function.
13395 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
13396 (TARGET_FUNCTION_ARG_PADDING): ...this.
13397 * doc/tm.texi: Regenerate.
13398 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
13399 instead of direction.
13400 (compute_argument_addresses): Likewise.
13401 (load_register_parameters): Likewise.
13402 (emit_library_call_value_1): Likewise.
13403 (store_one_arg): Use targetm.calls.function_arg_padding instead
13404 of FUNCTION_ARG_PADDING.
13405 (must_pass_in_stack_var_size_or_pad): Likewise.
13406 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
13407 (emit_group_store): Likewise.
13408 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
13409 instead of FUNCTION_ARG_PADDING.
13410 (emit_push_insn): Likewise, and propagate enum change throughout
13412 * function.h (direction): Delete.
13413 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
13415 * function.c (assign_parm_find_stack_rtl): Likewise.
13416 (assign_parm_setup_block_p): Likewise.
13417 (assign_parm_setup_block): Likewise.
13418 (gimplify_parameters): Likewise.
13419 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
13420 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
13422 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
13423 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13424 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
13425 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
13426 (aarch64_function_arg_padding): ...this new function.
13427 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
13428 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13429 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
13430 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13431 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
13432 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13433 (arm_pad_arg_upward): Replace with...
13434 (arm_function_arg_padding): ...this new function.
13435 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
13437 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
13438 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
13439 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13440 (ia64_hpux_function_arg_padding): Replace with...
13441 (ia64_function_arg_padding): ...this new function. Use pad_direction
13442 instead of direction. Check for TARGET_HPUX.
13443 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
13444 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13445 (iq2000_function_arg_padding): New function.
13446 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
13447 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
13448 (mips_function_arg_padding): ...this new function.
13449 (mips_pad_reg_upward): Update accordingly.
13450 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13451 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
13452 targetm.calls.function_arg_padding.
13453 (FUNCTION_ARG_PADDING): Delete.
13454 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13455 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
13456 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13457 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
13458 (nios2_block_reg_padding): Return pad_direction instead of direction.
13459 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
13460 instead of direction.
13461 (nios2_function_arg_padding): Likewise. Make static.
13462 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13463 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
13464 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
13465 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
13466 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13467 (pa_function_arg_padding): Make static. Return pad_direction instead
13469 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
13470 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13471 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
13472 instead of direction. Use targetm.calls.function_arg_padding.
13473 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
13474 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13475 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
13476 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
13477 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
13479 (function_arg_padding): Rename to...
13480 (rs6000_function_arg_padding): ...this. Make static. Return
13481 pad_direction instead of direction.
13482 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
13483 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
13484 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13485 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
13486 instead of direction. Use targetm.calls.function_arg_padding.
13487 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
13488 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13489 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
13490 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
13491 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13492 (function_arg_padding): Rename to...
13493 (rs6000_function_arg_padding): ...this. Make static. Return
13494 pad_direction instead of direction.
13495 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
13496 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
13497 * config/s390/s390.c (s390_function_arg_padding): New function.
13498 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13499 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
13500 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
13501 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13502 (function_arg_padding): Rename to...
13503 (sparc_function_arg_padding): ...this. Make static. Return
13504 pad_direction instead of direction.
13505 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
13506 * config/spu/spu.c (spu_function_arg_padding): New function.
13507 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13508 * system.h (FUNCTION_ARG_PADDING): Poison.
13510 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13511 Alan Hayward <alan.hayward@arm.com>
13512 David Sherwood <david.sherwood@arm.com>
13514 * target.def (modes_tieable_p): New hook.
13515 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
13516 (TARGET_MODES_TIEABLE_P): ...this.
13517 * doc/tm.texi.in: Regenerate.
13518 * hooks.h (hook_bool_mode_mode_true): Declare.
13519 * hooks.c (hook_bool_mode_mode_true): New function.
13520 * combine.c (subst): Use targetm.modes_tieable_p instead of
13522 * dse.c (find_shift_sequence): Likewise.
13523 * expmed.c (extract_low_bits): Likewise.
13524 * lower-subreg.c: Include target.h.
13525 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
13527 * rtlanal.c (rtx_cost): Likewise.
13528 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
13529 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
13530 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
13531 (TARGET_MODES_TIEABLE_P): Redefine.
13532 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
13533 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
13534 (TARGET_MODES_TIEABLE_P): Redefine.
13535 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
13536 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
13537 (arc_modes_tieable_p): New function.
13538 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
13539 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
13540 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
13541 (arm_modes_tieable_p): Make static.
13542 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
13543 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
13544 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
13545 (TARGET_MODES_TIEABLE_P): Redefine.
13546 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
13547 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
13548 (TARGET_MODES_TIEABLE_P): Redefine.
13549 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
13550 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
13551 (cr16_modes_tieable_p): New function.
13552 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
13553 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
13554 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
13555 (TRULY_NOOP_TRUNCATION): Update comment.
13556 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
13557 (TRULY_NOOP_TRUNCATION): Update comment.
13558 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
13559 (frv_modes_tieable_p): New function.
13560 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
13561 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
13562 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
13563 (TARGET_MODES_TIEABLE_P): Redefine.
13564 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
13565 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
13566 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
13567 (TARGET_MODES_TIEABLE_P): Redefine.
13568 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
13569 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
13570 (ia64_modes_tieable_p): New function.
13571 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
13572 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
13573 (iq2000_modes_tieable_p): New function.
13574 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
13575 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
13576 (lm32_modes_tieable_p): New function.
13577 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
13578 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
13579 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
13580 (TARGET_MODES_TIEABLE_P): Redefine.
13581 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
13582 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
13583 (m32r_modes_tieable_p): New function.
13584 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
13585 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
13586 (m68k_modes_tieable_p): New function.
13587 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
13588 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
13589 (mcore_modes_tieable_p): New function.
13590 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
13591 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
13593 (TARGET_MODES_TIEABLE_P): Redefine.
13594 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
13595 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
13596 * config/mips/mips.c (mips_modes_tieable_p): Make static.
13597 (TARGET_MODES_TIEABLE_P): Redefine.
13598 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
13599 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
13600 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
13601 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
13602 (mn10300_modes_tieable_p): ...this and make static.
13603 (TARGET_MODES_TIEABLE_P): Redefine.
13604 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
13605 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
13606 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
13607 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
13608 (msp430_modes_tieable_p): Make static.
13609 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
13610 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
13611 (TARGET_MODES_TIEABLE_P): Redefine.
13612 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
13613 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
13614 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
13615 (TARGET_MODES_TIEABLE_P): Redefine.
13616 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
13617 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
13618 * config/pa/pa.c (pa_modes_tieable_p): Make static.
13619 (TARGET_MODES_TIEABLE_P): Redefine.
13620 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
13621 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
13622 (pdp11_modes_tieable_p): New function.
13623 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
13624 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
13625 (rs6000_modes_tieable_p): New function.
13626 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
13627 * config/powerpcspe/powerpcspe.md: Update comment.
13628 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
13629 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
13630 (TARGET_MODES_TIEABLE_P): Redefine.
13631 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
13632 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
13633 (rl78_modes_tieable_p): New function.
13634 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
13635 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
13636 (rs6000_modes_tieable_p): New function.
13637 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
13638 * config/rs6000/rs6000.md: Update comment.
13639 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
13640 * config/rx/rx.c (rx_modes_tieable_p): New function.
13641 (TARGET_MODES_TIEABLE_P): Redefine.
13642 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
13643 * config/s390/s390.c (s390_modes_tieable_p): New function.
13644 (TARGET_MODES_TIEABLE_P): Redefine.
13645 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
13646 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
13647 (sh_modes_tieable_p): New function.
13648 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
13649 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
13650 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
13651 (sparc_modes_tieable_p): Make static.
13652 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
13653 * config/spu/spu.c (spu_modes_tieable_p): New function.
13654 (TARGET_MODES_TIEABLE_P): Redefine.
13655 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
13656 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
13657 (TARGET_MODES_TIEABLE_P): Redefine.
13658 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
13659 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
13660 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
13661 * config/v850/v850.c (v850_modes_tieable_p): New function.
13662 (TARGET_MODES_TIEABLE_P): Redefine.
13663 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
13664 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
13665 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
13666 (visium_modes_tieable_p): New function.
13667 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
13668 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
13669 (xtensa_modes_tieable_p): New function.
13670 * system.h (MODES_TIEABLE_P): Poison.
13672 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13673 Alan Hayward <alan.hayward@arm.com>
13674 David Sherwood <david.sherwood@arm.com>
13676 * target.def (hard_regno_mode_ok): New hook.
13677 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
13678 (TARGET_HARD_REGNO_MODE_OK): ...this.
13679 * doc/tm.texi.in: Regenerate.
13680 * hooks.h (hook_bool_uint_mode_true): Declare.
13681 * hooks.c (hook_bool_uint_mode_true): New function.
13682 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
13683 HARD_REGNO_MODE_OK.
13684 * genpreds.c (write_insn_preds_c): Add an include of target.h.
13685 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
13686 instead of HARD_REGNO_MODE_OK.
13687 * caller-save.c: Include target.h.
13688 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
13689 HARD_REGNO_MODE_OK.
13690 * combine.c (can_combine_p): Likewise.
13691 (combinable_i3pat): Likewise.
13692 (can_change_dest_mode): Likewise.
13693 * expr.c (init_expr_target): Likewise.
13694 (convert_move): Likewise.
13695 (convert_modes): Likewise.
13696 * ira.c (setup_prohibited_class_mode_regs): Likewise.
13697 (setup_prohibited_mode_move_regs): Likewise.
13698 * ira.h (target_ira): Likewise.
13699 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13700 * lra-constraints.c (process_alt_operands): Likewise.
13701 (split_reg): Likewise.
13702 * recog.c (peep2_find_free_register): Likewise.
13703 * ree.c (combine_reaching_defs): Likewise.
13704 * regcprop.c (maybe_mode_change): Likewise.
13705 * reginfo.c (init_reg_sets_1): Likewise.
13706 (choose_hard_reg_mode): Likewise.
13707 (simplifiable_subregs): Likewise.
13708 * regrename.c (check_new_reg_p): Likewise.
13709 * reload.c (find_valid_class): Likewise.
13710 (find_valid_class_1): Likewise.
13711 (reload_inner_reg_of_subreg): Likewise.
13712 (push_reload): Likewise.
13713 (combine_reloads): Likewise.
13714 (find_dummy_reload): Likewise.
13715 (find_reloads): Likewise.
13716 * reload1.c (find_reg): Likewise.
13717 (set_reload_reg): Likewise.
13718 (allocate_reload_reg): Likewise.
13719 (choose_reload_regs): Likewise.
13720 (reload_adjust_reg_for_temp): Likewise.
13721 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
13722 (simplify_subreg_regno): Likewise.
13723 * sel-sched.c (init_regs_for_mode): Likewise.
13724 * varasm.c (make_decl_rtl): Likewise.
13725 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
13726 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
13727 HARD_REGNO_MODE_OK.
13728 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
13729 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
13730 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13731 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
13732 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
13733 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13734 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
13735 (arc_mode_class): Delete.
13736 (HARD_REGNO_MODE_OK): Delete.
13737 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13738 (arc_hard_regno_mode_ok): Rename old array to...
13739 (arc_hard_regno_mode_ok_modes): ...this.
13740 (arc_conditional_register_usage): Update accordingly.
13741 (arc_mode_class): Make static.
13742 (arc_hard_regno_mode_ok): New function.
13743 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
13744 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
13745 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13746 (arm_hard_regno_mode_ok): Make static.
13747 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
13748 HARD_REGNO_MODE_OK.
13749 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
13750 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
13751 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
13753 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13754 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
13755 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
13756 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
13757 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
13758 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13759 * config/bfin/predicates.md (valid_reg_operand): Use
13760 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
13761 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
13762 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
13763 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13764 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
13765 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
13766 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13767 (cr16_hard_regno_mode_ok): Make static and return a bool.
13768 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
13769 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13770 (cris_hard_regno_mode_ok): New function.
13771 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
13772 (epiphany_mode_class): Delete.
13773 (HARD_REGNO_MODE_OK): Delete.
13774 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
13775 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13776 (hard_regno_mode_ok): Rename to...
13777 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
13778 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
13779 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
13780 HARD_REGNO_MODE_OK.
13781 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
13782 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
13783 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13784 (frv_hard_regno_mode_ok): Make static and return a bool.
13785 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
13786 HARD_REGNO_MODE_OK.
13787 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
13788 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
13789 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
13790 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
13792 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13793 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
13794 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
13795 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
13797 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13798 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
13799 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13800 (ia64_hard_regno_mode_ok): New function.
13801 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
13802 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13803 (iq2000_hard_regno_mode_ok): New function.
13804 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
13805 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13806 (lm32_hard_regno_mode_ok): New function.
13807 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
13808 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
13809 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
13810 instead of HARD_REGNO_MODE_OK.
13811 (m32c_hard_regno_ok): Rename to...
13812 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
13813 (m32c_cannot_change_mode_class): Update accordingly.
13814 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13815 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
13816 (m32r_mode_class): Delete.
13817 (HARD_REGNO_MODE_OK): Delete.
13818 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13819 (m32r_hard_regno_mode_ok): Rename to...
13820 (m32r_hard_regno_modes): ...this.
13821 (m32r_mode_class): Make static.
13822 (m32r_hard_regno_mode_ok): New function.
13823 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
13824 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
13825 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13826 (m68k_hard_regno_mode_ok): Make static.
13827 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
13828 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13829 (mcore_hard_regno_mode_ok): New function.
13830 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
13831 (HARD_REGNO_MODE_OK): Delete.
13832 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
13834 (microblaze_hard_regno_mode_ok_p): ...this and make static.
13835 (microblaze_hard_regno_mode_ok): New function.
13836 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13837 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
13838 (mips_hard_regno_mode_ok): Delete.
13839 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
13840 (mips_hard_regno_mode_ok_p): ...this and make static.
13841 (mips_hard_regno_mode_ok_p): Rename to...
13842 (mips_hard_regno_mode_ok_uncached): ...this.
13843 (mips_hard_regno_mode_ok): New function.
13844 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
13845 of HARD_REGNO_MODE_OK.
13846 (mips_option_override): Update after above name changes.
13847 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13848 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
13849 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
13850 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
13851 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
13852 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13853 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
13854 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
13855 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
13856 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13857 (msp430_hard_regno_mode_ok): Make static and return a bool.
13858 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
13859 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
13860 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
13862 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13863 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
13864 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
13865 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
13866 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
13867 (PA_HARD_REGNO_MODE_OK): ...this
13868 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
13869 (PA_HARD_REGNO_MODE_OK): ...this.
13870 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13871 (pa_hard_regno_mode_ok): New function.
13872 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
13873 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13874 (pdp11_hard_regno_mode_ok): New function.
13875 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
13876 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
13878 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
13880 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13881 (rs6000_hard_regno_mode_ok): Rename to...
13882 (rs6000_hard_regno_mode_ok_uncached): ...this.
13883 (rs6000_init_hard_regno_mode_ok): Update accordingly.
13884 (rs6000_hard_regno_mode_ok): New function.
13885 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
13886 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
13887 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
13888 (riscv_hard_regno_mode_ok): ...this and make static.
13889 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13890 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
13891 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
13892 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13893 (rl78_hard_regno_mode_ok): Make static and return bool.
13894 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
13895 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
13897 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
13898 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13899 (rs6000_hard_regno_mode_ok): Rename to...
13900 (rs6000_hard_regno_mode_ok_uncached): ...this.
13901 (rs6000_init_hard_regno_mode_ok): Update accordingly.
13902 (rs6000_hard_regno_mode_ok): New function.
13903 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
13904 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
13905 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13906 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
13907 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
13908 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
13909 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13910 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
13911 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
13912 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13913 (sh_hard_regno_mode_ok): Make static.
13914 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
13915 instead of HARD_REGNO_MODE_OK.
13916 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
13917 (sparc_mode_class): Delete.
13918 (HARD_REGNO_MODE_OK): Delete.
13919 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13920 (hard_regno_mode_classes): Make static.
13921 (sparc_mode_class): Likewise.
13922 (sparc_hard_regno_mode_ok): New function.
13923 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
13924 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
13925 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
13927 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13928 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
13929 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
13930 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
13931 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
13932 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13933 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
13934 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
13935 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13936 (visium_hard_regno_mode_ok): New function.
13937 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
13938 instead of HARD_REGNO_MODE_OK.
13939 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
13940 (HARD_REGNO_MODE_OK): Delete.
13941 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
13942 (xtensa_hard_regno_mode_ok_p): ...this and make static.
13943 (xtensa_option_override): Update accordingly.
13944 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13945 (xtensa_hard_regno_mode_ok): New function.
13946 * system.h (HARD_REGNO_MODE_OK): Poison.
13948 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13949 Alan Hayward <alan.hayward@arm.com>
13950 David Sherwood <david.sherwood@arm.com>
13952 * target.def (hard_regno_call_part_clobbered): New hook.
13953 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
13954 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
13955 * doc/tm.texi: Regenerate.
13956 * hooks.h (hook_bool_uint_mode_false): Declare.
13957 * hooks.c (hook_bool_uint_mode_false): New function.
13958 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13959 * cselib.c (cselib_process_insn): Use
13960 targetm.hard_regno_call_part_clobbered instead of
13961 HARD_REGNO_CALL_PART_CLOBBERED.
13962 * ira-conflicts.c (ira_build_conflicts): Likewise.
13963 * ira-costs.c (ira_tune_allocno_costs): Likewise.
13964 * lra-constraints.c (need_for_call_save_p): Likewise.
13965 * lra-lives.c: Include target.h.
13966 (check_pseudos_live_through_calls): Use
13967 targetm.hard_regno_call_part_clobbered instead of
13968 HARD_REGNO_CALL_PART_CLOBBERED.
13969 * regcprop.c: Include target.h.
13970 (copyprop_hardreg_forward_1): Use
13971 targetm.hard_regno_call_part_clobbered instead of
13972 HARD_REGNO_CALL_PART_CLOBBERED.
13973 * reginfo.c (choose_hard_reg_mode): Likewise.
13974 * regrename.c (check_new_reg_p): Likewise.
13975 * reload.c (find_equiv_reg): Likewise.
13976 * reload1.c (emit_reload_insns): Likewise.
13977 * sched-deps.c (deps_analyze_insn): Likewise.
13978 * sel-sched.c (init_regs_for_mode): Likewise.
13979 (mark_unavailable_hard_regs): Likewise.
13980 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
13981 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13982 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
13984 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13985 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13986 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
13988 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
13990 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13991 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13992 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
13994 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13995 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13996 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
13998 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13999 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
14001 * config/powerpcspe/powerpcspe.c
14002 (rs6000_hard_regno_call_part_clobbered): New function.
14003 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14004 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14005 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
14007 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14008 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14009 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
14011 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14012 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14013 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
14015 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14016 Alan Hayward <alan.hayward@arm.com>
14017 David Sherwood <david.sherwood@arm.com>
14019 * rtl.h (subreg_memory_offset): Declare.
14020 * emit-rtl.c (subreg_memory_offset): New function.
14021 * expmed.c (store_bit_field_1): Use it.
14022 * expr.c (undefined_operand_subword_p): Likewise.
14023 * simplify-rtx.c (simplify_subreg): Likewise.
14025 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
14027 PR rtl-optimization/57448
14030 * optabs.c (expand_atomic_load): Place compiler memory barriers if
14031 using atomic_load pattern.
14032 (expand_atomic_store): Likewise.
14034 2017-09-04 Jakub Jelinek <jakub@redhat.com>
14037 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
14038 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
14039 handling. Use replace_call_with_value with NULL instead of
14040 gsi_replace, unlink_stmt_vdef and release_defs.
14042 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
14045 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
14047 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14050 * rtl.h (emit_library_call_value_1): Declare.
14051 (emit_library_call): Replace declaration with a series of overloads.
14052 Remove the parameter count argument.
14053 (emit_library_call_value): Likewise.
14054 * calls.c (emit_library_call_value_1): Make global. Replace varargs
14055 with an "rtx_mode_t *".
14056 (emit_library_call_value): Delete.
14057 (emit_library_call): Likewise.
14058 * asan.c (asan_emit_stack_protection): Update calls accordingly.
14059 (asan_emit_allocas_unpoison): Likewise.
14060 * builtins.c (expand_builtin_powi): Likewise.
14061 (expand_asan_emit_allocas_unpoison): Likewise.
14062 * cfgexpand.c (expand_main_function): Likewise.
14063 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
14064 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
14065 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
14066 * config/arm/arm.c (arm_trampoline_init): Likewise.
14067 (arm_call_tls_get_addr): Likewise.
14068 (arm_expand_divmod_libfunc): Likewise.
14069 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
14070 (smulsi3_highpart): Likewise.
14071 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
14072 (c6x_expand_compare): Likewise.
14073 (c6x_expand_movmem): Likewise.
14074 * config/frv/frv.c (frv_trampoline_init): Likewise.
14075 * config/i386/i386.c (ix86_trampoline_init): Likewise.
14076 (ix86_expand_divmod_libfunc): Likewise.
14077 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
14078 (ia64_expand_compare): Likewise.
14079 (ia64_profile_hook): Likewise.
14080 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
14081 (nonlocal_goto): Likewise.
14082 (restore_stack_nonlocal): Likewise.
14083 * config/m32r/m32r.c (block_move_call): Likewise.
14084 (m32r_trampoline_init): Likewise.
14085 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
14086 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
14087 (m68k_call_m68k_read_tp): Likewise.
14088 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
14089 (microblaze_expand_divide): Likewise.
14090 * config/mips/mips.h (mips_args): Likewise.
14091 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
14092 (MIPS_ICACHE_SYNC): Likewise.
14093 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
14094 (nios2_trampoline_init): Likewise.
14095 * config/pa/pa.c (hppa_tls_call): Likewise.
14096 (pa_trampoline_init): Likewise.
14097 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
14098 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
14099 (expand_strn_compare): Likewise.
14100 (rs6000_generate_compare): Likewise.
14101 (rs6000_expand_float128_convert): Likewise.
14102 (output_profile_hook): Likewise.
14103 (rs6000_trampoline_init): Likewise.
14104 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
14105 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
14106 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
14107 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
14108 (rs6000_generate_compare): Likewise.
14109 (rs6000_expand_float128_convert): Likewise.
14110 (output_profile_hook): Likewise.
14111 (rs6000_trampoline_init): Likewise.
14112 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
14113 * config/sh/sh.c (sh_trampoline_init): Likewise.
14114 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
14115 (sparc_emit_float_lib_cmp): Likewise.
14116 (sparc32_initialize_trampoline): Likewise.
14117 (sparc64_initialize_trampoline): Likewise.
14118 (sparc_profile_hook): Likewise.
14119 * config/spu/spu.c (ea_load_store): Likewise.
14120 * config/spu/spu.md (floatunssidf2): Likewise.
14121 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
14122 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
14123 * config/visium/visium.c (expand_block_move_4): Likewise.
14124 (expand_block_move_2): Likewise.
14125 (expand_block_move_1): Likewise.
14126 (expand_block_set_4): Likewise.
14127 (expand_block_set_2): Likewise.
14128 (expand_block_set_1): Likewise.
14129 (visium_trampoline_init): Likewise.
14130 (visium_profile_hook): Likewise.
14131 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
14132 (xtensa_setup_frame_addresses): Likewise.
14133 (xtensa_trampoline_init): Likewise.
14134 * except.c (sjlj_emit_function_enter): Likewise.
14135 (sjlj_emit_function_exit): Likewise.
14136 * explow.c (allocate_dynamic_stack_space): Likewise.
14137 (probe_stack_range): Likewise.
14138 * expr.c (convert_mode_scalar): Likewise.
14139 * optabs.c (expand_binop): Likewise.
14140 (expand_twoval_binop_libfunc): Likewise.
14141 (expand_unop): Likewise.
14142 (prepare_cmp_insn): Likewise.
14143 (prepare_float_lib_cmp): Likewise.
14144 (expand_float): Likewise.
14145 (expand_fix): Likewise.
14146 (expand_fixed_convert): Likewise.
14147 (maybe_emit_sync_lock_test_and_set): Likewise.
14148 (expand_atomic_compare_and_swap): Likewise.
14149 (expand_mem_thread_fence): Likewise.
14150 (expand_atomic_fetch_op): Likewise.
14152 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
14154 * doc/generic.texi (OpenACC): Adjust URL.
14155 * doc/invoke.texi (C Dialect Options): Ditto.
14157 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
14159 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
14160 predicate for operand 1. Add (m,<S>) constraint.
14161 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
14162 Prevent memory operand 1 with register operand 2.
14164 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
14166 PR rtl-optimization/82024
14167 * combine.c (try_combine): If the combination result is a PARALLEL,
14168 and we only need to retain the SET in there that would be placed
14169 at I2, check that we can place that at I3 instead, before doing so.
14171 2017-09-01 Jakub Jelinek <jakub@redhat.com>
14174 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
14176 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
14177 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
14180 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
14181 Jeff Law <law@redhat.com>
14183 * varasm.c (bss_initializer_p): Do not put constants into .bss
14184 (categorize_decl_for_section): Handle bss_initializer_p returning
14185 false when DECL_INITIAL is NULL.
14187 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14190 * config/s390/s390.c (s390_can_inline_p): New function.
14192 2017-09-01 Jeff Law <law@redhat.com>
14194 PR tree-optimization/82052
14195 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
14196 Always initialize the returned slot after a hash table miss
14197 when INSERT is true.
14199 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
14201 * config/s390/s390.md (mem_signal_fence): Remove.
14202 * doc/md.texi (mem_signal_fence): Remove.
14203 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
14205 * target-insns.def (mem_signal_fence): Remove.
14207 2017-09-01 Jakub Jelinek <jakub@redhat.com>
14210 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
14213 * asan.c (create_odr_indicator): Strip name encoding from assembler
14214 name before appending it after __odr_asan_.
14216 2017-09-01 Martin Liska <mliska@suse.cz>
14218 PR tree-optimization/82059
14219 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
14220 frequency only when an edge is redirected.
14222 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14224 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
14225 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
14226 (arc_conditional_register_usage): Remove ARC600 lp_count
14228 (arc_file_start): Emit Tag_ARC_CPU_variation.
14229 (arc_can_use_doloop_p): New conditions to use ZOLs.
14230 (hwloop_fail): New function.
14231 (hwloop_optimize): Likewise.
14232 (hwloop_pattern_reg): Likewise.
14233 (arc_doloop_hooks): New struct, to be used with reorg_loops.
14234 (arc_reorg_loops): New function, calls reorg_loops.
14235 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
14236 (arc600_corereg_hazard): Remove ZOL checking, case handled by
14238 (arc_loop_hazard): Remove function, functionality moved into
14240 (arc_hazard): Remove arc_loop_hazard call.
14241 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
14242 into hwloop_optimize.
14243 (arc_label_align): Remove ZOL handling.
14244 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
14245 * config/arc/arc.md (doloop_begin): Remove pattern.
14246 (doloop_begin_i): Likewise.
14247 (doloop_end_i): Likewise.
14248 (doloop_fallback): Likewise.
14249 (doloop_fallback_m): Likewise.
14250 (doloop_end): Reimplement expand.
14251 (arc_lp): New pattern for LP instruction.
14252 (loop_end): New pattern.
14253 (loop_fail): Likewise.
14254 (decrement_and_branch_until_zero): Likewise.
14255 * config/arc/arc.opt (mlpc-width): New option.
14256 * doc/invoke.texi (mlpc-width): Document option.
14258 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14260 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
14261 (arc_ccfsm_advance): Fix checking for delay slots.
14262 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
14264 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14266 * config/arc/arc.md (movqi_insn): Add stores to save constant long
14268 (movhi_insn): Update store instruction constraint which are saving
14269 6-bit short immediates.
14270 (movsi_insn): Consider also short scaled load operations.
14271 (zero_extendhisi2_i): Use Usd constraint instead of T.
14272 (extendhisi2_i): Add q constraint.
14273 (arc_clzsi2): Add type and length attributes.
14274 (arc_ctzsi2): Likewise.
14275 * config/arc/constraints.md (Usc): Update constraint, the
14276 assembler can parse two relocations for a single instruction.
14278 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14280 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
14281 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
14283 2017-08-31 Olivier Hainque <hainque@adacore.com>
14285 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
14286 match as powerpc-wrs-vxworks*.
14288 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
14290 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
14291 register constraint for by-element operand.
14292 (aarch64_mls_elt_merge<mode>): Likewise.
14294 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14296 * config/arc/arc.c (arc_can_follow_jump): Check for short
14299 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14301 * config.gcc: Use g.opt for arc.
14302 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
14303 functionality moved to ...
14304 (legitimate_scaled_address_p): New function, ...here.
14305 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
14306 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
14307 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
14309 (arc_override_options): Handle G option.
14310 (arc_output_pic_addr_const): Correct function definition.
14311 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
14312 (arc_decl_anon_ns_mem_p): Delete.
14313 (arc_in_small_data_p): Overhaul this function to take into
14314 consideration the value given via G option.
14315 (arc_rewrite_small_data_1): Renamed and corrected old
14316 arc_rewrite_small_data function.
14317 (arc_rewrite_small_data): New function.
14318 (small_data_pattern): Don't use pic_offset_table_rtx.
14319 * config/arc/arc.h (CC1_SPEC): Recognize G option.
14320 * config/arc/simdext.md (movmisalignv2hi): Use
14321 prepare_move_operands function.
14323 (movmisalign*): Likewise.
14324 * doc/invoke.texi (ARC options): Document -G option.
14326 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14328 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
14330 * config/arc/arc.c (arc_print_operand): Output scalled address for
14331 sdata whenever is possible.
14332 (arc_in_small_data_p): Allow sdata for 64bit datum when double
14333 load/stores are available.
14334 (compact_sda_memory_operand): Check for the alignment required by
14335 code density instructions.
14336 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
14338 * config/arc/constraints.md (Usd): Update constraint.
14339 (Us0): New constraint.
14340 (Usc): Update constraint.
14342 2017-08-31 Richard Biener <rguenther@suse.de>
14344 PR middle-end/82054
14345 * dwarf2out.c (dwarf2out_early_global_decl): Process each
14346 function only once.
14348 2017-08-31 Tamar Christina <tamar.christina@arm.com>
14350 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
14351 Resize type_signature.
14353 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14354 Alan Hayward <alan.hayward@arm.com>
14355 David Sherwood <david.sherwood@arm.com>
14357 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
14358 subregs whose inner modes can be stored in GPRs.
14359 (aarch64_classify_index): Likewise.
14361 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14362 Alan Hayward <alan.hayward@arm.com>
14363 David Sherwood <david.sherwood@arm.com>
14365 * config/aarch64/iterators.md (V_cmp_result): Rename to...
14366 (V_INT_EQUIV): ...this.
14367 (v_cmp_result): Rename to...
14368 (v_int_equiv): ...this.
14369 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
14370 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
14371 (copysign<mode>3): Likewise.
14372 (aarch64_simd_bsl<mode>_internal): Likewise.
14373 (aarch64_simd_bsl<mode>): Likewise.
14374 (vec_cmp<mode><mode>): Likewise.
14375 (vcond<mode><mode>): Likewise.
14376 (vcond<v_cmp_mixed><mode>): Likewise.
14377 (vcondu<mode><v_cmp_mixed>): Likewise.
14378 (aarch64_cm<optab><mode>): Likewise.
14379 (aarch64_cmtst<mode>): Likewise.
14380 (aarch64_fac<optab><mode>): Likewise.
14381 (vec_perm_const<mode>): Likewise.
14382 (vcond_mask_<mode><v_cmp_result>): Rename to...
14383 (vcond_mask_<mode><v_int_equiv>): ...this.
14384 (vec_cmp<mode><v_cmp_result>): Rename to...
14385 (vec_cmp<mode><v_int_equiv>): ...this.
14387 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14388 Alan Hayward <alan.hayward@arm.com>
14389 David Sherwood <david.sherwood@arm.com>
14391 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
14393 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
14394 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
14395 (UNSPEC_LD4_DREG): New unspecs.
14396 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
14397 (aarch64_ld2<mode>_dreg_be): Replace with...
14398 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
14400 (aarch64_ld3<mode>_dreg_le)
14401 (aarch64_ld3<mode>_dreg_be): Replace with...
14402 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
14404 (aarch64_ld4<mode>_dreg_le)
14405 (aarch64_ld4<mode>_dreg_be): Replace with...
14406 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
14409 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14411 PR tree-optimization/81987
14412 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
14413 insert an initializer in a location not dominated by the stride
14416 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
14418 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
14419 on the entire header of the finally block in the fallthru case.
14421 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
14423 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
14425 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
14427 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
14428 rs6000_emit_move_from_cr and call renamed function.
14429 (rs6000_emit_prologue): Call renamed functions.
14430 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
14431 movesi_from_cr, remove volatile CRs.
14433 2017-08-30 Jon Beniston <jon@beniston.com>
14434 Richard Biener <rguenther@suse.de>
14436 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
14437 of VECTOR_MODE_P check.
14438 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
14439 element vector types.
14441 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14443 * df.h (df_read_modify_subreg_p): Remove in favor of...
14444 * rtl.h (read_modify_subreg_p): ...this new function. Take a
14445 const_rtx instead of an rtx.
14446 * cprop.c (local_cprop_find_used_regs): Update accordingly.
14447 * df-problems.c (df_word_lr_mark_ref): Likewise.
14448 * ira-lives.c (mark_pseudo_reg_live): Likewise.
14449 (mark_pseudo_reg_dead): Likewise.
14450 (mark_ref_dead): Likewise.
14451 * reginfo.c (init_subregs_of_mode): Likewise.
14452 * sched-deps.c (sched_analyze_1): Likewise.
14453 * df-scan.c (df_def_record_1): Likewise.
14454 (df_uses_record): Likewise.
14455 (df_read_modify_subreg_p): Remove in favor of...
14456 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
14457 const_rtx instead of an rtx.
14459 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14460 Alan Hayward <alan.hayward@arm.com>
14461 David Sherwood <david.sherwood@arm.com>
14463 * rtl.h (partial_subreg_p): New function.
14464 * caller-save.c (save_call_clobbered_regs): Use it.
14465 * calls.c (expand_call): Likewise.
14466 * combine.c (combinable_i3pat): Likewise.
14467 (simplify_set): Likewise.
14468 (make_extraction): Likewise.
14469 (make_compound_operation_int): Likewise.
14470 (gen_lowpart_or_truncate): Likewise.
14471 (force_to_mode): Likewise.
14472 (make_field_assignment): Likewise.
14473 (reg_truncated_to_mode): Likewise.
14474 (record_truncated_value): Likewise.
14475 (move_deaths): Likewise.
14476 * cse.c (record_jump_cond): Likewise.
14477 (cse_insn): Likewise.
14478 * cselib.c (cselib_lookup_1): Likewise.
14479 * expmed.c (extract_bit_field_using_extv): Likewise.
14480 * function.c (assign_parm_setup_reg): Likewise.
14481 * ifcvt.c (noce_convert_multiple_sets): Likewise.
14482 * ira-build.c (create_insn_allocnos): Likewise.
14483 * lra-coalesce.c (merge_pseudos): Likewise.
14484 * lra-constraints.c (match_reload): Likewise.
14485 (simplify_operand_subreg): Likewise.
14486 (curr_insn_transform): Likewise.
14487 * lra-lives.c (process_bb_lives): Likewise.
14488 * lra.c (new_insn_reg): Likewise.
14489 (lra_substitute_pseudo): Likewise.
14490 * regcprop.c (mode_change_ok): Likewise.
14491 (maybe_mode_change): Likewise.
14492 (copyprop_hardreg_forward_1): Likewise.
14493 * reload.c (push_reload): Likewise.
14494 (find_reloads): Likewise.
14495 (find_reloads_subreg_address): Likewise.
14496 * reload1.c (alter_reg): Likewise.
14497 (eliminate_regs_1): Likewise.
14498 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14500 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
14502 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
14503 back to if statements, including unpack.
14505 2017-08-30 Martin Liska <mliska@suse.cz>
14507 PR inline-asm/82001
14508 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
14510 (func_checker::compare_asm_inputs_outputs): ... this function.
14511 (func_checker::compare_gimple_asm): Use the function to compare
14512 also ASM constrains.
14513 * ipa-icf-gimple.h: Rename the function.
14515 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14516 Alan Hayward <alan.hayward@arm.com>
14517 David Sherwood <david.sherwood@arm.com>
14519 * coretypes.h (complex_mode): New type.
14520 * gdbhooks.py (build_pretty_printer): Handle it.
14521 * machmode.h (complex_mode): New class.
14522 (complex_mode::includes_p): New function.
14523 (is_complex_int_mode): Likewise.
14524 (is_complex_float_mode): Likewise.
14525 * genmodes.c (get_mode_class): Handle complex mode classes.
14526 * function.c (expand_function_end): Use is_complex_int_mode.
14528 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14529 Alan Hayward <alan.hayward@arm.com>
14530 David Sherwood <david.sherwood@arm.com>
14532 * coretypes.h (scalar_mode_pod): New typedef.
14533 * gdbhooks.py (build_pretty_printer): Handle it.
14534 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
14535 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
14536 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
14537 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
14538 as_a <scalar_mode>.
14540 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14541 Alan Hayward <alan.hayward@arm.com>
14542 David Sherwood <david.sherwood@arm.com>
14544 * machmode.h (mode_for_vector): Take a scalar_mode instead
14546 * stor-layout.c (mode_for_vector): Likewise.
14547 * explow.c (promote_mode): Use as_a <scalar_mode>.
14548 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
14550 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14551 Alan Hayward <alan.hayward@arm.com>
14552 David Sherwood <david.sherwood@arm.com>
14554 * target.def (preferred_simd_mode): Take a scalar_mode
14555 instead of a machine_mode.
14556 * targhooks.h (default_preferred_simd_mode): Likewise.
14557 * targhooks.c (default_preferred_simd_mode): Likewise.
14558 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
14559 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
14560 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
14561 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
14562 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
14563 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
14564 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
14565 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
14567 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
14568 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
14569 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
14570 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
14571 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
14572 * doc/tm.texi: Regenerate.
14573 * optabs-query.c (can_vec_mask_load_store_p): Return false for
14576 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14577 Alan Hayward <alan.hayward@arm.com>
14578 David Sherwood <david.sherwood@arm.com>
14580 * target.def (scalar_mode_supported_p): Take a scalar_mode
14581 instead of a machine_mode.
14582 * targhooks.h (default_scalar_mode_supported_p): Likewise.
14583 * targhooks.c (default_scalar_mode_supported_p): Likewise.
14584 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
14585 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
14586 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
14587 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
14588 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
14589 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
14590 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
14591 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
14592 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
14593 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
14594 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
14595 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
14597 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
14598 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
14599 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
14600 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
14601 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
14603 * doc/tm.texi: Regenerate.
14605 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14606 Alan Hayward <alan.hayward@arm.com>
14607 David Sherwood <david.sherwood@arm.com>
14609 * coretypes.h (opt_scalar_mode): New typedef.
14610 * gdbhooks.py (build_pretty_printers): Handle it.
14611 * machmode.h (mode_iterator::get_2xwider): Add overload for
14613 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
14615 * expr.c (convert_mode_scalar): Likewise.
14616 * omp-low.c (omp_clause_aligned_alignment): Likewise.
14617 * optabs.c (expand_float): Likewise.
14618 (expand_fix): Likewise.
14619 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14621 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14622 Alan Hayward <alan.hayward@arm.com>
14623 David Sherwood <david.sherwood@arm.com>
14625 * optabs.c (expand_float): Explicitly check for scalars before
14626 using a branching expansion.
14627 (expand_fix): Likewise.
14629 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14630 Alan Hayward <alan.hayward@arm.com>
14631 David Sherwood <david.sherwood@arm.com>
14633 * expr.c (convert_mode): Split scalar handling out into...
14634 (convert_mode_scalar): ...this new function. Treat the modes
14637 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14638 Alan Hayward <alan.hayward@arm.com>
14639 David Sherwood <david.sherwood@arm.com>
14641 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
14643 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
14645 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14646 Alan Hayward <alan.hayward@arm.com>
14647 David Sherwood <david.sherwood@arm.com>
14649 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
14650 rather than a machine_mode.
14651 (fixed_from_string): Likewise.
14652 (fixed_convert): Likewise.
14653 (fixed_convert_from_int): Likewise.
14654 (fixed_convert_from_real): Likewise.
14655 (real_convert_from_fixed): Likewise.
14656 * fixed-value.c (fixed_from_double_int): Likewise.
14657 (fixed_from_string): Likewise.
14658 (fixed_convert): Likewise.
14659 (fixed_convert_from_int): Likewise.
14660 (fixed_convert_from_real): Likewise.
14661 (real_convert_from_fixed): Likewise.
14662 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
14664 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14665 Alan Hayward <alan.hayward@arm.com>
14666 David Sherwood <david.sherwood@arm.com>
14668 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
14669 of separate mode class checks. Do not allow vector modes here.
14670 (immed_wide_int_const): Use as_a <scalar_mode>.
14671 * explow.c (trunc_int_for_mode): Likewise.
14672 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
14673 (wi::shwi): Likewise.
14674 (wi::min_value): Likewise.
14675 (wi::max_value): Likewise.
14676 * dwarf2out.c (loc_descriptor): Likewise.
14677 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
14678 for CONST_WIDE_INT.
14680 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14681 Alan Hayward <alan.hayward@arm.com>
14682 David Sherwood <david.sherwood@arm.com>
14684 * tree.h (SCALAR_TYPE_MODE): New macro.
14685 * expr.c (expand_expr_addr_expr_1): Use it.
14686 (expand_expr_real_2): Likewise.
14687 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
14688 (fold_convert_const_fixed_from_int): Likewise.
14689 (fold_convert_const_fixed_from_real): Likewise.
14690 (native_encode_fixed): Likewise
14691 (native_encode_complex): Likewise
14692 (native_encode_vector): Likewise.
14693 (native_interpret_fixed): Likewise.
14694 (native_interpret_real): Likewise.
14695 (native_interpret_complex): Likewise.
14696 (native_interpret_vector): Likewise.
14697 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
14698 (simd_clone_adjust_argument_types): Likewise.
14699 (simd_clone_init_simd_arrays): Likewise.
14700 (simd_clone_adjust): Likewise.
14701 * stor-layout.c (layout_type): Likewise.
14702 * tree.c (build_minus_one_cst): Likewise.
14703 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
14704 * tree-inline.c (estimate_move_cost): Likewise.
14705 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
14706 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
14707 (vectorizable_reduction): Likewise.
14708 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
14709 (vect_recog_mixed_size_cond_pattern): Likewise.
14710 (check_bool_pattern): Likewise.
14711 (adjust_bool_pattern): Likewise.
14712 (search_type_for_mask_1): Likewise.
14713 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
14714 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14715 (vectorizable_load): Likewise.
14716 (vectorizable_store): Likewise.
14717 * ubsan.c (ubsan_encode_value): Likewise.
14718 * varasm.c (output_constant): Likewise.
14720 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14721 Alan Hayward <alan.hayward@arm.com>
14722 David Sherwood <david.sherwood@arm.com>
14724 * coretypes.h (scalar_mode): New class.
14725 * machmode.h (scalar_mode): Likewise.
14726 (scalar_mode::includes_p): New function.
14727 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
14728 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
14729 * genmodes.c (get_mode_class): Handle remaining scalar modes.
14730 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
14731 * expmed.c (store_bit_field_1): Likewise.
14732 (extract_bit_field_1): Likewise.
14733 * expr.c (write_complex_part): Likewise.
14734 (read_complex_part): Likewise.
14735 (emit_move_complex_push): Likewise.
14736 (expand_expr_real_2): Likewise.
14737 * function.c (assign_parm_setup_reg): Likewise.
14738 (assign_parms_unsplit_complex): Likewise.
14739 * optabs.c (expand_binop): Likewise.
14740 * rtlanal.c (subreg_get_info): Likewise.
14741 * simplify-rtx.c (simplify_immed_subreg): Likewise.
14742 * varasm.c (output_constant_pool_2): Likewise.
14744 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14745 Alan Hayward <alan.hayward@arm.com>
14746 David Sherwood <david.sherwood@arm.com>
14748 * expmed.c (extract_high_half): Use scalar_int_mode and remove
14750 (expmed_mult_highpart_optab): Likewise.
14751 (expmed_mult_highpart): Likewise.
14753 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14754 Alan Hayward <alan.hayward@arm.com>
14755 David Sherwood <david.sherwood@arm.com>
14757 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
14758 instead of a machine_mode.
14759 (builtin_memset_read_str): Likewise.
14760 * builtins.c (c_readstr): Likewise.
14761 (builtin_memcpy_read_str): Likewise.
14762 (builtin_strncpy_read_str): Likewise.
14763 (builtin_memset_read_str): Likewise.
14764 (builtin_memset_gen_str): Likewise.
14765 (expand_builtin_signbit): Use scalar_int_mode for local variables.
14766 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
14767 instead of a machine_mode.
14768 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
14770 (make_extraction): Likewise.
14771 (try_widen_shift_mode): Take and return scalar_int_modes instead
14773 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
14774 a scalar_int_mode instead of a machine_mode.
14775 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
14776 (avr_addr_space_pointer_mode): Likewise.
14777 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
14778 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
14779 (msp430_unwind_word_mode): Likewise.
14780 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
14781 (spu_addr_space_pointer_mode): Likewise.
14782 (spu_addr_space_address_mode): Likewise.
14783 (spu_libgcc_cmp_return_mode): Likewise.
14784 (spu_libgcc_shift_count_mode): Likewise.
14785 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
14786 (rl78_addr_space_pointer_mode): Likewise.
14787 (fl78_unwind_word_mode): Likewise.
14788 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
14790 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
14791 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
14792 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
14793 (mips_valid_pointer_mode): Likewise.
14794 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
14795 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
14796 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
14798 (ft32_addr_space_address_mode): Likewise.
14799 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
14800 scalar_int_mode instead of a machine_mode.
14801 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
14803 (m32c_addr_space_address_mode): Likewise.
14804 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
14805 (rs6000_eh_return_filter_mode): Likewise.
14806 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
14807 (rs6000_eh_return_filter_mode): Likewise.
14808 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
14809 (s390_libgcc_shift_count_mode): Likewise.
14810 (s390_unwind_word_mode): Likewise.
14811 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
14813 * target.def (mode_rep_extended): Likewise.
14814 (valid_pointer_mode): Likewise.
14815 (addr_space.valid_pointer_mode): Likewise.
14816 (eh_return_filter_mode): Return a scalar_int_mode rather than
14818 (libgcc_cmp_return_mode): Likewise.
14819 (libgcc_shift_count_mode): Likewise.
14820 (unwind_word_mode): Likewise.
14821 (addr_space.pointer_mode): Likewise.
14822 (addr_space.address_mode): Likewise.
14823 * doc/tm.texi: Regenerate.
14824 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
14826 (do_jump): Use scalar_int_mode for local variables.
14827 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
14828 rather than a machine_mode.
14829 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
14830 (scompare_loc_descriptor_wide): Likewise.
14831 (scompare_loc_descriptor_narrow): Likewise.
14832 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
14834 * except.c (sjlj_emit_dispatch_table): Likewise.
14835 (expand_builtin_eh_copy_values): Likewise.
14836 * explow.c (convert_memory_address_addr_space_1): Likewise.
14837 Take a scalar_int_mode rather than a machine_mode.
14838 (convert_memory_address_addr_space): Take a scalar_int_mode rather
14839 than a machine_mode.
14840 (memory_address_addr_space): Use scalar_int_mode for local variables.
14841 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
14842 rather than a machine_mode.
14843 * expmed.c (mask_rtx): Likewise.
14844 (init_expmed_one_conv): Likewise.
14845 (expand_mult_highpart_adjust): Likewise.
14846 (extract_high_half): Likewise.
14847 (expmed_mult_highpart_optab): Likewise.
14848 (expmed_mult_highpart): Likewise.
14849 (expand_smod_pow2): Likewise.
14850 (expand_sdiv_pow2): Likewise.
14851 (emit_store_flag_int): Likewise.
14852 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
14854 (extract_low_bits): Likewise.
14855 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
14857 * expr.c (pieces_addr::adjust): Likewise.
14858 (can_store_by_pieces): Likewise.
14859 (store_by_pieces): Likewise.
14860 (clear_by_pieces_1): Likewise.
14861 (expand_expr_addr_expr_1): Likewise.
14862 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
14863 (expand_expr_real_1): Likewise.
14864 (try_casesi): Likewise.
14865 * final.c (shorten_branches): Likewise.
14866 * fold-const.c (fold_convert_const_int_from_fixed): Change the
14867 type of "mode" to machine_mode.
14868 * internal-fn.c (expand_arith_overflow_result_store): Take a
14869 scalar_int_mode rather than a machine_mode.
14870 (expand_mul_overflow): Use scalar_int_mode for local variables.
14871 * loop-doloop.c (doloop_modify): Likewise.
14872 (doloop_optimize): Likewise.
14873 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
14874 than a machine_mode.
14875 (expand_doubleword_shift_condmove): Likewise.
14876 (expand_doubleword_shift): Likewise.
14877 (expand_doubleword_clz): Likewise.
14878 (expand_doubleword_popcount): Likewise.
14879 (expand_doubleword_parity): Likewise.
14880 (expand_absneg_bit): Use scalar_int_mode for local variables.
14881 (prepare_float_lib_cmp): Likewise.
14882 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
14883 rather than a machine_mode.
14884 (convert_memory_address_addr_space): Likewise.
14885 (get_mode_bounds): Likewise.
14886 (get_address_mode): Return a scalar_int_mode rather than a
14888 * rtlanal.c (get_address_mode): Likewise.
14889 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
14890 than a machine_mode.
14891 * targhooks.c (default_mode_rep_extended): Likewise.
14892 (default_valid_pointer_mode): Likewise.
14893 (default_addr_space_valid_pointer_mode): Likewise.
14894 (default_eh_return_filter_mode): Return a scalar_int_mode rather
14895 than a machine_mode.
14896 (default_libgcc_cmp_return_mode): Likewise.
14897 (default_libgcc_shift_count_mode): Likewise.
14898 (default_unwind_word_mode): Likewise.
14899 (default_addr_space_pointer_mode): Likewise.
14900 (default_addr_space_address_mode): Likewise.
14901 * targhooks.h (default_eh_return_filter_mode): Likewise.
14902 (default_libgcc_cmp_return_mode): Likewise.
14903 (default_libgcc_shift_count_mode): Likewise.
14904 (default_unwind_word_mode): Likewise.
14905 (default_addr_space_pointer_mode): Likewise.
14906 (default_addr_space_address_mode): Likewise.
14907 (default_mode_rep_extended): Take a scalar_int_mode rather than
14909 (default_valid_pointer_mode): Likewise.
14910 (default_addr_space_valid_pointer_mode): Likewise.
14911 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
14913 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
14914 rather than a machine_mode.
14915 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
14916 for local variables.
14917 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14918 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
14919 than a machine_mode.
14921 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14922 Alan Hayward <alan.hayward@arm.com>
14923 David Sherwood <david.sherwood@arm.com>
14925 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
14926 the mode argument to scalar_int_mode.
14927 (do_jump_by_parts_zero_rtx): Likewise.
14928 (do_jump_by_parts_equality_rtx): Likewise.
14929 (do_jump_by_parts_greater): Take a mode argument.
14930 (do_jump_by_parts_equality): Likewise.
14931 (do_jump_1): Update calls accordingly.
14933 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14934 Alan Hayward <alan.hayward@arm.com>
14935 David Sherwood <david.sherwood@arm.com>
14937 * is-a.h (safe_dyn_cast): New function.
14938 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
14939 (jump_table_for_label): Likewise.
14940 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
14941 instead of an rtx_insn *.
14942 (shorten_branches): Use dyn_cast instead of LABEL_P and
14943 JUMP_TABLE_DATA_P. Use jump_table_for_label and
14944 rtx_jump_table_data::get_data_mode.
14945 (final_scan_insn): Likewise.
14947 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14948 Alan Hayward <alan.hayward@arm.com>
14949 David Sherwood <david.sherwood@arm.com>
14951 * combine.c (try_combine): Use is_a <scalar_int_mode> when
14952 trying to combine a full-register integer set with a subreg
14955 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14956 Alan Hayward <alan.hayward@arm.com>
14957 David Sherwood <david.sherwood@arm.com>
14959 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
14960 that is always either address_mode or pointer_mode.
14962 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14963 Alan Hayward <alan.hayward@arm.com>
14964 David Sherwood <david.sherwood@arm.com>
14966 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
14967 when the two are known to be equal.
14969 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14970 Alan Hayward <alan.hayward@arm.com>
14971 David Sherwood <david.sherwood@arm.com>
14973 * simplify-rtx.c (simplify_const_unary_operation): Use
14974 is_a <scalar_int_mode> instead of checking for a nonzero
14975 precision. Forcibly convert op_mode to a scalar_int_mode
14976 in that case. More clearly differentiate the operand and
14977 result modes and use the former when deciding what the value
14978 of a count-bits operation should be. Use is_int_mode instead
14979 of checking for a MODE_INT. Remove redundant check for whether
14980 this mode has a zero precision.
14982 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14983 Alan Hayward <alan.hayward@arm.com>
14984 David Sherwood <david.sherwood@arm.com>
14986 * optabs.c (widen_leading): Change the type of the mode argument
14987 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
14988 (widen_bswap): Likewise.
14989 (expand_parity): Likewise.
14990 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
14991 (expand_ffs): Likewise.
14992 (epand_unop): Check for scalar integer modes before calling the
14995 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14996 Alan Hayward <alan.hayward@arm.com>
14997 David Sherwood <david.sherwood@arm.com>
14999 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
15001 (expand_expr_real_1): Update call accordingly.
15003 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15004 Alan Hayward <alan.hayward@arm.com>
15005 David Sherwood <david.sherwood@arm.com>
15007 * expmed.c (store_bit_field_using_insv): Add op0_mode and
15008 value_mode arguments. Use scalar_int_mode internally.
15009 (store_bit_field_1): Rename the new integer mode from imode
15010 to op0_mode and use it instead of GET_MODE (op0). Update calls
15011 to store_split_bit_field, store_bit_field_using_insv and
15012 store_fixed_bit_field.
15013 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
15014 Use scalar_int_mode internally. Use a bit count rather than a mode
15015 when calculating the largest bit size for get_best_mode.
15016 Update calls to store_split_bit_field and store_fixed_bit_field_1.
15017 (store_fixed_bit_field_1): Add mode and value_mode arguments.
15018 Remove assertion that OP0 has a scalar integer mode.
15019 (store_split_bit_field): Add op0_mode and value_mode arguments.
15020 Update calls to extract_fixed_bit_field.
15021 (extract_bit_field_using_extv): Add an op0_mode argument.
15022 Use scalar_int_mode internally.
15023 (extract_bit_field_1): Rename the new integer mode from imode to
15024 op0_mode and use it instead of GET_MODE (op0). Update calls to
15025 extract_split_bit_field, extract_bit_field_using_extv and
15026 extract_fixed_bit_field.
15027 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
15028 to extract_split_bit_field and extract_fixed_bit_field_1.
15029 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
15030 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
15031 on the target mode.
15032 (extract_split_bit_field): Add an op0_mode argument. Update call
15033 to extract_fixed_bit_field.
15035 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15036 Alan Hayward <alan.hayward@arm.com>
15037 David Sherwood <david.sherwood@arm.com>
15039 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
15040 * explow.c (hard_function_value): Likewise.
15041 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
15042 convert_to_mode call outside the loop.
15043 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
15044 for the mode iterator. Require the mode specified by max_pieces
15046 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
15048 (copy_blkmode_to_reg): Likewise.
15049 (set_storage_via_setmem): Likewise.
15050 * optabs.c (prepare_cmp_insn): Likewise.
15051 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15052 * stor-layout.c (finish_bitfield_representative): Likewise.
15054 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15055 Alan Hayward <alan.hayward@arm.com>
15056 David Sherwood <david.sherwood@arm.com>
15058 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
15059 * expr.c (convert_move): Use them.
15060 (convert_modes): Likewise.
15061 (store_expr_with_bounds): Likewise.
15063 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15064 Alan Hayward <alan.hayward@arm.com>
15065 David Sherwood <david.sherwood@arm.com>
15067 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
15068 parameter for the mode of "x". Remove the "known_x", "known_mode"
15069 and "known_ret" arguments. Change the type of the mode argument
15070 to scalar_int_mode.
15071 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
15072 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
15073 (reg_num_sign_bit_copies_for_combine): Likewise.
15074 * rtlanal.c (nonzero_bits1): Likewise.
15075 (num_sign_bit_copies1): Likewise.
15076 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
15077 (reg_num_sign_bit_copies_general): Likewise.
15078 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
15079 (reg_nonzero_bits_general): Likewise.
15081 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15082 Alan Hayward <alan.hayward@arm.com>
15083 David Sherwood <david.sherwood@arm.com>
15085 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
15086 than in subroutines. Return 1 for non-integer modes.
15087 (cached_num_sign_bit_copies): Change the type of the mode parameter
15088 to scalar_int_mode.
15089 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
15090 classes. Handle CONST_INT_P first and then check whether X also
15091 has a scalar integer mode. Check the same thing for inner registers
15092 of a SUBREG and for values that are being extended or truncated.
15094 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15095 Alan Hayward <alan.hayward@arm.com>
15096 David Sherwood <david.sherwood@arm.com>
15098 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
15099 in subroutines. Return the mode mask for non-integer modes.
15100 (cached_nonzero_bits): Change the type of the mode parameter
15101 to scalar_int_mode.
15102 (nonzero_bits1): Likewise. Remove early exit for other mode
15103 classes. Handle CONST_INT_P first and then check whether X
15104 also has a scalar integer mode.
15106 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15107 Alan Hayward <alan.hayward@arm.com>
15108 David Sherwood <david.sherwood@arm.com>
15110 * expr.c (widest_int_mode_for_size): Make the comment match the code.
15111 Return a scalar_int_mode and assert that the size is greater than
15113 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
15114 (op_by_pieces_d::op_by_pieces_d): Likewise.
15115 (op_by_pieces_d::run): Likewise.
15116 (can_store_by_pieces): Likewise.
15118 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15119 Alan Hayward <alan.hayward@arm.com>
15120 David Sherwood <david.sherwood@arm.com>
15122 * combine.c (extract_left_shift): Add a mode argument and update
15124 (make_compound_operation_int): Change the type of the mode parameter
15125 to scalar_int_mode and update the call to extract_left_shift.
15127 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15128 Alan Hayward <alan.hayward@arm.com>
15129 David Sherwood <david.sherwood@arm.com>
15131 * combine.c (simplify_and_const_int): Change the type of the mode
15132 parameter to scalar_int_mode.
15133 (simplify_and_const_int_1): Likewise. Update recursive call.
15135 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15136 Alan Hayward <alan.hayward@arm.com>
15137 David Sherwood <david.sherwood@arm.com>
15139 * combine.c (simplify_compare_const): Check that the mode is a
15140 scalar_int_mode (rather than VOIDmode) before testing its
15142 (simplify_comparison): Move COMPARISON_P handling out of the
15143 loop and restrict the latter part of the loop to scalar_int_modes.
15144 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
15145 and when considering SUBREG_REGs. Use is_int_mode instead of
15146 checking GET_MODE_CLASS against MODE_INT.
15148 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15149 Alan Hayward <alan.hayward@arm.com>
15150 David Sherwood <david.sherwood@arm.com>
15152 * combine.c (try_widen_shift_mode): Move check for equal modes to...
15153 (simplify_shift_const_1): ...here. Use scalar_int_mode for
15154 shift_unit_mode and for modes involved in scalar shifts.
15156 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15157 Alan Hayward <alan.hayward@arm.com>
15158 David Sherwood <david.sherwood@arm.com>
15160 * combine.c (force_int_to_mode): New function, split out from...
15161 (force_to_mode): ...here. Keep xmode up-to-date and use it
15162 instead of GET_MODE (x).
15164 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15165 Alan Hayward <alan.hayward@arm.com>
15166 David Sherwood <david.sherwood@arm.com>
15168 * optabs-query.h (extraction_insn::struct_mode): Change type to
15169 opt_scalar_int_mode and update comment.
15170 (extraction_insn::field_mode): Change type to scalar_int_mode.
15171 (extraction_insn::pos_mode): Likewise.
15172 * combine.c (make_extraction): Update accordingly.
15173 * optabs-query.c (get_traditional_extraction_insn): Likewise.
15174 (get_optab_extraction_insn): Likewise.
15175 * recog.c (simplify_while_replacing): Likewise.
15176 * expmed.c (narrow_bit_field_mem): Change the type of the mode
15177 parameter to opt_scalar_int_mode.
15179 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15180 Alan Hayward <alan.hayward@arm.com>
15181 David Sherwood <david.sherwood@arm.com>
15183 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
15184 to a scalar_int_mode instead of a machine_mode.
15185 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
15186 (get_best_mode): Return a boolean and use a pointer argument to store
15187 the selected mode. Replace the limit mode parameter with a bit limit.
15188 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
15189 for the values returned by bit_field_mode_iterator::next_mode.
15190 (store_bit_field): Update call to get_best_mode.
15191 (store_fixed_bit_field): Likewise.
15192 (extract_fixed_bit_field): Likewise.
15193 * expr.c (optimize_bitfield_assignment_op): Likewise.
15194 * fold-const.c (optimize_bit_field_compare): Likewise.
15195 (fold_truth_andor_1): Likewise.
15196 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
15197 Update for new type of m_mode.
15198 (get_best_mode): As above.
15200 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15201 Alan Hayward <alan.hayward@arm.com>
15202 David Sherwood <david.sherwood@arm.com>
15204 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
15205 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
15206 (store_bit_field): Check is_a <scalar_int_mode> before calling
15207 strict_volatile_bitfield_p.
15208 (extract_bit_field): Likewise.
15210 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15211 Alan Hayward <alan.hayward@arm.com>
15212 David Sherwood <david.sherwood@arm.com>
15214 * target.def (cstore_mode): Return a scalar_int_mode.
15215 * doc/tm.texi: Regenerate.
15216 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
15217 * targhooks.h (default_cstore_mode): Likewise.
15218 * targhooks.c (default_cstore_mode): Likewise, using a forced
15220 * expmed.c (emit_cstore): Expect the target of the cstore to be
15223 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15224 Alan Hayward <alan.hayward@arm.com>
15225 David Sherwood <david.sherwood@arm.com>
15227 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
15229 (niter_desc): Likewise mode.
15230 (iv_analyze): Add a mode parameter.
15232 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
15233 and change its type to scalar_int_mode.
15234 * loop-iv.c: Update commentary at head of file.
15235 (iv_constant): Pass the mode paraeter before the rtx it describes
15236 and change its type to scalar_int_mode. Remove VOIDmode handling.
15237 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
15238 (iv_extend): Likewise.
15239 (shorten_into_mode): Likewise.
15240 (iv_add): Use scalar_int_mode.
15241 (iv_mult): Likewise.
15242 (iv_shift): Likewise.
15243 (canonicalize_iv_subregs): Likewise.
15244 (get_biv_step_1): Pass the outer_mode parameter before the rtx
15245 it describes and change its mode to scalar_int_mode. Also change
15246 the type of the returned inner_mode to scalar_int_mode.
15247 (get_biv_step): Likewise, turning outer_mode from a pointer
15248 into a direct parameter. Update call to get_biv_step_1.
15249 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
15250 iv_constant and get_biv_step.
15251 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
15252 and change its type to scalar_int_mode. Don't initialise iv->mode
15253 to VOIDmode and remove later checks for its still being VOIDmode.
15254 Update calls to iv_analyze_op and iv_analyze_expr. Check
15255 is_a <scalar_int_mode> when changing the mode under consideration.
15256 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
15257 Update call to iv_analyze_expr.
15258 (iv_analyze_op): Add a mode parameter. Reject subregs whose
15259 inner register is not also a scalar_int_mode. Update call to
15261 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
15262 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
15263 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
15264 separate mode class checks. Update calls to iv_analyze. Remove
15265 fix-up of VOIDmodes after iv_analyze_biv.
15266 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
15267 don't have a scalar_int_mode. Update call to biv_p.
15269 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15270 Alan Hayward <alan.hayward@arm.com>
15271 David Sherwood <david.sherwood@arm.com>
15273 * cfgexpand.c (convert_debug_memory_address): Use
15274 as_a <scalar_int_mode>.
15275 * combine.c (expand_compound_operation): Likewise.
15276 (make_extraction): Likewise.
15277 (change_zero_ext): Likewise.
15278 (simplify_comparison): Likewise.
15279 * cse.c (cse_insn): Likewise.
15280 * dwarf2out.c (minmax_loc_descriptor): Likewise.
15281 (mem_loc_descriptor): Likewise.
15282 (loc_descriptor): Likewise.
15283 * expmed.c (init_expmed_one_mode): Likewise.
15284 (synth_mult): Likewise.
15285 (emit_store_flag_1): Likewise.
15286 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
15287 of a comparison with size.
15288 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
15289 (reduce_to_bit_field_precision): Likewise.
15290 * function.c (expand_function_end): Likewise.
15291 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
15292 * loop-doloop.c (doloop_modify): Likewise.
15293 * optabs.c (expand_binop): Likewise.
15294 (expand_unop): Likewise.
15295 (expand_copysign_absneg): Likewise.
15296 (prepare_cmp_insn): Likewise.
15297 (maybe_legitimize_operand): Likewise.
15298 * recog.c (const_scalar_int_operand): Likewise.
15299 * rtlanal.c (get_address_mode): Likewise.
15300 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15301 (simplify_cond_clz_ctz): Likewise.
15302 * tree-nested.c (get_nl_goto_field): Likewise.
15303 * tree.c (build_vector_type_for_mode): Likewise.
15304 * var-tracking.c (use_narrower_mode): Likewise.
15306 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15307 Alan Hayward <alan.hayward@arm.com>
15308 David Sherwood <david.sherwood@arm.com>
15310 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
15311 * builtins.c (expand_builtin_signbit): Use it.
15312 * cfgexpand.c (expand_debug_expr): Likewise.
15313 * dojump.c (do_jump): Likewise.
15314 (do_compare_and_jump): Likewise.
15315 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
15316 * expmed.c (make_tree): Likewise.
15317 * expr.c (expand_expr_real_2): Likewise.
15318 (expand_expr_real_1): Likewise.
15319 (try_casesi): Likewise.
15320 * fold-const-call.c (fold_const_call_ss): Likewise.
15321 * fold-const.c (unextend): Likewise.
15322 (extract_muldiv_1): Likewise.
15323 (fold_single_bit_test): Likewise.
15324 (native_encode_int): Likewise.
15325 (native_encode_string): Likewise.
15326 (native_interpret_int): Likewise.
15327 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
15328 * internal-fn.c (expand_addsub_overflow): Likewise.
15329 (expand_neg_overflow): Likewise.
15330 (expand_mul_overflow): Likewise.
15331 (expand_arith_overflow): Likewise.
15332 * match.pd: Likewise.
15333 * stor-layout.c (layout_type): Likewise.
15334 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
15335 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
15336 * tree-ssanames.c (get_range_info): Likewise.
15337 * tree-switch-conversion.c (array_value_type) Likewise.
15338 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
15339 (vect_recog_divmod_pattern): Likewise.
15340 (vect_recog_mixed_size_cond_pattern): Likewise.
15341 * tree-vrp.c (extract_range_basic): Likewise.
15342 (simplify_float_conversion_using_ranges): Likewise.
15343 * tree.c (int_fits_type_p): Likewise.
15344 * ubsan.c (instrument_bool_enum_load): Likewise.
15345 * varasm.c (mergeable_string_section): Likewise.
15346 (narrowing_initializer_constant_valid_p): Likewise.
15347 (output_constant): Likewise.
15349 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15350 Alan Hayward <alan.hayward@arm.com>
15351 David Sherwood <david.sherwood@arm.com>
15353 * machmode.h (NARROWEST_INT_MODE): New macro.
15354 * expr.c (alignment_for_piecewise_move): Use it instead of
15355 GET_CLASS_NARROWEST_MODE (MODE_INT).
15356 (push_block): Likewise.
15357 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
15359 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15361 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15362 Alan Hayward <alan.hayward@arm.com>
15363 David Sherwood <david.sherwood@arm.com>
15365 * postreload.c (move2add_valid_value_p): Change the type of the
15366 mode parameter to scalar_int_mode.
15367 (move2add_use_add2_insn): Add a mode parameter and use it instead
15369 (move2add_use_add3_insn): Likewise.
15370 (reload_cse_move2add): Update accordingly.
15372 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15373 Alan Hayward <alan.hayward@arm.com>
15374 David Sherwood <david.sherwood@arm.com>
15376 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
15378 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
15379 * optabs.c (expand_unop): Likewise.
15381 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15382 Alan Hayward <alan.hayward@arm.com>
15383 David Sherwood <david.sherwood@arm.com>
15385 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
15386 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
15387 (popcount_loc_descriptor): Likewise.
15388 (bswap_loc_descriptor): Likewise.
15389 (rotate_loc_descriptor): Likewise.
15390 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
15391 calling the functions above.
15393 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15394 Alan Hayward <alan.hayward@arm.com>
15395 David Sherwood <david.sherwood@arm.com>
15397 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
15399 (try_combine): Likewise.
15400 (simplify_if_then_else): Likewise.
15401 * cse.c (cse_insn): Likewise.
15402 * dwarf2out.c (mem_loc_descriptor): Likewise.
15403 * emit-rtl.c (gen_lowpart_common): Likewise.
15404 * simplify-rtx.c (simplify_truncation): Likewise.
15405 (simplify_binary_operation_1): Likewise.
15406 (simplify_const_relational_operation): Likewise.
15407 (simplify_ternary_operation): Likewise.
15408 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
15410 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15411 Alan Hayward <alan.hayward@arm.com>
15412 David Sherwood <david.sherwood@arm.com>
15414 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
15415 * reload.c (push_reload): Likewise.
15416 (find_reloads): Likewise.
15418 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15419 Alan Hayward <alan.hayward@arm.com>
15420 David Sherwood <david.sherwood@arm.com>
15422 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
15423 (make_compound_operation_int): Likewise.
15424 (change_zero_ext): Likewise.
15425 * expr.c (convert_move): Likewise.
15426 (convert_modes): Likewise.
15427 * fwprop.c (forward_propagate_subreg): Likewise.
15428 * loop-iv.c (get_biv_step_1): Likewise.
15429 * optabs.c (widen_operand): Likewise.
15430 * postreload.c (move2add_valid_value_p): Likewise.
15431 * recog.c (simplify_while_replacing): Likewise.
15432 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15433 (simplify_binary_operation_1): Likewise. Remove redundant
15434 mode equality check.
15436 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15437 Alan Hayward <alan.hayward@arm.com>
15438 David Sherwood <david.sherwood@arm.com>
15440 * combine.c (combine_simplify_rtx): Add checks for
15441 is_a <scalar_int_mode>.
15442 (simplify_if_then_else): Likewise.
15443 (make_field_assignment): Likewise.
15444 (simplify_comparison): Likewise.
15445 * ifcvt.c (noce_try_bitop): Likewise.
15446 * loop-invariant.c (canonicalize_address_mult): Likewise.
15447 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15449 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15450 Alan Hayward <alan.hayward@arm.com>
15451 David Sherwood <david.sherwood@arm.com>
15453 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
15454 is_a <scalar_int_mode> instead of != BLKmode.
15456 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15457 Alan Hayward <alan.hayward@arm.com>
15458 David Sherwood <david.sherwood@arm.com>
15460 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
15461 instead of != VOIDmode.
15462 * combine.c (if_then_else_cond): Likewise.
15463 (change_zero_ext): Likewise.
15464 * dwarf2out.c (mem_loc_descriptor): Likewise.
15465 (loc_descriptor): Likewise.
15466 * rtlanal.c (canonicalize_condition): Likewise.
15467 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
15469 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15470 Alan Hayward <alan.hayward@arm.com>
15471 David Sherwood <david.sherwood@arm.com>
15473 * simplify-rtx.c (simplify_binary_operation_1): Use
15474 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
15476 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15477 Alan Hayward <alan.hayward@arm.com>
15478 David Sherwood <david.sherwood@arm.com>
15480 * wide-int.h (int_traits<unsigned char>) New class.
15481 (int_traits<unsigned short>) Likewise.
15482 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
15483 Use GET_MODE_UNIT_PRECISION and remove redundant test for
15485 * combine.c (set_nonzero_bits_and_sign_copies): Use
15486 is_a <scalar_int_mode>.
15487 (find_split_point): Likewise.
15488 (combine_simplify_rtx): Likewise.
15489 (simplify_logical): Likewise.
15490 (expand_compound_operation): Likewise.
15491 (expand_field_assignment): Likewise.
15492 (make_compound_operation): Likewise.
15493 (extended_count): Likewise.
15494 (change_zero_ext): Likewise.
15495 (simplify_comparison): Likewise.
15496 * dwarf2out.c (scompare_loc_descriptor): Likewise.
15497 (ucompare_loc_descriptor): Likewise.
15498 (minmax_loc_descriptor): Likewise.
15499 (mem_loc_descriptor): Likewise.
15500 (loc_descriptor): Likewise.
15501 * expmed.c (init_expmed_one_mode): Likewise.
15502 * lra-constraints.c (lra_constraint_offset): Likewise.
15503 * optabs.c (prepare_libcall_arg): Likewise.
15504 * postreload.c (move2add_note_store): Likewise.
15505 * reload.c (operands_match_p): Likewise.
15506 * rtl.h (load_extend_op): Likewise.
15507 * rtlhooks.c (gen_lowpart_general): Likewise.
15508 * simplify-rtx.c (simplify_truncation): Likewise.
15509 (simplify_unary_operation_1): Likewise.
15510 (simplify_binary_operation_1): Likewise.
15511 (simplify_const_binary_operation): Likewise.
15512 (simplify_const_relational_operation): Likewise.
15513 (simplify_subreg): Likewise.
15514 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15515 * var-tracking.c (adjust_mems): Likewise.
15516 (prepare_call_arguments): Likewise.
15518 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15519 Alan Hayward <alan.hayward@arm.com>
15520 David Sherwood <david.sherwood@arm.com>
15522 * machmode.h (is_int_mode): New fuction.
15523 * combine.c (find_split_point): Use it.
15524 (combine_simplify_rtx): Likewise.
15525 (simplify_if_then_else): Likewise.
15526 (simplify_set): Likewise.
15527 (simplify_shift_const_1): Likewise.
15528 (simplify_comparison): Likewise.
15529 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
15530 * cse.c (notreg_cost): Likewise.
15531 (cse_insn): Likewise.
15532 * cselib.c (cselib_lookup_1): Likewise.
15533 * dojump.c (do_jump_1): Likewise.
15534 (do_compare_rtx_and_jump): Likewise.
15535 * dse.c (get_call_args): Likewise.
15536 * dwarf2out.c (rtl_for_decl_init): Likewise.
15537 (native_encode_initializer): Likewise.
15538 * expmed.c (emit_store_flag_1): Likewise.
15539 (emit_store_flag): Likewise.
15540 * expr.c (convert_modes): Likewise.
15541 (store_field): Likewise.
15542 (expand_expr_real_1): Likewise.
15543 * fold-const.c (fold_read_from_constant_string): Likewise.
15544 * gimple-ssa-sprintf.c (get_format_string): Likewise.
15545 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
15546 * optabs.c (expand_binop): Likewise.
15547 (expand_unop): Likewise.
15548 (expand_abs_nojump): Likewise.
15549 (expand_one_cmpl_abs_nojump): Likewise.
15550 * simplify-rtx.c (mode_signbit_p): Likewise.
15551 (val_signbit_p): Likewise.
15552 (val_signbit_known_set_p): Likewise.
15553 (val_signbit_known_clear_p): Likewise.
15554 (simplify_relational_operation_1): Likewise.
15555 * tree.c (vector_type_mode): Likewise.
15557 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15558 Alan Hayward <alan.hayward@arm.com>
15559 David Sherwood <david.sherwood@arm.com>
15561 * machmode.h (smallest_mode_for_size): Fix formatting.
15562 (smallest_int_mode_for_size): New function.
15563 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
15564 instead of smallest_mode_for_size.
15565 * combine.c (make_extraction): Likewise.
15566 * config/arc/arc.c (arc_expand_movmem): Likewise.
15567 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
15568 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15569 * config/s390/s390.c (s390_expand_insv): Likewise.
15570 * config/sparc/sparc.c (assign_int_registers): Likewise.
15571 * config/spu/spu.c (spu_function_value): Likewise.
15572 (spu_function_arg): Likewise.
15573 * coverage.c (get_gcov_type): Likewise.
15574 (get_gcov_unsigned_t): Likewise.
15575 * dse.c (find_shift_sequence): Likewise.
15576 * expmed.c (store_bit_field_1): Likewise.
15577 * expr.c (convert_move): Likewise.
15578 (store_field): Likewise.
15579 * internal-fn.c (expand_arith_overflow): Likewise.
15580 * optabs-query.c (get_best_extraction_insn): Likewise.
15581 * optabs.c (expand_twoval_binop_libfunc): Likewise.
15582 * stor-layout.c (layout_type): Likewise.
15583 (initialize_sizetypes): Likewise.
15584 * targhooks.c (default_get_mask_mode): Likewise.
15585 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
15587 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15588 Alan Hayward <alan.hayward@arm.com>
15589 David Sherwood <david.sherwood@arm.com>
15591 * machmode.h (opt_mode::else_blk): New function.
15592 (int_mode_for_mode): Declare.
15593 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
15594 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
15596 * cfgexpand.c (expand_debug_expr): Likewise.
15597 * combine.c (gen_lowpart_or_truncate): Likewise.
15598 (gen_lowpart_for_combine): Likewise.
15599 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
15600 * config/avr/avr.c (avr_to_int_mode): Likewise.
15601 (avr_out_plus_1): Likewise.
15602 (avr_out_plus): Likewise.
15603 (avr_out_round): Likewise.
15604 * config/i386/i386.c (ix86_split_to_parts): Likewise.
15605 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15606 (s390_expand_vcond): Likewise.
15607 * config/spu/spu.c (spu_split_immediate): Likewise.
15608 (spu_expand_mov): Likewise.
15609 * dse.c (get_stored_val): Likewise.
15610 * expmed.c (store_bit_field_1): Likewise.
15611 (convert_extracted_bit_field): Use int_mode_for_mode instead of
15613 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
15614 (extract_low_bits): Likewise.
15615 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
15616 handling rather than repeating the check.
15617 (emit_group_store): Likewise.
15618 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
15619 * optabs.c (expand_absneg_bit): Likewise.
15620 (expand_copysign_absneg): Likewise.
15621 (expand_copysign_bit): Likewise.
15622 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
15623 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
15624 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
15625 * var-tracking.c (prepare_call_arguments): Likewise.
15626 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
15627 int_mode_for_mode instead of mode_for_size.
15628 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15630 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15631 Alan Hayward <alan.hayward@arm.com>
15632 David Sherwood <david.sherwood@arm.com>
15634 * machmode.h (int_mode_for_size): New function.
15635 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
15636 instead of mode_for_size.
15637 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
15639 * combine.c (expand_field_assignment): Use int_mode_for_size
15640 instead of mode_for_size.
15641 (make_extraction): Likewise.
15642 (simplify_shift_const_1): Likewise.
15643 (simplify_comparison): Likewise.
15644 * dojump.c (do_jump): Likewise.
15645 * dwarf2out.c (mem_loc_descriptor): Likewise.
15646 * emit-rtl.c (init_derived_machine_modes): Likewise.
15647 * expmed.c (flip_storage_order): Likewise.
15648 (convert_extracted_bit_field): Likewise.
15649 * expr.c (copy_blkmode_from_reg): Likewise.
15650 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
15651 * internal-fn.c (expand_mul_overflow): Likewise.
15652 * lower-subreg.c (simple_move): Likewise.
15653 * optabs-libfuncs.c (init_optabs): Likewise.
15654 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15655 * tree.c (vector_type_mode): Likewise.
15656 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
15657 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
15658 * tree-vect-generic.c (expand_vector_parallel): Likewise.
15659 * tree-vect-stmts.c (vectorizable_load): Likewise.
15660 (vectorizable_store): Likewise.
15662 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15663 Alan Hayward <alan.hayward@arm.com>
15664 David Sherwood <david.sherwood@arm.com>
15666 * coretypes.h (pod_mode): New type.
15667 (scalar_int_mode_pod): New typedef.
15668 * machmode.h (pod_mode): New class.
15669 (int_n_data_t::m): Change type to scalar_int_mode_pod.
15670 * genmodes.c (emit_mode_int_n): Update accordingly.
15671 * lower-subreg.h (target_lower_subreg): Change type to
15672 scalar_int_mode_pod.
15673 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
15674 scalar_int_mode_pod.
15676 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15677 Alan Hayward <alan.hayward@arm.com>
15678 David Sherwood <david.sherwood@arm.com>
15680 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
15681 machine_mode to scalar_int_mode.
15682 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
15683 (rs6000_option_override_internal): Remove cast to int.
15684 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
15685 machine_mode to scalar_int_mode.
15686 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
15687 (rs6000_option_override_internal): Remove cast to int.
15688 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
15689 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
15691 * config/s390/s390.c (s390_expand_builtin): Likewise.
15692 * coretypes.h (scalar_int_mode): New type.
15693 (opt_scalar_int_mode): New typedef.
15694 * machmode.h (scalar_int_mode): New class.
15695 (scalar_int_mode::includes_p): New function.
15696 (byte_mode): Change type to scalar_int_mode.
15697 (word_mode): Likewise.
15698 (ptr_mode): Likewise.
15699 * emit-rtl.c (byte_mode): Likewise.
15700 (word_mode): Likewise.
15701 (ptr_mode): Likewise.
15702 (init_derived_machine_modes): Update accordingly.
15703 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
15704 and MODE_PARTIAL_INT.
15705 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
15706 opt_scalar_int_mode.
15708 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15709 Alan Hayward <alan.hayward@arm.com>
15710 David Sherwood <david.sherwood@arm.com>
15712 * target.def (libgcc_floating_mode_supported_p): Take a
15714 * doc/tm.texi: Regenerate.
15715 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
15717 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
15718 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
15721 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15722 Alan Hayward <alan.hayward@arm.com>
15723 David Sherwood <david.sherwood@arm.com>
15725 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
15726 * doc/tm.texi: Regenerate.
15727 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
15728 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
15729 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
15730 * targhooks.h (default_floatn_mode): Likewise.
15731 * targhooks.c (default_floatn_mode): Likewise.
15732 * tree.c (build_common_tree_nodes): Update accordingly.
15734 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15735 Alan Hayward <alan.hayward@arm.com>
15736 David Sherwood <david.sherwood@arm.com>
15738 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
15739 (mode_iterator::iterate_p): Likewise.
15740 (mode_iterator::get_wider): Likewise.
15741 * expr.c (init_expr_target): Use opt_scalar_float_mode.
15743 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15744 Alan Hayward <alan.hayward@arm.com>
15745 David Sherwood <david.sherwood@arm.com>
15747 * coretypes.h (opt_scalar_float_mode): New typedef.
15748 * machmode.h (float_mode_for_size): New function.
15749 * emit-rtl.c (double_mode): Delete.
15750 (init_emit_once): Use float_mode_for_size.
15751 * stor-layout.c (layout_type): Likewise.
15752 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
15754 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15755 Alan Hayward <alan.hayward@arm.com>
15756 David Sherwood <david.sherwood@arm.com>
15758 * output.h (assemble_real): Take a scalar_float_mode.
15759 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
15760 * config/arm/arm.md (consttable_4): Likewise.
15761 (consttable_8): Likewise.
15762 (consttable_16): Likewise.
15763 * config/mips/mips.md (consttable_float): Likewise.
15764 * config/s390/s390.c (s390_output_pool_entry): Likewise.
15765 * varasm.c (assemble_real): Take a scalar_float_mode.
15766 (output_constant_pool_2): Update accordingly.
15767 (output_constant): Likewise.
15769 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15770 Alan Hayward <alan.hayward@arm.com>
15771 David Sherwood <david.sherwood@arm.com>
15773 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
15774 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
15775 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
15776 (native_encode_real): Likewise.
15777 (native_interpret_real): Likewise.
15778 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
15779 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15781 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15782 Alan Hayward <alan.hayward@arm.com>
15783 David Sherwood <david.sherwood@arm.com>
15785 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
15786 <scalar_float_mode>. Simplify.
15787 (gen_extend_conv_libfunc): Likewise.
15789 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15790 Alan Hayward <alan.hayward@arm.com>
15791 David Sherwood <david.sherwood@arm.com>
15793 * coretypes.h (scalar_float_mode): New type.
15794 * machmode.h (mode_traits::from_int): Use machine_mode if
15795 USE_ENUM_MODES is defined.
15796 (is_a): New function.
15798 (dyn_cast): Likewise.
15799 (scalar_float_mode): New class.
15800 (scalar_float_mode::includes_p): New function.
15801 (is_float_mode): Likewise.
15802 * gdbhooks.py (MachineModePrinter): New class.
15803 (build_pretty_printer): Use it for scalar_float_mode.
15804 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
15805 (format_helper::format_helper): Turn into a template.
15806 * genmodes.c (get_mode_class): New function.
15807 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
15808 or machine_mode if none.
15809 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
15810 as_a <scalar_float_mode>.
15811 * dwarf2out.c (mem_loc_descriptor): Likewise.
15812 (insert_float): Likewise.
15813 (add_const_value_attribute): Likewise.
15814 * simplify-rtx.c (simplify_immed_subreg): Likewise.
15815 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
15816 (expand_unop): Update accordingly.
15817 (expand_abs_nojump): Likewise.
15818 (expand_copysign_absneg): Take a scalar_float_mode.
15819 (expand_copysign_bit): Likewise.
15820 (expand_copysign): Update accordingly.
15822 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15823 Alan Hayward <alan.hayward@arm.com>
15824 David Sherwood <david.sherwood@arm.com>
15826 * coretypes.h (opt_mode): New class.
15827 * machmode.h (opt_mode): Likewise.
15828 (opt_mode::else_void): New function.
15829 (opt_mode::require): Likewise.
15830 (opt_mode::exists): Likewise.
15831 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
15832 (GET_MODE_2XWIDER_MODE): Likewise.
15833 (mode_iterator::get_wider): Update accordingly.
15834 (mode_iterator::get_2xwider): Likewise.
15835 (mode_iterator::get_known_wider): Likewise, turning into a template.
15836 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
15837 forcing a wider mode to exist.
15838 * config/cr16/cr16.h (LONG_REG_P): Likewise.
15839 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15840 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
15841 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
15842 * lower-subreg.c (init_lower_subreg): Likewise.
15843 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
15844 on the final iteration.
15845 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
15846 a wider mode exists before asking for a move pattern.
15847 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
15848 forcing a wider mode to exist.
15849 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
15850 returning false if no such mode exists.
15851 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
15852 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
15853 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
15854 Avoid checking for a MODE_INT if we already know the mode is not a
15856 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
15857 forcing a wider mode to exist.
15858 (expmed_mult_highpart_optab): Likewise.
15859 (expmed_mult_highpart): Likewise.
15860 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
15862 * lto-streamer-in.c (lto_input_mode_table): Likewise.
15863 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
15864 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15865 * internal-fn.c (expand_mul_overflow): Update use of
15866 GET_MODE_2XWIDER_MODE.
15867 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15868 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
15869 GET_MODE_WIDER_MODE.
15870 (convert_plusminus_to_widen): Likewise.
15871 * tree-switch-conversion.c (array_value_type): Likewise.
15872 * var-tracking.c (emit_note_insn_var_location): Likewise.
15873 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15874 Return false inside rather than outside the loop if no wider mode
15876 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
15877 and GET_MODE_2XWIDER_MODE
15878 (can_compare_p): Use else_void.
15879 * gdbhooks.py (OptMachineModePrinter): New class.
15880 (build_pretty_printer): Use it for opt_mode.
15882 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15883 Alan Hayward <alan.hayward@arm.com>
15884 David Sherwood <david.sherwood@arm.com>
15886 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
15887 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
15889 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15890 Alan Hayward <alan.hayward@arm.com>
15891 David Sherwood <david.sherwood@arm.com>
15893 * machmode.h (mode_traits): New structure.
15894 (get_narrowest_mode): New function.
15895 (mode_iterator::start): Likewise.
15896 (mode_iterator::iterate_p): Likewise.
15897 (mode_iterator::get_wider): Likewise.
15898 (mode_iterator::get_known_wider): Likewise.
15899 (mode_iterator::get_2xwider): Likewise.
15900 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
15901 (FOR_EACH_MODE): Likewise.
15902 (FOR_EACH_MODE_FROM): Likewise.
15903 (FOR_EACH_MODE_UNTIL): Likewise.
15904 (FOR_EACH_WIDER_MODE): Likewise.
15905 (FOR_EACH_2XWIDER_MODE): Likewise.
15906 * builtins.c (expand_builtin_strlen): Use new mode iterators.
15907 * combine.c (simplify_comparison): Likewise
15908 * config/i386/i386.c (type_natural_mode): Likewise.
15909 * cse.c (cse_insn): Likewise.
15910 * dse.c (find_shift_sequence): Likewise.
15911 * emit-rtl.c (init_derived_machine_modes): Likewise.
15912 (init_emit_once): Likewise.
15913 * explow.c (hard_function_value): Likewise.
15914 * expmed.c (extract_fixed_bit_field_1): Likewise.
15915 (extract_bit_field_1): Likewise.
15916 (expand_divmod): Likewise.
15917 (emit_store_flag_1): Likewise.
15918 * expr.c (init_expr_target): Likewise.
15919 (convert_move): Likewise.
15920 (alignment_for_piecewise_move): Likewise.
15921 (widest_int_mode_for_size): Likewise.
15922 (emit_block_move_via_movmem): Likewise.
15923 (copy_blkmode_to_reg): Likewise.
15924 (set_storage_via_setmem): Likewise.
15925 (compress_float_constant): Likewise.
15926 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15927 * optabs-query.c (get_best_extraction_insn): Likewise.
15928 * optabs.c (expand_binop): Likewise.
15929 (expand_twoval_unop): Likewise.
15930 (expand_twoval_binop): Likewise.
15931 (widen_leading): Likewise.
15932 (widen_bswap): Likewise.
15933 (expand_parity): Likewise.
15934 (expand_unop): Likewise.
15935 (prepare_cmp_insn): Likewise.
15936 (prepare_float_lib_cmp): Likewise.
15937 (expand_float): Likewise.
15938 (expand_fix): Likewise.
15939 (expand_sfix_optab): Likewise.
15940 * postreload.c (move2add_use_add2_insn): Likewise.
15941 * reg-stack.c (reg_to_stack): Likewise.
15942 * reginfo.c (choose_hard_reg_mode): Likewise.
15943 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15944 * stor-layout.c (mode_for_size): Likewise.
15945 (smallest_mode_for_size): Likewise.
15946 (mode_for_vector): Likewise.
15947 (finish_bitfield_representative): Likewise.
15948 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
15949 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
15950 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15951 * var-tracking.c (prepare_call_arguments): Likewise.
15953 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15954 Alan Hayward <alan.hayward@arm.com>
15955 David Sherwood <david.sherwood@arm.com>
15957 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
15958 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
15959 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
15960 * machmode.h (mode_size): Move earlier in file.
15961 (mode_precision): Likewise.
15962 (mode_inner): Likewise.
15963 (mode_nunits): Likewise.
15964 (mode_unit_size): Likewise.
15965 (unit_unit_precision): Likewise.
15966 (mode_wider): Likewise.
15967 (mode_2xwider): Likewise.
15968 (machine_mode): New class.
15969 (mode_to_bytes): New function.
15970 (mode_to_bits): Likewise.
15971 (mode_to_precision): Likewise.
15972 (mode_to_inner): Likewise.
15973 (mode_to_unit_size): Likewise.
15974 (mode_to_unit_precision): Likewise.
15975 (mode_to_nunits): Likewise.
15976 (GET_MODE_SIZE): Use mode_to_bytes.
15977 (GET_MODE_BITSIZE): Use mode_to_bits.
15978 (GET_MODE_PRECISION): Use mode_to_precision.
15979 (GET_MODE_INNER): Use mode_to_inner.
15980 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
15981 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
15982 (GET_MODE_NUNITS): Use mode_to_nunits.
15983 * system.h (ALWAYS_INLINE): New macro.
15984 * config/powerpcspe/powerpcspe-c.c
15985 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
15986 int for arg1_mode and arg2_mode.
15988 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15989 Alan Hayward <alan.hayward@arm.com>
15990 David Sherwood <david.sherwood@arm.com>
15992 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
15993 Prefix mode names with E_ in case statements.
15994 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15995 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
15996 (aarch64_split_simd_move): Likewise.
15997 (aarch64_gen_storewb_pair): Likewise.
15998 (aarch64_gen_loadwb_pair): Likewise.
15999 (aarch64_gen_store_pair): Likewise.
16000 (aarch64_gen_load_pair): Likewise.
16001 (aarch64_get_condition_code_1): Likewise.
16002 (aarch64_constant_pool_reload_icode): Likewise.
16003 (get_rsqrte_type): Likewise.
16004 (get_rsqrts_type): Likewise.
16005 (get_recpe_type): Likewise.
16006 (get_recps_type): Likewise.
16007 (aarch64_gimplify_va_arg_expr): Likewise.
16008 (aarch64_simd_container_mode): Likewise.
16009 (aarch64_emit_load_exclusive): Likewise.
16010 (aarch64_emit_store_exclusive): Likewise.
16011 (aarch64_expand_compare_and_swap): Likewise.
16012 (aarch64_gen_atomic_cas): Likewise.
16013 (aarch64_emit_bic): Likewise.
16014 (aarch64_emit_atomic_swap): Likewise.
16015 (aarch64_emit_atomic_load_op): Likewise.
16016 (aarch64_evpc_trn): Likewise.
16017 (aarch64_evpc_uzp): Likewise.
16018 (aarch64_evpc_zip): Likewise.
16019 (aarch64_evpc_ext): Likewise.
16020 (aarch64_evpc_rev): Likewise.
16021 (aarch64_evpc_dup): Likewise.
16022 (aarch64_gen_ccmp_first): Likewise.
16023 (aarch64_gen_ccmp_next): Likewise.
16024 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
16025 (alpha_emit_xfloating_libcall): Likewise.
16026 (emit_insxl): Likewise.
16027 (alpha_arg_type): Likewise.
16028 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
16029 (arc_preferred_simd_mode): Likewise.
16030 (arc_secondary_reload): Likewise.
16031 (get_arc_condition_code): Likewise.
16032 (arc_print_operand): Likewise.
16033 (arc_legitimate_constant_p): Likewise.
16034 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16035 * config/arc/arc.md (casesi_load): Likewise.
16036 (casesi_compact_jump): Likewise.
16037 * config/arc/predicates.md (proper_comparison_operator): Likewise.
16038 (cc_use_register): Likewise.
16039 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16040 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
16041 (arm_init_iwmmxt_builtins): Likewise.
16042 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
16043 (neon_expand_vector_init): Likewise.
16044 (arm_attr_length_move_neon): Likewise.
16045 (maybe_get_arm_condition_code): Likewise.
16046 (arm_emit_vector_const): Likewise.
16047 (arm_preferred_simd_mode): Likewise.
16048 (arm_output_iwmmxt_tinsr): Likewise.
16049 (thumb1_output_casesi): Likewise.
16050 (thumb2_output_casesi): Likewise.
16051 (arm_emit_load_exclusive): Likewise.
16052 (arm_emit_store_exclusive): Likewise.
16053 (arm_expand_compare_and_swap): Likewise.
16054 (arm_evpc_neon_vuzp): Likewise.
16055 (arm_evpc_neon_vzip): Likewise.
16056 (arm_evpc_neon_vrev): Likewise.
16057 (arm_evpc_neon_vtrn): Likewise.
16058 (arm_evpc_neon_vext): Likewise.
16059 (arm_validize_comparison): Likewise.
16060 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
16061 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
16062 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
16063 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
16064 (c6x_preferred_simd_mode): Likewise.
16065 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
16066 (epiphany_rtx_costs): Likewise.
16067 * config/epiphany/predicates.md (proper_comparison_operator):
16069 * config/frv/frv.c (condexec_memory_operand): Likewise.
16070 (frv_emit_move): Likewise.
16071 (output_move_single): Likewise.
16072 (output_condmove_single): Likewise.
16073 (frv_hard_regno_mode_ok): Likewise.
16074 (frv_matching_accg_mode): Likewise.
16075 * config/h8300/h8300.c (split_adds_subs): Likewise.
16076 (h8300_rtx_costs): Likewise.
16077 (h8300_print_operand): Likewise.
16078 (compute_mov_length): Likewise.
16079 (output_logical_op): Likewise.
16080 (compute_logical_op_length): Likewise.
16081 (compute_logical_op_cc): Likewise.
16082 (h8300_shift_needs_scratch_p): Likewise.
16083 (output_a_shift): Likewise.
16084 (compute_a_shift_length): Likewise.
16085 (compute_a_shift_cc): Likewise.
16086 (expand_a_rotate): Likewise.
16087 (output_a_rotate): Likewise.
16088 * config/i386/i386.c (classify_argument): Likewise.
16089 (function_arg_advance_32): Likewise.
16090 (function_arg_32): Likewise.
16091 (function_arg_64): Likewise.
16092 (function_value_64): Likewise.
16093 (ix86_gimplify_va_arg): Likewise.
16094 (ix86_legitimate_constant_p): Likewise.
16095 (put_condition_code): Likewise.
16096 (split_double_mode): Likewise.
16097 (ix86_avx256_split_vector_move_misalign): Likewise.
16098 (ix86_expand_vector_logical_operator): Likewise.
16099 (ix86_split_idivmod): Likewise.
16100 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
16101 (ix86_build_const_vector): Likewise.
16102 (ix86_build_signbit_mask): Likewise.
16103 (ix86_match_ccmode): Likewise.
16104 (ix86_cc_modes_compatible): Likewise.
16105 (ix86_expand_branch): Likewise.
16106 (ix86_expand_sse_cmp): Likewise.
16107 (ix86_expand_sse_movcc): Likewise.
16108 (ix86_expand_int_sse_cmp): Likewise.
16109 (ix86_expand_vec_perm_vpermi2): Likewise.
16110 (ix86_expand_vec_perm): Likewise.
16111 (ix86_expand_sse_unpack): Likewise.
16112 (ix86_expand_int_addcc): Likewise.
16113 (ix86_split_to_parts): Likewise.
16114 (ix86_vectorize_builtin_gather): Likewise.
16115 (ix86_vectorize_builtin_scatter): Likewise.
16116 (avx_vpermilp_parallel): Likewise.
16117 (inline_memory_move_cost): Likewise.
16118 (ix86_tieable_integer_mode_p): Likewise.
16119 (x86_maybe_negate_const_int): Likewise.
16120 (ix86_expand_vector_init_duplicate): Likewise.
16121 (ix86_expand_vector_init_one_nonzero): Likewise.
16122 (ix86_expand_vector_init_one_var): Likewise.
16123 (ix86_expand_vector_init_concat): Likewise.
16124 (ix86_expand_vector_init_interleave): Likewise.
16125 (ix86_expand_vector_init_general): Likewise.
16126 (ix86_expand_vector_set): Likewise.
16127 (ix86_expand_vector_extract): Likewise.
16128 (emit_reduc_half): Likewise.
16129 (ix86_emit_i387_round): Likewise.
16130 (ix86_mangle_type): Likewise.
16131 (ix86_expand_round_sse4): Likewise.
16132 (expand_vec_perm_blend): Likewise.
16133 (canonicalize_vector_int_perm): Likewise.
16134 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
16135 (expand_vec_perm_1): Likewise.
16136 (expand_vec_perm_interleave3): Likewise.
16137 (expand_vec_perm_even_odd_pack): Likewise.
16138 (expand_vec_perm_even_odd_1): Likewise.
16139 (expand_vec_perm_broadcast_1): Likewise.
16140 (ix86_vectorize_vec_perm_const_ok): Likewise.
16141 (ix86_expand_vecop_qihi): Likewise.
16142 (ix86_expand_mul_widen_hilo): Likewise.
16143 (ix86_expand_sse2_abs): Likewise.
16144 (ix86_expand_pextr): Likewise.
16145 (ix86_expand_pinsr): Likewise.
16146 (ix86_preferred_simd_mode): Likewise.
16147 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
16148 * config/i386/sse.md (*andnot<mode>3): Likewise.
16149 (<mask_codefor><code><mode>3<mask_name>): Likewise.
16150 (*<code><mode>3): Likewise.
16151 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
16152 (ia64_expand_atomic_op): Likewise.
16153 (ia64_arg_type): Likewise.
16154 (ia64_mode_to_int): Likewise.
16155 (ia64_scalar_mode_supported_p): Likewise.
16156 (ia64_vector_mode_supported_p): Likewise.
16157 (expand_vec_perm_broadcast): Likewise.
16158 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
16159 (iq2000_function_arg_advance): Likewise.
16160 (iq2000_function_arg): Likewise.
16161 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
16162 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
16163 (m68k_libcall_value): Likewise.
16164 (m68k_function_value): Likewise.
16165 (sched_attr_op_type): Likewise.
16166 * config/mcore/mcore.c (mcore_output_move): Likewise.
16167 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
16169 (microblaze_function_arg): Likewise.
16170 * config/mips/mips.c (mips16_build_call_stub): Likewise.
16171 (mips_print_operand): Likewise.
16172 (mips_mode_ok_for_mov_fmt_p): Likewise.
16173 (mips_vector_mode_supported_p): Likewise.
16174 (mips_preferred_simd_mode): Likewise.
16175 (mips_expand_vpc_loongson_even_odd): Likewise.
16176 (mips_expand_vec_unpack): Likewise.
16177 (mips_expand_vi_broadcast): Likewise.
16178 (mips_expand_vector_init): Likewise.
16179 (mips_expand_vec_reduc): Likewise.
16180 (mips_expand_msa_cmp): Likewise.
16181 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
16182 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
16183 (cc_flags_for_mode): Likewise.
16184 * config/msp430/msp430.c (msp430_print_operand): Likewise.
16185 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
16186 (nds32_output_casesi_pc_relative): Likewise.
16187 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16188 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
16189 (nvptx_gen_unpack): Likewise.
16190 (nvptx_gen_pack): Likewise.
16191 (nvptx_gen_shuffle): Likewise.
16192 (nvptx_gen_wcast): Likewise.
16193 (nvptx_preferred_simd_mode): Likewise.
16194 * config/pa/pa.c (pa_secondary_reload): Likewise.
16195 * config/pa/predicates.md (base14_operand): Likewise.
16196 * config/powerpcspe/powerpcspe-c.c
16197 (altivec_resolve_overloaded_builtin): Likewise.
16198 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
16200 (rs6000_preferred_simd_mode): Likewise.
16201 (output_vec_const_move): Likewise.
16202 (rs6000_expand_vector_extract): Likewise.
16203 (rs6000_split_vec_extract_var): Likewise.
16204 (reg_offset_addressing_ok_p): Likewise.
16205 (rs6000_legitimate_offset_address_p): Likewise.
16206 (rs6000_legitimize_address): Likewise.
16207 (rs6000_emit_set_const): Likewise.
16208 (rs6000_const_vec): Likewise.
16209 (rs6000_emit_move): Likewise.
16210 (spe_build_register_parallel): Likewise.
16211 (rs6000_darwin64_record_arg_recurse): Likewise.
16212 (swap_selector_for_mode): Likewise.
16213 (spe_init_builtins): Likewise.
16214 (paired_init_builtins): Likewise.
16215 (altivec_init_builtins): Likewise.
16216 (do_load_for_compare): Likewise.
16217 (rs6000_generate_compare): Likewise.
16218 (rs6000_expand_float128_convert): Likewise.
16219 (emit_load_locked): Likewise.
16220 (emit_store_conditional): Likewise.
16221 (rs6000_output_function_epilogue): Likewise.
16222 (rs6000_handle_altivec_attribute): Likewise.
16223 (rs6000_function_value): Likewise.
16224 (emit_fusion_gpr_load): Likewise.
16225 (emit_fusion_p9_load): Likewise.
16226 (emit_fusion_p9_store): Likewise.
16227 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
16228 (fusion_gpr_mem_load): Likewise.
16229 (fusion_addis_mem_combo_load): Likewise.
16230 (fusion_addis_mem_combo_store): Likewise.
16231 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
16232 (fusion_gpr_mem_load): Likewise.
16233 (fusion_addis_mem_combo_load): Likewise.
16234 (fusion_addis_mem_combo_store): Likewise.
16235 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16237 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
16238 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16239 (rs6000_preferred_simd_mode): Likewise.
16240 (output_vec_const_move): Likewise.
16241 (rs6000_expand_vector_extract): Likewise.
16242 (rs6000_split_vec_extract_var): Likewise.
16243 (reg_offset_addressing_ok_p): Likewise.
16244 (rs6000_legitimate_offset_address_p): Likewise.
16245 (rs6000_legitimize_address): Likewise.
16246 (rs6000_emit_set_const): Likewise.
16247 (rs6000_const_vec): Likewise.
16248 (rs6000_emit_move): Likewise.
16249 (rs6000_darwin64_record_arg_recurse): Likewise.
16250 (swap_selector_for_mode): Likewise.
16251 (paired_init_builtins): Likewise.
16252 (altivec_init_builtins): Likewise.
16253 (rs6000_expand_float128_convert): Likewise.
16254 (emit_load_locked): Likewise.
16255 (emit_store_conditional): Likewise.
16256 (rs6000_output_function_epilogue): Likewise.
16257 (rs6000_handle_altivec_attribute): Likewise.
16258 (rs6000_function_value): Likewise.
16259 (emit_fusion_gpr_load): Likewise.
16260 (emit_fusion_p9_load): Likewise.
16261 (emit_fusion_p9_store): Likewise.
16262 * config/rx/rx.c (rx_gen_move_template): Likewise.
16263 (flags_from_mode): Likewise.
16264 * config/s390/predicates.md (s390_alc_comparison): Likewise.
16265 (s390_slb_comparison): Likewise.
16266 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
16267 (s390_vector_mode_supported_p): Likewise.
16268 (s390_cc_modes_compatible): Likewise.
16269 (s390_match_ccmode_set): Likewise.
16270 (s390_canonicalize_comparison): Likewise.
16271 (s390_emit_compare_and_swap): Likewise.
16272 (s390_branch_condition_mask): Likewise.
16273 (s390_rtx_costs): Likewise.
16274 (s390_secondary_reload): Likewise.
16275 (__SECONDARY_RELOAD_CASE): Likewise.
16276 (s390_expand_cs): Likewise.
16277 (s390_preferred_simd_mode): Likewise.
16278 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
16279 * config/sh/sh.c (sh_print_operand): Likewise.
16280 (dump_table): Likewise.
16281 (sh_secondary_reload): Likewise.
16282 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16283 * config/sh/sh.md (casesi_worker_1): Likewise.
16284 (casesi_worker_2): Likewise.
16285 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
16286 (fcc_comparison_operator): Likewise.
16287 * config/sparc/sparc.c (sparc_expand_move): Likewise.
16288 (emit_soft_tfmode_cvt): Likewise.
16289 (sparc_preferred_simd_mode): Likewise.
16290 (output_cbranch): Likewise.
16291 (sparc_print_operand): Likewise.
16292 (sparc_expand_vec_perm_bmask): Likewise.
16293 (vector_init_bshuffle): Likewise.
16294 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
16295 (spu_vector_mode_supported_p): Likewise.
16296 (spu_expand_insv): Likewise.
16297 (spu_emit_branch_or_set): Likewise.
16298 (spu_handle_vector_attribute): Likewise.
16299 (spu_builtin_splats): Likewise.
16300 (spu_builtin_extract): Likewise.
16301 (spu_builtin_promote): Likewise.
16302 (spu_expand_sign_extend): Likewise.
16303 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
16304 (tilegx_simd_int): Likewise.
16305 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
16306 (tilepro_simd_int): Likewise.
16307 * config/v850/v850.c (const_double_split): Likewise.
16308 (v850_print_operand): Likewise.
16309 (ep_memory_offset): Likewise.
16310 * config/vax/vax.c (vax_rtx_costs): Likewise.
16311 (vax_output_int_move): Likewise.
16312 (vax_output_int_add): Likewise.
16313 (vax_output_int_subtract): Likewise.
16314 * config/visium/predicates.md (visium_branch_operator): Likewise.
16315 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
16316 (visium_print_operand_address): Likewise.
16317 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16318 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
16319 (xtensa_expand_conditional_branch): Likewise.
16320 (xtensa_copy_incoming_a7): Likewise.
16321 (xtensa_output_literal): Likewise.
16322 * dfp.c (decimal_real_maxval): Likewise.
16323 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
16325 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16326 Alan Hayward <alan.hayward@arm.com>
16327 David Sherwood <david.sherwood@arm.com>
16329 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
16330 (mode_nunits_inline): Likewise.
16331 (mode_inner_inline): Likewise.
16332 (mode_unit_size_inline): Likewise.
16333 (mode_unit_precision_inline): Likewise.
16334 (emit_insn_modes_h): Likewise. Also emit a #define of the
16336 (emit_mode_wider): Add an E_ prefix to mode names.
16337 (emit_mode_complex): Likewise.
16338 (emit_mode_inner): Likewise.
16339 (emit_mode_adjustments): Likewise.
16340 (emit_mode_int_n): Likewise.
16341 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
16342 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
16343 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
16344 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
16345 (CRC32_BUILTIN, ENTRY): Likewise.
16346 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
16347 (aarch64_pop_regs): Likewise.
16348 (aarch64_process_components): Likewise.
16349 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
16350 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
16351 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
16352 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
16353 * config/arm/arm.c (arm_init_libfuncs): Likewise.
16354 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
16356 * config/i386/i386-builtin.def (pcmpestr): Likewise.
16357 (pcmpistr): Likewise.
16358 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
16359 * config/mmix/mmix.c (mmix_output_condition): Likewise.
16360 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
16362 * config/rl78/rl78.c (mduc_regs): Likewise.
16363 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16364 (htm_expand_builtin): Likewise.
16365 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
16366 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
16367 * config/xtensa/xtensa.c (print_operand): Likewise.
16368 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
16369 (NUM_MODE_VECTOR_INT): Likewise.
16370 * genoutput.c (null_operand): Likewise.
16371 (output_operand_data): Likewise.
16372 * genrecog.c (print_parameter_value): Likewise.
16373 * lra.c (debug_operand_data): Likewise.
16375 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16377 * dwarf2out.c (scompare_loc_descriptor_wide)
16378 (scompare_loc_descriptor_narrow): New functions, split out from...
16379 (scompare_loc_descriptor): ...here.
16380 * expmed.c (emit_store_flag_int): New function, split out from...
16381 (emit_store_flag): ...here.
16383 2017-08-30 Richard Biener <rguenther@suse.de>
16385 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
16386 (dwarf2out_early_finish): Move setting of AT_pubnames from
16387 early debug output to early finish.
16389 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16391 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
16392 and -mdata-region to the assembler.
16394 2017-08-30 Richard Biener <rguenther@suse.de>
16396 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
16398 (gen_subprogram_die): Add DW_AT_object_pointer only early.
16399 (dwarf2out_early_global_decl): Only generate a DIE for the
16400 abstract origin if it doesn't already exist or is a declaration DIE.
16401 (resolve_addr): Do not add the linkage name twice when
16402 generating a stub DIE for the DW_TAG_GNU_call_site target.
16404 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16406 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16407 Use machine_mode rather than int for arg1_mode.
16409 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
16412 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
16413 that the second argument of the built-in functions to unpack
16414 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
16415 switch statement instead a lot of if statements.
16416 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
16417 Allow 64-bit values to be in Altivec registers as well as
16418 traditional floating point registers.
16419 (pack<mode>, FMOVE128_VSX iterator): Likewise.
16421 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
16423 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
16424 MAX_REGS_PER_ADDRESS == 1.
16426 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
16428 * config/i386/i386.opt (flag_fentry): Do not init to -1.
16429 * config/i386/i386.c (ix86_option_override_internal): Simplify
16430 setting of opts->x_flag_entry.
16432 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16433 Jakub Jelinek <jakub@redhat.com>
16434 Richard Biener <rguenther@suse.de>
16436 PR tree-optimization/81503
16437 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
16438 folded constant fits in the target type; reorder tests for clarity.
16440 2017-08-29 Martin Liska <mliska@suse.cz>
16442 * passes.def: Include pass_lower_switch.
16443 * stmt.c (dump_case_nodes): Remove and move to
16444 tree-switch-conversion.
16445 (case_values_threshold): Likewise.
16446 (expand_switch_as_decision_tree_p): Likewise.
16447 (emit_case_decision_tree): Likewise.
16448 (expand_case): Likewise.
16449 (balance_case_nodes): Likewise.
16450 (node_has_low_bound): Likewise.
16451 (node_has_high_bound): Likewise.
16452 (node_is_bounded): Likewise.
16453 (emit_case_nodes): Likewise.
16454 (struct simple_case_node): New struct.
16455 (add_case_node): Remove.
16456 (emit_case_dispatch_table): Use vector instead of case_list.
16457 (reset_out_edges_aux): Remove.
16458 (compute_cases_per_edge): Likewise.
16459 (expand_case): Build list of simple_case_node.
16460 (expand_sjlj_dispatch_table): Use it.
16461 * tree-switch-conversion.c (struct case_node): Moved from
16462 stmt.c and adjusted.
16463 (emit_case_nodes): Likewise.
16464 (node_has_low_bound): Likewise.
16465 (node_has_high_bound): Likewise.
16466 (node_is_bounded): Likewise.
16467 (case_values_threshold): Likewise.
16468 (reset_out_edges_aux): Likewise.
16469 (compute_cases_per_edge): Likewise.
16470 (add_case_node): Likewise.
16471 (dump_case_nodes): Likewise.
16472 (balance_case_nodes): Likewise.
16473 (expand_switch_as_decision_tree_p): Likewise.
16474 (emit_jump): Likewise.
16475 (emit_case_decision_tree): Likewise.
16476 (try_switch_expansion): Likewise.
16477 (do_jump_if_equal): Likewise.
16478 (emit_cmp_and_jump_insns): Likewise.
16479 (fix_phi_operands_for_edge): New function.
16480 (record_phi_operand_mapping): Likewise.
16481 (class pass_lower_switch): New pass.
16482 (pass_lower_switch::execute): New function.
16483 (make_pass_lower_switch): Likewise.
16484 (conditional_probability):
16485 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
16486 * tree-pass.h: Add make_pass_lower_switch.
16488 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16491 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
16494 2017-08-29 Richard Biener <rguenther@suse.de>
16496 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
16497 we do not add a DW_AT_inline attribute twice.
16498 (gen_subprogram_die): Remove code setting DW_AT_inline on
16499 DECL_ABSTRACT_P nodes.
16501 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
16503 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
16504 calls to internal functions.
16505 (gimplify_modify_expr): Likewise.
16506 * tree-call-cdce.c (use_internal_fn): Likewise.
16507 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
16508 (convert_to_divmod): Set the nothrow flag.
16509 * tree-if-conv.c (predicate_mem_writes): Likewise.
16510 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
16511 (vectorizable_call): Likewise.
16512 (vectorizable_store): Likewise.
16513 (vectorizable_load): Likewise.
16514 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
16515 (vect_recog_mask_conversion_pattern): Likewise.
16517 2017-08-29 Martin Liska <mliska@suse.cz>
16520 * gcc.c (driver_handle_option): Add new argument.
16521 * opts-common.c (handle_option): Pass
16522 target_option_override_hook.
16523 * opts-global.c (lang_handle_option): Add new option.
16524 (set_default_handlers): Add new argument.
16525 (decode_options): Likewise.
16526 * opts.c (target_handle_option): Likewise.
16527 (common_handle_option): Call target_option_override_hook.
16528 * opts.h (struct cl_option_handler_func): Add hook for
16529 target option override.
16530 (struct cl_option_handlers): Likewise.
16531 (set_default_handlers): Add new argument.
16532 (decode_options): Likewise.
16533 (common_handle_option): Likewise.
16534 (target_handle_option): Likewise.
16535 * toplev.c (toplev::main): Pass targetm.target_option.override
16538 2017-08-29 Richard Biener <rguenther@suse.de>
16539 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
16541 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
16542 life to the active subtree.
16544 2017-08-28 Jeff Law <law@redhat.com>
16546 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
16547 derive_equivalences.
16548 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
16549 Code moved into....
16550 (edge_info::derive_equivalences): New private member function
16552 * tree-ssa-dom.c (class edge_info): Changed from a struct
16553 to a class. Add ctor/dtor, methods and data members.
16554 (edge_info::edge_info): Renamed from allocate_edge_info.
16555 Initialize additional members.
16556 (edge_info::~edge_info): New.
16557 (free_dom_edge_info): Delete the edge info.
16558 (record_edge_info): Use new class & associated member functions.
16559 Tighten forms for testing for edge equivalences.
16560 (record_temporary_equivalences): Iterate over the simple
16561 equivalences rather than assuming there's only one per edge.
16562 (cprop_into_successor_phis): Iterate over the simple
16563 equivalences rather than assuming there's only one per edge.
16564 (optimize_stmt): Use operand_equal_p rather than pointer
16565 equality for mini-DSE code.
16567 2017-08-28 Nathan Sidwell <nathan@acm.org>
16569 * gcc.c (execute): Fold SIGPIPE handling into switch
16570 statement. Adjust internal error message.
16572 2017-08-28 Richard Biener <rguenther@suse.de>
16575 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
16576 Do nothing for removed DIEs.
16578 2017-08-28 Richard Biener <rguenther@suse.de>
16580 PR tree-optimization/81977
16581 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
16584 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
16587 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
16588 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
16589 using targetm.gen_mem_thread_fence.
16591 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
16594 * config/i386/i386.md (*<btsc><mode>): Change operand 2
16595 predicate to register_operand. Reorder operands.
16596 (*btr<mode>): Ditto.
16597 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
16598 (*btr<mode>_mask): Ditto.
16600 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
16602 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
16603 * config/rs6000/xmmintrin.h: New file.
16604 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
16606 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16609 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
16610 parameter and_insn and return it.
16611 (recombine_lvx_pattern): Insert a copy to ensure availability of
16612 the base register of the copied masking operation at the point of
16613 the instruction replacement.
16614 (recombine_stvx_pattern): Likewise.
16616 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
16618 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
16619 undocumented switches.
16620 (-mpower9-dform-vector): Likewise.
16621 (-mpower9-dform): Likewise.
16622 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
16623 comments to delete references to -mpower9-dform* switches.
16624 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
16625 Delete reference to -mpower9-dform* switches, test for
16626 -mpower9-vector instead.
16627 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
16628 (OTHER_P9_VECTOR_MASKS): Likewise.
16629 (POWERPC_MASKS): Likewise.
16630 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
16631 tests against -mpower9-dform* to -mpower9-vector. Delete code
16632 that checked for -mpower9-dform* consistancy with other options.
16633 Add test for -mpower9-misc to enable other power9 switches.
16634 (rs6000_init_hard_regno_mode_ok): Likewise.
16635 (rs6000_option_override_internal): Likewise.
16636 (rs6000_emit_prologue): Likewise.
16637 (rs6000_emit_epilogue): Likewise.
16638 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
16639 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
16640 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
16642 (emit_fusion_p9_store): Likewise.
16643 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
16644 resetting these macros if the assembler does not support ISA 3.0
16646 (TARGET_P9_DFORM_VECTOR): Likewise.
16647 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
16648 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
16650 2017-08-25 Alan Modra <amodra@gmail.com>
16652 PR rtl-optimization/81747
16653 * cse.c (cse_extended_basic_block): Don't attempt to record
16654 equivalences for degenerate conditional branches that branch
16655 to their fall-through.
16657 2017-08-24 Martin Sebor <msebor@redhat.com>
16659 PR middle-end/81908
16660 * gimple-fold.c (size_must_be_zero_p): New function.
16661 (gimple_fold_builtin_memory_op): Call it.
16663 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
16665 * config/rs6000/mm_malloc.h: New file.
16667 2017-08-24 Bin Cheng <bin.cheng@arm.com>
16669 PR tree-optimization/81913
16670 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
16671 analysis when either IVs in condition can wrap.
16673 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
16675 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
16676 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
16678 2017-08-24 Richard Biener <rguenther@suse.de>
16681 * targhooks.c (default_target_can_inline_p): Properly
16682 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
16683 is present and always compare.
16684 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
16685 infer -mfpmath=sse from TARGET_SSE_P.
16686 (ix86_can_inline_p): Properly use target_option_default_node when
16687 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
16689 2017-08-24 Richard Biener <rguenther@suse.de>
16692 * dwarf2out.c (output_die): Handle flag_generate_offload like
16694 (output_comp_unit): Likewise.
16695 (gen_array_type_die): Likewise.
16696 (dwarf2out_early_finish): Likewise.
16697 (note_variable_value_in_expr): Likewise.
16698 (dwarf2out_finish): Likewise. Adjust assert.
16699 * cgraphunit.c (symbol_table::compile): Move setting of
16700 flag_generate_offload earlier ...
16701 (symbol_table::finalize_compilation_unit): ... here, before
16702 early debug finalization.
16704 2017-08-24 Richard Biener <rguenther@suse.de>
16706 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
16707 and ipa-fnsummary.h.
16708 (ix86_can_inline_p): When ix86_fpmath flags do not match
16709 check whether the callee uses FP math at all.
16711 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
16713 PR middle-end/81931
16714 * tree-ssanames.c (get_nonzero_bits): Use element_precision
16715 instead of TYPE_PRECISION.
16717 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
16718 Alan Hayward <alan.hayward@arm.com>
16719 David Sherwood <david.sherwood@arm.com>
16721 * combine.c (make_extraction): Use subreg_offset_from_lsb.
16723 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
16725 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
16727 (ix86_frame::stack_realign_allocate): New field.
16728 (struct machine_frame_state): Modify comments.
16729 (machine_frame_state::sp_realigned_fp_end): New field.
16730 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
16731 layout calculation.
16732 (sp_valid_at): Add assertion to assure no attempt to access invalid
16733 offset of a realigned stack.
16734 (fp_valid_at): Likewise.
16735 (choose_baseaddr): Modify comments.
16736 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
16737 ix86_expand_prologue.
16738 (ix86_expand_prologue): Modify stack realignment and allocation.
16739 (ix86_expand_epilogue): Modify comments.
16740 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
16741 avx2_runtime, avx512f, and avx512f_runtime.
16743 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
16745 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
16746 (mstackrealign): Do not init to -1.
16747 * config/i386/i386.c (ix86_option_override_internal):
16748 Check opts_set, not opts when setting default value of
16749 opts->x_ix86_force_align_arg_pointer.
16751 2017-08-23 Richard Biener <rguenther@suse.de>
16753 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
16754 lang_hooks.decl_printable_name.
16755 * print-rtl-function.c (print_rtx_function): Likewise.
16756 * tree-pretty-print.c (dump_function_header): Likewise.
16758 2017-08-23 Richard Biener <rguenther@suse.de>
16761 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
16762 -g0 at compile-time.
16764 2017-08-23 Tamar Christina <tamar.christina@arm.com>
16766 PR middle-end/19706
16767 * doc/sourcebuild.texi (Other hardware attributes):
16770 2017-08-23 Tamar Christina <tamar.christina@arm.com>
16772 PR middle-end/19706
16773 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
16774 Fix single-use check.
16776 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
16778 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
16780 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
16782 * doc/install.texi: Modify to add more details on running selected
16785 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
16787 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
16788 is combined with -mabi=ms.
16789 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
16792 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16794 PR tree-optimization/81488
16795 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
16796 and cached_basis fields.
16797 (MAX_SPREAD): New constant.
16798 (alloc_cand_and_find_basis): Initialize new fields.
16799 (clear_visited): New function.
16800 (create_phi_basis_1): Rename from create_phi_basis, set visited
16801 and cached_basis fields.
16802 (create_phi_basis): New wrapper function.
16803 (phi_add_costs_1): Rename from phi_add_costs, add spread
16804 parameter, set visited field, short-circuit when limits reached.
16805 (phi_add_costs): New wrapper function.
16806 (record_phi_increments_1): Rename from record_phi_increments, set
16808 (record_phi_increments): New wrapper function.
16809 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
16810 (phi_incr_cost): New wrapper function.
16811 (all_phi_incrs_profitable_1): Rename from
16812 all_phi_incrs_profitable, set visited field.
16813 (all_phi_incrs_profitable): New wrapper function.
16815 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
16816 Alan Hayward <alan.hayward@arm.com>
16817 David Sherwood <david.sherwood@arm.com>
16819 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
16820 that takes the outer and inner modes.
16821 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
16822 comparison as the canonical test for a paradoxical subreg.
16823 * combine.c (simplify_set): Use paradoxical_subreg_p.
16824 (make_extraction): Likewise.
16825 (force_to_mode): Likewise.
16826 (rtx_equal_for_field_assignment_p): Likewise.
16827 (gen_lowpart_for_combine): Likewise.
16828 (simplify_comparison): Likewise.
16829 * cse.c (equiv_constant): Likewise.
16830 * expmed.c (store_bit_field_1): Likewise.
16831 * final.c (alter_subreg): Likewise.
16832 * fwprop.c (propagate_rtx): Likewise.
16833 (forward_propagate_subreg): Likewise.
16834 * ira-conflicts.c (ira_build_conflicts): Likewise.
16835 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
16836 * lra-constraints.c (curr_insn_transform): Likewise.
16837 (split_reg): Likewise.
16838 * lra-eliminations.c (move_plus_up): Likewise.
16839 (lra_eliminate_regs_1): Likewise.
16840 * recog.c (general_operand): Likewise.
16841 * ree.c (combine_reaching_defs): Likewise.
16842 * reload.c (push_reload): Likewise.
16843 (find_reloads): Likewise.
16844 * reload1.c (elimination_effects): Likewise.
16845 (compute_reload_subreg_offset): Likewise.
16846 (choose_reload_regs): Likewise.
16847 * rtlanal.c (subreg_lsb_1): Likewise.
16848 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16849 (simplify_subreg): Likewise.
16850 * var-tracking.c (track_loc_p): Likewise.
16851 * emit-rtl.c (byte_lowpart_offset): Likewise.
16852 (paradoxical_subreg_p): Delete out-of-line definition.
16854 2017-08-22 Jeff Law <law@redhat.com>
16856 PR tree-optimization/81741
16857 PR tree-optimization/71947
16858 * tree-ssa-dom.c: Include tree-inline.h.
16859 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
16860 equivalences if one is more expensive to compute than the other.
16861 * tree-ssa-scopedtables.h (class const_or_copies): Make
16862 record_const_or_copy_raw method private.
16863 (class avail_exprs_stack): New method simplify_binary_operation.
16864 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
16865 avail_exprs_stack::simplify_binary_operation as needed.
16866 (avail_exprs_stack::simplify_binary_operation): New function.
16868 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
16870 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
16871 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
16872 (DOT_SYMBOLS): Likewise.
16873 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
16874 (RELOCATABLE_NEEDS_FIXUP): Likewise.
16875 (RS6000_ABI_NAME): Likewise.
16876 (TARGET_CMODEL): Likewise.
16877 (TOC_SECTION_ASM_OP): Likewise.
16878 (SET_CMODEL): New macro.
16879 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
16881 2017-08-22 Richard Biener <rguenther@suse.de>
16883 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
16884 to allow for free-lang-data replacements similar to verify_type_variant.
16886 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
16887 Alan Hayward <alan.hayward@arm.com>
16888 David Sherwood <david.sherwood@arm.com>
16890 * config/aarch64/aarch64.md (casesi): Use DImode rather than
16891 VOIDmode for the LABEL_REF.
16893 2017-08-22 Richard Biener <rguenther@suse.de>
16895 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
16897 2017-08-22 Richard Biener <rguenther@suse.de>
16899 * common.opt (feliminate-dwarf2-dups): Ignore.
16900 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
16901 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
16902 same_die_p_wrap, compute_section_prefix,
16903 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
16904 (comdat_symbol_id, comdat_symbol_number): Likewise.
16905 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
16907 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
16908 (output_die): Mark unreachable path unreachable.
16909 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
16910 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
16911 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
16912 (dwarf2out_early_finish): Likewise.
16914 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
16916 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
16918 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
16921 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
16922 not VAR_P. Filter attribute warnings with OPT_Wattributes.
16923 (avr_attribute_table) <io, io_low, address>: Initialize
16924 .decl_required with true.
16926 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
16928 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
16929 undocumented debugging options.
16930 (-mvsx-scalar-double): Likewise.
16931 (-mallow-df-permute): Likewise.
16932 (-mvectorize-builtins): Likewise.
16933 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16934 (rs6000_builtin_vectorized_function): Likewise.
16935 (rs6000_builtin_md_vectorized_function): Likewise.
16936 (rs6000_opt_vars): Likewise.
16938 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
16941 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
16942 (*btrq_imm): Rename from *btrq.
16943 (*btcq_imm): Rename from *btcq.
16944 (btsc): New code attribute.
16945 (*<btsc><mode>): New insn pattern.
16946 (*btr<mode>): Ditto.
16947 (*<btsc><mode>_mask): New insn_and_split pattern.
16948 (*btr<mode>_mask): Ditto.
16950 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16951 Alan Hayward <alan.hayward@arm.com>
16952 David Sherwood <david.sherwood@arm.com>
16954 * function.c (pad_below): Simplify padding calculation.
16956 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16957 Alan Hayward <alan.hayward@arm.com>
16958 David Sherwood <david.sherwood@arm.com>
16960 * target.def (function_prologue): Remove frame size argument.
16961 (function_epilogue): Likewise.
16962 * doc/tm.texi: Regenerate.
16963 * final.c (final_start_function): Update call to function_prologue.
16964 (final_end_function): Update call to function_epilogue.
16965 (default_function_pro_epilogue): Remove frame size argument.
16966 * output.h (default_function_pro_epilogue): Likewise.
16967 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
16968 (arm_output_function_prologue): Likewise.
16969 * config/frv/frv.c (frv_function_prologue): Likewise.
16970 (frv_function_epilogue): Likewise.
16971 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
16972 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
16973 (ia64_output_function_epilogue): Likewise.
16974 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
16975 (m32r_output_function_epilogue): Likewise.
16976 * config/microblaze/microblaze.c (microblaze_function_prologue)
16977 (microblaze_function_epilogue): Likewise.
16978 * config/mips/mips.c (mips_output_function_prologue): Likewise.
16979 (mips_output_function_epilogue): Likewise.
16980 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
16981 (mmix_target_asm_function_epilogue): Likewise.
16982 * config/msp430/msp430.c (msp430_start_function): Likewise.
16983 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
16984 (nds32_asm_function_epilogue): Likewise.
16985 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
16986 * config/pa/pa.c (pa_output_function_prologue): Likewise.
16987 (pa_output_function_epilogue): Likewise.
16988 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
16989 (rs6000_output_function_epilogue): Likewise.
16990 * config/rl78/rl78.c (rl78_start_function): Likewise.
16991 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
16992 (rs6000_output_function_epilogue): Likewise.
16993 * config/rx/rx.c (rx_output_function_prologue): Likewise.
16994 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
16995 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
16996 (sparc_asm_function_epilogue): Likewise.
16998 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17000 * tree.h (type_has_mode_precision_p): New function.
17001 * convert.c (convert_to_integer_1): Use it.
17002 * expr.c (expand_expr_real_2): Likewise.
17003 (expand_expr_real_1): Likewise.
17004 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
17005 * match.pd: Likewise.
17006 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
17007 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
17008 * tree-tailcall.c (process_assignment): Likewise.
17009 * tree-vect-loop.c (vectorizable_reduction): Likewise.
17010 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
17011 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
17012 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17013 (vectorizable_assignment): Likewise.
17014 (vectorizable_shift): Likewise.
17015 (vectorizable_operation): Likewise.
17016 * tree-vrp.c (register_edge_assert_for_2): Likewise.
17018 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
17020 * match.pd: Add pow (C, x) simplification.
17022 2017-08-21 Richard Biener <rguenther@suse.de>
17024 PR tree-optimization/81900
17025 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
17026 for blocks with abnormal predecessors.
17027 (compute_antic): Do not set visited flag prematurely.
17029 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
17032 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
17034 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17036 * stor-layout.h (vector_type_mode): Move to...
17037 * tree.h (vector_type_mode): ...here.
17038 * stor-layout.c (vector_type_mode): Move to...
17039 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
17041 2017-08-21 Richard Biener <rguenther@suse.de>
17043 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
17044 register_external_die hooks.
17045 (debug_false_tree_charstarstar_uhwistar): Declare.
17046 (debug_nothing_tree_charstar_uhwi): Likewise.
17047 * debug.c (do_nothing_debug_hooks): Adjust.
17048 (debug_false_tree_charstarstar_uhwistar): New do nothing.
17049 (debug_nothing_tree_charstar_uhwi): Likewise.
17050 * dbxout.c (dbx_debug_hooks): Adjust.
17051 (xcoff_debug_hooks): Likewise.
17052 * sdbout.c (sdb_debug_hooks): Likewise.
17053 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
17054 * dwarf2out.c (macinfo_label_base): New global.
17055 (dwarf2out_register_external_die): New function for the
17056 register_external_die hook.
17057 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
17058 (dwarf2_debug_hooks): Use them.
17059 (dwarf2_lineno_debug_hooks): Adjust.
17060 (struct die_struct): Add with_offset flag.
17061 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
17062 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
17063 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
17064 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
17065 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
17066 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
17067 defining section names for the early LTO debug variants.
17068 (reset_indirect_string): New helper.
17069 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
17070 (print_dw_val): Add support for offsetted symbol references.
17071 (get_ultimate_context): Split out from is_cxx.
17072 (is_cxx): Use get_ultimate_context.
17073 (is_fortran): Add decl overload.
17074 (compute_comp_unit_symbol): Split out worker from
17075 compute_section_prefix.
17076 (compute_section_prefix): Call compute_comp_unit_symbol and
17077 set comdat_type_p here.
17078 (output_die): Skip DIE symbol output for the LTO added one.
17079 Handle DIE symbol references with offset.
17080 (output_comp_unit): Guard section name mangling properly.
17081 For LTO debug sections emit a symbol at the section beginning
17082 which we use to refer to its DIEs.
17083 (add_abstract_origin_attribute): For DIEs registered via
17084 dwarf2out_register_external_die directly refer to the early
17085 DIE rather than indirectly through the shadow one we created.
17086 Remove obsolete call to dwarf2out_abstract_function for
17087 non-function/block origins.
17088 (gen_array_type_die): When generating early LTO debug do
17089 not emit DW_AT_string_length.
17090 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
17091 late when in LTO. As suggested place a gcc_unreachable for
17092 the DECL_ABSTRACT_P case.
17093 (gen_subprogram_die): Avoid another specification DIE
17094 for early built declarations/definitions for the late LTO case.
17095 (gen_variable_die): Add type references for late duplicated VLA dies
17097 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
17098 we have the abstract instance already.
17099 (process_scope_var): Adjust decl DIE contexts in LTO which
17100 first puts them in limbo.
17101 (gen_decl_die): Do not generate type DIEs late apart from
17102 types for VLAs or for decls we do not yet have a DIE. Do not
17103 call dwarf2out_abstract_function late.
17104 (dwarf2out_early_global_decl): Make sure to create DIEs
17105 for abstract instances of a decl first.
17106 (dwarf2out_late_global_decl): Adjust comment.
17107 (output_macinfo_op): With multiple macro sections use
17108 macinfo_label_base to distinguish labels.
17109 (output_macinfo): Likewise. Update macinfo_label_base.
17110 Pass in the line info label.
17111 (note_variable_value_in_expr): When generating LTO resolve
17112 all variable values here by generating DIEs as needed.
17113 (init_sections_and_labels): Add early LTO debug flag parameter
17114 and generate different sections and names if set. Add generation
17115 counter for the labels so we can have multiple of them.
17116 (reset_dies): Helper to allow DIEs to be output multiple times.
17117 (dwarf2out_finish): When outputting DIEs to the fat part of an
17118 LTO object first reset DIEs.
17119 (dwarf2out_early_finish): Output early DIEs when generating LTO.
17120 (modified_type_die): Check for decl_ultimate_origin being self
17122 (gen_type_die_with_usage): Likewise.
17123 (gen_typedef_die): Allow decl_ultimate_origin being self.
17124 (set_decl_abstract_flags): Remove.
17125 (set_block_abstract_flags): Likewise.
17126 (dwarf2out_abstract_function): Treat the early generated DIEs
17127 as the abstract copy and only add DW_AT_inline and
17128 DW_AT_artificial here and call set_decl_origin_self.
17129 If the DIE has an abstract origin don't do anything.
17130 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
17131 if we have none yet (Go fails to build one, PR78628).
17132 (variably_modified_type_p): Prevent endless recursion for Ada
17133 cyclic pointer types.
17134 * lto-streamer-in.c: Include debug.h.
17135 (dref_queue): New global.
17136 (lto_read_tree_1): Stream in DIE references.
17137 (lto_input_tree): Register DIE references.
17138 (input_function): Stream DECL_DEBUG_ARGS.
17139 * lto-streamer-out.c: Include debug.h.
17140 (lto_write_tree_1): Output DIE references.
17141 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
17142 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
17143 (output_function): Stream DECL_DEBUG_ARGS.
17144 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
17145 Stream DECL_ABSTRACT_ORIGIN.
17146 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
17147 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
17148 DECL_CONTEXT for file-scope decls.
17149 * lto-streamer.h (struct dref_entry): Declare.
17150 (dref_queue): Likewise.
17151 * cfgexpand.c (pass_expand::execute): Do not call the
17152 outlining_inline_function hook here.
17153 * lto-wrapper.c (debug_obj): New global.
17154 (tool_cleanup): Unlink it if required.
17155 (debug_objcopy): New function.
17156 (run_gcc): Handle early debug sections in the IL files by
17157 extracting them to separate files, partially linkin them and
17158 feeding the result back as result to the linker.
17159 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
17160 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
17161 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
17162 sections into a separate segment.
17163 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
17165 (darwin_asm_dwarf_section): Likewise.
17166 (darwin_asm_output_dwarf_offset): Likewise.
17167 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
17169 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17170 Alan Hayward <alan.hayward@arm.com>
17171 David Sherwood <david.sherwood@arm.com>
17173 * read-md.h (md_reader::record_potential_iterator_use): Replace
17174 pointer argument with an rtx and an index.
17175 * read-rtl.c (iterator_group::apply_iterator): Likewise.
17176 (apply_mode_iterator): Likewise.
17177 (apply_code_iterator): Likewise.
17178 (apply_int_iterator): Likewise.
17179 (apply_subst_iterator): Likewise.
17180 (record_iterator_use): Likewise.
17181 (record_attribute_use): Likewise.
17182 (md_reader::record_potential_iterator_use): Likewise. Update calls
17183 to record_iterator_use and apply_iterator.
17184 (iterator_use): Replace ptr with x and index.
17185 (attribute_use): Likewise.
17186 (apply_attribute_uses): Update calls to apply_iterator.
17187 (apply_iterators): Likewise. Update initialization of iterator_use.
17188 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
17189 and record_potential_iterator_use.
17190 (rtx_reader::read_rtx_operand): Likewise.
17192 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17193 Alan Hayward <alan.hayward@arm.com>
17194 David Sherwood <david.sherwood@arm.com>
17196 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
17199 2017-08-21 Richard Biener <rguenther@suse.de>
17201 PR middle-end/81884
17202 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
17203 at struct end conservatively when comparing common bases.
17205 2017-08-21 Richard Biener <rguenther@suse.de>
17207 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
17208 (mem_ref_in_stmt): Remove.
17209 (determine_max_movement): Use ref index to get at the reference.
17210 (invariantness_dom_walker::before_dom_children): Deal with
17211 lim data already initialized.
17212 (gather_mem_refs_stmt): Initialize lim data and record ref index.
17214 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
17216 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
17217 (TARGET_ISA_ROUND): Ditto.
17218 (TARGET_ROUND): Ditto.
17219 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
17220 * config/i386/i386.md: Ditto.
17221 * config/i386/sse.md: Ditto.
17222 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
17223 with OPTION_MASK_ISA_SSE4_1.
17225 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
17228 * doc/extend.texi (x86 Built-in Functions): Correct the name of
17229 __builtin_ia32_lzcnt_u16.
17231 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
17234 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
17235 (rs6000_set_current_function): Rewrite function to use it.
17237 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
17240 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
17241 and TYPE_WARN_IF_NOT_ALIGN.
17242 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
17243 (handle_warn_if_not_align): New.
17244 (place_union_field): Call handle_warn_if_not_align.
17245 (place_field): Call handle_warn_if_not_align.
17246 Copy TYPE_WARN_IF_NOT_ALIGN.
17247 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
17248 (layout_type): Likewise.
17249 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
17251 (tree_decl_common): Add warn_if_not_align.
17252 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
17253 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
17254 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
17255 (DECL_WARN_IF_NOT_ALIGN): Likewise.
17256 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
17257 * doc/extend.texi: Document warn_if_not_aligned attribute.
17258 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
17260 2017-08-17 Martin Liska <mliska@suse.cz>
17263 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
17264 (get_data_dependence): Use it as pointer type.
17265 (distribute_loop): Likewise.
17267 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17269 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
17270 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
17271 (p8_vmrgow_<mode>_direct): New define_insn.
17272 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
17273 handle endianness for vmrgew and vmrgow permute patterns.
17275 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
17277 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
17278 * config/rs6000/rs6000-cpus.def: Remove comment.
17279 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
17280 (POWERPC_MASKS): Likewise.
17281 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
17282 use of TARGET_VSX_TIMODE.
17283 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
17284 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
17285 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
17286 (rs6000_option_override_internal): Remove dead code.
17287 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
17288 (rs6000_legitimize_reload_address): Likewise.
17289 (rs6000_legitimate_address_p): Likewise.
17290 (rs6000_opt_masks): Delete "vsx-timode".
17291 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
17292 from function comment.
17293 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
17294 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
17295 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
17297 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
17298 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
17299 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
17300 (VSX_TI): Likewise.
17302 (define_peephole2): Likewise.
17304 2017-08-17 Martin Sebor <msebor@redhat.com>
17307 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
17308 past the end of an array.
17309 (test_pp_format): Add test cases.
17311 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
17313 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
17314 missing ECF_NOTHROW flags.
17316 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
17319 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
17320 operands residing in integer registers.
17321 (*vsx_le_perm_load_<mode>): Likewise.
17322 (*vsx_le_perm_store_<mode>): Likewise.
17323 (define_peephole2): Add peepholes to optimize the above.
17325 2017-08-17 Marek Polacek <polacek@redhat.com>
17327 PR middle-end/81814
17328 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
17329 to mimic what shorten_compare did. Change the return type to bool.
17330 (fold_cond_expr_with_comparison): Update call to
17331 operand_equal_for_comparison_p.
17332 (fold_ternary_loc): Likewise.
17334 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
17336 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
17338 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
17340 2017-08-17 Richard Biener <rguenther@suse.de>
17342 * tree-ssa-structalias.c (solve_graph): When propagating
17343 to successors update the graphs succ edges and avoid duplicate work.
17345 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
17348 * config/i386/i386.c (ix86_option_override_internal): Save target
17349 specific options after ix86_stack_protector_guard_reg was changed.
17351 2017-08-17 Richard Biener <rguenther@suse.de>
17353 PR tree-optimization/81827
17354 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
17355 (new_var_info): Initialize it conservatively.
17356 (get_call_vi): Mark register vars.
17357 (new_scalar_tmp_constraint_exp): Likewise.
17358 (handle_rhs_call): Likewise.
17359 (handle_const_call): Likewise.
17360 (create_function_info_for): Likewise.
17361 (solve_constraints): Sort varinfos to separate register from
17362 non-register vars to pack points-to solution bitmaps during
17365 2017-08-17 Marek Polacek <polacek@redhat.com>
17367 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
17369 2017-08-17 Richard Biener <rguenther@suse.de>
17371 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
17372 to true when overflow is undefined and we saturated the result.
17374 2017-08-17 Alan Modra <amodra@gmail.com>
17377 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
17378 Don't use store multiple if only one reg needs saving.
17379 (interesting_frame_related_regno): New function.
17380 (rs6000_frame_related): Don't emit frame info for regs that
17382 (rs6000_emit_epilogue): Likewise.
17384 2017-08-16 Nathan Sidwell <nathan@acm.org>
17386 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
17387 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
17388 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
17389 * tree.c (free_lang_data_in_type): Use else-if chain. Always
17390 clear TYPE_LANG_1. Remove obsolete member-function stripping.
17391 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
17392 (verify_type): Adjust for TYPE_BINFO move.
17393 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
17394 process TYPE_BINFO directly.
17395 (hash_tree): Likewise.
17396 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17398 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17401 2017-08-16 David Malcolm <dmalcolm@redhat.com>
17403 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
17405 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
17408 * config/i386/i386.md (*anddi_1_btr): Change predicates of
17409 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
17410 Add ix86_binary_operator_ok to insn constraint.
17411 (*iordi_1_bts): Ditto.
17412 (*xordi_1_btc): Ditto.
17413 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
17414 Update corresponding peephole2 pattern.
17418 2017-08-16 Bin Cheng <bin.cheng@arm.com>
17420 PR tree-optimization/81832
17421 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
17422 copy loop header which has IFN_LOOP_DIST_ALIAS call.
17424 2017-08-16 Marek Polacek <polacek@redhat.com>
17427 * fold-const.c (fold_indirect_ref_1): Restore original behavior
17428 regarding size_zero_node.
17430 2017-08-16 Martin Liska <mliska@suse.cz>
17433 * config.gcc: Respect previously set extra_objs in case
17436 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
17438 PR tree-optimization/81835
17439 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
17440 the phi SSA_NAME. Check that the condition in a COND_EXPR does
17441 not depend on the phi.
17443 2017-08-16 Alan Modra <amodra@gmail.com>
17445 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
17448 2017-08-16 Alan Modra <amodra@gmail.com>
17450 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
17451 (save_reg_p): ..into this. Update all callers.
17452 (first_reg_to_save): Simplify.
17454 2017-08-16 Alan Modra <amodra@gmail.com>
17456 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
17459 2017-08-15 Joseph Myers <joseph@codesourcery.com>
17463 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
17464 constant count if that count is less than 32.
17466 2017-08-15 Nathan Sidwell <nathan@acm.org>
17468 * gcc.c (execute): Emit friendlier message if inferior is killed
17469 by an external cause.
17471 2017-08-15 Richard Biener <rguenther@suse.de>
17473 PR tree-optimization/81790
17474 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
17475 CONSTRUCTORs from simplifying and VN.
17477 2017-08-14 Martin Sebor <msebor@redhat.com>
17479 * builtin-attrs.def: Add comments.
17481 2017-08-14 Martin Sebor <msebor@redhat.com>
17484 * doc/extend.texi (attribute nonstring): Document new attribute.
17486 2017-08-14 Martin Sebor <msebor@redhat.com>
17489 * tree-diagnostic.c (default_tree_printer): Handle %G.
17490 * gimple-pretty-print.h (percent_G_format): Declare new function.
17491 * gimple-pretty-print.c (percent_G_format): Define.
17492 * tree-pretty-print.c (percent_K_format): Add argument.
17494 2017-08-14 Martin Sebor <msebor@redhat.com>
17496 PR translation/79998
17497 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
17498 Remove a stray space.
17500 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
17503 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
17504 (*iordi_1_bts): Ditto.
17505 (*xordi_1_btc): Ditto.
17507 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17510 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
17511 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17513 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
17514 quoted strings, and make more translator-friendly.
17515 (darwin_rs6000_override_options): Likewise.
17516 (rs6000_option_override_internal): Likewise.
17517 (rs6000_return_in_memory): Fix overlong line.
17518 (init_cmulative_args): Use quoted strings, and make more
17519 translator-friendly.
17520 (rs6000_pass_by_reference): Fix overlong line.
17521 (def_builtin): Use quoted strings.
17522 (altivec_expand_predicate_builtin): Use quoted strings, and make
17523 more translator-friendly.
17524 (htm_expand_builtin): Use quoted strings.
17525 (cpu_expand_builtin): Use quoted strings, and make more
17526 translator-friendly.
17527 (altivec_expand_builtin): Likewise.
17528 (paired_expand_predicate_builtin): Likewise.
17529 (rs6000_invalid_builtin): Likewise.
17530 (builtin_function_type): Use quoted strings.
17531 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
17532 more translator-friendly.
17533 (rs6000_trampoline_init): Likewise.
17534 (rs6000_handle_altivec_attribute): Likewise.
17535 (rs6000_inner_target_options): Use quoted strings.
17536 (rs6000_disable_incompatible_switches): Likewise.
17537 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
17538 strings, and make more translator-friendly.
17539 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
17541 2017-08-14 Bin Cheng <bin.cheng@arm.com>
17543 PR tree-optimization/81799
17544 * tree-loop-distribution.c (version_loop_by_alias_check): Force
17545 cond_expr to simple gimple operand.
17547 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
17549 PR middle-end/46932
17550 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
17552 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
17556 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
17557 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
17558 TARGET_GASISR_PROLOGUES.
17559 * config/avr/avr.c (avr_option_override): Same.
17560 (avr_pass_pre_proep::execute): Same.
17562 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
17565 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
17566 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
17568 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
17570 * config/i386/i386.md (*load_tp_<mode>): Redefine as
17571 define_insn_and_split. Split to a memory load from 0 in
17572 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
17573 using PTR mode iterator.
17574 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
17575 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
17576 (*add_tp_<mode>): Redefine as define_insn_and_split.
17577 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
17578 address space. Merge with *add_tp_x32 using PTR mode iterator.
17579 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
17580 Split to an add with a memory load from 0 in
17581 DEFAULT_TLS_SEG_REG address space.
17583 2017-08-12 Andrew Pinski <apinski@cavium.com>
17585 * config/aarch64/aarch64-option-extensions.def (rdma):
17586 Fix feature string to what Linux prints out in /proc/cpuinfo.
17588 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
17591 * dwarf2out.c (modified_type_die): For C typedef types that have
17592 an ultimate origin, process the ultimate origin instead of the
17594 (gen_typedef_die): Assert that input DECLs have no ultimate
17596 (gen_type_die_with_usage): For typedef variants that have an
17597 ultimate origin, just call gen_decl_die on the original DECL.
17598 (process_scope_var): Avoid creating DIEs for local typedefs and
17599 concrete static variables.
17601 2017-08-12 Alan Modra <amodra@gmail.com>
17605 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
17606 match gnu-user.h startfile.
17607 (ENDFILE_LINUX_SPEC): Similarly.
17609 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
17612 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
17614 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
17616 2017-08-11 Tamar Christina <tamar.christina@arm.com>
17617 * config/aarch64/aarch64.md (mov<mode>): Change.
17618 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
17619 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
17620 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
17622 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
17624 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
17625 for storage order barriers.
17627 2017-08-11 Martin Liska <mliska@suse.cz>
17630 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
17631 variables of void type.
17633 2017-08-11 Martin Liska <mliska@suse.cz>
17635 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
17636 TARGET_SUPPORTS_ALIASES.
17637 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
17638 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
17639 (optimize_weakref): Likewise.
17640 * symtab.c (symtab_node::noninterposable_alias): Likewise.
17641 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
17642 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
17644 2017-08-11 Martin Liska <mliska@suse.cz>
17647 * config/i386/i386.c (make_resolver_func): Do complete
17648 refactoring of the function.
17650 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
17653 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
17654 * config/i386/i386.c (ix86_stack_protect_guard): Use
17655 ix86_stack_protect_guard_symbol_str to generate varible declaration.
17656 * doc/invoke.texi (x86 Options): Document
17657 -mstack-protector-guard-symbol= option.
17659 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
17661 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
17662 * config/i386/i386.c (ix86_split_stack_guard): New function.
17663 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
17664 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
17665 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
17666 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
17667 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
17668 (split_stack_space_check): Call ix86_split_stack_guard.
17670 2017-08-10 Martin Sebor <msebor@redhat.com>
17672 * print-tree.c (print_node): Print location using the established
17674 Replace spaces with colons.
17675 (debug_raw, debug): Ditto.
17677 2017-08-10 Martin Sebor <msebor@redhat.com>
17680 * pretty-print.c (pp_format): Correct the handling of %s precision.
17682 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
17685 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
17687 (ix86_finalize_stack_frame_flags): This. Also clear
17688 frame_pointer_needed if -fno-omit-frame-pointer is used without
17690 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
17691 with ix86_finalize_stack_frame_flags.
17692 (ix86_expand_epilogue): Likewise.
17693 (ix86_expand_split_stack_prologue): Likewise.
17694 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
17696 2017-08-10 Martin Liska <mliska@suse.cz>
17699 * c-attribs.c (handle_target_attribute):
17700 Report warning for an empty string argument of target attribute.
17702 2017-08-09 Jakub Jelinek <jakub@redhat.com>
17705 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
17707 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
17708 or DECL_NONLOCAL labels.
17709 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
17710 or DECL_NONLOCAL labels here.
17712 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
17714 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
17715 to indicate when early gimple folding has been disabled.
17716 (rs6000_gimple_fold_builtin): Add debug content.
17717 (rs6000_invalid_builtin): Fix whitespace.
17718 (rs6000_expand_builtin): Fix whitespace.
17719 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
17721 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
17724 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
17725 SAVE_MULTIPLE if not all the registers that saves, should be saved.
17727 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
17729 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
17730 (qdf24xx): Likewise.
17731 * config/aarch64/aarch64.md: Include falkor.md.
17732 * config/aarch64/falkor.md: New.
17734 2017-08-09 Marek Polacek <polacek@redhat.com>
17737 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
17738 * diagnostic.c (emit_diagnostic): Add a comment.
17739 (emit_diagnostic_valist): New function.
17741 2017-08-09 Marek Polacek <polacek@redhat.com>
17744 * input.c (make_location): New overload.
17745 * input.h (make_location): Declare.
17747 2017-08-08 Alan Modra <amodra@gmail.com>
17748 H.J. Lu <hongjiu.lu@intel.com>
17751 * gcc.c (NO_PIE_SPEC): Delete.
17752 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
17754 (LINK_PIE_SPEC): ..to here.
17755 (LINK_COMMAND_SPEC): Support -no-pie.
17756 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
17757 chain of crtbegin*.o selection, update for PIE_SPEC changes and
17759 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
17760 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
17761 (ENDFILE_CRTEND_SPEC): Similarly.
17763 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
17766 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
17767 (mstack-protector-guard-offset=): Ditto.
17768 * config/i386/i386.c (ix86_option_override): Handle
17769 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
17771 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
17772 ix86_stack_protect_guard_offset variables.
17773 (TARGET_STACK_PROTECT_GUARD): Always define.
17774 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
17775 and -mstack-protector-guard-offset= options.
17777 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17779 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
17780 boundary case for the last candidate.
17782 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17784 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
17785 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
17787 2017-08-08 Tamar Christina <tamar.christina@arm.com>
17789 PR middle-end/19706
17790 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
17791 * config/aarch64/aarch64-builtins.c
17792 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
17793 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
17794 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
17796 2017-08-08 Tamar Christina <tamar.christina@arm.com>
17797 Andrew Pinski <pinskia@gmail.com>
17799 PR middle-end/19706
17800 * internal-fn.def (XORSIGN): New.
17801 * optabs.def (xorsign_optab): New.
17802 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
17803 (convert_expand_mult_copysign): New.
17804 (pass_optimize_widening_mul::execute): Call
17805 convert_expand_mult_copysign.
17807 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17809 PR tree-optimization/81354
17810 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
17811 Insert on edges rather than explicitly creating landing pads.
17812 (analyze_candidates_and_replace): Commit edge inserts.
17814 2017-08-08 Richard Biener <rguenther@suse.de>
17816 PR middle-end/81719
17817 * tree-ssa-loop-niter.c: Include tree-dfa.h.
17818 (expand_simple_operations): Also look through ADDR_EXPRs with
17819 MEM_REF bases treating them as POINTER_PLUS_EXPR.
17821 2017-08-08 Richard Biener <rguenther@suse.de>
17823 PR tree-optimization/81723
17824 * tree-vect-slp.c (struct bst_traits): New hash traits.
17825 (bst_fail): New global.
17826 (vect_build_slp_tree_2): New worker, split out from ...
17827 (vect_build_slp_tree): ... this now wrapping it with using
17828 bst_fail set to cache SLP tree build fails. Properly handle
17830 (vect_analyze_slp_instance): Allocate and free bst_fail.
17832 2017-08-08 Martin Liska <mliska@suse.cz>
17835 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
17836 LABEL_DECLs that can be from a different function.
17838 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17840 PR tree-optimization/81744
17841 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
17842 loop's number of iterations.
17844 2017-08-08 Martin Liska <mliska@suse.cz>
17846 * asan.c: Include header files.
17847 * attribs.c (build_decl_attribute_variant): New function moved
17849 (build_type_attribute_qual_variant): Likewise.
17850 (cmp_attrib_identifiers): Likewise.
17851 (simple_cst_list_equal): Likewise.
17852 (omp_declare_simd_clauses_equal): Likewise.
17853 (attribute_value_equal): Likewise.
17854 (comp_type_attributes): Likewise.
17855 (build_type_attribute_variant): Likewise.
17856 (lookup_ident_attribute): Likewise.
17857 (remove_attribute): Likewise.
17858 (merge_attributes): Likewise.
17859 (merge_type_attributes): Likewise.
17860 (merge_decl_attributes): Likewise.
17861 (merge_dllimport_decl_attributes): Likewise.
17862 (handle_dll_attribute): Likewise.
17863 (attribute_list_equal): Likewise.
17864 (attribute_list_contained): Likewise.
17865 * attribs.h (lookup_attribute): New function moved from tree.[ch].
17866 (lookup_attribute_by_prefix): Likewise.
17867 * bb-reorder.c: Include header files.
17868 * builtins.c: Likewise.
17869 * calls.c: Likewise.
17870 * cfgexpand.c: Likewise.
17871 * cgraph.c: Likewise.
17872 * cgraphunit.c: Likewise.
17873 * convert.c: Likewise.
17874 * dwarf2out.c: Likewise.
17875 * final.c: Likewise.
17876 * fold-const.c: Likewise.
17877 * function.c: Likewise.
17878 * gimple-expr.c: Likewise.
17879 * gimple-fold.c: Likewise.
17880 * gimple-pretty-print.c: Likewise.
17881 * gimple.c: Likewise.
17882 * gimplify.c: Likewise.
17883 * hsa-common.c: Likewise.
17884 * hsa-gen.c: Likewise.
17885 * internal-fn.c: Likewise.
17886 * ipa-chkp.c: Likewise.
17887 * ipa-cp.c: Likewise.
17888 * ipa-devirt.c: Likewise.
17889 * ipa-fnsummary.c: Likewise.
17890 * ipa-inline.c: Likewise.
17891 * ipa-visibility.c: Likewise.
17893 * lto-cgraph.c: Likewise.
17894 * omp-expand.c: Likewise.
17895 * omp-general.c: Likewise.
17896 * omp-low.c: Likewise.
17897 * omp-offload.c: Likewise.
17898 * omp-simd-clone.c: Likewise.
17899 * opts-global.c: Likewise.
17900 * passes.c: Likewise.
17901 * predict.c: Likewise.
17902 * sancov.c: Likewise.
17903 * sanopt.c: Likewise.
17904 * symtab.c: Likewise.
17905 * toplev.c: Likewise.
17906 * trans-mem.c: Likewise.
17907 * tree-chkp.c: Likewise.
17908 * tree-eh.c: Likewise.
17909 * tree-into-ssa.c: Likewise.
17910 * tree-object-size.c: Likewise.
17911 * tree-parloops.c: Likewise.
17912 * tree-profile.c: Likewise.
17913 * tree-ssa-ccp.c: Likewise.
17914 * tree-ssa-live.c: Likewise.
17915 * tree-ssa-loop.c: Likewise.
17916 * tree-ssa-sccvn.c: Likewise.
17917 * tree-ssa-structalias.c: Likewise.
17918 * tree-ssa.c: Likewise.
17919 * tree-streamer-in.c: Likewise.
17920 * tree-vectorizer.c: Likewise.
17921 * tree-vrp.c: Likewise.
17922 * tsan.c: Likewise.
17923 * ubsan.c: Likewise.
17924 * varasm.c: Likewise.
17925 * varpool.c: Likewise.
17926 * tree.c: Remove functions moved to attribs.[ch].
17927 * tree.h: Likewise.
17928 * config/aarch64/aarch64.c: Add attrs.h header file.
17929 * config/alpha/alpha.c: Likewise.
17930 * config/arc/arc.c: Likewise.
17931 * config/arm/arm.c: Likewise.
17932 * config/avr/avr.c: Likewise.
17933 * config/bfin/bfin.c: Likewise.
17934 * config/c6x/c6x.c: Likewise.
17935 * config/cr16/cr16.c: Likewise.
17936 * config/cris/cris.c: Likewise.
17937 * config/darwin.c: Likewise.
17938 * config/epiphany/epiphany.c: Likewise.
17939 * config/fr30/fr30.c: Likewise.
17940 * config/frv/frv.c: Likewise.
17941 * config/ft32/ft32.c: Likewise.
17942 * config/h8300/h8300.c: Likewise.
17943 * config/i386/winnt.c: Likewise.
17944 * config/ia64/ia64.c: Likewise.
17945 * config/iq2000/iq2000.c: Likewise.
17946 * config/lm32/lm32.c: Likewise.
17947 * config/m32c/m32c.c: Likewise.
17948 * config/m32r/m32r.c: Likewise.
17949 * config/m68k/m68k.c: Likewise.
17950 * config/mcore/mcore.c: Likewise.
17951 * config/microblaze/microblaze.c: Likewise.
17952 * config/mips/mips.c: Likewise.
17953 * config/mmix/mmix.c: Likewise.
17954 * config/mn10300/mn10300.c: Likewise.
17955 * config/moxie/moxie.c: Likewise.
17956 * config/msp430/msp430.c: Likewise.
17957 * config/nds32/nds32-isr.c: Likewise.
17958 * config/nds32/nds32.c: Likewise.
17959 * config/nios2/nios2.c: Likewise.
17960 * config/nvptx/nvptx.c: Likewise.
17961 * config/pa/pa.c: Likewise.
17962 * config/pdp11/pdp11.c: Likewise.
17963 * config/powerpcspe/powerpcspe.c: Likewise.
17964 * config/riscv/riscv.c: Likewise.
17965 * config/rl78/rl78.c: Likewise.
17966 * config/rx/rx.c: Likewise.
17967 * config/s390/s390.c: Likewise.
17968 * config/sh/sh.c: Likewise.
17969 * config/sol2.c: Likewise.
17970 * config/sparc/sparc.c: Likewise.
17971 * config/spu/spu.c: Likewise.
17972 * config/stormy16/stormy16.c: Likewise.
17973 * config/tilegx/tilegx.c: Likewise.
17974 * config/tilepro/tilepro.c: Likewise.
17975 * config/v850/v850.c: Likewise.
17976 * config/vax/vax.c: Likewise.
17977 * config/visium/visium.c: Likewise.
17978 * config/xtensa/xtensa.c: Likewise.
17980 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
17983 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
17984 constraints since the -mupper-regs-* switches have been
17986 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
17987 into a vector from a double word element that was extracted from
17988 another vector, and eliminate extra XXPERMDI instructions.
17989 (vsx_concat_<mode>_2): Likewise.
17990 (vsx_concat_<mode>_3): Likewise.
17991 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
17992 concat to allow optimizing inserts from previous extracts.
17994 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
17996 * config/i386/i386.c (ix86_stack_protect_guard): Generate
17997 memory reference to a SSP offset in TLS address space.
17998 (ix86_print_operand) <case '@'>: Remove.
17999 (ix86_print_operand_punct_valid_p): Remove '@' code.
18000 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
18001 UNSPEC_SP_TLS_TEST.
18002 (stack_tls_protect_set_<mode>): Remove.
18003 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
18004 (stack_tls_protect_test_<mode>): Remove.
18005 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
18007 2017-08-07 Olivier Hainque <hainque@adacore.com>
18010 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
18012 2017-08-07 Douglas Rupp <rupp@adacore.com>
18014 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
18015 variable was referenced as multidir in command.
18017 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18020 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
18023 2017-08-07 Martin Liska <mliska@suse.cz>
18025 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
18026 * config/rl78/rl78.c: Add include of attribs.h.
18027 * config/sh/sh.c: Likewise.
18028 * config/v850/v850.c: Likewise.
18030 2017-08-07 Tom de Vries <tom@codesourcery.com>
18032 PR middle-end/78266
18033 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
18035 2017-08-07 Martin Liska <mliska@suse.cz>
18037 * config/mips/mips.c: Include attribs.h.
18039 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
18042 * doc/invoke.texi: Document change in behvaior for -Ofast for
18045 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
18047 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
18049 (aarch64_pop_regs): Likewise.
18050 (aarch64_gen_load_pair): Likewise.
18051 (aarch64_save_callee_saves): Likewise.
18052 (aarch64_restore_callee_saves): Likewise.
18054 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18056 * config/i386/i386.c: Revert the last change.
18058 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18061 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
18063 (ix86_finalize_stack_frame_flags): This. Also clear
18064 frame_pointer_needed if -fno-omit-frame-pointer is used without
18066 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
18067 with ix86_finalize_stack_frame_flags.
18068 (ix86_expand_epilogue): Likewise.
18069 (ix86_expand_split_stack_prologue): Likewise.
18071 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18074 * config/i386/i386.c (get_builtin_code_for_version): Set priority
18075 to P_AES for Westmere.
18077 2017-08-07 Jonathan Yong <10walls@gmail.com>
18079 * config/i386/mingw.opt (fset-stack-executable): Removed.
18080 * config/i386/cygming.opt (fset-stack-executable): Moved
18082 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
18084 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
18086 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
18088 2017-08-07 Marek Polacek <polacek@redhat.com>
18090 PR middle-end/81737
18091 * fold-const.c (fold_indirect_ref_1): Check type_domain.
18093 2017-08-07 Martin Liska <mliska@suse.cz>
18095 * attribs.h (canonicalize_attr_name): New function.
18096 (cmp_attribs): Move from c-format.c and adjusted.
18097 (is_attribute_p): Moved from tree.h.
18098 * tree-inline.c: Add new includes.
18099 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
18100 (private_is_attribute_p): Remove.
18101 (private_lookup_attribute): Likewise.
18102 (private_lookup_attribute_by_prefix): Simplify.
18103 (remove_attribute): Use is_attribute_p.
18104 * tree.h: Remove removed declarations.
18106 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18108 PR middle-end/81698
18109 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
18110 instead of computing it in the function. Formatting fix.
18111 (expand_case): Don't rely on default_edge being the first edge,
18112 clear it if removing it, pass default_edge to
18113 emit_case_dispatch_table.
18114 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
18117 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
18119 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
18120 insn in the function, emit NOP after the insn.
18122 2017-08-06 Tom de Vries <tom@codesourcery.com>
18124 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
18127 2017-08-06 Tom de Vries <tom@codesourcery.com>
18129 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
18132 2017-08-04 Yury Gribov <tetra2005@gmail.com>
18134 PR tree-optimization/57371
18135 * match.pd: New pattern.
18137 2017-08-04 Marek Polacek <polacek@redhat.com>
18139 PR middle-end/81695
18140 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
18141 perform the computation in offset_int.
18143 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18145 PR tree-optimization/81136
18146 * tree-vectorizer.h: Include tree-hash-traits.h.
18147 (vec_base_alignments): New typedef.
18148 (vec_info): Add a base_alignments field.
18149 (vect_record_base_alignments): Declare.
18150 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
18152 (DR_IS_CONDITIONAL_IN_STMT): New macro.
18153 (create_data_ref): Add an is_conditional_in_stmt argument.
18154 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
18155 the is_conditional_in_stmt field.
18156 (data_ref_loc): Add an is_conditional_in_stmt field.
18157 (get_references_in_stmt): Set the is_conditional_in_stmt field.
18158 (find_data_references_in_stmt): Update call to create_data_ref.
18159 (graphite_find_data_references_in_stmt): Likewise.
18160 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
18161 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
18162 (vect_record_base_alignment): New function.
18163 (vect_record_base_alignments): Likewise.
18164 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
18165 for nested statements even if we fail to compute a misalignment.
18166 Use pooled base alignments for unconditional references.
18167 (vect_find_same_alignment_drs): Compare base addresses instead
18169 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
18170 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
18172 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18174 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
18175 Add an explicit name for the enum. Use auto_vec for slp_instances
18176 and grouped_stores.
18177 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
18179 (_bb_vec_info): Add a constructor and destructor.
18180 (vinfo_for_stmt): Return NULL for uids of -1 as well.
18181 (destroy_loop_vec_info): Delete.
18182 (vect_destroy_datarefs): Likewise.
18183 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
18184 (vec_info::vec_info): New function.
18185 (vec_info::~vec_info): Likewise.
18186 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
18187 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
18188 destroy_loop_vec_info.
18189 * tree-vect-loop.c (new_loop_vec_info): Replace with...
18190 (_loop_vec_info::_loop_vec_info): ...this.
18191 (destroy_loop_vec_info): Replace with...
18192 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
18193 the stmt_vec_infos. Leave handling of vec_info information to its
18194 destructor. Remove explicit vector releases.
18195 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
18196 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
18197 * tree-vect-slp.c (new_bb_vec_info): Replace with...
18198 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
18199 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
18200 (destroy_bb_vec_info): Replace with...
18201 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
18202 information to its destructor.
18203 (vect_slp_analyze_bb_1): Use new and delete instead of
18204 new_bb_vec_info and destroy_bb_vec_info.
18205 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
18208 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18210 * tree-data-ref.h (subscript): Add access_fn field.
18211 (data_dependence_relation): Add could_be_independent_p.
18212 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
18213 (same_access_functions): Move to tree-data-ref.c.
18214 * tree-data-ref.c (ref_contains_union_access_p): New function.
18215 (access_fn_component_p): Likewise.
18216 (access_fn_components_comparable_p): Likewise.
18217 (dr_analyze_indices): Add a reference to access_fn_component_p.
18218 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
18220 (constant_access_functions): Likewise.
18221 (add_other_self_distances): Likewise.
18222 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
18223 (initialize_data_dependence_relation): Use XCNEW and remove
18224 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
18225 of access functions that have the same type. Allow the
18226 subsequence to end with different bases in some circumstances.
18227 Record the chosen access functions in SUB_ACCESS_FN.
18228 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
18229 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
18230 (subscript_dependence_tester_1): Likewise dra and drb.
18231 (build_classic_dist_vector): Update calls accordingly.
18232 (subscript_dependence_tester): Likewise.
18233 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
18234 DDR_COULD_BE_INDEPENDENT_P.
18235 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
18236 comp_alias_ddrs instead of may_alias_ddrs.
18237 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
18239 (vect_analyze_data_ref_dependence): Use it if
18240 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
18241 distance vectors if that fails.
18242 (dependence_distance_ge_vf): New function.
18243 (vect_prune_runtime_alias_test_list): Use it. Don't clear
18244 LOOP_VINFO_MAY_ALIAS_DDRS.
18246 2017-08-04 Richard Biener <rguenther@suse.de>
18248 PR middle-end/81705
18249 * fold-const.c (fold_binary_loc): Properly restrict
18250 minus_var0 && minus_var1 case when associating undefined overflow
18253 2017-08-04 Tom de Vries <tom@codesourcery.com>
18255 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
18257 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18259 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18260 Don't start diagnostic messages with a capital letter.
18261 * config/rs6000/rs6000.c (rs6000_option_override_internal):
18263 (rs6000_invalid_builtin): Likewise.
18264 (rs6000_trampoline_init): Likewise.
18266 2017-08-03 Jakub Jelinek <jakub@redhat.com>
18269 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
18270 after setting changeable df flags.
18272 2017-08-03 Richard Biener <rguenther@suse.de>
18274 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
18275 up if the use is in USE - X.
18277 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
18279 * toplev.c (dumpfile.h): New include.
18280 (internal_error_reentered): New static function. Use it...
18281 (internal_error_function): ...here to handle reentered internal_error.
18283 2017-08-03 Richard Biener <rguenther@suse.de>
18285 PR middle-end/81148
18286 * fold-const.c (split_tree): Add minus_var and minus_con
18287 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
18288 here but always use minus_*.
18289 (associate_trees): Assert we never associate with MINUS_EXPR
18290 and NULL first operand. Do not recurse for PLUS_EXPR operands
18291 when associating as MINUS_EXPR either.
18292 (fold_binary_loc): Track minus_var and minus_con.
18294 2017-08-03 Tom de Vries <tom@codesourcery.com>
18297 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
18298 ACCEL_COMPILER, apply finish_options on
18299 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
18301 2017-08-03 Tom de Vries <tom@codesourcery.com>
18304 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
18305 function_entry_patch_area_size > 0.
18307 2017-08-03 Jakub Jelinek <jakub@redhat.com>
18310 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
18311 instead of 10??LU, perform unit multiplication in wide_int,
18312 don't change alloc_object_size_limit if the limit is larger
18315 PR tree-optimization/81655
18316 PR tree-optimization/81588
18317 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
18318 the case when ranges[i].low and high are 1 for unsigned type with
18321 PR middle-end/81052
18322 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
18323 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
18325 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18327 * tree-vrp.h: Add include guard.
18329 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
18332 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
18333 (ud2): New insn pattern.
18334 * config/i386/i386.c (ix86_expand_epilogue):
18335 For naked functions, generate ud2 instead of trap insn.
18337 2017-08-02 Marek Polacek <polacek@redhat.com>
18340 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
18342 2017-08-02 Tom de Vries <tom@codesourcery.com>
18343 Cesar Philippidis <cesar@codesourcery.com>
18345 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
18346 Add missing edge probabilities.
18348 2017-08-02 Tamar Christina <tamar.christina@arm.com>
18350 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
18351 Correct endianness.
18353 2017-08-02 Jakub Jelinek <jakub@redhat.com>
18355 PR middle-end/79499
18356 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
18357 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
18358 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
18359 prologue_seq sequences - if any.
18361 2017-08-02 Richard Biener <rguenther@suse.de>
18363 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
18364 via vectors if supported, integer extracts via punning if supported
18365 or otherwise vector extracts.
18367 2017-08-02 Richard Biener <rguenther@suse.de>
18369 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
18371 (bitmap_insert_into_set): ... this.
18373 2017-08-02 Richard Biener <rguenther@suse.de>
18375 PR tree-optimization/81633
18377 2015-08-17 Alan Hayward <alan.hayward@arm.com>
18379 PR tree-optimization/71752
18380 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
18382 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
18384 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
18385 (machine_function::call_ms2sysv_pad_out): Remove field.
18386 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
18387 (ix86_compute_frame_layout): Likewise.
18389 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
18392 * config/i386/i386.c (ix86_set_func_type): Disallow naked
18393 attribute with interrupt attribute.
18395 2017-08-01 Andrew Pinski <apinski@cavium.com>
18397 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
18398 BIT_INSERT_EXPR's operand 1
18399 to see if the types precision matches.
18401 2017-08-01 Martin Liska <mliska@suse.cz>
18403 PR middle-end/70140
18404 * builtins.c (expand_builtin_memcpy_args): Remove.
18405 (expand_builtin_memcpy): Call newly added function
18406 expand_builtin_memory_copy_args.
18407 (expand_builtin_memcpy_with_bounds): Likewise.
18408 (expand_builtin_mempcpy): Remove last argument.
18409 (expand_builtin_mempcpy_with_bounds): Likewise.
18410 (expand_builtin_memory_copy_args): New function created from
18411 expand_builtin_mempcpy_args with small modifications.
18412 (expand_builtin_mempcpy_args): Remove.
18413 (expand_builtin_stpcpy): Remove unused argument.
18414 (expand_builtin): Likewise.
18415 (expand_builtin_with_bounds): Likewise.
18417 2017-08-01 Martin Liska <mliska@suse.cz>
18420 Make mempcpy more optimal (PR middle-end/70140).
18422 2017-08-01 Jakub Jelinek <jakub@redhat.com>
18425 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
18426 __builtin_vec_cmpne verify both arguments are compatible vectors
18427 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
18428 verify arg1_type is a pointer or array type. For __builtin_vec_st,
18429 move computation of aligned to after checking the argument types.
18433 * config/rs6000/vsx.md (vextract_fp_from_shorth,
18434 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
18437 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
18438 Doug Rupp <rupp@adacore.com>
18439 Olivier Hainque <hainque@adacore.com>
18441 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
18442 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
18444 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
18445 for TARGET_OS_CPP_BUILTIN.
18446 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
18447 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
18449 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
18450 passing required abi options to the assembler for EABI configurations.
18451 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
18452 of .text.hot and .text.unlikely sections for kernel modules when
18453 using ARM style exceptions.
18454 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
18455 options. Add EXTRA_CC1_SPEC.
18456 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
18458 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
18460 (ARM_TARGET2_DWARF_FORMAT): Define.
18461 * config/arm/t-vxworks: Adjust multilib control to removal of the
18462 Diab command line options.
18464 2017-08-01 Martin Liska <mliska@suse.cz>
18466 PR gcov-profile/81561
18467 * gcov.c (unblock): Make unblocking safe as we need to preserve
18468 index correspondence of blocks and block_lists.
18470 2017-08-01 Richard Biener <rguenther@suse.de>
18472 PR tree-optimization/81181
18473 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
18474 (compute_antic): ... end of iteration here.
18476 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
18478 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
18479 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
18480 (ftree-slp-vectorize): Likewise.
18481 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
18482 can no longer be set independent of flag_tree_loop_vectorize.
18483 * omp-general.c (emp_max_vf): Likewise.
18484 * opts.c (enable_fdo_optimizations): Remove references to
18485 flag_tree_vectorize, these are now implicit.
18486 (common_handle_option): Remove handling for OPT_ftree_vectorize,
18487 and leave it for the options machinery.
18489 2017-08-01 Martin Liska <mliska@suse.cz>
18491 PR middle-end/70140
18492 * builtins.c (expand_builtin_memcpy_args): Remove.
18493 (expand_builtin_memcpy): Call newly added function
18494 expand_builtin_memory_copy_args.
18495 (expand_builtin_memcpy_with_bounds): Likewise.
18496 (expand_builtin_mempcpy): Remove last argument.
18497 (expand_builtin_mempcpy_with_bounds): Likewise.
18498 (expand_builtin_memory_copy_args): New function created from
18499 expand_builtin_mempcpy_args with small modifications.
18500 (expand_builtin_mempcpy_args): Remove.
18501 (expand_builtin_stpcpy): Remove unused argument.
18502 (expand_builtin): Likewise.
18503 (expand_builtin_with_bounds): Likewise.
18505 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18508 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
18509 print "ds:" only for immediates in generic address space.
18511 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18514 * config/i386/i386.c (ix86_funciton_naked): New prototype.
18515 (ix86_function_ok_for_sibcall): Return false for naked functions.
18517 2017-08-01 Richard Biener <rguenther@suse.de>
18519 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
18520 (compute_antic): Seed worklist with exit block predecessors.
18521 * cfganal.c (dfs_find_deadend): For a cycle return the source
18522 of the edge closing it.
18524 2017-08-01 Tamar Christina <tamar.christina@arm.com>
18526 * config/aarch64/aarch64.c
18527 (aarch64_can_const_movi_rtx_p): Move 0 check.
18529 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18531 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
18532 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
18534 * match.pd: Ditto in address comparison pattern.
18536 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18538 PR tree-optimization/81627
18539 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
18540 closed ssa form for store-store chain.
18542 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18544 PR tree-optimization/81620
18545 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
18546 for store-store chain.
18548 2017-08-01 Jakub Jelinek <jakub@redhat.com>
18550 PR tree-optimization/81588
18551 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
18552 ranges[i].in_p, invert comparison code ccode. For >/>=,
18553 swap rhs1 and rhs2 and comparison code unconditionally,
18554 for </<= don't do that. Don't swap rhs1/rhs2 again if
18555 ranges[i].in_p, instead invert comparison code ccode if
18556 opcode or oe->rank is BIT_IOR_EXPR.
18559 * optabs.def (vec_extract_optab, vec_init_optab): Change from
18560 a direct optab to conversion optab.
18561 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
18562 with GET_MODE_INNER as last argument instead of optab_handler.
18563 * expmed.c (extract_bit_field_1): Likewise. Use vector from
18564 vector extraction if possible and optab is available.
18565 * expr.c (store_constructor): Use convert_optab_handler instead
18566 of optab_handler. Use vector initialization from smaller
18567 vectors if possible and optab is available.
18568 * tree-vect-stmts.c (vectorizable_load): Likewise.
18569 * doc/md.texi (vec_extract, vec_init): Document that the optabs
18570 now have two modes.
18571 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
18572 of vec_init from half-sized vectors with the same element mode.
18573 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
18574 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
18575 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
18576 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
18577 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
18578 after mode in gen_vec_extract* calls.
18579 (vec_extract<mode>): Renamed to ...
18580 (vec_extract<mode><ssescalarmodelower>): ... this.
18581 (vec_extract<mode><ssehalfvecmodelower>): New expander.
18582 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
18583 element mode after mode in gen_vec_init* calls.
18584 (VEC_INIT_HALF_MODE): New mode iterator.
18585 (vec_init<mode>): Renamed to ...
18586 (vec_init<mode><ssescalarmodelower>): ... this.
18587 (vec_init<mode><ssehalfvecmodelower>): New expander.
18588 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
18589 (vec_extractv2sfsf): ... this.
18590 (vec_initv2sf): Renamed to ...
18591 (vec_initv2sfsf): ... this.
18592 (vec_extractv2si): Renamed to ...
18593 (vec_extractv2sisi): ... this.
18594 (vec_initv2si): Renamed to ...
18595 (vec_initv2sisi): ... this.
18596 (vec_extractv4hi): Renamed to ...
18597 (vec_extractv4hihi): ... this.
18598 (vec_initv4hi): Renamed to ...
18599 (vec_initv4hihi): ... this.
18600 (vec_extractv8qi): Renamed to ...
18601 (vec_extractv8qiqi): ... this.
18602 (vec_initv8qi): Renamed to ...
18603 (vec_initv8qiqi): ... this.
18604 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
18605 (vec_init<mode>): Renamed to ...
18606 (vec_init<mode><VEC_base_l>): ... this.
18607 (vec_extract<mode>): Renamed to ...
18608 (vec_extract<mode><VEC_base_l>): ... this.
18609 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
18610 (vec_initv2sfsf): ... this.
18611 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
18612 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
18613 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
18614 element mode after mode in gen_vec_init* calls.
18615 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
18616 (vec_init<mode><Vel>): ... this.
18617 (vec_extract<mode>): Renamed to ...
18618 (vec_extract<mode><Vel>): ... this.
18619 * config/aarch64/iterators.md (Vel): New mode attribute.
18620 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
18621 Add element mode after mode in gen_vec_extract* calls.
18622 * config/s390/vector.md (non_vec_l): New mode attribute.
18623 (vec_extract<mode>): Renamed to ...
18624 (vec_extract<mode><non_vec_l>): ... this.
18625 (vec_init<mode>): Renamed to ...
18626 (vec_init<mode><non_vec_l>): ... this.
18627 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
18628 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
18630 * config/arm/iterators.md (V_elem_l): New mode attribute.
18631 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
18632 (vec_extract<mode><V_elem_l>): ... this.
18633 (vec_extractv2di): Renamed to ...
18634 (vec_extractv2didi): ... this.
18635 (vec_init<mode>): Renamed to ...
18636 (vec_init<mode><V_elem_l>): ... this.
18637 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
18638 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
18639 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
18640 Add element mode after gen_vec_extract* calls.
18641 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
18642 (vec_init<mode><unitmode>): ... this.
18643 (vec_extract<mode>): Renamed to ...
18644 (vec_extract<mode><unitmode>): ... this.
18645 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
18646 (vec_init<mode><unitmode>): ... this.
18647 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
18648 (vec_initv2sfsf): ... this.
18649 (vec_extractv2sf): Renamed to ...
18650 (vec_extractv2sfsf): ... this.
18651 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
18652 Add element mode after gen_vec_extract* calls.
18653 * config/mips/mips.md (unitmode): New mode iterator.
18654 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
18655 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
18656 * config/spu/spu.md (inner_l): New mode attribute.
18657 (vec_init<mode>): Renamed to ...
18658 (vec_init<mode><inner_l>): ... this.
18659 (vec_extract<mode>): Renamed to ...
18660 (vec_extract<mode><inner_l>): ... this.
18661 * config/sparc/sparc.md (veltmode): New mode iterator.
18662 (vec_init<VMALL:mode>): Renamed to ...
18663 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
18664 * config/ia64/vect.md (vec_initv2si): Renamed to ...
18665 (vec_initv2sisi): ... this.
18666 (vec_initv2sf): Renamed to ...
18667 (vec_initv2sfsf): ... this.
18668 (vec_extractv2sf): Renamed to ...
18669 (vec_extractv2sfsf): ... this.
18670 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
18671 (vec_init<mode>): Renamed to ...
18672 (vec_init<mode><VEC_base_l>): ... this.
18673 (vec_extract<mode>): Renamed to ...
18674 (vec_extract<mode><VEC_base_l>): ... this.
18675 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
18676 (vec_initv2sfsf): ... this.
18677 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
18678 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
18679 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
18680 gen_vec_init* calls.
18682 2017-08-01 Richard Biener <rguenther@suse.de>
18684 PR tree-optimization/81297
18685 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
18686 TREE_OVERFLOW from INTEGER_CSTs.
18688 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
18690 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
18692 2017-07-31 Carl Love <cel@us.ibm.com>
18694 * config/rs6000/rs6000-c: Add support for built-in functions
18695 vector signed char vec_xl_be (signed long long, signed char *);
18696 vector unsigned char vec_xl_be (signed long long, unsigned char *);
18697 vector signed int vec_xl_be (signed long long, signed int *);
18698 vector unsigned int vec_xl_be (signed long long, unsigned int *);
18699 vector signed long long vec_xl_be (signed long long, signed long long *);
18700 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
18701 vector signed short vec_xl_be (signed long long, signed short *);
18702 vector unsigned short vec_xl_be (signed long long, unsigned short *);
18703 vector double vec_xl_be (signed long long, double *);
18704 vector float vec_xl_be (signed long long, float *);
18705 * config/rs6000/altivec.h (vec_xl_be): Add #define.
18706 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
18707 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
18709 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
18710 (altivec_expand_builtin): Add switch statement to call
18711 altivec_expand_xl_be for each builtin.
18712 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
18713 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
18714 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
18715 __builtin_vsx_le_be_v16qi.
18716 * doc/extend.texi: Update the built-in documentation file for the
18717 new built-in functions.
18719 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
18722 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
18724 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
18726 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18728 * config.gcc: Add z14.
18729 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
18730 CPU model numbers for z13s and z14.
18731 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
18733 * config/s390/s390-opts.h (enum processor_type): Rename
18734 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
18735 * config/s390/s390.c (processor_table): Add field for CPU name to
18736 be passed to Binutils.
18737 (s390_asm_output_machine_for_arch): Use the new field in
18738 processor_table for Binutils.
18739 (s390_expand_builtin): Replace arch12 with z14.
18740 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
18741 (s390_get_sched_attrmask): Likewise.
18742 (s390_get_unit_mask): Likewise.
18743 * config/s390/s390.opt: Add z14 to processor_type enum.
18745 2017-07-31 Martin Jambor <mjambor@suse.cz>
18748 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
18749 regardless of optimization level.
18751 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
18752 Martin Liska <mliska@suse.cz>
18754 * predict.def: Remove old comment and adjust probability.
18755 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
18756 PREDICT statements.
18758 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
18761 * config/i386/i386.c (ix86_function_naked): New function.
18762 (ix86_can_use_return_insn_p): Return false for naked functions.
18763 (ix86_expand_prologue): Skip prologue for naked functions.
18764 (ix86_expand_epilogue): Skip epilogue for naked functions
18765 and emit trap instruction.
18766 (ix86_warn_func_return): New function.
18767 (ix86_attribute_table): Add "naked" attribute specification.
18768 (TARGET_WARN_FUNC_RETURN): Define.
18769 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
18771 2017-07-31 Martin Liska <mliska@suse.cz>
18773 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
18774 (dump_gimple_bb_header): Always dump BB info.
18775 (pp_cfg_jump): Do not append info about BB when dumping a jump.
18777 2017-07-31 Martin Liska <mliska@suse.cz>
18780 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
18781 also with current_function_decl non-null equality.
18783 2017-07-31 Jakub Jelinek <jakub@redhat.com>
18786 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
18787 change type to the element type, instead add eltype variable and
18788 use it where we are interested in the element type.
18790 PR tree-optimization/81603
18791 * ipa-polymorphic-call.c
18792 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
18793 offset arithmetic in offset_int, bail out if the resulting bit offset
18794 doesn't fit into shwi.
18796 2017-07-31 Martin Liska <mliska@suse.cz>
18798 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
18799 (gimplify_save_expr): Fix comment.
18801 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
18804 * config/i386/i386.c (ix86_function_arg): Update arguments for
18806 (ix86_compute_frame_layout): Set the initial stack offset to
18807 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
18808 INCOMING_FRAME_SP_OFFSET.
18809 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
18810 stack before exception handler returns.
18811 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
18812 the 'ERROR_CODE' for exception handler.
18814 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
18816 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
18817 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
18818 (ASM_OUTPUT_REG_POP): Ditto.
18819 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
18820 instead of asm_fprintf to output pure string.
18822 2017-07-29 Jakub Jelinek <jakub@redhat.com>
18824 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
18825 to imported_module_or_decl hook.
18826 (debug_nothing_tree_tree_tree_bool): Remove.
18827 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
18828 * debug.c (do_nothing_debug_hooks): Use
18829 debug_nothing_tree_tree_tree_bool_bool instead of
18830 debug_nothing_tree_tree_tree_bool.
18831 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
18832 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
18833 * sdbout.c (sdb_debug_hooks): Likewise.
18834 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
18835 (gen_namespace_die): Add DW_AT_export_symbols attribute if
18837 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
18838 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
18839 attribute, don't add anything.
18841 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18843 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
18844 (fold_build2_stat_loc): Likewise.
18845 (fold_build3_stat_loc): Likewise.
18846 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
18847 (fold_build1_loc): Remove macro.
18848 (fold_build2_loc): Likewise.
18849 (fold_build3_loc): Likewise.
18851 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18853 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
18854 (gimple_build_debug_bind_source_stat): Likewise.
18855 * gimple.h (gimple_build_debug_bind): Remove macro.
18856 (gimple_build_debug_bind_source): Likewise.
18858 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18860 * bitmap.c (bitmap_alloc): Adjust.
18861 (bitmap_gc_alloc): Likewise.
18862 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
18864 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18866 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
18867 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
18868 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
18869 (bitmap_gc_alloc_stat): Likewise.
18870 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
18872 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18874 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
18875 * rtl.h (shallow_copy_rtx): Remove macro.
18877 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18879 * emit-rtl.c (gen_raw_REG): Adjust.
18880 * gengenrtl.c (gendef): Likewise.
18881 * rtl.c (rtx_alloc_stat): Remove _stat from name.
18882 * rtl.h (rtx_alloc): Remove macro.
18884 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18886 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
18887 (build_tree_list_stat): Likewise.
18888 * tree.h (build_tree_list): Remove macro.
18889 (build_tree_list_vec): Likewise.
18891 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18893 * tree.c (make_vector_stat): Remove _stat from name.
18894 (build_vector_stat): Likewise.
18895 * tree.h (make_vector_stat): Remove macro.
18896 (build_vector_stat): Likewise.
18898 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18900 * tree.h (build_var_debug_value): Remove prototype.
18902 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18904 * tree.c (tree_cons_stat): Remove _stat from name.
18905 * tree.h (tree_cons): Remove macro.
18907 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18909 * tree.c (build_vl_exp_stat): Remove _stat from name.
18910 * tree.h (build_vl_exp): Remove macro.
18912 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18914 * tree.c (build_decl_stat): Remove _stat from name.
18915 * tree.h (build_decl): Remove macro.
18917 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18919 * gimple.c (gimple_build_with_ops_stat): Adjust.
18920 (gimple_alloc_stat): Remove _stat from name.
18921 * gimple.h (gimple_alloc): Remove macro.
18923 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18925 * tree.c (make_tree_vec_stat): Remove _stat from name.
18926 (grow_tree_vec_stat): Likewise.
18927 * tree.h (make_tree_vec_stat): Adjust prototype.
18928 (grow_tree_vec_stat): Likewise.
18929 (make_tree_vec): Remove macro.
18930 (grow_tree_vec): Likewise.
18932 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18934 * fold-const.c (fold_build1_stat_loc): Adjust.
18935 (fold_build2_stat_loc): Likewise.
18936 (fold_build3_stat_loc): Likewise.
18937 * tree.c (build0_stat): Remove _stat from name.
18938 (build1_stat): Likewise.
18939 (build2_stat): Likewise.
18940 (build3_stat): Likewise.
18941 (build4_stat): Likewise.
18942 (build5_stat): Likewise.
18943 * tree.h (build1_loc): Remove macro, and rename _stat function
18945 (build2_loc): Likewise.
18946 (build3_loc): Likewise.
18947 (build4_loc): Likewise.
18948 (build5_loc): Likewise.
18950 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18952 * tree.c (make_int_cst_stat): Remove _stat from name.
18953 * tree.h (make_int_cst_stat): Adjust prototype.
18954 (make_int_cst): Remove macro.
18956 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18958 * tree.c (make_tre_binfo_stat): Remove _stat from name.
18959 * tree.h (make_tree_binfo_stat): Adjust prototype.
18960 (make_tree_binfo): Remove.
18962 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18964 * tree.c (copy_node_stat): Rename to copy_node.
18965 (build_distinct_type_copy): Adjust.
18966 * tree.h (copy_node_stat): Adjust prototype.
18967 (copy_node): Remove macro.
18969 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18971 * tree.c (make_node_stat): rename to make_node.
18972 (build_tree_list_stat): Adjust.
18973 (build0_stat): Likewise.
18974 (build2_stat): Likewise.
18975 (build3_stat): Likewise.
18976 (build4_stat): Likewise.
18977 (build5_stat): Likewise.
18978 (build_decl_stat): Likewise.
18979 * tree.h (make_node_stat): Adjust prototype.
18980 (make_node): remove macro.
18982 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
18984 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
18985 (PPC_FEATURE2_SCV): Likewise.
18986 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
18988 2017-07-28 Tamar Christina <tamar.christina@arm.com>
18990 * config/aarch64/aarch64.c
18991 (aarch64_internal_mov_immediate): Add new special pattern.
18992 * config/aarch64/aarch64.md (*movdi_aarch64):
18993 Add reg/32bit const mov case.
18995 2017-07-28 Tamar Christina <tamar.christina@arm.com>
18996 Richard Sandiford <richard.sandiford@linaro.org>
18998 * config/aarch64/aarch64.md (mov<mode>): Generalize.
18999 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
19000 Add integer and movi cases.
19001 (movi-split-hf-df-sf split, fp16): New.
19002 (enabled): Added TARGET_FP_F16INST.
19003 * config/aarch64/iterators.md (GPF_HF): New.
19004 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
19006 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19008 * config/aarch64/aarch64.c
19009 (aarch64_simd_container_mode): Add prototype.
19010 (aarch64_expand_mov_immediate): Add HI support.
19011 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
19012 (aarch64_can_const_movi_rtx_p): New.
19013 (aarch64_preferred_reload_class):
19014 Remove restrictions of using FP registers for certain SIMD operations.
19015 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
19016 (aarch64_valid_floating_const): Add integer move validation.
19017 (aarch64_simd_imm_scalar_p): Remove.
19018 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
19019 (aarch64_legitimate_constant_p): Expand list of supported cases.
19020 * config/aarch64/aarch64-protos.h
19021 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
19022 (aarch64_reinterpret_float_as_int): New.
19023 (aarch64_simd_imm_scalar_p): Remove.
19024 * config/aarch64/constraints.md (Uvi): New.
19025 (Dd): Split into Ds and new Dd.
19026 * config/aarch64/aarch64.md (*movsi_aarch64):
19028 (*movdi_aarch64): Add SIMD mov case.
19030 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19032 * tree-predcom.c: (struct chain): Handle store-store chain in which
19033 stores for elimination only store loop invariant values.
19034 (execute_pred_commoning_chain): Ditto.
19035 (prepare_initializers_chain_store_elim): Ditto.
19036 (prepare_finalizers): Ditto.
19037 (is_inv_store_elimination_chain): New function.
19038 (initialize_root_vars_store_elim_1): New function.
19040 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19042 * tree-predcom.c: Revise general description of the pass.
19043 (enum chain_type): New enum type for store elimination.
19044 (struct chain): New field supporting store elimination.
19045 (struct component): Ditto.
19046 (dump_chain): Dump store-stores chain.
19047 (release_chain): Release resources.
19048 (split_data_refs_to_components): Compute and create component
19049 contains only stores for elimination.
19050 (get_chain_last_ref_at): New function.
19051 (make_invariant_chain): Initialization.
19052 (make_rooted_chain): Specify chain type in parameter and record it.
19053 (add_looparound_copies): Skip for store-stores chain.
19054 (determine_roots_comp): Compute type of chain and pass it to
19056 (initialize_root_vars_store_elim_2): New function.
19057 (finalize_eliminated_stores): New function.
19058 (remove_stmt): Handle store for elimination.
19059 (execute_pred_commoning_chain): Execute predictive commoning on
19060 store-store chains.
19061 (determine_unroll_factor): Skip unroll for store-stores chain.
19062 (prepare_initializers_chain_store_elim): New function.
19063 (prepare_initializers_chain): Hanlde store-store chain.
19064 (prepare_finalizers_chain, prepare_finalizers): New function.
19065 (tree_predictive_commoning_loop): Return integer value indicating
19066 if loop is unrolled or lcssa form is corrupted.
19067 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
19069 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19071 * tree-predcom.c (initialize_root): Delete.
19072 (execute_pred_commoning_chain): Initialize root vars and replace
19073 reference of non-combined chain directly, rather than call above
19076 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19078 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
19079 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
19081 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19083 * tree-predcom.c (struct chain): New field init_seq.
19084 (release_chain): Release init_seq.
19085 (prepare_initializers_chain): Record intialization stmts in above
19087 (insert_init_seqs): New function.
19088 (tree_predictive_commoning_loop): Call insert_init_seqs.
19090 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19092 * tree-predcom.c (determine_roots_comp): Skip trivial components.
19094 2017-07-28 Richard Biener <rguenther@suse.de>
19096 * match.pd: Remove superfluous :c.
19097 * genmatch.c (simplify::id): Add member.
19098 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
19100 (current_id): New global.
19101 (dt_node::parent): Move from ...
19102 (dt_operand::parent): ... here. Add for_id member.
19103 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
19104 (decision_tree::find_node): Relax order requirement when
19105 merging DT_TRUE nodes to ones inbetween the current simplify
19106 and the one we try to merge with. Add diagnostic whenever
19107 we need to enforce pattern order by not merging.
19108 (decision_tree::insert): Set current_id.
19109 (decision_tree::print_node): Dump parent node and for_id.
19110 (parser::last_id): Add member.
19111 (parser::push_simplify): Assign unique id.
19112 (parser::parser): Initialize last_id.
19114 2017-07-28 Martin Liska <mliska@suse.cz>
19117 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
19118 gimple_build_debug_bind.
19120 2017-07-28 Richard Biener <rguenther@suse.de>
19122 PR tree-optimization/81502
19123 * match.pd: Add pattern combining BIT_INSERT_EXPR with
19125 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
19127 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
19128 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
19129 for BIT_FIELD_REF args.
19130 * fold-const.c (make_bit_field_ref): Likewise.
19131 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
19133 2017-07-28 Jakub Jelinek <jakub@redhat.com>
19136 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
19137 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
19138 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
19139 Or it into SANITIZER_UNDEFINED.
19140 * ubsan.c: Include gimple-fold.h and varasm.h.
19141 (ubsan_expand_ptr_ifn): New function.
19142 (instrument_pointer_overflow): New function.
19143 (maybe_instrument_pointer_overflow): New function.
19144 (instrument_object_size): Formatting fix.
19145 (pass_ubsan::execute): Call instrument_pointer_overflow
19146 and maybe_instrument_pointer_overflow.
19147 * internal-fn.c (expand_UBSAN_PTR): New function.
19148 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
19149 * sanitizer.def (__ubsan_handle_pointer_overflow,
19150 __ubsan_handle_pointer_overflow_abort): New builtins.
19151 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
19152 * internal-fn.def (UBSAN_PTR): New internal function.
19153 * opts.c (sanitizer_opts): Add pointer-overflow.
19154 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
19155 * fold-const.c (build_range_check): Compute pointer range check in
19156 integral type if pointer arithmetics would be needed. Formatting
19159 2017-07-28 Martin Liska <mliska@suse.cz>
19162 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
19163 parameters that are of a variable-length.
19165 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
19167 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
19169 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
19170 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
19171 (CRT_CALL_STATIC_FUNCTION): Likewise.
19172 (ASM_DEFAULT_SPEC): New define.
19173 (ASM_SPEC32): Likewise.
19174 (ASM_SPEC64): Likewise.
19175 (ASM_SPEC_COMMON): Likewise.
19176 (ASM_SPEC): Likewise.
19177 (INVALID_64BIT): Likewise.
19178 (LINK_OS_DEFAULT_SPEC): Likewise.
19179 (LINK_OS_SPEC32): Likewise.
19180 (LINK_OS_SPEC64): Likewise.
19181 (POWERPC_LINUX): Likewise.
19182 (PTRDIFF_TYPE): Likewise.
19183 (RESTORE_FP_PREFIX): Likewise.
19184 (RESTORE_FP_SUFFIX): Likewise.
19185 (SAVE_FP_PREFIX): Likewise.
19186 (SAVE_FP_SUFFIX): Likewise.
19187 (SIZE_TYPE): Likewise.
19188 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
19189 (TARGET_64BIT): Likewise.
19190 (TARGET_64BIT): Likewise.
19191 (TARGET_AIX): Likewise.
19192 (WCHAR_TYPE_SIZE): Likewise.
19193 (WCHAR_TYPE): Undefine.
19194 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
19195 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
19196 (CPP_OS_RTEMS_SPEC): Delete.
19197 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
19198 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
19200 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
19202 2017-07-28 Jakub Jelinek <jakub@redhat.com>
19204 PR tree-optimization/81578
19205 * tree-parloops.c (build_new_reduction): Bail out if
19206 reduction_code isn't one of the standard OpenMP reductions.
19207 Move the details printing after that decision.
19209 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19211 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
19212 related to reload_in_progress.
19213 (splat_input_operand): Likewise.
19214 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
19216 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
19218 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
19219 (TARGET_INSTANTIATE_DECLS): Likewise.
19220 (legitimate_indexed_address_p): Delete reload_in_progress code.
19221 (rs6000_debug_legitimate_address_p): Likewise.
19222 (rs6000_eliminate_indexed_memrefs): Likewise.
19223 (rs6000_emit_le_vsx_store): Likewise.
19224 (rs6000_emit_move_si_sf_subreg): Likewise.
19225 (rs6000_emit_move): Likewise.
19226 (register_to_reg_type): Likewise.
19227 (rs6000_pre_atomic_barrier): Likewise.
19228 (rs6000_machopic_legitimize_pic_address): Likewise.
19229 (rs6000_allocate_stack_temp): Likewise.
19230 (rs6000_address_for_fpconvert): Likewise.
19231 (rs6000_address_for_altivec): Likewise.
19232 (rs6000_secondary_memory_needed_rtx): Delete function.
19233 (rs6000_check_sdmode): Likewise.
19234 (rs6000_alloc_sdmode_stack_slot): Likewise.
19235 (rs6000_instantiate_decls): Likewise.
19236 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
19237 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
19238 Delete reload_in_progress.
19239 (*vec_reload_and_plus_<mptrsize>): Likewise.
19240 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
19241 (vsx_div_v2di): Likewise.
19242 (vsx_udiv_v2di): Likewise.
19244 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19246 * config/rs6000/rs6000.opt (mlra): Replace with stub.
19247 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
19248 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
19249 (rs6000_debug_reg_global): Delete print of LRA status.
19250 (rs6000_option_override_internal): Delete dead LRA related code.
19251 (rs6000_lra_p): Delete function.
19252 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
19254 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
19256 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
19257 * config/riscv/rtems.h: New file.
19259 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19260 Sudakshina Das <sudi.das@arm.com>
19262 * config/aarch64/aarch64.md
19263 (define_split for and<mode>3nr_compare): Move
19264 non aarch64_logical_operand to a register.
19265 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
19266 register immediate operand to a register.
19267 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
19269 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19271 PR middle-end/81564
19272 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
19274 2017-07-27 Richard Biener <rguenther@suse.de>
19276 PR tree-optimization/81573
19277 PR tree-optimization/81494
19278 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
19279 multi defuse cycle case.
19281 2017-07-27 Richard Biener <rguenther@suse.de>
19283 PR tree-optimization/81571
19284 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
19287 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
19289 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
19290 earlier and only if MASK_FPU is set. Adjust formatting.
19292 2017-07-27 Martin Liska <mliska@suse.cz>
19294 * opt-functions.awk: Add validation of value of Init.
19295 * optc-gen.awk: Pass new argument.
19297 2017-07-27 Martin Liska <mliska@suse.cz>
19299 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
19300 Fix wrong condition.
19302 2017-07-27 Martin Liska <mliska@suse.cz>
19304 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
19305 BBs and edges seen by autoFDO.
19307 2017-07-27 Richard Biener <rguenther@suse.de>
19309 PR tree-optimization/81502
19310 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
19311 with incompatible but same sized type.
19312 (execute_update_addresses_taken): Likewise.
19314 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
19316 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
19317 flag_tree_loop_vectorize rather than flag_tree_vectorize.
19319 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19322 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
19323 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
19324 Change s_operand to memory_operand.
19326 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
19328 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
19329 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
19330 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
19331 Emit instructions rather than returning an expression. Handle TFmode
19332 and KFmode by casting to TImode.
19333 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
19334 (rs6000_emit_le_vsx_store): Likewise.
19335 * config/rs6000/vsx.md (VSX_TI): New iterator.
19336 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
19337 (*vsx_le_undo_permute_<mode>): Likewise.
19338 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
19339 emit the split sequence.
19340 (*vsx_le_perm_store_<mode>): Likewise.
19342 2017-07-27 Jakub Jelinek <jakub@redhat.com>
19344 PR tree-optimization/81555
19345 PR tree-optimization/81556
19346 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
19347 if true, force CHANGED for the recursive invocation.
19348 (reassociate_bb): Remember original length of ops array, pass
19349 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
19351 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
19352 attributes for noipa attribute. For naked attribute use
19353 lookup_attribute first before lookup_attribute_spec.
19354 * final.c (rest_of_handle_final): Disable IPA RA for functions with
19356 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
19357 for functions with noipa attribute.
19358 (cgraph_externally_visible_p): Return true for functions with noipa
19360 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
19361 for functions with noipa attribute.
19362 * doc/extend.texi: Document noipa function attribute.
19363 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
19364 also for functions with noipa attribute.
19365 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
19367 2017-07-26 Andrew Pinski <apinski@cavium.com>
19369 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
19370 vec_unalign_load_cost and vec_unalign_store_cost.
19372 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
19374 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
19375 -mvsx-small-integer option.
19376 (ISA_3_0_MASKS_IEEE): Likewise.
19377 (OTHER_VSX_VECTOR_MASKS): Likewise.
19378 (POWERPC_MASKS): Likewise.
19379 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
19380 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
19381 code, only testing for DImode being allowed in non-VSX floating
19383 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
19384 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
19386 (rs6000_option_override_internal): Delete -mvsx-small-integer.
19387 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
19388 TARGET_P8_VECTOR test.
19389 (rs6000_secondary_reload_simple_move): Likewise.
19390 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
19391 since TARGET_P9_VECTOR was already tested.
19392 (rs6000_opt_masks): Remove -mvsx-small-integer.
19393 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
19394 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19396 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
19397 test for TARGET_VEXTRACTUB was used, and that uses
19399 (p9 extract splitter): Likewise.
19400 (vsx_extract_<mode>_di_p9): Likewise.
19401 (vsx_extract_<mode>_store_p9): Likewise.
19402 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
19403 for TARGET_P9_VECTOR was used. Delete code that is now dead with
19404 the elimination of TARGET_VSX_SMALL_INTEGER.
19405 (vsx_extract_<mode>_p8): Likewise.
19406 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
19407 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
19408 (vsx_set_<mode>_p9): Likewise.
19409 (vsx_set_v4sf_p9): Likewise.
19410 (vsx_set_v4sf_p9_zero): Likewise.
19411 (vsx_insert_extract_v4sf_p9): Likewise.
19412 (vsx_insert_extract_v4sf_p9_2): Likewise.
19413 * config/rs6000/rs6000.md (sign extend splitter): Change
19414 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
19415 (floatsi<mode>2_lfiwax_mem): Likewise.
19416 (floatunssi<mode>2_lfiwzx_mem): Likewise.
19417 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
19418 since a test for TARGET_P9_VECTOR was used.
19419 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19420 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
19421 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19422 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
19423 TARGET_P8_VECTOR test.
19424 (fix_trunc<mode>si2_stfiwx): Likewise.
19425 (fix_trunc<mode>si2_internal): Likewise.
19426 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
19427 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19429 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19430 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
19431 TARGET_P8_VECTOR test.
19432 (fixuns_trunc<mode>si2_stfiwx): Likewise.
19433 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
19434 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19436 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19437 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
19438 since a test for TARGET_P9_VECTOR was used.
19439 (splitter for loading small constants): Likewise.
19441 2017-07-26 Andrew Pinski <apinski@cavium.com>
19443 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
19446 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
19449 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
19450 (fp_valid_at): Likewise.
19452 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
19454 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
19455 (qdf24xx_addrcost_table): Likewise.
19456 (cortexa57_tunings): Update to use generic_branch_cost.
19457 (cortexa72_tunings): Likewise.
19458 (cortexa73_tunings): Likewise.
19459 (qdf24xx_tunings): Likewise.
19461 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
19463 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
19464 (thunderx2t99_branch_cost): Likewise.
19465 (cortexa35_tunings): Update to use generic_branch_cost.
19466 (cortexa53_tunings): Likewise.
19467 (cortexa57_tunings): Likewise.
19468 (cortexa72_tunings): Likewise.
19469 (cortexa73_tunings): Likewise.
19470 (thunderx2t99_tunings): Likewise.
19472 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19474 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
19475 (sparc_option_override): Honour MASK_FSMULD.
19476 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
19477 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
19478 * config/sparc/sparc.opt (mfsmuld): New option.
19479 * doc/invoke.texi (mfsmuld): Document option.
19481 2017-07-26 Marek Polacek <polacek@redhat.com>
19483 PR middle-end/70992
19484 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
19486 2017-07-26 Richard Biener <rguenther@suse.de>
19488 * gimple-match-head.c (do_valueize): Return OP if valueize
19490 (get_def): New helper to get at the def stmt of a SSA name
19491 if valueize allows.
19492 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
19493 do_valueize to get at the def stmt.
19494 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
19496 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
19498 PR middle-end/46932
19499 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
19501 2017-07-26 Martin Liska <mliska@suse.cz>
19504 * function.c (expand_function_start): Make expansion of
19505 nonlocal_goto_save_area after parm_birth_insn.
19507 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19509 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
19510 from all CPU target flags enable members.
19512 2017-07-26 Richard Biener <rguenther@suse.de>
19514 * genmatch.c (dt_simplify::gen): Make iterator vars const.
19515 (decision_tree::gen): Make 'type' const.
19516 (write_predicate): Likewise.
19518 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
19520 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
19521 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
19522 (rs6000_option_override_internal): Likewise.
19523 (rs6000_expand_vector_set): Likewise.
19524 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
19525 (TARGET_UPPER_REGS_SF): Likewise.
19526 (TARGET_UPPER_REGS_DI): Likewise.
19527 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
19528 (TARGET_DIRECT_MOVE_64BIT): Likewise.
19529 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
19530 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19531 (Splitters for DI constants in Altivec registers): Likewise.
19532 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
19533 (vsx_set_v4sf_p9): Likewise.
19534 (vsx_set_v4sf_p9_zero): Likewise.
19535 (vsx_insert_extract_v4sf_p9): Likewise.
19536 (vsx_insert_extract_v4sf_p9_2): Likewise.
19538 2017-07-25 Carl Love <cel@us.ibm.com>
19540 * doc/extend.texi: Update the built-in documentation file for the
19541 existing built-in functions
19542 vector signed char vec_cnttz (vector signed char);
19543 vector unsigned char vec_cnttz (vector unsigned char);
19544 vector signed short vec_cnttz (vector signed short);
19545 vector unsigned short vec_cnttz (vector unsigned short);
19546 vector signed int vec_cnttz (vector signed int);
19547 vector unsigned int vec_cnttz (vector unsigned int);
19548 vector signed long long vec_cnttz (vector signed long long);
19549 vector unsigned long long vec_cnttz (vector unsigned long long);
19551 2017-07-25 Andrew Pinski <apinski@cavium.com>
19553 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
19554 accesses where the use is for the first operand of a BIT_INSERT.
19556 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19559 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
19560 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
19562 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19564 * config/i386/gstabs.h: Delete.
19565 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
19567 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
19569 * config/i386/i386.c (ix86_decompose_address): Do not check for
19570 register RTX when looking at index_reg or base_reg.
19571 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
19573 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
19575 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
19576 to update EH info here.
19578 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19580 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
19582 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19584 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
19586 2017-07-25 Torsten Duwe <duwe@suse.de>
19588 * common.opt: Introduce -fpatchable-function-entry
19589 command line option, and its variables function_entry_patch_area_size
19590 and function_entry_patch_area_start.
19591 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
19592 including a two-value parser.
19593 * target.def (print_patchable_function_entry): New target hook.
19594 * targhooks.h (default_print_patchable_function_entry): New function.
19595 * targhooks.c (default_print_patchable_function_entry): Likewise.
19596 * toplev.c (process_options): Switch off IPA-RA if
19597 patchable function entries are being generated.
19598 * varasm.c (assemble_start_function): Look at the
19599 patchable-function-entry command line switch and current
19600 function attributes and maybe generate NOP instructions by
19601 calling the print_patchable_function_entry hook.
19602 * doc/extend.texi: Document patchable_function_entry attribute.
19603 * doc/invoke.texi: Document -fpatchable_function_entry
19604 command line option.
19605 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
19607 * doc/tm.texi: Re-generate.
19609 2017-07-25 Jakub Jelinek <jakub@redhat.com>
19612 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
19613 TARGET_AVX512DQ rather than TARGET_AVX512BW.
19615 2017-07-25 Tamar Christina <tamar.christina@arm.com>
19617 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
19619 2017-07-25 Richard Biener <rguenther@suse.de>
19621 PR tree-optimization/81455
19622 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
19623 not walk in cycles when looking for guards.
19625 2017-07-25 Richard Biener <rguenther@suse.de>
19627 PR tree-optimization/81529
19628 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
19629 when optimizing backedge uses.
19631 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
19633 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
19635 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
19636 to dl_section_ref. On AIX, append an expression to subtract
19637 the size of the section length to dl_section_ref.
19639 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
19641 * configure.ac: If any of the config.* scripts fail, exit 1.
19642 * configure: Regenerate.
19644 2017-07-25 Richard Biener <rguenther@suse.de>
19646 PR middle-end/81546
19647 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
19648 of immediate uses, be more verbose on errors.
19650 2017-07-25 Richard Biener <rguenther@suse.de>
19652 PR tree-optimization/81510
19653 * tree-vect-loop.c (vect_is_simple_reduction): When the
19654 reduction stmt is not inside the loop bail out.
19656 2017-07-25 Richard Biener <rguenther@suse.de>
19658 PR tree-optimization/81303
19659 * tree-vect-loop-manip.c (vect_loop_versioning): Build
19660 profitability check against LOOP_VINFO_NITERSM1.
19662 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19664 * domwalk.c (cmp_bb_postorder): Simplify.
19665 (sort_bbs_postorder): New function. Use it...
19666 (dom_walker::walk): ...here to optimize common cases.
19668 2017-07-25 Martin Liska <mliska@suse.cz>
19671 * ipa-visibility.c (function_and_variable_visibility): Make the
19672 redirection just on target that supports aliasing.
19673 Fix GNU coding style.
19675 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
19678 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
19680 * config/arm/rtems.h: Likewise.
19681 * config/bfin/rtems.h: Likewise.
19682 * config/i386/rtemself.h: Likewise.
19683 * config/lm32/rtems.h: Likewise.
19684 * config/m32c/rtems.h: Likewise.
19685 * config/m68k/rtemself.h: Likewise.
19686 * config/microblaze/rtems.h: Likewise.
19687 * config/mips/rtems.h: Likewise.
19688 * config/moxie/rtems.h: Likewise.
19689 * config/nios2/rtems.h: Likewise.
19690 * config/powerpcspe/rtems.h: Likewise.
19691 * config/rs6000/rtems.h: Likewise.
19692 * config/rtems.h: Likewise.
19693 * config/sh/rtems.h: Likewise.
19694 * config/sh/rtemself.h: Likewise.
19695 * config/sparc/rtemself.h: Likewise.
19697 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
19700 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
19701 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
19702 * tree-ssa-structalias.c (alias_get_name): Same.
19704 2017-07-25 Bin Cheng <bin.cheng@arm.com>
19707 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
19708 instructions if no du chain is found.
19710 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
19712 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
19714 2017-07-25 Richard Biener <rguenther@suse.de>
19716 PR middle-end/81505
19717 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
19720 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
19722 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
19723 upper-regs options.
19724 (ISA_2_7_MASKS_SERVER): Likewise.
19725 (ISA_3_0_MASKS_IEEE): Likewise.
19726 (OTHER_P8_VECTOR_MASKS): Likewise.
19727 (OTHER_VSX_VECTOR_MASKS): Likewise.
19728 (POWERPC_MASKS): Likewise.
19729 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
19730 duplicate list of options.
19731 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
19732 explicit -mupper-regs options.
19733 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
19734 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
19735 alias for -mupper-regs-df.
19736 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19737 (rs6000_init_hard_regno_mode_ok): Likewise.
19738 (rs6000_option_override_internal): Likewise.
19739 (rs6000_opt_masks): Likewise.
19740 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
19741 options in terms of whether -mvsx or -mpower8-vector was used.
19742 (TARGET_UPPER_REGS_DI): Likewise.
19743 (TARGET_UPPER_REGS_SF): Likewise.
19744 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
19745 -mupper-regs-* options.
19747 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
19749 * passes.c (emergency_dump_function): Print some empty lines and a
19750 header before the RTL dump.
19752 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
19754 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
19756 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
19759 * config/aarch64/aarch64.c (aarch64_classify_symbol):
19760 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
19762 2017-07-24 Carl Love <cel@us.ibm.com>
19764 * config/rs6000/rs6000-c.c: Add support for built-in functions
19765 vector float vec_extract_fp32_from_shorth (vector unsigned short);
19766 vector float vec_extract_fp32_from_shortl (vector unsigned short);
19767 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
19768 vec_extract_fp_from_shortl): Add defines for the two builtins.
19769 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
19770 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
19772 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
19773 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
19774 * doc/extend.texi: Update the built-in documentation file for the
19775 new built-in function.
19777 2017-07-24 Jakub Jelinek <jakub@redhat.com>
19780 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
19782 * doc/generic.texi: Likewise.
19783 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
19784 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
19786 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
19788 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
19789 (aarch64_mls_elt_merge<mode>): Likewise.
19791 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
19793 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
19794 having __cxa_atexit.
19796 2017-07-23 Michael Collison <michael.collison@arm.com>
19798 * config/arm/arm.c (arm_option_override): Deprecate
19799 use of -mstructure-size-boundary.
19800 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
19801 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
19803 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19806 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
19807 Reduce cost estimate for direct moves.
19809 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
19812 * config/i386/i386.c (ix86_option_override_internal): Disable
19813 BMI, BMI2 and TBM instructions for -m16.
19815 2017-07-21 Carl Love <cel@us.ibm.com>
19817 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19818 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19819 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19820 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19821 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19822 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19823 VMULOSW): New enum "unspec" values.
19824 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19825 altivec_vmulosw): New patterns.
19826 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19827 VMULOSW): Add definitions.
19829 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
19831 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
19832 (qdf24xx): Likewise.
19833 * config/aarch64/aarch64-options-extensions.def (rdma); New.
19834 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
19835 (AARCH64_FL_V8_1): Renumber.
19836 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
19837 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
19838 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
19839 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
19840 rdma to feature modifiers list.
19842 2017-07-21 Yury Gribov <tetra2005@gmail.com>
19844 PR middle-end/56727
19845 * ipa-visibility (function_and_variable_visibility): Convert
19846 recursive PLT call to direct call if appropriate.
19848 2017-07-21 Andrew Pinski <apinski@cavium.com>
19850 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
19851 operand 1 to see if the types precision matches.
19852 * fold-const.c (operand_equal_p): Likewise.
19854 2017-07-21 Richard Biener <rguenther@suse.de>
19856 PR tree-optimization/81303
19857 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
19858 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
19859 (vect_peeling_hash_get_lowest_cost): Adjust.
19860 (vect_enhance_data_refs_alignment): Likewise. Use
19861 vect_get_peeling_costs_all_drs to compute the penalty for no
19862 peeling to match up costs.
19864 2017-07-21 Richard Biener <rguenther@suse.de>
19866 PR tree-optimization/81500
19867 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
19868 we didn't identify a reduction path.
19870 2017-07-21 Tom de Vries <tom@codesourcery.com>
19871 Cesar Philippidis <cesar@codesourcery.com>
19873 PR gcov-profile/81442
19874 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
19877 2017-07-21 Tom de Vries <tom@codesourcery.com>
19880 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
19882 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
19883 nvptx_override_options_after_change.
19885 2017-07-21 Ulrich Drepper <drepper@redhat.com>
19887 * dwarf2out.c (output_file_names): Avoid double testing for
19888 dwarf_version >= 5.
19890 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
19892 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
19894 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
19896 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
19898 (try_crossjump_to_edge): Do not punt on partitioned functions.
19900 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
19902 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
19903 Put all BBs reachable only via paths crossing cold region to cold
19905 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
19907 2016-07-21 Richard Biener <rguenther@suse.de>
19909 PR tree-optimization/81303
19910 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
19911 into account prologue and epilogue iterations when raising
19912 min_profitable_iters to sth at least covering one vector iteration.
19914 2017-07-21 Tamar Christina <tamar.christina@arm.com>
19916 * config/arm/arm.c (arm_test_cpu_arch_dat):
19919 2017-07-20 Nathan Sidwell <nathan@acm.org>
19921 Remove TYPE_METHODS.
19922 * tree.h (TYPE_METHODS): Delete.
19923 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
19924 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
19925 (dbxout_type_methods): Scan TYPE_FIELDS.
19926 (dbxout_type): Don't check TYPE_METHODS here.
19927 * function.c (use_register_for_decl): Always ignore register for
19928 class types when not optimizing.
19929 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
19930 * tree.c (free_lang_data_in_type): Stitch out member functions and
19931 templates from TYPE_FIELDS.
19932 (build_distinct_type_copy, verify_type_variant,
19933 verify_type): Member fns are on TYPE_FIELDS.
19934 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
19935 * tree-pretty-print.c (dump_generic_node): Likewise.
19937 2017-07-20 Jakub Jelinek <jakub@redhat.com>
19940 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
19941 V2TImode and V4TImode.
19942 (ix86_expand_vector_extract): Likewise.
19943 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
19944 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
19945 (ssescalarmode): Handle V4TImode and V2TImode.
19946 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
19947 (*vec_extractv2ti, *vec_extractv4ti): New insns.
19948 (VEXTRACTI128_MODE): New mode iterator.
19949 (splitter for *vec_extractv?ti first element): New.
19950 (VEC_INIT_MODE): New mode iterator.
19951 (vec_init<mode>): Consolidate 3 expanders into one using
19952 VEC_INIT_MODE mode iterator.
19954 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
19956 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
19957 non_spilled_static_chain_regno_p.
19959 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
19961 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
19963 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
19965 * bb-reorder.c (connect_traces): Allow copying of blocks within
19968 2017-07-20 Richard Biener <rguenther@suse.de>
19970 * gimple.h (gimple_phi_result): Add gphi * overload.
19971 (gimple_phi_result_ptr): Likewise.
19972 (gimple_phi_arg): Likewise. Adjust index assert to only
19973 allow actual argument accesses rather than all slots available
19975 (gimple_phi_arg_def): Add gphi * overload.
19976 * tree-phinodes.c (make_phi_node): Initialize only actual
19978 (resize_phi_node): Clear memory not covered by old node,
19979 do not initialize excess argument slots.
19980 (reserve_phi_args_for_new_edge): Initialize new argument slot
19983 2017-07-20 Bin Cheng <bin.cheng@arm.com>
19985 PR tree-optimization/81388
19987 2016-07-21 Bin Cheng <bin.cheng@arm.com>
19989 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
19990 by removing computation of may_be_zero.
19992 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
19993 Tom de Vries <tom@codesourcery.com>
19995 PR middle-end/81030
19996 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
19997 when gimple level profile disagrees with what RTL expander did.
19999 2017-07-20 Richard Biener <rguenther@suse.de>
20001 PR tree-optimization/61171
20002 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
20003 (vect_analyze_stmt): Add slp instance parameter.
20004 (vectorizable_reduction): Likewise.
20005 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
20006 (vect_is_simple_reduction): Deal with chains not detected
20007 as SLP reduction chain, specifically not properly associated
20008 chains containing a mix of plus/minus.
20009 (get_reduction_op): Remove.
20010 (get_initial_defs_for_reduction): Simplify, pass in whether
20011 this is a reduction chain, pass in the SLP node for the PHIs.
20012 (vect_create_epilog_for_reduction): Get the SLP instance as
20014 (vectorizable_reduction): Get the SLP instance as arg.
20015 During analysis remember the SLP node with the PHIs in the
20016 instance. Simplify getting at the vectorized reduction PHIs.
20017 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
20018 through SLP instance.
20019 (vect_slp_analyze_operations): Likewise.
20020 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
20021 (vect_transform_stmt): Likewise.
20023 2017-07-20 Tom de Vries <tom@codesourcery.com>
20025 PR tree-optimization/81489
20026 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
20027 read of phi arg location to before loop that modifies phi.
20029 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20031 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
20034 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20036 PR middle-end/81331
20037 * except.c (execute): Fix ordering issue.
20039 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20041 PR rtl-optimization/81423
20042 * combine.c (make_compound_operation_int): Don't try to optimize
20043 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
20045 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20047 PR rtl-optimization/81423
20048 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
20049 with a constant that is -1 in the truncated to mode.
20051 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20053 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
20054 (determine_unlikely_bbs): ... here.
20055 * predict.h (propagate_unlikely_bbs_forward): Declare.
20056 * cfgexpand.c (pass_expand::execute): Use it.
20057 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
20059 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
20060 propagate_unlikely_bbs_forward.
20062 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20064 PR middle-end/81331
20065 * except.c (maybe_add_nop_after_section_switch): New function.
20068 2017-07-19 Tom de Vries <tom@codesourcery.com>
20070 * gimple.h (gimple_phi_set_arg): Make assert more strict.
20072 2017-07-19 Tom de Vries <tom@codesourcery.com>
20074 * gimple.h (gimple_phi_arg): Make assert more strict.
20076 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
20078 * config.gcc (powerpc*-*-*): Add mmintrin.h.
20079 * config/rs6000/mmintrin.h: New file.
20080 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
20082 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20084 PR tree-optimization/81346
20085 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
20087 2017-07-19 Tom de Vries <tom@codesourcery.com>
20089 * config/nvptx/nvptx.md (VECIM): Add V2DI.
20091 2017-07-19 Tom de Vries <tom@codesourcery.com>
20093 * config/nvptx/nvptx-modes.def: Add V2DImode.
20094 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
20095 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
20096 (nvptx_output_mov_insn): Handle lack of mov.b128.
20097 (nvptx_print_operand): Handle 'H' and 'L' codes.
20098 (nvptx_vector_mode_supported): Allow V2DImode.
20099 (nvptx_preferred_simd_mode): New function.
20100 (nvptx_data_alignment): New function.
20101 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
20102 nvptx_preferred_simd_mode.
20103 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
20105 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
20107 2017-07-19 Tom de Vries <tom@codesourcery.com>
20109 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
20110 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
20111 (nvptx_vector_mode_supported): New function. Allow V2SImode.
20112 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
20113 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
20114 (mov<VECIM>_insn): New define_insn.
20115 (define_expand "mov<VECIM>): New define_expand.
20117 2017-07-19 Tom de Vries <tom@codesourcery.com>
20119 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
20121 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20123 PR tree-optimization/81346
20124 * fold-const.h (fold_div_compare, range_check_type): Declare.
20125 * fold-const.c (range_check_type): New function.
20126 (build_range_check): Use range_check_type.
20127 (fold_div_compare): No longer static, rewritten into
20128 a match.pd helper function.
20129 (fold_comparison): Don't call fold_div_compare here.
20130 * match.pd (X / C1 op C2): New optimization using fold_div_compare
20131 as helper function.
20133 2017-07-19 Nathan Sidwell <nathan@acm.org>
20135 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
20136 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
20137 * tree.c (find_decls_types_r, verify_type): Use
20138 TYPE_{MIN,MAX}_VALUE_RAW.
20139 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
20140 (hash_tree): Likewise.
20141 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20143 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20146 2017-07-18 Tom de Vries <tom@codesourcery.com>
20148 PR middle-end/81464
20149 * omp-expand.c (expand_omp_for_static_chunk): Handle
20150 equal-argument loop exit phi.
20152 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
20155 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
20156 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
20157 operand 2 predicate.
20158 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
20159 operand 2 predicate.
20160 (ror,rol -> rorx splitters): Use const_int_operand as
20161 operand 2 predicate.
20163 2017-06-18 Richard Biener <rguenther@suse.de>
20165 PR tree-optimization/81410
20166 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
20167 the gap in the ! slp_perm SLP case after each group.
20169 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20171 PR middle-end/81463
20172 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
20175 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20177 PR middle-end/81462
20178 * predict.c (set_even_probabilities): Cleanup; do not affect
20179 probabilities that are already known.
20180 (combine_predictions_for_bb): Call even when count is set.
20182 2017-07-18 Nathan Sidwell <nathan@acm.org>
20184 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
20187 2017-07-18 Bin Cheng <bin.cheng@arm.com>
20190 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
20191 optimization for loop niter analysis.
20193 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
20196 * config/avr/avr.c (avr_optimize_casesi): Don't use
20197 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
20199 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
20201 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
20202 body_cost_vec from _vect_peel_extended_info.
20203 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
20204 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
20207 2017-07-18 Bin Cheng <bin.cheng@arm.com>
20209 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
20211 2017-07-18 Richard Biener <rguenther@suse.de>
20213 PR tree-optimization/80620
20214 PR tree-optimization/81403
20215 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
20216 info when re-using a VN table entry.
20218 2017-07-18 Richard Biener <rguenther@suse.de>
20220 PR tree-optimization/81418
20221 * tree-vect-loop.c (vectorizable_reduction): Properly compute
20222 vectype_in. Verify that with lane-reducing reduction operations
20223 we have a single def-use cycle.
20225 2017-07-17 Carl Love <cel@us.ibm.com>
20227 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
20229 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20230 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20231 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20232 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20233 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20234 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20235 VMULOSW): New enum "unspec" values.
20236 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20237 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20238 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20239 altivec_vmulosw): New patterns.
20240 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20241 VMULOSW): Add definitions.
20243 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
20245 * config/alpha/alpha.c: Include predict.h.
20247 2017-07-17 Yury Gribov <tetra2005@gmail.com>
20249 * tree-vrp.c (compare_assert_loc): Fix comparison function
20250 to return predictable results.
20252 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20254 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
20256 (subdi3): Likewise.
20257 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
20258 * doc/invoke.texi (mexpand-adddi): Update text.
20260 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20262 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
20263 that also clobbers the CC register. The old expand code is moved
20265 (*arc_clzsi2): ... here.
20266 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
20267 the CC register. The old expand code is moved to ...
20268 (arc_ctzsi2): ... here.
20270 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20272 * config/arc/arc.opt (mindexed-loads): Use initial value
20273 TARGET_INDEXED_LOADS_DEFAULT.
20274 (mauto-modify-reg): Use initial value
20275 TARGET_AUTO_MODIFY_REG_DEFAULT.
20276 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
20277 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
20278 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
20279 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
20281 2017-07-17 Martin Liska <mliska@suse.cz>
20284 * opts.c (finish_options): Do not allow -fgnu-tm
20285 w/ -fsanitize={kernel-,}address. Say sorry.
20287 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20290 * tree-loop-distribution.c (classify_partition): Only assert on
20291 numer of iterations.
20292 (merge_dep_scc_partitions): Delete prameter. Update function call.
20293 (distribute_loop): Remove code handling loop with unknown niters.
20294 (pass_loop_distribution::execute): Skip loop with unknown niters.
20296 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20299 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
20300 function sort_partitions_by_post_order.
20302 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20304 PR tree-optimization/81374
20305 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
20306 the max index of basic blocks, rather than number of basic blocks.
20308 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20310 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
20312 (arc_legitimate_pic_operand_p): Likewise.
20313 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
20315 (arc_needs_pcl_p): Likewise.
20316 (arc_legitimate_pc_offset_p): Likewise.
20317 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
20318 function is also used in constrains.md.
20319 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
20320 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
20321 PLUS. Only return true/false in known cases, otherwise assert.
20322 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
20323 is already called in arc_legitimate_constant_p.
20324 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
20326 (LEGITIMATE_PIC_OPERAND_P): Use
20327 arc_raw_symbolic_reference_mentioned_p function.
20328 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
20333 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20334 Andrew Burgess <andrew.burgess@embecosm.com>
20336 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
20337 (arc_return_address_register): New function.
20338 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
20339 (arc_handle_fndecl_attribute): Add naked attribute.
20340 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
20341 (TARGET_WARN_FUNC_RETURN): Likewise.
20342 (arc_allocate_stack_slots_for_args): New function.
20343 (arc_warn_func_return): Likewise.
20344 (machine_function): Change type fn_type.
20345 (arc_compute_function_type): Consider new naked function type,
20346 change function return type.
20347 (arc_must_save_register): Adapt to handle new
20348 arc_compute_function_type's return type.
20349 (arc_expand_prologue): Likewise.
20350 (arc_expand_epilogue): Likewise.
20351 (arc_return_address_regs): Delete.
20352 (arc_return_address_register): New function.
20353 (arc_epilogue_uses): Use above function.
20354 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
20355 (arc_function_type): Change encoding, add naked type.
20356 (ARC_INTERRUPT_P): Change to handle the new encoding.
20357 (ARC_FAST_INTERRUPT_P): Likewise.
20358 (ARC_NORMAL_P): Define.
20359 (ARC_NAKED_P): Likewise.
20360 (arc_compute_function_type): Delete prototype.
20361 * config/arc/arc.md (in_ret_delay_slot): Use
20362 arc_return_address_register function.
20363 (simple_return): Likewise.
20364 (p_return_i): Likewise.
20366 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20368 PR tree-optimization/81428
20369 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
20370 can't be built for those types.
20372 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
20374 Remove stuff dead since r239246.
20376 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
20377 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
20378 (avr_inform_devices): Remove dead stuff.
20380 2017-07-17 Tamar Christina <tamar.christina@arm.com>
20382 * config/arm/arm_neon.h: Fix softp typo.
20384 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20386 PR tree-optimization/81365
20387 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
20388 aggregate moves onto bb predecessor edges, make sure there are no
20389 loads that could alias the lhs in between the start of bb and the
20392 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
20395 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
20396 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
20397 [LSHIFTRT, outer_code = TRUNCATE]: Same.
20399 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20401 PR tree-optimization/81396
20402 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
20403 (init_symbolic_number): Initialize it to 1.
20404 (perform_symbolic_merge): Add n_ops from both operands into the new
20406 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
20407 without base_addr as useless if they need more than one operation.
20408 (bswap_replace): Handle !bswap case for NULL base_addr.
20410 2017-07-17 Tom de Vries <tom@codesourcery.com>
20413 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
20416 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
20418 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
20419 conditional builtin define __FIX_LEON3FT_B2BST.
20421 2017-07-17 Daniel Cederman <cederman@gaisler.com>
20423 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
20424 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
20427 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
20429 PR rtl-optimization/81424
20430 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
20431 to remove potential trapping from operands if -fnon-call-exceptions.
20433 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20435 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
20436 profile_proability for scalling.
20437 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
20439 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20441 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
20443 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20445 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
20446 fixpoint arithmetics.
20448 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20450 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
20451 fixpoint arithmetics.
20453 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20455 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
20456 fixpoint arithmetics.
20458 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20460 * profile-count.h (profile_probability::from_reg_br_prob_note,
20461 profile_probability::to_reg_br_prob_note): New functions.
20462 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
20463 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
20464 * predict.c (probability_reliable_p): Update.
20465 (edge_probability_reliable_p): Update.
20466 (br_prob_note_reliable_p): Update.
20467 (invert_br_probabilities): Update.
20468 (add_reg_br_prob_note): New function.
20469 (combine_predictions_for_insn): Update.
20470 * asan.c (asan_clear_shadow): Update.
20471 * cfgbuild.c (compute_outgoing_frequencies): Update.
20472 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
20473 (update_br_prob_note): Update.
20474 (rtl_verify_edges): Update.
20475 (purge_dead_edges): Update.
20476 (fixup_reorder_chain): Update.
20477 * emit-rtl.c (try_split): Update.
20478 * ifcvt.c (cond_exec_process_insns): Update.
20479 (cond_exec_process_if_block): Update.
20480 (dead_or_predicable): Update.
20481 * internal-fn.c (expand_addsub_overflow): Update.
20482 (expand_neg_overflow): Update.
20483 (expand_mul_overflow): Update.
20484 * loop-doloop.c (doloop_modify): Update.
20485 * loop-unroll.c (compare_and_jump_seq): Update.
20486 * optabs.c (emit_cmp_and_jump_insn_1): Update.
20487 * predict.h: Update.
20488 * reorg.c (mostly_true_jump): Update.
20490 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
20491 * config/alpha/alpha.c (emit_unlikely_jump): Update.
20492 * config/arc/arc.c: (emit_unlikely_jump): Update.
20493 * config/arm/arm.c: (emit_unlikely_jump): Update.
20494 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
20495 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
20496 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
20497 (ix86_print_operand): Update.
20498 (ix86_split_fp_branch): Update.
20499 (predict_jump): Update.
20500 * config/ia64/ia64.c (ia64_print_operand): Update.
20501 * config/mmix/mmix.c (mmix_print_operand): Update.
20502 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
20503 (rs6000_expand_split_stack_prologue): Update.
20504 * config/rs6000/rs6000.c: Update.
20505 * config/s390/s390.c (s390_expand_vec_strlen): Update.
20506 (s390_expand_vec_movstr): Update.
20507 (s390_expand_cs_tdsi): Update.
20508 (s390_expand_split_stack_prologue): Update.
20509 * config/sh/sh.c (sh_print_operand): Update.
20510 (expand_cbranchsi4): Update.
20511 (expand_cbranchdi4): Update.
20512 * config/sparc/sparc.c (output_v9branch): Update.
20513 * config/spu/spu.c (get_branch_target): Update.
20514 (ea_load_store_inline): Update.
20515 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
20516 * config/tilepro/tilepro.c: Update.
20518 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
20520 * gimplify.c (mostly_copy_tree_r): Revert latest change.
20521 (gimplify_save_expr): Likewise.
20523 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20525 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
20527 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20529 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
20531 * timevar.def (TV_IPA_FNSUMMARY): Define.
20533 2017-07-16 Daniel Cederman <cederman@gaisler.com>
20535 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
20536 to back store errata sensitive sequence from being generated.
20537 (sqrtdf2_fix): Likewise.
20539 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20541 * tree-ssa-threadupdate.c (compute_path_counts,
20542 update_joiner_offpath_counts): Use profile_probability.
20544 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
20547 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20549 * config/arm/arm-c.c (arm_cpu_builtins): Define
20550 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20552 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
20554 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20555 array entries to represent __ieee128 versions of the
20556 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
20557 scalar_extract_sig, and scalar_insert_exp built-in functions.
20558 (altivec_resolve_overloaded_builtin): Add special case handling
20559 for the __builtin_scalar_insert_exp function, as represented by
20560 the P9V_BUILTIN_VEC_VSIEDP constant.
20561 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
20562 exponent support for __ieee128 argument.
20563 (VSESQP): Add scalar extract signature support for __ieee128
20565 (VSTDCNQP): Add scalar test negative support for __ieee128
20567 (VSIEQP): Add scalar insert exponent support for __int128 argument
20568 with __ieee128 result.
20569 (VSIEQPF): Add scalar insert exponent support for __ieee128
20570 argument with __ieee128 result.
20571 (VSTDCQP): Add scalar test data class support for __ieee128
20573 (VSTDCNQP): Add overload support for scalar test negative with
20574 __ieee128 argument.
20575 (VSTDCQP): Add overload support for scalar test data class
20576 __ieee128 argument.
20577 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
20578 UNSPEC_VSX_SXSIGDP.
20579 (UNSPEC_VSX_SIEXPQP): New constant.
20580 (xsxexpqp): New insn for VSX scalar extract exponent quad
20582 (xsxsigqp): New insn for VSX scalar extract significand quad
20584 (xsiexpqpf): New insn for VSX scalar insert exponent quad
20585 precision with floating point argument.
20586 (xststdcqp): New expand for VSX scalar test data class quad
20588 (xststdcnegqp): New expand for VSX scalar test negative quad
20590 (xststdcqp): New insn to match expansions for VSX scalar test data
20591 class quad precision and VSX scalar test negative quad precision.
20592 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
20593 special case operand checking to enforce that second operand of
20594 VSX scalar test data class with quad precision argument is a 7-bit
20596 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20597 prototypes and descriptions of __ieee128 versions of
20598 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
20599 scalar_test_data_class, and scalar_test_neg built-in functions.
20601 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20603 PR tree-optimization/81162
20604 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
20605 replace a negate with an add.
20607 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
20609 * doc/invoke.texi (arm/-mcpu): Document +crypto.
20611 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20613 * config/arm/arm-c.c (arm_cpu_builtins): Define
20614 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20616 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20618 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
20619 (armv8-r): Set ARM Cortex-R52 as default CPU.
20620 * config/arm/arm-tables.opt: Regenerate.
20621 * config/arm/arm-tune.md: Regenerate.
20622 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
20624 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
20625 extension for -mcpu=cortex-r52.
20627 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20629 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
20630 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
20631 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
20632 (fp-armv8): Define it as FP_ARMv8 only.
20633 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
20634 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
20636 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
20637 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
20638 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
20639 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
20640 than TARGET_FPU_ARMV8.
20641 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
20642 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
20643 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
20645 * config/arm/neon.md (neon_vrint): Likewise.
20646 (neon_vcvt): Likewise.
20647 (neon_<fmaxmin_op><mode>): Likewise.
20648 (<fmaxmin><mode>3): Likewise.
20649 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
20650 * config/arm/predicates.md (arm_cond_move_operator): Check against
20651 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
20653 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
20655 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
20656 to top of function.
20658 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20660 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
20661 loop in comment with memset.
20663 2017-07-14 Martin Liska <mliska@suse.cz>
20665 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
20666 * dwarf2out.c (is_java): Remove the function.
20667 (output_pubname): Remove usage of the function.
20668 (lower_bound_default): Remove usage of DW_LANG_Java.
20669 (gen_compile_unit_die): Likewise.
20670 * gcc.c: Remove compiler defaults for .java and .zip files.
20671 * gimple-expr.c (remove_suffix): Change as there's no longer
20672 extension than 4-letter one.
20673 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
20674 (gimplify_save_expr): Likewise.
20675 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
20676 as it's possible even for other languages than Java.
20677 * langhooks.h (struct lang_hooks): Remove Java from a comment.
20678 * lto-opts.c (lto_write_options): Remove reference to Java.
20679 * opts.c (strip_off_ending): Update file extension handling.
20680 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
20681 * tree-eh.c (lower_resx): Likewise.
20682 * tree.c (free_lang_data_in_type): Remove dead code.
20683 (find_decls_types_r): Likewise.
20684 (build_common_builtin_nodes): Remove Java from a comment.
20685 (verify_type): Remove dead code.
20686 * varasm.c (assemble_external): Remove Java from a comment.
20688 2017-07-14 Martin Liska <mliska@suse.cz>
20690 * opts.c (finish_options): Add quotes.
20691 (common_handle_option): Likewise.
20693 2017-07-14 Martin Liska <mliska@suse.cz>
20695 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
20696 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
20697 Remove N_SO_PASCAL.
20698 * dwarf2out.c (lower_bound_default): Do not handle
20700 (gen_compile_unit_die): Likewise.
20701 * gcc.c: Remove default extension binding for GNU Pascal.
20702 * stmt.c: Remove Pascal language from a comment.
20703 * xcoffout.c: Likewise.
20705 2017-07-13 David Malcolm <dmalcolm@redhat.com>
20708 * diagnostic-show-locus.c (fixit_cmp): New function.
20709 (layout::layout): Sort m_fixit_hints.
20710 (column_range::column_range): Assert that the values are valid.
20711 (struct char_span): New struct.
20712 (correction::overwrite): New method.
20713 (struct source_line): New struct.
20714 (line_corrections::add_hint): Add assertions. Reimplement memcpy
20715 calls in terms of classes source_line and char_span, and
20716 correction::overwrite.
20717 (selftest::test_overlapped_fixit_printing_2): New function.
20718 (selftest::diagnostic_show_locus_c_tests): Call it.
20720 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
20722 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
20723 early if there is no lhs.
20725 2017-07-13 Martin Liska <mliska@suse.cz>
20727 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
20728 (gen_reference_type_die): Likewise.
20729 * stor-layout.c: Remove Pascal-related comment.
20731 2017-07-13 Martin Liska <mliska@suse.cz>
20733 * opts.c (finish_options): Add quotes to error messages.
20734 (parse_sanitizer_options): Likewise.
20736 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
20738 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
20740 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
20742 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
20744 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
20746 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
20748 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
20750 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
20753 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
20754 provides the hardware capability bits, define the macro
20755 __BUILTIN_CPU_SUPPORTS__.
20756 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
20757 if GLIBC does not provide the hardware capability bits. Add a
20758 gcc_unreachable call if the built-in cpu function is neither
20759 __builtin_cpu_is nor __builtin_cpu_supports.
20760 (rs6000_get_function_versions_dispatcher): Change the warning
20761 that an old GLIBC is used which does not export the capability
20762 bits to be an error.
20763 * doc/extend.texi (target_clones attribute): Document the
20764 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
20765 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
20766 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
20767 the macros defined by GCC if the newer GLIBC is available.
20769 2017-07-12 Jeff Law <law@redhat.com>
20771 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
20772 remaining includes slightly.
20773 * config/riscv/riscv-builtins.c: Include profile-count.h.
20775 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
20778 * config/avr/avr.c (avr_set_current_function): In diagnostic
20779 messages: Quote keywords and (parts of) identifiers.
20780 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
20783 2017-07-12 Carl Love <cel@us.ibm.com>
20785 * config/rs6000/rs6000-c.c: Add support for built-in functions
20786 vector bool char vec_revb (vector bool char);
20787 vector bool short vec_revb (vector short char);
20788 vector bool int vec_revb (vector bool int);
20789 vector bool long long vec_revb (vector bool long long);
20790 * doc/extend.texi: Update the built-in documentation file for the
20791 new built-in functions.
20793 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20795 * config/s390/s390.md: Remove movcc splitter.
20797 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20799 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
20800 load/store on condition.
20802 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
20805 * config/avr/avr.c (avr_encode_section_info)
20806 [progmem && !TREE_READONLY]: Error if progmem object needs
20809 2017-07-11 Michael Collison <michael.collison@arm.com>
20811 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
20814 2017-07-11 Carl Love <cel@us.ibm.com>
20816 * config/rs6000/rs6000-c.c: Add support for builtins
20817 vector unsigned int vec_parity_lsbb (vector signed int);
20818 vector unsigned int vec_parity_lsbb (vector unsigned int);
20819 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
20820 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
20821 vector unsigned long long vec_parity_lsbb (vector signed long long);
20822 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
20823 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
20824 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
20825 * doc/extend.texi: Update the built-in documentation file for the
20826 new built-in functions.
20828 2017-07-11 David Malcolm <dmalcolm@redhat.com>
20830 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
20831 (layout::m_primary_loc): New field.
20832 (layout::layout): Initialize new field. Move location filtering
20833 logic from here to...
20834 (layout::maybe_add_location_range): ...this new method. Add
20835 support for filtering to just the lines already specified by other
20837 (layout::will_show_line_p): New method.
20838 (gcc_rich_location::add_location_if_nearby): New method.
20839 (selftest::test_add_location_if_nearby): New test function.
20840 (selftest::diagnostic_show_locus_c_tests): Call it.
20841 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
20844 2017-07-11 Tom de Vries <tom@codesourcery.com>
20846 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
20847 (bb_first_real_insn): New function.
20848 (nvptx_single): Add extra initialization of broadcasted condition
20851 2017-07-11 Nathan Sidwell <nathan@acm.org>
20853 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
20855 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
20857 * doc/extend.texi (AVR Function Attributes): Remove weblink to
20858 Binutils doc as TEXI will mess them up.
20859 * doc/invoke.texi (AVR Options): Same here.
20861 2017-07-11 Daniel Cederman <cederman@gaisler.com>
20863 * config/sparc/sparc.opt (mfix-ut700): New option.
20864 (mfix-gr712rc): Likewise.
20865 (sparc_fix_b2bst): New variable.
20866 * doc/invoke.texi (SPARC options): Document them.
20867 (ARM options): Fix warnings.
20868 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
20869 instructions to prevent sequences that can trigger the store-store
20870 errata for certain LEON3FT processors.
20871 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
20872 (sparc_option_override): Set sparc_fix_b2bst appropriately.
20873 * config/sparc/sparc.md (fix_b2bst): New attribute.
20874 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
20876 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
20879 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
20881 (*rsqrtsf2_sse): Ditto.
20883 (div<mode>3): Macroize insn from divdf3 and divsf3
20884 using MODEF mode iterator.
20886 2017-07-10 Martin Sebor <msebor@redhat.com>
20888 PR tree-optimization/80397
20889 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
20890 instead of testing for equality to INTEGER_TYPE.
20892 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
20894 * config.gcc: Remove uclibc from arc target spec.
20896 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
20898 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
20900 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20903 * lto-wrapper.c (remove_option): New function.
20904 (merge_and_complain): Merge PIC/PIE options more realistically.
20906 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
20908 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
20912 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
20913 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
20914 * config.in: Regenerate.
20915 * configure: Regenerate.
20916 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
20917 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
20918 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
20919 (TARGET_GASISR_PROLOGUES): ...target mask.
20920 * common/config/avr/avr-common.c
20921 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
20922 Set -mgas-isr-prologues.
20923 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
20924 INSERT_PASS_BEFORE for it.
20925 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
20926 * config/avr/avr.c (avr_option_override)
20927 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
20928 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
20929 (avr_attribute_table) <no_gccisr>: Add new function attribute.
20930 (avr_set_current_function) <is_no_gccisr>: Init machine field.
20931 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
20933 (make_avr_pass_pre_proep): New function.
20934 (emit_push_sfr) <treg>: Add argument to function and use it
20935 instead of TMP_REG.
20936 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
20937 and set machine->gasisr.yes.
20938 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
20939 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
20940 __gcc_isr.n_pushed to .L__stack_usage.
20941 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
20942 (avr_asm_final_postscan_insn): ...this new static function.
20943 * config/avr/avr.h (machine_function)
20944 <is_no_gccisr, use_L__stack_usage>: New fields.
20945 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
20946 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
20947 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
20948 (gasisr, *gasisr): New expander and insn.
20949 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
20950 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
20951 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
20953 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
20955 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
20958 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
20960 Move jump-tables out of .text again.
20963 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
20964 (ASM_OUTPUT_ADDR_VEC): New function.
20965 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
20966 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
20967 INSN_ADDRESSes as asm comment.
20968 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
20969 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
20970 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
20971 * config/avr/avr.md (*tablejump): Adjust comment.
20972 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
20973 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
20975 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
20976 (avr_output_addr_vec): New proto.
20977 (avr_log_t) <insn_addresses>: New field.
20979 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
20982 * config/i386/i386.c (ix86_function_arg_advance): Set
20983 outgoing_args_on_stack to true if there are outgoing arguments
20985 (ix86_function_arg): Likewise.
20986 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
20987 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
20988 * config/i386/i386.h (machine_function): Add
20989 outgoing_args_on_stack.
20991 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
20993 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
20994 supporting pthreds.
20995 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
20997 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
20999 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
21000 (REAL_H): Remove $(MACHMODE_H).
21001 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
21003 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
21004 $(MACHMODE_H) and double-int.h.
21005 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
21007 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
21010 2017-07-07 Andrew Pinski <apinski@cavium.com>
21012 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
21013 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
21015 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21017 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
21018 Add warning if GCC was not configured to link against a GLIBC that
21019 exports the hardware capability bits.
21020 (make_resolver_func): Make resolver function private and not a
21021 COMDAT function. Create the name with clone_function_name instead
21022 of make_unique_name.
21025 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
21026 correct operand in doing the split.
21028 2017-07-07 Carl Love <cel@us.ibm.com>
21030 * config/rs6000/rs6000-c: Add support for built-in function
21031 vector unsigned short vec_pack_to_short_fp32 (vector float,
21033 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
21034 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
21035 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
21036 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
21037 (convert_4f32_8i16): Add define_expand.
21038 * doc/extend.texi: Update the built-in documentation file for the
21039 new built-in function.
21041 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21043 * config/sparc/m8.md: New file.
21044 * config/sparc/sparc.md: Include m8.md.
21046 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21048 * config/sparc/sparc.opt: New option -mvis4b.
21049 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
21050 (sparc_option_override): Handle VIS4B.
21051 (enum sparc_builtins): Define
21052 SPARC_BUILTIN_DICTUNPACK{8,16,32},
21053 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
21054 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
21055 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
21056 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
21057 (check_constant_argument): New function.
21058 (sparc_vis_init_builtins): Define builtins
21059 __builtin_vis_dictunpack{8,16,32},
21060 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
21061 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
21062 __builtin_vis_fpcmpde{8,16,32}shl and
21063 __builtin_vis_fpcmpur{8,16,32}shl.
21064 (sparc_expand_builtin): Check that the constant operands to
21065 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
21066 constant and in range.
21067 * config/sparc/sparc-c.c (sparc_target_macros): Handle
21069 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
21070 (SPARC_IMM5_P): Likewise.
21071 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
21072 (enabled): Handle vis4b.
21073 (UNSPEC_DICTUNPACK): New unspec.
21074 (UNSPEC_FPCMPSHL): Likewise.
21075 (UNSPEC_FPUCMPSHL): Likewise.
21076 (UNSPEC_FPCMPDESHL): Likewise.
21077 (UNSPEC_FPCMPURSHL): Likewise.
21078 (cpu_feature): New CPU feature `vis4b'.
21079 (dictunpack{8,16,32}): New insns.
21080 (FPCSMODE): New mode iterator.
21081 (fpcscond): New code iterator.
21082 (fpcsucond): Likewise.
21083 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
21084 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
21085 (fpcmpde{8,16,32}{si,di}shl): Likewise.
21086 (fpcmpur{8,16,32}{si,di}shl): Likewise.
21087 * config/sparc/constraints.md: Define constraints `q' for unsigned
21088 2-bit integer constants and `t' for unsigned 5-bit integer
21090 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
21092 (imm5_operand_dictunpack16): Likewise.
21093 (imm5_operand_dictunpack32): Likewise.
21094 (imm2_operand): Likewise.
21095 * doc/invoke.texi (SPARC Options): Document -mvis4b.
21096 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
21097 ditunpack* and fpcmp*shl builtins.
21099 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21101 * config.gcc: Handle m8 in --with-{cpu,tune} options.
21102 * config.in: Add HAVE_AS_SPARC6 define.
21103 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
21105 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
21107 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
21108 (CPP_CPU_SPEC): Handle m8.
21109 (ASM_CPU_SPEC): Likewise.
21110 * config/sparc/sparc-opts.h (enum processor_type): Add
21112 * config/sparc/sparc.c (m8_costs): New struct.
21113 (sparc_option_override): Handle TARGET_CPU_m8.
21114 (sparc32_initialize_trampoline): Likewise.
21115 (sparc64_initialize_trampoline): Likewise.
21116 (sparc_issue_rate): Likewise.
21117 (sparc_register_move_cost): Likewise.
21118 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
21119 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
21120 (ASM_CPU64_DEFAULT_SPEC): Likewise.
21121 (CPP_CPU_SPEC): Handle M8.
21122 (ASM_CPU_SPEC): Likewise.
21123 (AS_M8_FLAG): Define.
21124 * config/sparc/sparc.md: Add m8 to the cpu attribute.
21125 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
21126 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
21128 * configure: Regenerate.
21129 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
21132 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21134 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
21136 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
21137 ("*movdi_insn_sp32"): Do not set v3pipe.
21138 ("*movsi_insn"): Likewise.
21139 ("*movdi_insn_sp64"): Likewise.
21140 ("*movsf_insn"): Likewise.
21141 ("*movdf_insn_sp32"): Likewise.
21142 ("*movdf_insn_sp64"): Likewise.
21143 ("*zero_extendsidi2_insn_sp64"): Likewise.
21144 ("*sign_extendsidi2_insn"): Likewise.
21145 ("*mov<VM32:mode>_insn"): Likewise.
21146 ("*mov<VM64:mode>_insn_sp64"): Likewise.
21147 ("*mov<VM64:mode>_insn_sp32"): Likewise.
21148 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
21149 ("<vlop:code><VL:mode>3"): Likewise.
21150 ("*not_<vlop:code><VL:mode>3"): Likewise.
21151 ("*nand<VL:mode>_vis"): Likewise.
21152 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
21153 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
21154 ("one_cmpl<VL:mode>2"): Likewise.
21155 ("faligndata<VM64:mode>_vis"): Likewise.
21156 ("alignaddrsi_vis"): Likewise.
21157 ("alignaddrdi_vis"): Likweise.
21158 ("alignaddrlsi_vis"): Likewise.
21159 ("alignaddrldi_vis"): Likewise.
21160 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
21161 ("bmaskdi_vis"): Likewise.
21162 ("bmasksi_vis"): Likewise.
21163 ("bshuffle<VM64:mode>_vis"): Likewise.
21164 ("cmask8<P:mode>_vis"): Likewise.
21165 ("cmask16<P:mode>_vis"): Likewise.
21166 ("cmask32<P:mode>_vis"): Likewise.
21167 ("pdistn<P:mode>_vis"): Likewise.
21168 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
21170 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21172 * config/sparc/sparc.md ("subtype"): New insn attribute.
21173 ("*wrgsr_sp64"): Set insn subtype.
21174 ("*rdgsr_sp64"): Likewise.
21175 ("alignaddrsi_vis"): Likewise.
21176 ("alignaddrdi_vis"): Likewise.
21177 ("alignaddrlsi_vis"): Likewise.
21178 ("alignaddrldi_vis"): Likewise.
21179 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
21180 ("fexpand_vis"): Likewise.
21181 ("fpmerge_vis"): Likewise.
21182 ("faligndata<VM64:mode>_vis"): Likewise.
21183 ("bshuffle<VM64:mode>_vis"): Likewise.
21184 ("cmask8<P:mode>_vis"): Likewise.
21185 ("cmask16<P:mode>_vis"): Likewise.
21186 ("cmask32<P:mode>_vis"): Likewise.
21187 ("fchksm16_vis"): Likewise.
21188 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
21189 ("fmean16_vis"): Likewise.
21190 ("fp<plusminus_insn>64_vis"): Likewise.
21191 ("<plusminus_insn>v8qi3"): Likewise.
21192 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
21193 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
21194 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
21195 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
21196 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
21197 ("*movqi_insn"): Likewise.
21198 ("*movhi_insn"): Likewise.
21199 ("*movsi_insn"): Likewise.
21200 ("movsi_pic_gotdata_op"): Likewise.
21201 ("*movdi_insn_sp32"): Likewise.
21202 ("*movdi_insn_sp64"): Likewise.
21203 ("movdi_pic_gotdata_op"): Likewise.
21204 ("*movsf_insn"): Likewise.
21205 ("*movdf_insn_sp32"): Likewise.
21206 ("*movdf_insn_sp64"): Likewise.
21207 ("*zero_extendhisi2_insn"): Likewise.
21208 ("*zero_extendqihi2_insn"): Likewise.
21209 ("*zero_extendqisi2_insn"): Likewise.
21210 ("*zero_extendqidi2_insn"): Likewise.
21211 ("*zero_extendhidi2_insn"): Likewise.
21212 ("*zero_extendsidi2_insn_sp64"): Likewise.
21213 ("ldfsr"): Likewise.
21214 ("prefetch_64"): Likewise.
21215 ("prefetch_32"): Likewise.
21216 ("tie_ld32"): Likewise.
21217 ("tie_ld64"): Likewise.
21218 ("*tldo_ldub_sp32"): Likewise.
21219 ("*tldo_ldub1_sp32"): Likewise.
21220 ("*tldo_ldub2_sp32"): Likewise.
21221 ("*tldo_ldub_sp64"): Likewise.
21222 ("*tldo_ldub1_sp64"): Likewise.
21223 ("*tldo_ldub2_sp64"): Likewise.
21224 ("*tldo_ldub3_sp64"): Likewise.
21225 ("*tldo_lduh_sp32"): Likewise.
21226 ("*tldo_lduh1_sp32"): Likewise.
21227 ("*tldo_lduh_sp64"): Likewise.
21228 ("*tldo_lduh1_sp64"): Likewise.
21229 ("*tldo_lduh2_sp64"): Likewise.
21230 ("*tldo_lduw_sp32"): Likewise.
21231 ("*tldo_lduw_sp64"): Likewise.
21232 ("*tldo_lduw1_sp64"): Likewise.
21233 ("*tldo_ldx_sp64"): Likewise.
21234 ("*mov<VM32:mode>_insn"): Likewise.
21235 ("*mov<VM64:mode>_insn_sp64"): Likewise.
21236 ("*mov<VM64:mode>_insn_sp32"): Likewise.
21238 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21240 * config/sparc/sparc.md ("type"): New insn type viscmp.
21241 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
21243 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
21244 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
21245 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
21246 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
21248 ("n7_vis_logical_11cycle"): Likewise.
21249 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
21250 * config/sparc/niagara2.md ("niag3_vis": Likewise.
21251 * config/sparc/niagara.md ("niag_vis"): Likewise.
21252 * config/sparc/ultra3.md ("us3_fga"): Likewise.
21253 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
21255 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21257 * config/sparc/sparc.md: New instruction type `bmask'.
21258 (bmaskdi_vis): Use the `bmask' type.
21259 (bmasksi_vis): Likewise.
21260 * config/sparc/ultra3.md (us3_array): Likewise.
21261 * config/sparc/niagara7.md (n7_array): Likewise.
21262 * config/sparc/niagara4.md (n4_array): Likewise.
21263 * config/sparc/niagara2.md (niag2_vis): Likewise.
21264 (niag3_vis): Likewise.
21265 * config/sparc/niagara.md (niag_vis): Likewise.
21267 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21269 * ipa-comdats.c: Remove optimize check from gate.
21270 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
21271 for functions not optimized.
21272 (ipa_fn_summary_read): Skip optimize check.
21273 (ipa_fn_summary_write): Likewise.
21274 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
21276 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
21278 (can_inline_edge_p): Check flag_pcc_struct_return for match.
21279 (check_callers): Give up on caller which is not optimized.
21280 (inline_small_functions): Likewise.
21281 (ipa_inline): Do not give up when not optimizing.
21282 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
21283 away unoptimizes cdtors.
21284 (whole_program_function_and_variable_visibility): Do
21285 ipa_discover_readonly_nonaddressable_vars in LTO mode.
21286 * ipa.c (process_references): Do not check optimize.
21287 (symbol_table::remove_unreachable_nodes): Update optimize check.
21288 (set_writeonly_bit): Update optimize check.
21289 (pass_ipa_cdtor_merge::gate): Do not check optimize.
21290 (pass_ipa_single_use::gate): Remove.
21292 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21294 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
21295 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
21296 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
21297 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
21298 permute_load, permute_store, adjust_extract, adjust_splat,
21299 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
21300 replace_swap_with_copy, dump_swap_insn_table,
21301 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
21302 recombine_lvx_pattern, recombine_stvx_pattern,
21303 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
21304 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
21305 to file rs6000-p8swap.c.
21306 * config/rs6000/rs6000-p8swap.c: New file.
21307 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
21308 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
21309 and rs6000*-*-* targets.
21311 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21313 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
21315 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21317 * lto-wrapper.c (merge_and_complain): Do not merge
21318 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
21319 fsigned_zeros, ftrapping_math, fwrapv.
21320 (append_compiler_options): Do not track these options.
21321 (append_linker_options): Likewie
21323 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21325 * cgraphunit.c (cgraph_node::finalize_function): When
21326 !flag_toplevel_reorde set no_reorder flag.
21327 (varpool_node::finalize_decl): Likewise.
21328 (symbol_table::compile): Drop no toplevel reorder path.
21330 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21332 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
21333 edges; zero probability is not better than uninitialized.
21335 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
21337 * asan.h (asan_sanitize_allocas_p): Declare.
21338 * asan.c (asan_sanitize_allocas_p): New function.
21339 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
21340 (handle_builtin_alloca): Likewise.
21341 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
21342 if !asan_sanitize_allocas_p.
21343 * params.def (asan-instrument-allocas): Add new option.
21344 * params.h (ASAN_PROTECT_ALLOCAS): Define.
21345 * opts.c (common_handle_option): Disable allocas sanitization for
21348 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
21350 * asan.c: Include gimple-fold.h.
21351 (get_last_alloca_addr): New function.
21352 (handle_builtin_stackrestore): Likewise.
21353 (handle_builtin_alloca): Likewise.
21354 (asan_emit_allocas_unpoison): Likewise.
21355 (get_mem_refs_of_builtin_call): Add new parameter, remove const
21356 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
21357 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
21358 (instrument_builtin_call): Pass gimple iterator to
21359 get_mem_refs_of_builtin_call.
21360 (last_alloca_addr): New global.
21361 * asan.h (asan_emit_allocas_unpoison): Declare.
21362 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
21363 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
21364 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
21365 if function calls alloca.
21366 * gimple-fold.c (replace_call_with_value): Remove static keyword.
21367 * gimple-fold.h (replace_call_with_value): Declare.
21368 * internal-fn.c: Include asan.h.
21369 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
21370 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
21372 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21374 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
21375 (C_SELFTEST_FLAGS): New.
21376 (CPP_SELFTEST_FLAGS): New.
21377 (SELFTEST_DEPS): New, from deps of s-selftest.
21378 (C_SELFTEST_DEPS): New, from deps of s-selftest.
21379 (CPP_SELFTEST_DEPS): New.
21380 (selftest): Add dependency on s-selftest-c++.
21381 (s-selftest): Rename to...
21382 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
21383 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
21384 than SELFTEST_FLAGS.
21385 (selftest-gdb): Rename to...
21386 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
21388 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
21389 (selftest-valgrind): Rename to...
21390 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
21392 (selftest-valgrind): Reintroduce as an alias for
21393 selftest-c-valgrind.
21394 (s-selftest-c++): New.
21395 (selftest-c++-gdb): New.
21396 (selftest-c++-valgrind): New.
21398 2017-07-06 Olivier Hainque <hainque@adacore.com>
21400 * gcc.c (process_command): When deciding if undefined variables
21401 should be ignored when processing specs, accept "gcc -v" as well.
21403 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21405 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
21406 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
21408 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21410 * config/arm/arm-cpus.in (armv8-r): Add new entry.
21411 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
21412 * config/arm/arm-tables.opt: Regenerate.
21413 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
21415 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
21417 2017-07-06 Carl Love <cel@us.ibm.com>
21419 * ChangeLog: Clean up from mid air collision
21421 2017-07-06 Carl Love <cel@us.ibm.com>
21423 * config/rs6000/rs6000-c.c: Add support for built-in functions
21424 vector signed int vec_subc (vector signed int, vector signed int);
21425 vector signed __int128 vec_subc (vector signed __int128,
21426 vector signed __int128);
21427 vector unsigned __int128 vec_subc (vector unsigned __int128,
21428 vector unsigned __int128);
21429 vector signed int vec_sube (vector signed int, vector signed int,
21430 vector signed int);
21431 vector unsigned int vec_sube (vector unsigned int,
21432 vector unsigned int,
21433 vector unsigned int);
21434 vector signed __int128 vec_sube (vector signed __int128,
21435 vector signed __int128,
21436 vector signed__int128);
21437 vector unsigned __int128 vec_sube (vector unsigned __int128,
21438 vector unsigned __int128,
21439 vector unsigned __int128);
21440 vector signed int vec_subec (vector signed int, vector signed int,
21441 vector signed int);
21442 vector unsigned int vec_subec (vector unsigned int,
21443 vector unsigned int,
21444 vector unsigned int);
21445 vector signed __int128 vec_subec (vector signed __int128,
21446 vector signed __int128,
21447 vector signed__int128);
21448 vector unsigned __int128 vec_subec (vector unsigned __int128,
21449 vector unsigned __int128,
21450 vector unsigned __int128);
21451 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
21452 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
21453 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
21454 BU_ALTIVEC_OVERLOAD_X definitions.
21455 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
21456 * doc/extend.texi: Update the built-in documentation file for the new
21457 built-in functions.
21459 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21462 * diagnostic-show-locus.c (layout::layout): Use start and finish
21463 spelling location for the start and finish of each range.
21464 * genmatch.c (linemap_client_expand_location_to_spelling_point):
21465 Add unused aspect param.
21466 * input.c (expand_location_1): Add "aspect" param, and use it
21467 to access the correct part of the location.
21468 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
21470 (expand_location_to_spelling_point): Likewise.
21471 (linemap_client_expand_location_to_spelling_point): Add "aspect"
21472 param, and pass it to expand_location_1.
21474 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
21476 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
21477 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
21478 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
21479 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
21480 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
21481 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
21482 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
21483 _mm_maskz_getmant_ss): New intrinsics.
21484 (__builtin_ia32_getexpss128_mask): Changed to ...
21485 __builtin_ia32_getexpss128_round ... this.
21486 (__builtin_ia32_getexpsd128_mask): Changed to ...
21487 __builtin_ia32_getexpsd128_round ... this.
21488 * config/i386/i386-builtin-types.def
21489 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
21490 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
21491 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
21492 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
21493 __builtin_ia32_getmantss_mask_round): New builtins.
21494 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
21495 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
21496 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
21497 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
21498 * config/i386/sse.md
21499 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
21500 avx512f_sgetexp<mode><mask_scalar_name>
21501 <round_saeonly_scalar_name> ... this.
21502 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
21503 %0, %1, %2<round_saeonly_op3>}): Changed to ...
21504 vgetexp<ssescalarmodesuffix>
21505 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21506 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
21507 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
21508 avx512f_vgetmant<mode><mask_scalar_name>
21509 <round_saeonly_scalar_name> ... this.
21510 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
21511 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
21512 vgetmant<ssescalarmodesuffix>
21513 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
21514 %0<mask_scalar_operand4>, %1, %2
21515 <round_saeonly_scalar_mask_op4>, %3} ... this.
21516 * config/i386/subst.md (mask_scalar_operand4,
21517 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
21518 round_saeonly_scalar_nimm_predicate): New subst attributes.
21520 2017-07-06 Julia Koval <julia.koval@intel.com>
21522 * config/i386/i386.c (ix86_erase_embedded_rounding):
21523 Remove code for old rounding pattern.
21525 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
21527 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
21529 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
21531 * doc/sourcebuild.texi (Test Directives, Variants of
21532 dg-require-support): Add documentation for dg-require-stack-check.
21534 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
21536 * config/i386/subst.md (mask_scalar, round_scalar,
21537 round_saeonly_scalar): New meta-templates.
21538 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
21539 round_scalar_mask_operand3, round_scalar_mask_op3,
21540 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
21541 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
21542 round_saeonly_scalar_constraint,
21543 round_saeonly_scalar_prefix): New subst attribute.
21544 * config/i386/sse.md
21545 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
21546 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
21547 <round_scalar_name> ... this.
21548 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
21549 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
21550 <round_scalar_name> ... this.
21551 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
21552 <sse>_vm<code><mode>3<mask_scalar_name>
21553 <round_saeonly_scalar_name> ... this.
21554 (v<plusminus_mnemonic><ssescalarmodesuffix>
21555 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21556 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21557 v<plusminus_mnemonic><ssescalarmodesuffix>
21558 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21559 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21560 (v<multdiv_mnemonic><ssescalarmodesuffix>
21561 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21562 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21563 v<multdiv_mnemonic><ssescalarmodesuffix>
21564 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21565 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21566 (v<maxmin_float><ssescalarmodesuffix>
21567 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
21568 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
21569 v<maxmin_float><ssescalarmodesuffix>
21570 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21571 %0<mask_scalar_operand3>, %1, %<iptr>2
21572 <round_saeonly_scalar_mask_op3>} ... this.
21574 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
21576 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
21577 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
21579 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
21580 Alan Hayward <alan.hayward@arm.com>
21581 David Sherwood <david.sherwood@arm.com>
21583 * combine.c (simplify_if_then_else): Remove "enum" before
21585 * compare-elim.c (can_eliminate_compare): Likewise.
21586 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
21588 (aarch64_lookup_simd_builtin_type): Likewise.
21589 (aarch64_simd_builtin_type): Likewise.
21590 (aarch64_init_simd_builtin_types): Likewise.
21591 (aarch64_simd_expand_args): Likewise.
21592 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
21594 (aarch64_reverse_mask): Likewise.
21595 (aarch64_simd_emit_reg_reg_move): Likewise.
21596 (aarch64_gen_adjusted_ldpstp): Likewise.
21597 (aarch64_ccmp_mode_to_code): Likewise.
21598 (aarch64_operands_ok_for_ldpstp): Likewise.
21599 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
21600 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
21602 (aarch64_min_divisions_for_recip_mul): Likewise.
21603 (aarch64_reassociation_width): Likewise.
21604 (aarch64_get_condition_code_1): Likewise.
21605 (aarch64_simd_emit_reg_reg_move): Likewise.
21606 (aarch64_simd_attr_length_rglist): Likewise.
21607 (aarch64_reverse_mask): Likewise.
21608 (aarch64_operands_ok_for_ldpstp): Likewise.
21609 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
21610 (aarch64_gen_adjusted_ldpstp): Likewise.
21611 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
21613 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
21614 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
21615 (arm_lookup_simd_builtin_type): Likewise.
21616 (arm_simd_builtin_type): Likewise.
21617 (arm_init_simd_builtin_types): Likewise.
21618 (arm_expand_builtin_args): Likewise.
21619 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
21620 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
21621 (ft32_setup_incoming_varargs): Likewise.
21622 (ft32_function_arg): Likewise.
21623 (ft32_function_arg_advance): Likewise.
21624 (ft32_pass_by_reference): Likewise.
21625 (ft32_arg_partial_bytes): Likewise.
21626 (ft32_valid_pointer_mode): Likewise.
21627 (ft32_addr_space_pointer_mode): Likewise.
21628 (ft32_addr_space_legitimate_address_p): Likewise.
21629 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
21631 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
21632 (ix86_emit_outlined_ms2sysv_restore): Likewise.
21633 (iamcu_alignment): Likewise.
21634 (canonicalize_vector_int_perm): Likewise.
21635 (ix86_noce_conversion_profitable_p): Likewise.
21636 (ix86_mpx_bound_mode): Likewise.
21637 (ix86_operands_ok_for_move_multiple): Likewise.
21638 * config/microblaze/microblaze-protos.h
21639 (microblaze_expand_conditional_branch_reg): Likewise.
21640 * config/microblaze/microblaze.c
21641 (microblaze_expand_conditional_branch_reg): Likewise.
21642 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
21644 (rs6000_reassociation_width): Likewise.
21645 (rs6000_invalid_binary_op): Likewise.
21646 (fusion_p9_p): Likewise.
21647 (emit_fusion_p9_load): Likewise.
21648 (emit_fusion_p9_store): Likewise.
21649 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
21651 (riscv_hard_regno_mode_ok_p): Likewise.
21652 (riscv_address_insns): Likewise.
21653 (riscv_split_symbol): Likewise.
21654 (riscv_legitimize_move): Likewise.
21655 (riscv_function_value): Likewise.
21656 (riscv_hard_regno_nregs): Likewise.
21657 (riscv_expand_builtin): Likewise.
21658 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
21659 (riscv_build_integer): Likewise.
21660 (riscv_split_integer): Likewise.
21661 (riscv_legitimate_constant_p): Likewise.
21662 (riscv_cannot_force_const_mem): Likewise.
21663 (riscv_regno_mode_ok_for_base_p): Likewise.
21664 (riscv_valid_base_register_p): Likewise.
21665 (riscv_valid_offset_p): Likewise.
21666 (riscv_valid_lo_sum_p): Likewise.
21667 (riscv_classify_address): Likewise.
21668 (riscv_legitimate_address_p): Likewise.
21669 (riscv_address_insns): Likewise.
21670 (riscv_load_store_insns): Likewise.
21671 (riscv_force_binary): Likewise.
21672 (riscv_split_symbol): Likewise.
21673 (riscv_force_address): Likewise.
21674 (riscv_legitimize_address): Likewise.
21675 (riscv_move_integer): Likewise.
21676 (riscv_legitimize_const_move): Likewise.
21677 (riscv_legitimize_move): Likewise.
21678 (riscv_address_cost): Likewise.
21679 (riscv_subword): Likewise.
21680 (riscv_output_move): Likewise.
21681 (riscv_canonicalize_int_order_test): Likewise.
21682 (riscv_emit_int_order_test): Likewise.
21683 (riscv_function_arg_boundary): Likewise.
21684 (riscv_pass_mode_in_fpr_p): Likewise.
21685 (riscv_pass_fpr_single): Likewise.
21686 (riscv_pass_fpr_pair): Likewise.
21687 (riscv_get_arg_info): Likewise.
21688 (riscv_function_arg): Likewise.
21689 (riscv_function_arg_advance): Likewise.
21690 (riscv_arg_partial_bytes): Likewise.
21691 (riscv_function_value): Likewise.
21692 (riscv_pass_by_reference): Likewise.
21693 (riscv_setup_incoming_varargs): Likewise.
21694 (riscv_print_operand): Likewise.
21695 (riscv_elf_select_rtx_section): Likewise.
21696 (riscv_save_restore_reg): Likewise.
21697 (riscv_for_each_saved_reg): Likewise.
21698 (riscv_register_move_cost): Likewise.
21699 (riscv_hard_regno_mode_ok_p): Likewise.
21700 (riscv_hard_regno_nregs): Likewise.
21701 (riscv_class_max_nregs): Likewise.
21702 (riscv_memory_move_cost): Likewise.
21703 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
21704 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
21705 (rl78_addr_space_address_mode): Likewise.
21706 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21708 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
21709 (rs6000_reassociation_width): Likewise.
21710 (rs6000_invalid_binary_op): Likewise.
21711 (fusion_p9_p): Likewise.
21712 (emit_fusion_p9_load): Likewise.
21713 (emit_fusion_p9_store): Likewise.
21714 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
21715 (ok_for_simple_move_operands): Likewise.
21716 (ok_for_simple_move_strict_operands): Likewise.
21717 (ok_for_simple_arith_logic_operands): Likewise.
21718 (visium_legitimize_reload_address): Likewise.
21719 (visium_select_cc_mode): Likewise.
21720 (output_cbranch): Likewise.
21721 (visium_split_double_move): Likewise.
21722 (visium_expand_copysign): Likewise.
21723 (visium_expand_int_cstore): Likewise.
21724 (visium_expand_fp_cstore): Likewise.
21725 * config/visium/visium.c (visium_pass_by_reference): Likewise.
21726 (visium_function_arg): Likewise.
21727 (visium_function_arg_advance): Likewise.
21728 (visium_libcall_value): Likewise.
21729 (visium_setup_incoming_varargs): Likewise.
21730 (visium_legitimate_constant_p): Likewise.
21731 (visium_legitimate_address_p): Likewise.
21732 (visium_legitimize_address): Likewise.
21733 (visium_secondary_reload): Likewise.
21734 (visium_register_move_cost): Likewise.
21735 (visium_memory_move_cost): Likewise.
21736 (prepare_move_operands): Likewise.
21737 (ok_for_simple_move_operands): Likewise.
21738 (ok_for_simple_move_strict_operands): Likewise.
21739 (ok_for_simple_arith_logic_operands): Likewise.
21740 (visium_function_value_1): Likewise.
21741 (rtx_ok_for_offset_p): Likewise.
21742 (visium_legitimize_reload_address): Likewise.
21743 (visium_split_double_move): Likewise.
21744 (visium_expand_copysign): Likewise.
21745 (visium_expand_int_cstore): Likewise.
21746 (visium_expand_fp_cstore): Likewise.
21747 (visium_split_cstore): Likewise.
21748 (visium_select_cc_mode): Likewise.
21749 (visium_split_cbranch): Likewise.
21750 (output_cbranch): Likewise.
21751 (visium_print_operand_address): Likewise.
21752 * expmed.c (flip_storage_order): Likewise.
21753 * expmed.h (emit_cstore): Likewise.
21754 (flip_storage_order): Likewise.
21755 * genrecog.c (validate_pattern): Likewise.
21756 * hsa-gen.c (gen_hsa_addr): Likewise.
21757 * internal-fn.c (expand_arith_overflow): Likewise.
21758 * ira-color.c (allocno_copy_cost_saving): Likewise.
21759 * lra-assigns.c (find_hard_regno_for_1): Likewise.
21760 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
21761 (process_invariant_for_inheritance): Likewise.
21762 * lra-eliminations.c (move_plus_up): Likewise.
21763 * omp-low.c (lower_oacc_reductions): Likewise.
21764 * simplify-rtx.c (simplify_subreg): Likewise.
21765 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
21766 (TARGET_CHKP_BOUND_MODE): Likewise..
21767 * targhooks.c (default_chkp_bound_mode): Likewise.
21768 (default_setup_incoming_vararg_bounds): Likewise.
21769 * targhooks.h (default_chkp_bound_mode): Likewise.
21770 (default_setup_incoming_vararg_bounds): Likewise.
21771 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
21772 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
21773 (have_whole_vector_shift): Likewise.
21774 * tree-vect-stmts.c (vectorizable_load): Likewise.
21775 * doc/tm.texi: Regenerate.
21777 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
21779 Graceful degrade if Binutils PR21472 is not available.
21782 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
21783 .rodata in flash test fails.
21784 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
21785 * confgure: Regenerate.
21786 * config.in: Regenerate.
21787 * config/avr/avr.c (avr_asm_named_section)
21788 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
21789 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
21790 (avr_asm_init_sections): Same.
21792 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21794 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
21795 (fma<VH:mode>4_intrinsic): Likewise.
21796 (*fmsub<VCVTF:mode>4): Likewise.
21797 (*fmsub<VH:mode>4_intrinsic): Likewise.
21799 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
21802 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
21803 Don't depend on "optimize > 0".
21804 (out_movhi_r_mr, out_movqi_mr_r): Same.
21805 (out_movhi_mr_r, out_movqi_r_mr): Same.
21806 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
21807 io_address_operand on "optimize > 0".
21809 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21811 * tree-loop-distribution.c: Add general explanantion on the pass.
21812 (generate_loops_for_partition): Mark distributed loop.
21813 (pg_add_dependence_edges): New parameter. Handle alias data
21814 dependence specially and record it in the parameter if asked.
21815 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
21816 (init_partition_graph_vertices, add_partition_graph_edge): New.
21817 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
21818 (free_partition_graph_vdata, build_partition_graph): New.
21819 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
21820 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
21821 (data_ref_segment_size, latch_dominated_by_data_ref): New.
21822 (compute_alias_check_pairs, version_loop_by_alias_check): New.
21823 (version_for_distribution_p, finalize_partitions): New.
21824 (distribute_loop): Handle alias data dependence specially. Factor
21825 out loop fusion code as functions and call these functions.
21827 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21829 * tree-loop-distribution.c (classify_partition): New parameter and
21830 better handle reduction statement.
21831 (rdg_build_partitions): Revise comment.
21832 (distribute_loop): Compute statements in all partitions and pass it
21833 to classify_partition.
21835 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21837 * tree-loop-distribution.c (enum partition_type): New.
21838 (struct partition): New field type.
21839 (partition_merge_into): Add parameter. Update partition type.
21840 (data_dep_in_cycle_p, update_type_for_merge): New functions.
21841 (build_rdg_partition_for_vertex): Compute partition type.
21842 (rdg_build_partitions): Dump partition type.
21843 (distribute_loop): Update calls to partition_merge_into.
21845 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21847 * tree-loop-distribution.c (struct ddr_hasher): New.
21848 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
21850 (classify_partition): Call get_data_dependence.
21851 (pg_add_dependence_edges): Ditto.
21852 (distribute_loop): Release data dependence hash table.
21854 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21856 * tree-loop-distribution.c (ref_base_address): Delete.
21857 (similar_memory_accesses): Rename ...
21858 (share_memory_accesses): ... to this. Check if partitions access
21859 the same memory reference.
21860 (distribute_loop): Call share_memory_accesses.
21862 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21864 * tree-loop-distribution.c (struct partition): New field recording
21865 its data reference.
21866 (partition_alloc, partition_free): Init and release data refs.
21867 (partition_merge_into): Merge data refs.
21868 (build_rdg_partition_for_vertex): Collect data refs for partition.
21869 (pg_add_dependence_edges): Change parameters from vector to bitmap.
21871 (distribute_loop): Remve data refs from vertice data of partition
21874 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21876 * tree-loop-distribution.c (params.h): Include header file.
21877 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
21878 (datarefs_vec): New global var.
21879 (create_rdg_vertices): Use datarefs_vec directly.
21880 (free_rdg): Don't free data references.
21881 (build_rdg): Update use. Don't free data references.
21882 (distribute_loop): Compute global variable for data references.
21883 Bail out if there are too many data references.
21885 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21887 * tree-loop-distribution.c (loop_nest): New global var.
21888 (build_rdg): Use loop directly, rather than loop nest.
21889 (pg_add_dependence_edges): Remove loop nest parameter. Use global
21891 (distribute_loop): Compute global variable loop nest. Update use.
21893 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21895 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
21896 (partition_merge_into): New parameter. Dump reason for fusion.
21897 (distribute_loop): Update use of partition_merge_into.
21899 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21901 * tree-loop-distribution.c (bb_top_order_index): New.
21902 (bb_top_order_index_size, bb_top_order_cmp): New.
21903 (stmts_from_loop): Use topological order.
21904 (pass_loop_distribution::execute): Compute and release topological
21905 order for basic blocks.
21907 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21909 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
21912 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21914 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
21915 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
21916 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
21917 * internal-fn.def (LOOP_DIST_ALIAS): New.
21918 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
21919 (fold_loop_internal_call): ... this.
21920 (vect_loop_dist_alias_call): New function.
21921 (set_uid_loop_bbs): Call fold_loop_internal_call.
21922 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
21925 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
21928 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
21929 Require dead FLAGS_REG at the beginning of a peephole.
21931 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
21934 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
21935 arguments in the call to __builtin_ia32_sbb_u32.
21936 (_subborrow_u64): Swap _X and _Y arguments in the call to
21937 __builtin_ia32_sbb_u64.
21939 2017-07-04 Jakub Jelinek <jakub@redhat.com>
21942 * tree-vrp.c (compare_assert_loc): Turn into a function template
21943 with stable template parameter. Only test if a->e is NULL,
21944 !a->e == !b->e has been verified already. Use e == NULL or
21945 e != NULL instead of e or ! e tests. If stable is true, don't use
21946 iterative_hash_expr, on the other side allow a or b or both NULL
21947 and sort the NULLs last.
21948 (process_assert_insertions): Sort using compare_assert_loc<false>
21949 instead of compare_assert_loc, later sort using
21950 compare_assert_loc<true> before calling process_assert_insertions_for
21951 in a loop. Use break instead of continue once seen NULL pointer.
21953 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21955 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
21956 Cortex-R7 and Cortex-R8 processors.
21958 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
21960 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
21961 uninitialized while src is not.
21963 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
21965 * common/config/arm/arm-common.c: Adjust include path for
21967 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
21968 (arm-cpu.h): Create in build directory. Adjust dependency rules.
21969 (arm-cpu-data.h): Likewise.
21970 (arm-cpu-cdata.h): Likewise.
21971 * config/arm/arm-cpu.h: Delete.
21972 * config/arm/arm-cpu-cdata.h: Delete.
21973 * config/arm/arm-cpu-data.h: Delete.
21975 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
21977 * config/arm/arm-cpus.in (cortex-a55): New.
21978 (cortex-a75): Likewise.
21979 (cortex-a75.cortex-a55): Likewise.
21980 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
21982 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
21983 * config/arm/arm-cpu-cdata.h: Regenerate.
21984 * config/arm/arm-cpu-data.h: Regenerate.
21985 * config/arm/arm-cpu.h: Regenerate.
21986 * config/arm/arm-tables.opt: Regenerate.
21987 * config/arm/arm-tune.md: Regenerate.
21989 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
21991 * haifa-sched.c (sched_create_recovery_edges): Update profile.
21993 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
21995 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
21998 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
22000 PR tree-optimization/81292
22001 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
22002 full_string_p, also call adjust_related_strinfos if the adjustment
22003 is simple, otherwise invalidate related strinfos.
22005 2017-07-04 Martin Liska <mliska@suse.cz>
22008 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
22009 newly created variable as DECL_IGNORED_P.
22011 2017-07-04 Martin Liska <mliska@suse.cz>
22014 * ipa-inline.c (inline_small_functions):
22015 Use xstrdup_for_dump.
22017 2017-07-04 Tom de Vries <tom@codesourcery.com>
22019 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
22021 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22024 * config/darwin.c (darwin_function_switched_text_sections):
22027 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
22029 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
22031 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22033 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
22035 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22037 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
22038 min_profitable_iters, and th as inclusive lower bounds.
22039 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
22040 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
22041 for min_profitable_iters and min_profitable_estimate.
22042 (vect_transform_loop): Treat th as an inclusive lower bound.
22043 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22045 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22048 * config/darwin.c (darwin_function_switched_text_sections):
22049 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
22050 in two pieces, and suppress the use of buf.
22052 2017-07-03 Nathan Sidwell <nathan@acm.org>
22054 * hash-table.h (hash_table_mod1): Fix indentation.
22056 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22058 PR middle-end/81290
22059 * predict.c (force_edge_cold): Be more careful about propagation
22061 * profile-count.h (profile_probability::guessed,
22062 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
22064 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
22066 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
22068 * doc/invoke.texi (rcpc architecture extension): Document it.
22070 2017-07-03 Richard Biener <rguenther@suse.de>
22072 PR tree-optimization/60510
22073 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
22074 the scalar reduction PHI and use it.
22075 (vectorizable_reduction): Properly guard the single_defuse_cycle
22076 path for non-SLP reduction chains where we cannot use it.
22077 Rework reduc_def/index and vector type deduction. Rework
22078 vector operand gathering during reduction op code-gen.
22079 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
22080 chains dissolve the chain and leave it to non-SLP reduction
22083 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22085 * tree-data-ref.h (dr_alignment): Declare.
22086 * tree-data-ref.c (dr_alignment): New function.
22087 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
22088 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
22090 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
22092 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22094 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
22095 and base_misalignment fields.
22096 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
22097 * tree-data-ref.c: Include builtins.h.
22098 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
22099 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
22100 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
22101 * tree-vect-data-refs.c: Include tree-cfg.h.
22102 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
22103 fields instead of calculating an alignment here.
22104 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
22105 innermost_loop_behavior fields.
22107 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22109 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
22111 (DR_STEP_ALIGNMENT): New macro.
22112 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
22113 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
22114 (create_data_ref): Print it.
22115 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
22116 to tell whether the step preserves vector (mis)alignment.
22117 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
22118 Move the check for an integer step and generalise to all INTEGER_CST.
22119 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
22120 Print the outer step alignment.
22122 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22124 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
22125 with offset_alignment.
22126 (DR_ALIGNED_TO): Delete.
22127 (DR_OFFSET_ALIGNMENT): New macro.
22128 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
22129 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
22130 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
22131 (create_data_ref): Likewise.
22132 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
22133 (vect_analyze_data_refs): Likewise.
22134 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
22135 creating dummy innermost behavior.
22137 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22139 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
22140 with a "innermost_loop_behavior *" and refeence tree.
22141 * tree-data-ref.c (dr_analyze_innermost): Likewise.
22142 (create_data_ref): Update call accordingly.
22143 * tree-predcom.c (find_looparound_phi): Likewise.
22145 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22147 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
22148 fields with dr_wrt_vec_loop.
22149 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
22150 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
22151 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
22152 (vect_dr_behavior): New function.
22153 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
22154 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
22155 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
22156 track whether the step preserves the misalignment.
22157 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
22158 Use vect_dr_behavior.
22159 (vect_setup_realignment): Update call accordingly.
22160 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
22161 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
22162 call to vect_create_addr_base_for_vector_ref.
22163 (vect_create_cond_for_align_checks): Likewise.
22164 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
22165 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
22166 (vect_recog_mask_conversion_pattern): Likewise.
22167 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
22168 (new_stmt_vec_info): Remove redundant zeroing.
22170 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22172 * common/config/arm/arm-common.c (arm_be8_option): New function.
22173 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
22174 (ISA_ARMv6): Add isa_bit_be8.
22175 * config/arm/arm.h (arm_be8_option): Add prototype.
22176 (BE8_SPEC_FUNCTION): New define.
22177 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
22178 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
22179 (mlittle-endian): Similarly.
22180 (mbe8, mbe32): New options.
22181 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
22182 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
22184 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22186 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
22188 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22190 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
22191 (cleanup_tree_cfg_bb): Use it.
22192 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
22194 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
22196 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22199 * loop-doloop.c (add_test): Update profile.
22201 2017-07-03 Martin Liska <mliska@suse.cz>
22204 * sanopt.c (rewrite_usage_of_param): New function.
22205 (sanitize_rewrite_addressable_params): Likewise.
22206 (pass_sanopt::execute): Call rewrite_usage_of_param.
22208 2017-07-03 Richard Biener <rguenther@suse.de>
22210 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
22211 back to using VIEW_CONVERT_EXPR.
22213 2017-07-03 Martin Liska <mliska@suse.cz>
22216 * doc/extend.texi: Document when a resolver function is
22217 generated for target_clones.
22219 2017-07-03 Martin Liska <mliska@suse.cz>
22221 * asan.c (asan_emit_stack_protection): Unpoison just red zones
22222 and shadow memory of auto variables which are subject of
22223 use-after-scope sanitization.
22224 (asan_expand_mark_ifn): Add do set only when is_poison.
22226 2016-07-03 Richard Biener <rguenther@suse.de>
22228 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
22230 (vect_force_simple_reduction): Record reduction def -> phi mapping.
22231 (vectorizable_reduction): Perform reduction PHI creation when
22232 visiting a reduction PHI and adjust and simplify code generation
22233 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
22234 (vect_transform_loop): Visit reduction PHIs.
22235 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
22236 defs into the SLP tree.
22237 (vect_build_slp_tree): Reduction defs terminate the recursion.
22238 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
22240 (vect_get_vec_defs_for_stmt_copy): Export.
22241 (vect_get_vec_defs): Likewise.
22242 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
22244 (vect_get_vec_defs_for_stmt_copy): Declare.
22245 (vect_get_vec_defs): Likewise.
22247 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22249 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
22250 parameter with a "loop" parameter and use it instead of the
22251 loop containing DR_STMT. Don't check simple_iv when doing
22252 BB analysis. Describe the two analysis modes in the comment.
22254 2017-07-03 Tom de Vries <tom@codesourcery.com>
22256 PR tree-optimization/69468
22257 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
22258 (find_same_succ_bb): Handle ignore_edge_flags.
22260 2017-07-03 Tom de Vries <tom@codesourcery.com>
22262 PR tree-optimization/81192
22263 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
22265 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
22267 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
22269 2017-07-03 Tom de Vries <tom@codesourcery.com>
22271 PR tree-optimization/81192
22272 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
22273 BB_SAME_SUCC (bb) == NULL.
22275 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22277 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
22280 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22282 * dumpfile.c: Include profile-count.h
22283 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
22285 (insert_cond_bb): Update profile.
22286 * tree-cfg.h (insert_cond_bb): Update prototype.
22287 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
22288 * tree-dump.c: Do not include tree-cfg.
22290 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22292 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
22294 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22296 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
22299 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22301 * tree-complex.c (expand_complex_div_wide): update profile.
22303 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22304 Alan Hayward <alan.hayward@arm.com>
22305 David Sherwood <david.sherwood@arm.com>
22307 * Makefile.in (MACHMODE_H): Remove insn-modes.h
22308 (CORETYPES_H): New define.
22309 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
22310 (insn-modes-inline.h, s-modes-inline-h): New rules.
22311 (generated_files): Add insn-modes-inline.h.
22312 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
22313 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
22314 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
22315 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
22316 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
22317 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
22318 (build/gencodes.o, build/genconditions.o): Likewise.
22319 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
22320 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
22321 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
22322 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
22323 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
22324 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
22325 * coretypes.h: Include everything up to real.h for generators.
22326 Include insn-modes.h first. Include wide-int-print.h after
22327 wide-int.h. Include insn-modes-inline.h and then machmode.h.
22328 * machmode.h: Don't include insn-modes.h here.
22329 * function-tests.c: Remove includes of signop.h, machmode.h,
22330 double-int.h and wide-int.h.
22332 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
22334 * optc-save-gen.awk: Likewise.
22335 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
22336 * godump.c: Remove include of wide-int-print.h.
22337 * pretty-print.h: Likewise.
22338 * wide-int-print.cc: Likewise.
22339 * wide-int.cc: Likewise.
22340 * hash-map-tests.c: Remove include of signop.h.
22341 * hash-set-tests.c: Likewise.
22342 * rtl-tests.c: Likewise.
22343 * mkconfig.sh: Remove include of machmode.h.
22344 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
22346 (emit_insn_modes_inline_h): ...this new function. Emit the code
22347 into an insn-modes-inline.h header file, adding appropriate
22348 include guards and end comments.
22349 (emit_insn_modes_c_header): Remove include of machmode.h.
22350 (emit_min_insn_modes_c_header): Include coretypes.h rather than
22352 (main): Handle -i flag and call emit_insn_modes_inline_h when
22355 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22357 * tree-ssa-strlen.c (strinfo): Rename the length field to
22358 nonzero_chars. Add a full_string_p field.
22359 (compare_nonzero_chars, zero_length_string_p): New functions.
22360 (get_addr_stridx): Add an offset_out parameter.
22361 Use compare_nonzero_chars.
22362 (get_stridx): Update accordingly. Use compare_nonzero_chars.
22363 (new_strinfo): Update after above changes to strinfo.
22364 (set_endptr_and_length): Set full_string_p.
22365 (get_string_length): Update after above changes to strinfo.
22366 (unshare_strinfo): Update call to new_strinfo.
22367 (maybe_invalidate): Likewise.
22368 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
22369 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
22370 as a uhwi instead of an shwi. Update after above changes to
22371 strinfo and new_strinfo.
22372 (zero_length_string): Assert that chainsi contains full strings.
22373 Use zero_length_string_p. Update call to new_strinfo.
22374 (adjust_related_strinfos): Update after above changes to strinfo.
22375 Copy full_string_p from origsi.
22376 (adjust_last_stmt): Use zero_length_string_p.
22377 (handle_builtin_strlen): Update after above changes to strinfo and
22378 new_strinfo. Install the lhs as the string length if the previous
22379 entry didn't describe a full string.
22380 (handle_builtin_strchr): Update after above changes to strinfo
22382 (handle_builtin_strcpy): Likewise.
22383 (handle_builtin_strcat): Likewise.
22384 (handle_builtin_malloc): Likewise.
22385 (handle_pointer_plus): Likewise.
22386 (handle_builtin_memcpy): Likewise. Track nonzero characters
22387 that aren't necessarily followed by a nul terminator.
22388 (handle_char_store): Likewise.
22390 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22392 PR tree-optimization/80769
22393 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
22394 for malloc and calloc. Document the new invariant that all related
22395 strinfos have delayed lengths or none do.
22396 (verify_related_strinfos): Move earlier in file.
22397 (set_endptr_and_length): New function, split out from...
22398 (get_string_length): ...here. Also set the lengths of related
22400 (zero_length_string): Assert that chainsi has known (rather than
22402 (adjust_related_strinfos): Likewise.
22404 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22406 PR tree-optimization/81136
22407 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
22408 assert that two references with the same misalignment have the same
22409 compile-time misalignment if those compile-time misalignments
22412 2017-07-01 Andi Kleen <ak@linux.intel.com>
22414 * print-tree.c (print_node): Print all attributes.
22416 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22418 * cfg.c (scale_bbs_frequencies): New function.
22419 * cfg.h (scale_bbs_frequencies): Declare it.
22420 * cfgloopanal.c (single_likely_exit): Cleanup.
22421 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
22423 (scale_loop_profile): Likewise.
22424 (loop_version): Likewise.
22425 (create_empty_loop_on_edge): Update.
22426 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
22427 scale_loop_frequencies, scale_loop_profile, loopify,
22428 loop_version): Update prototypes.
22429 * modulo-sched.c (sms_schedule): Update.
22430 * predict.c (unlikely_executed_edge_p): Also check probability.
22431 (probably_never_executed_edge_p): Fix typo.
22432 * tree-if-conv.c (version_loop_for_if_conversion): Update.
22433 * tree-parloops.c (gen_parallel_loop): Update.
22434 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
22435 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22436 * tree-ssa-loop-split.c (split_loop): Update.
22437 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
22438 * tree-vect-loop-manip.c (vect_do_peeling): Update.
22439 (vect_loop_versioning): Update.
22440 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22442 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22444 * trans-mem.c (split_bb_make_tm_edge): Update profile.
22446 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22448 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
22449 to keep profile consistent.
22451 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22453 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
22454 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
22455 * profile-count.h (max_safe_multiplier): Make unsigned.
22456 (profile_count::guessed_zero): New.
22458 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22460 * bb-reorder.c (fix_up_crossing_landing_pad,
22461 fix_crossing_conditional_branches): Use make_single_succ_edge
22462 to keep profile consistent.
22464 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22466 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
22469 2017-07-01 Jakub Jelinek <jakub@redhat.com>
22472 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
22473 the right scopes, make sure cond_jump isn't preserved between multiple
22474 iterations. Search for fallthru edge whenever there are 3+ edges and
22475 use find_fallthru_edge for it.
22477 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22479 Patch by Alexander Monakov <amonakov@ispras.ru>
22480 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
22481 probabilities consistently.
22483 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22485 * pa.c (pa_expand_compare_and_swap_loop): Update call of
22486 emit_cmp_and_jump_insns.
22488 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22491 * tree-inline.c (expand_call_inline): Combine profile statuses.
22493 2017-06-30 Andrew Pinski <apinski@cavium.com>
22495 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
22496 fold_stmt returned true.
22498 2017-06-30 Nathan Sidwell <nathan@acm.org>
22500 * ggc.h (empty_string): Delete.
22501 * cfgexpand.c (expand_asm_stmt): Use plain "".
22502 * optabs.c (expand_asm_memory_barrier): Likewise.
22503 * stringpool.c (empty_string): Delete.
22504 (digit_vector, digit_string): Delete.
22505 (ggc_alloc_string): Use plain "", don't optimize single digit
22506 strings. Use ggc_alloc_atomic.
22508 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
22510 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
22511 comparison set and one other set, use the cost of the non-comparison
22514 2017-06-30 Nathan Sidwell <nathan@acm.org>
22516 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
22519 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
22521 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
22522 loops. Remove now unneeded calls to gimple_switch_set_label() that
22523 just set removed labels to NULL_TREE.
22525 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
22527 * tree-ssanames.c (set_range_info_raw): Abstract from ...
22528 (set_range_info): ...here. Only call set_range_info_raw if domain
22530 (set_nonzero_bits): Call set_range_info_raw.
22531 * tree-ssanames.h (set_range_info_raw): New.
22533 2017-06-30 Jakub Jelinek <jakub@redhat.com>
22536 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
22537 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
22538 of nonimmediate_operand and <store_mask_constraint> instead of m
22539 for the input operand. For V8FI iterator, always split if input
22540 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
22541 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
22542 <store_mask_predicate> instead of register_operand and
22543 <store_mask_constraint> instead of v for the input operand. Make
22544 sure both operands aren't MEMs for if not <mask_applied>.
22546 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
22548 * lto-wrapper.c (copy_file) Close both file descriptors before
22551 2017-06-30 Martin Liska <mliska@suse.cz>
22554 * multiple_target.c (create_dispatcher_calls): Make ifunc
22555 also for function that don't have calls or are not referenced.
22557 2017-06-30 Richard Biener <rguenther@suse.de>
22559 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
22560 analyze the first scalar stmt. Move vector type computation
22561 for the BB case here from ...
22562 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
22563 live operation processing in the SLP case properly.
22565 2017-06-30 Richard Biener <rguenther@suse.de>
22567 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
22569 2017-06-30 Martin Liska <mliska@suse.cz>
22572 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
22573 before BUILT_IN_UNWIND_RESUME when ASAN is used.
22575 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
22577 * doc/invoke.texi (AArch64): Add missing options and remove redundant
22580 2017-06-30 Richard Biener <rguenther@suse.de>
22582 PR tree-optimization/81249
22583 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
22584 condition reduction result to original scalar type.
22586 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22588 * profile-count.h (enum profile_quality): Fix typos and whitespace
22591 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22593 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
22594 type for branch probabilities.
22596 2017-06-29 Julian Brown <julian@codesourcery.com>
22597 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22599 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
22600 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
22601 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
22602 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
22604 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22606 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
22607 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
22608 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
22609 CC usage from generic code to here.
22610 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
22611 CC usage into the target macros.
22613 2017-06-29 Maya Rashish <coypu@sdf.org>
22615 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
22618 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22620 * arm/arm-builtins.c: Include profile-count.h
22621 * except.c (sjlj_emit_function_enter): Use
22622 profile_probability::unlikely.
22624 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22626 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
22627 and tocrel_offset be pointer args rather than implicitly using
22629 (legitimate_constant_pool_address_p, rs6000_emit_move,
22630 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
22631 tocrel_offset and use in toc_relative_expr_p call.
22632 (print_operand, print_operand_address): Use static tocrel_base_oac
22633 and tocrel_offset_oac.
22634 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
22637 2017-06-29 Maya Rashish <coypu@sdf.org>
22639 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
22641 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
22643 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
22644 objects, take into account only the alignment of 'op0' and 'mode1' if
22647 2017-06-29 Steve Ellcey <sellcey@cavium.com>
22649 * ccmp.c (ccmp_tree_comparison_p): New function.
22650 (ccmp_candidate_p): Update to use above function.
22651 (get_compare_parts): New function.
22652 (expand_ccmp_next): Update to use new functions.
22653 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
22655 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
22656 take mode as argument.
22657 * ccmp.h (expand_ccmp_expr): Add mode as argument.
22658 * expr.c (expand_expr_real_1): Pass mode as argument.
22660 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
22662 * combine.c (combine_instructions): Print insns to dump_file, together
22665 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22667 * asan.c (asan_emit_stack_protection): Update.
22668 (create_cond_insert_point): Update.
22669 * auto-profile.c (afdo_propagate_circuit): Update.
22670 * basic-block.h (struct edge_def): Turn probability to
22671 profile_probability.
22672 (EDGE_FREQUENCY): Update.
22673 * bb-reorder.c (find_traces_1_round): Update.
22674 (better_edge_p): Update.
22675 (sanitize_hot_paths): Update.
22676 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
22677 (make_single_succ_edge): Update.
22678 (check_bb_profile): Update.
22679 (dump_edge_info): Update.
22680 (update_bb_profile_for_threading): Update.
22681 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
22682 probabilitycount to 0.
22683 * cfgbuild.c (compute_outgoing_frequencies): Update.
22684 * cfgcleanup.c (try_forward_edges): Update.
22685 (outgoing_edges_match): Update.
22686 (try_crossjump_to_edge): Update.
22687 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
22688 (expand_gimple_tailcall): Update.
22689 (construct_init_block): Use make_single_succ_edge.
22690 (construct_exit_block): Use make_single_succ_edge.
22691 * cfghooks.c (verify_flow_info): Update.
22692 (redirect_edge_succ_nodup): Update.
22693 (split_edge): Update.
22694 (account_profile_record): Update.
22695 * cfgloopanal.c (single_likely_exit): Update.
22696 * cfgloopmanip.c (scale_loop_profile): Update.
22697 (set_zero_probability): Remove.
22698 (duplicate_loop_to_header_edge): Update.
22699 * cfgloopmanip.h (loop_version): Update prototype.
22700 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
22701 (force_nonfallthru_and_redirect): Update.
22702 (update_br_prob_note): Update.
22703 (rtl_verify_edges): Update.
22704 (purge_dead_edges): Update.
22705 (rtl_lv_add_condition_to_bb): Update.
22706 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
22707 * cgraphunit.c (init_lowered_empty_function): Update.
22708 (cgraph_node::expand_thunk): Update.
22709 * cilk-common.c: Include profile-count.h
22710 * dojump.c (inv): Remove.
22711 (jumpifnot): Update.
22712 (jumpifnot_1): Update.
22713 (do_jump_1): Update.
22715 (do_jump_by_parts_greater_rtx): Update.
22716 (do_compare_rtx_and_jump): Update.
22717 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
22718 do_jump_1. do_compare_rtx_and_jump): Update prototype.
22719 * dwarf2cfi.c: Include profile-count.h
22720 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
22721 (sjlj_emit_dispatch_table): Likewise.
22722 * explow.c: Include profile-count.h
22723 * expmed.c (emit_store_flag_force): Update.
22724 (do_cmp_and_jump): Update.
22725 * expr.c (compare_by_pieces_d::generate): Update.
22726 (compare_by_pieces_d::finish_mode): Update.
22727 (emit_block_move_via_loop): Update.
22728 (store_expr_with_bounds): Update.
22729 (store_constructor): Update.
22730 (expand_expr_real_2): Update.
22731 (expand_expr_real_1): Update.
22732 * expr.h (try_casesi, try_tablejump): Update prototypes.
22733 * gimple-pretty-print.c (dump_probability): Update.
22734 (dump_profile): New.
22735 (dump_gimple_label): Update.
22736 (dump_gimple_bb_header): Update.
22737 * graph.c (draw_cfg_node_succ_edges): Update.
22738 * hsa-gen.c (convert_switch_statements): Update.
22739 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
22740 (find_if_case_1): Update.
22741 (find_if_case_2): Update.
22742 * internal-fn.c (expand_arith_overflow_result_store): Update.
22743 (expand_addsub_overflow): Update.
22744 (expand_neg_overflow): Update.
22745 (expand_mul_overflow): Update.
22746 (expand_vector_ubsan_overflow): Update.
22747 * ipa-cp.c (good_cloning_opportunity_p): Update.
22748 * ipa-split.c (split_function): Use make_single_succ_edge.
22749 * ipa-utils.c (ipa_merge_profiles): Update.
22750 * loop-doloop.c (add_test): Update.
22751 (doloop_modify): Update.
22752 * loop-unroll.c (compare_and_jump_seq): Update.
22753 (unroll_loop_runtime_iterations): Update.
22754 * lra-constraints.c (lra_inheritance): Update.
22755 * lto-streamer-in.c (input_cfg): Update.
22756 * lto-streamer-out.c (output_cfg): Update.
22757 * mcf.c (adjust_cfg_counts): Update.
22758 * modulo-sched.c (sms_schedule): Update.
22759 * omp-expand.c (expand_omp_for_init_counts): Update.
22760 (extract_omp_for_update_vars): Update.
22761 (expand_omp_ordered_sink): Update.
22762 (expand_omp_for_ordered_loops): Update.
22763 (expand_omp_for_generic): Update.
22764 (expand_omp_for_static_nochunk): Update.
22765 (expand_omp_for_static_chunk): Update.
22766 (expand_cilk_for): Update.
22767 (expand_omp_simd): Update.
22768 (expand_omp_taskloop_for_outer): Update.
22769 (expand_omp_taskloop_for_inner): Update.
22770 * omp-simd-clone.c (simd_clone_adjust): Update.
22771 * optabs.c (expand_doubleword_shift): Update.
22772 (expand_abs): Update.
22773 (emit_cmp_and_jump_insn_1): Update.
22774 (expand_compare_and_swap_loop): Update.
22775 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
22776 * predict.c (predictable_edge_p): Update.
22777 (edge_probability_reliable_p): Update.
22778 (set_even_probabilities): Update.
22779 (combine_predictions_for_insn): Update.
22780 (combine_predictions_for_bb): Update.
22781 (propagate_freq): Update.
22782 (estimate_bb_frequencies): Update.
22783 (force_edge_cold): Update.
22784 * profile-count.c (profile_count::dump): Add missing space into dump.
22785 (profile_count::debug): Add newline.
22786 (profile_count::differs_from_p): Explicitly convert to unsigned.
22787 (profile_count::stream_in): Update.
22788 (profile_probability::dump): New member function.
22789 (profile_probability::debug): New member function.
22790 (profile_probability::differs_from_p): New member function.
22791 (profile_probability::differs_lot_from_p): New member function.
22792 (profile_probability::stream_in): New member function.
22793 (profile_probability::stream_out): New member function.
22794 * profile-count.h (profile_count_quality): Rename to ...
22795 (profile_quality): ... this one.
22796 (profile_probability): New.
22797 (profile_count): Update.
22798 * profile.c (compute_branch_probabilities): Update.
22799 * recog.c (peep2_attempt): Update.
22800 * sched-ebb.c (schedule_ebbs): Update.
22801 * sched-rgn.c (find_single_block_region): Update.
22802 (compute_dom_prob_ps): Update.
22803 (schedule_region): Update.
22804 * sel-sched-ir.c (compute_succs_info): Update.
22805 * stmt.c (struct case_node): Update.
22806 (do_jump_if_equal): Update.
22807 (get_outgoing_edge_probs): Update.
22808 (conditional_probability): Update.
22809 (emit_case_dispatch_table): Update.
22810 (expand_case): Update.
22811 (expand_sjlj_dispatch_table): Update.
22812 (emit_case_nodes): Update.
22813 * targhooks.c: Update.
22814 * tracer.c (better_p): Update.
22815 (find_best_successor): Update.
22816 * trans-mem.c (expand_transaction): Update.
22817 * tree-call-cdce.c: Update.
22818 * tree-cfg.c (gimple_split_edge): Upate.
22819 (move_sese_region_to_fn): Upate.
22820 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
22821 * tree-eh.c (lower_resx): Upate.
22822 (cleanup_empty_eh_move_lp): Upate.
22823 * tree-if-conv.c (version_loop_for_if_conversion): Update.
22824 * tree-inline.c (copy_edges_for_bb): Update.
22825 (copy_cfg_body): Update.
22826 * tree-parloops.c (gen_parallel_loop): Update.
22827 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
22828 (gimple_gen_time_profiler): Update.
22829 * tree-ssa-dce.c (remove_dead_stmt): Update.
22830 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22831 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
22832 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
22833 (unloop_loops): Update.
22834 (try_peel_loop): Update.
22835 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22836 * tree-ssa-loop-split.c (connect_loops): Update.
22837 (split_loop): Update.
22838 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
22839 (hoist_guard): Update.
22840 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
22841 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
22842 (value_replacement): Update.
22843 * tree-ssa-reassoc.c (branch_fixup): Update.
22844 * tree-ssa-tail-merge.c (replace_block_by): Update.
22845 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
22846 (create_edge_and_update_destination_phis): Update.
22847 (compute_path_counts): Update.
22848 (recompute_probabilities): Update.
22849 (update_joiner_offpath_counts): Update.
22850 (freqs_to_counts_path): Update.
22851 (duplicate_thread_path): Update.
22852 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
22853 (struct switch_conv_info): Update.
22854 (gen_inbound_check): Update.
22855 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22856 (vect_do_peeling): Update.
22857 (vect_loop_versioning): Update.
22858 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22859 (optimize_mask_stores): Update.
22860 * ubsan.c (ubsan_expand_null_ifn): Update.
22861 * value-prof.c (gimple_divmod_fixed_value): Update.
22862 (gimple_divmod_fixed_value_transform): Update.
22863 (gimple_mod_pow2): Update.
22864 (gimple_mod_pow2_value_transform): Update.
22865 (gimple_mod_subtract): Update.
22866 (gimple_mod_subtract_transform): Update.
22867 (gimple_ic): Update.
22868 (gimple_stringop_fixed_value): Update.
22869 (gimple_stringops_transform): Update.
22870 * value-prof.h: Update.
22872 2017-06-29 Carl Love <cel@us.ibm.com>
22874 * config/rs6000/rs6000-c.c: Add support for built-in functions
22875 vector signed int vec_signed (vector float);
22876 vector signed long long vec_signed (vector double);
22877 vector signed int vec_signed2 (vector double, vector double);
22878 vector signed int vec_signede (vector double);
22879 vector signed int vec_signedo (vector double);
22880 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
22881 instruction generator.
22882 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
22883 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
22884 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
22886 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
22887 vunsignede_v2df): Add define_expands.
22888 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
22889 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
22890 VEC_UNSIGNEDO): Add definitions.
22891 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
22892 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
22893 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
22894 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
22895 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
22896 * config/rs6000/altivec.h (vec_signed, vec_signed2,
22897 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
22898 vec_unsignede, vec_unsignedo): Add builtin defines.
22899 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
22901 * doc/extend.texi: Update the built-in documentation file for the
22902 new built-in functions.
22904 2017-06-29 Richard Biener <rguenther@suse.de>
22906 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
22907 reduction chains to LOOP_VINFO_REDUCTIONS.
22908 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
22909 SLP reductions after processing reduction chains.
22911 2017-06-29 Nathan Sidwell <nathan@acm.org>
22913 * builtins.c (fold_builtin_FUNCTION): Use
22914 lang_hooks.decl_printable_name.
22916 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
22918 PR middle-end/81194
22919 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
22920 with only one label.
22921 * stmt.c (expand_case): Assert NCASES is greater than one.
22923 2017-06-29 Richard Biener <rguenther@suse.de>
22925 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
22927 (group_case_labels): Likewise.
22928 (find_taken_edge): Push sanity checking on val to workers...
22929 (find_taken_edge_cond_expr): ... here
22930 (find_taken_edge_switch_expr): ... and here, handle cases
22931 with just a default label.
22932 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
22933 (group_case_labels): Likewise.
22934 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
22935 group_case_labels does anything cleanup the CFG again.
22937 2017-06-29 Bin Cheng <bin.cheng@arm.com>
22939 PR tree-optimization/81196
22940 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
22941 exit condition comparing two IVs.
22943 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
22945 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
22946 profile to the dummy entry at the end of the list of architectures.
22947 * config/arm/arm-cpu-cdata.h: Regenerated.
22949 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22950 Michael Collison <michael.collison@arm.com>
22953 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
22955 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
22956 (*aarch64_reg_<mode>3_minus_mask): New pattern.
22957 (*aarch64_<optab>_reg_di3_mask2): New pattern.
22958 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
22959 of shift when the shift amount is masked with constant equal to
22960 the size of the mode.
22961 * config/aarch64/predicates.md (subreg_lowpart_operator): New
22964 2017-06-29 Martin Liska <mliska@suse.cz>
22966 * config/i386/i386.opt: Change range from [1,5] to [0,5].
22968 2017-06-29 Yury Gribov <tetra2005@gmail.com>
22971 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
22974 (edge_growth_cache_entry::edge_growth_cache_entry): New
22976 (reset_edge_growth_cache): Update to use constructor.
22978 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22980 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
22981 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
22982 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
22984 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
22986 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
22987 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
22989 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
22991 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
22992 (*-linux-uclibc*): Add t-uclibc tmake_file.
22993 * config/t-musl: New.
22994 * config/t-uclibc: New.
22996 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
22998 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
23000 (gen_comm_data): Emit architectural setting of arch_prof.
23001 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
23003 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
23004 (armv8-m.base, armv8-m.main): Likewise.
23005 * arm-protos.h (arm_build_target): Add profile field.
23006 (arch_option): Likewise.
23007 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
23009 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
23010 arm_active_target.profile.
23012 2017-06-28 Richard Biener <rguenther@suse.de>
23014 PR middle-end/81227
23015 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
23016 TYPE_OVERFLOW_WRAPS.
23017 * match.pd (negate_expr_p): Likewise.
23018 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
23019 fold_build2, not fold_binary.
23021 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23023 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
23024 Convert memory address to Pmode.
23025 (aarch64_print_operand): Assert MEM operands are always Pmode.
23027 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23030 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
23031 Remove redundant if.
23032 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
23033 * config/arm/aarch-common-protos.h
23034 (aarch_forward_to_shift_is_not_shifted_re): Remove.
23035 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
23037 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
23040 * multiple_target.c (create_dispatcher_calls): Set the default
23041 clone to be static, not public.
23043 2017-06-28 Richard Biener <rguenther@suse.de>
23045 * tree-vect-loop.c (vectorizable_reduction): Move special
23046 cond reduction IV var creation ...
23047 (vect_create_epilog_for_reduction): ... here. Remove induction_index
23048 parameter. Use STMT_VINFO_VECTYPE.
23049 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
23052 2017-06-28 Martin Liska <mliska@suse.cz>
23055 * ipa-visibility.c (non_local_p): Handle visibility.
23057 2017-06-28 Martin Liska <mliska@suse.cz>
23060 * common.opt: Add IntegerRange to various options.
23061 * opt-functions.awk (integer_range_info): New function.
23062 * optc-gen.awk: Add integer_range_info to cl_options struct.
23063 * opts-common.c (decode_cmdline_option): Handle
23064 CL_ERR_INT_RANGE_ARG.
23065 (cmdline_handle_error): Likewise.
23066 * opts.c (print_filtered_help): Show valid interval in
23067 when --help is provided.
23068 * opts.h (struct cl_option): Add range_min and range_max fields.
23069 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
23071 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
23073 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
23074 (x * C EQ/NE y * C): New transformation.
23076 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
23078 * genmultilib (combination_space): Accept '+' in option names.
23080 2017-06-28 Martin Liska <mliska@suse.cz>
23083 * asan.c (instrument_derefs): Bail out inner references
23084 that are hard register variables.
23086 2017-06-28 Jakub Jelinek <jakub@redhat.com>
23089 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
23090 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
23092 2017-06-28 Richard Biener <rguenther@suse.de>
23094 * tree-vectorizer.h (vect_get_vec_defs): Remove.
23095 (vect_get_slp_defs): Adjust.
23096 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
23098 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
23100 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
23101 get_initial_defs_for_reduction instead of vect_get_vec_defs.
23102 (vectorizable_reduction): Adjust.
23103 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
23105 (vect_get_slp_defs): Likewise.
23106 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
23107 (vectorizable_bswap): Adjust.
23108 (vectorizable_call): Likewise.
23109 (vectorizable_conversion): Likewise.
23110 (vectorizable_assignment): Likewise.
23111 (vectorizable_shift): Likewise.
23112 (vectorizable_operation): Likewise.
23113 (vectorizable_store): Likewise.
23114 (vectorizable_condition): Likewise.
23115 (vectorizable_comparison): Likewise.
23117 2017-06-28 Michael Collison <michael.collison@arm.com>
23120 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
23122 (arm_gen_movmemqi): Removed unused variable 'i'.
23123 Convert 'for' loop into 'while' loop.
23124 (arm_expand_prologue): Remove last unnecessary set of insn.
23125 (thumb_pop): Remove unused variable 'pushed_words'.
23126 (thumb_exit): Remove last unnecessary set of regs_to_pop.
23128 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23130 * config/s390/predicates.md: Use s390_rel_address_ok_p.
23131 * config/s390/s390-protos.h: Add prototype of
23132 s390_rel_address_ok_p.
23133 * config/s390/s390.c (s390_got_symbol): New function.
23134 (s390_rel_address_ok_p): New function.
23135 (legitimize_pic_address): Use s390_rel_address_ok_p.
23136 (s390_load_got): Use s390_got_symbol.
23137 (s390_option_override): Issue error if
23138 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
23139 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
23141 * config/s390/s390.opt: New option mpic-data-is-text-relative.
23143 2017-06-27 Andrew Pinski <apinski@cavium.com>
23145 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
23146 (X * copysign (1.0, X)): New pattern.
23147 (X * copysign (1.0, -X)): New pattern.
23148 (copysign (-1.0, CST)): New pattern.
23150 2017-06-27 Joseph Myers <joseph@codesourcery.com>
23152 * genmultilib (combination_space): Remove variable.
23153 Validate reuse rules against regular expression for any sequence
23154 of multilib options in any order.
23156 2017-06-27 Michael Collison <michael.collison@arm.com>
23158 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
23159 call aarch64_split_simd_combine.
23160 * (aarch64_combine_internal<mode>): Delete pattern.
23161 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
23162 Allow register and subreg operands.
23164 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23166 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
23167 specific need, just fallback on defaults.
23168 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
23170 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23171 Olivier Hainque <hainque@adacore.com>
23173 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
23175 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
23176 targets. Pick a default if no particular attempt applied.
23177 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
23180 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23182 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
23183 (x86_64-wrs-vxworks7): Likewise.
23185 2017-06-27 Marek Polacek <polacek@redhat.com>
23188 * ubsan.c (instrument_null): Check get_base_address's result for null.
23190 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
23192 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
23194 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
23196 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
23197 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
23198 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
23199 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
23200 New function types.
23201 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
23202 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
23203 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
23204 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
23205 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
23206 BUILT_IN_FEUPDATEENV): New builtins.
23207 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
23208 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
23209 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
23210 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
23212 (builtin_structptr_types): Adjust size.
23213 * tree.c (builtin_structptr_types): Add four entries.
23215 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23216 Olivier Hainque <hainque@adacore.com>
23218 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
23219 (TLS_SYM): New local macro, forcing reference to __tls__ on
23220 link command lines for VxWorks 7 RTPs, triggering initialization
23222 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
23223 OS features TLS support, true for RTPs on VxWorks 7.
23224 * config/vxworks.c (vxworks_override_options): Setup emutls
23227 2017-06-27 Jakub Jelinek <jakub@redhat.com>
23229 * predict.c (test_prediction_value_range): Use -1U instead of -1
23230 to avoid narrowing conversion warning.
23231 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
23232 to avoid narrowing conversion warning.
23233 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
23235 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
23237 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23239 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
23240 64bit configurations.
23241 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
23242 (SIZE_TYPE): Likewise.
23243 * config/vxworks.c (vxworks_emutls_var_fields): Use
23244 long_unsigned_type_node instead of unsigned_type_node as the offset
23245 field type, which is "pointer" mode in emutls.c.
23247 2017-06-27 Jakub Jelinek <jakub@redhat.com>
23250 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
23252 PR middle-end/81207
23253 * gimple-fold.c (replace_call_with_call_and_fold): Handle
23254 gimple_vuse copying separately from gimple_vdef copying.
23256 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23258 * value-prof.c (free_hist): Remove call to memset and the enclosing if
23261 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
23262 Olivier Hainque <hainque@adacore.com>
23264 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
23265 for all vxworks7 targets.
23266 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
23267 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
23268 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
23269 variations for VX6/VX7 and 32/64bits later on in ...
23270 (VXWORKS_LIB_SPEC): Leverage new macros.
23271 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
23272 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
23274 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
23276 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
23277 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
23279 2017-06-26 Carl Love <cel@us.ibm.com>
23281 * config/rs6000/rs6000-c.c: Add support for built-in functions
23282 vector bool char vec_reve (vector bool char);
23283 vector signed char vec_reve (vector signed char);
23284 vector unsigned char vec_reve (vector unsigned char);
23285 vector bool int vec_reve (vector bool int);
23286 vector signed int vec_reve (vector signed int);
23287 vector unsigned int vec_reve (vector unsigned int);
23288 vector bool long long vec_reve (vector bool long long);
23289 vector signed long long vec_reve (vector signed long long);
23290 vector unsigned long long vec_reve (vector unsigned long long);
23291 vector bool short vec_reve (vector bool short);
23292 vector signed short vec_reve (vector signed short);
23293 vector double vec_reve (vector double);
23294 vector float vec_reve (vector float);
23295 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
23296 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
23297 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
23298 (altivec_vreve): New pattern.
23299 * config/rs6000/altivec.h (vec_reve): New define.
23300 * doc/extend.texi (vec_rev): Update the built-in documentation file
23301 for the new built-in functions.
23303 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23305 PR tree-optimization/71815
23306 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
23308 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
23310 (slsr_process_phi): Likewise.
23311 (replace_uncond_cands_and_profitable_phis): Don't replace a
23312 multiply candidate with a stride of 1 (copy or cast).
23313 (phi_incr_cost): Call uses_consumed_by_stmt rather than
23315 (lowest_cost_path): Likewise.
23316 (total_savings): Likewise.
23318 2017-06-26 Richard Biener <rguenther@suse.de>
23321 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
23322 Use def_builtin_pure for all gather builtins.
23324 2017-06-26 Richard Biener <rguenther@suse.de>
23326 PR tree-optimization/81203
23327 * tree-tailcall.c (find_tail_calls): Do not move stmts into
23328 non-dominating BBs.
23330 2017-06-26 Marek Polacek <polacek@redhat.com>
23333 * doc/invoke.texi: Document -Wmultistatement-macros.
23335 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
23337 * doc/sourcebuild.texi (ARM-specific attributes): Document new
23338 arm_neon_ok_no_float_abi effective target.
23340 2017-06-26 Richard Biener <rguenther@suse.de>
23342 PR tree-optimization/80928
23343 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
23344 (copy_bbs): Set BB_DUPLICATED flag early.
23345 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
23347 (execute_on_shrinking_pred): Likewise.
23348 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
23349 BB_DUPLICATED blocks.
23350 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
23351 iterate over all PHIs considering removal of *gsi.
23353 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
23355 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
23358 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23360 * config/rs6000/rs6000-string.c: (expand_block_clear,
23361 do_load_for_compare, select_block_compare_mode,
23362 compute_current_alignment, expand_block_compare,
23363 expand_strncmp_align_check, expand_strn_compare,
23364 expand_block_move, rs6000_output_load_multiple)
23365 Move functions related to string/block move/compare
23366 to a separate file.
23367 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
23368 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
23369 for this function which is now used in two files.
23370 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
23371 * config.gcc: Add rs6000-string.o to extra_objs for
23372 targets powerpc*-*-* and rs6000*-*-*.
23374 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
23377 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
23378 32-bit, since indexed is not valid for DImode.
23379 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
23380 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
23381 (define_peephole2 for Altivec d-form load): Add 32-bit support.
23382 (define_peephole2 for Altivec d-form store): Likewise.
23385 * multiple_target.c (create_dispatcher_calls): Only create the
23386 dispatcher call if the function is the default clone of a
23387 versioned function.
23389 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
23391 PR middle-end/80902
23392 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
23393 a call, force the call to not be a tail call.
23395 2017-06-23 Jeff Law <law@redhat.com>
23397 * doc/contrib.texi: Add entry for Steven Pemberton's work on
23400 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
23402 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
23403 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
23404 handling for early expansion of vector shifts (sl,sr,sra,rl).
23405 (builtin_function_type): Add vector shift right instructions
23406 to the unsigned argument list.
23408 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
23410 rtl-optimizatoin/79286
23411 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
23412 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
23413 trap. PIC register plus a const unspec without offset can never trap.
23415 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
23417 * tree.h (builtin_structptr_type): New type.
23418 (builtin_structptr_types): Declare new array.
23419 * tree.c (builtin_structptr_types): New array.
23420 (free_lang_data, build_common_tree_nodes): Use it.
23422 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
23425 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
23428 2017-06-22 Matt Turner <mattst88@gmail.com>
23430 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
23431 Lake models to skylake case. Assume skylake for unknown
23432 models with clflushopt.
23434 2017-06-22 Jeff Law <law@redhat.com>
23436 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
23437 frame sizes that do not satisfy aarch64_uimm12_shift.
23439 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
23441 * profile-count.h (apply_probability,
23442 apply_scale, probability_in): Fix checks for zero.
23444 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23446 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
23447 * doc/cppdiropts.texi (-I @var{dir}): Document it.
23449 2016-06-22 Richard Biener <rguenther@suse.de>
23451 * tree-vect-loop.c (vect_model_reduction_cost): Handle
23452 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
23453 REDUC_MAX_EXPR support.
23454 (vectorizable_reduction): Likewise.
23455 (vect_create_epilog_for_reduction): Likewise.
23457 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
23459 * match.pd (A / (1 << B) -> A >> B): New.
23460 * generic-match-head.c: Include optabs-tree.h.
23461 * gimple-match-head.c: Likewise.
23462 * optabs-tree.h (target_supports_op_p): New.
23463 * optabs-tree.c (target_supports_op_p): New.
23465 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23467 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
23468 $gcc_cv_ld --help output.
23469 (gcc_cv_ld_demangle): Likewise.
23470 (gcc_cv_ld_eh_frame_hdr): Likewise.
23471 (gcc_cv_ld_pie): Likewise.
23472 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
23473 (gcc_cv_ld_buildid): Likewise.
23474 (gcc_cv_ld_sysroot): Likewise.
23475 (ld_bndplt_support): Likewise.
23476 (ld_pushpopstate_support): Likewise.
23477 * configure: Regenerate.
23478 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
23480 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23483 * config/i386/sse.md (round<mode>2): Renumber match_dup and
23484 operands indexes to avoid gap between operands and match_dups.
23486 2017-06-21 Andrew Pinski <apinski@cavium.com>
23488 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
23489 Increment Arith_shift and Arith_shift_reg by 1.
23490 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
23492 * config/aarch64/aarch64.c (thunderx_tunings): Enable
23493 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
23494 (aarch64_strip_extend): Add new argument and test for it.
23495 (aarch64_cheap_mult_shift_p): New function.
23496 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
23497 add a cost if it is true.
23498 Update calls to aarch64_strip_extend.
23499 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
23501 2017-06-21 Andrew Pinski <apinski@cavium.com>
23503 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
23505 (thunderxt88): Likewise.
23506 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
23507 (thunderx_prefetch_tune): New variable.
23508 (thunderx2t99_prefetch_tune): Update for the correct values.
23509 (thunderxt88_tunings): New variable.
23510 (thunderx_tunings): Use thunderx_prefetch_tune instead of
23511 generic_prefetch_tune.
23512 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
23514 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23516 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
23517 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
23518 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
23519 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
23520 (aarch64_atomic_cas<mode>, GPI): Likewise.
23522 2017-06-21 Martin Liska <mliska@suse.cz>
23524 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
23525 statements on cold and hot labels.
23526 * predict.c (tree_estimate_probability_bb): Remove the
23527 prediction from this place.
23529 2017-06-21 Martin Liska <mliska@suse.cz>
23531 PR tree-optimization/79489
23532 * gimplify.c (maybe_add_early_return_predict_stmt): New
23534 (gimplify_return_expr): Call the function.
23535 * predict.c (tree_estimate_probability_bb): Remove handling
23537 * predict.def: Update comment about early return predictor.
23538 * gimple-predict.h (is_gimple_predict): New function.
23539 * predict.def: Change default value of early return to 66.
23540 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
23542 * passes.def: Put pass_strip_predict_hints to the beginning of
23545 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
23547 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
23548 FUNCTION_DECL declarations.
23549 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
23551 (dwaf2out_decl): Likewise.
23552 * godump.c (go_early_global_decl): Skip call to the real debug hook
23553 for FUNCTION_DECL declarations.
23554 * passes.c (rest_of_decl_compilation): Skip call to the
23555 early_global_decl debug hook for FUNCTION_DECL declarations, unless
23556 -fdump-go-spec is passed.
23558 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23560 * config/i386/i386.c (struct builtin_isa): New field pure_p.
23561 Reorder for compactness.
23562 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
23563 (def_builtin_pure, def_builtin_pure2): New functions.
23564 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
23566 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23568 * match.pd (nop_convert): New predicate.
23569 ((A +- CST1) +- CST2): Allow some NOP conversions.
23571 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23574 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
23575 with ctors/dtors if GOVD_SHARED is set.
23577 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23579 * config/aarch64/aarch64.md (movti_aarch64):
23580 Emit mov rather than orr.
23581 (movtf_aarch64): Likewise.
23582 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
23583 Emit mov rather than orr.
23585 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23587 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
23588 Swap alternatives, make integer dup more expensive.
23590 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23592 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
23593 Return true for non-tls symbols.
23595 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
23597 * config/aarch64/aarch64-cores.def (cortex-a55): New.
23598 (cortex-a75): Likewise.
23599 (cortex-a75.cortex-a55): Likewise.
23600 * config/aarch64/aarch64-tune.md: Regenerate.
23601 * doc/invoke.texi (-mtune): Document new values for -mtune.
23603 2017-06-21 Tom de Vries <tom@codesourcery.com>
23605 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
23606 stack_size feature.
23607 (Effective-Target Keywords, Other attributes): Suggest using
23608 dg-add-options stack_size feature to get stack limit in stack_size
23609 effective target documentation.
23611 2017-06-21 Julian Brown <julian@codesourcery.com>
23612 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23614 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
23615 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
23616 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
23618 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
23619 attribute type list for neon_multiply.
23620 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
23621 attribute type list for neon_multiply.
23622 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
23623 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
23624 attribute type list for neon_multiply.
23625 * config/arm/types.md (crypto_pmull): Add.
23626 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
23627 attribute type list.
23629 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
23631 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
23634 2017-06-20 Jakub Jelinek <jakub@redhat.com>
23636 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
23637 to make sure not to dereference a NULL cost_classes_ptr pointer.
23639 2017-06-20 Carl Love <cel@us.ibm.com>
23641 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23642 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23643 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23644 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23645 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23646 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23647 VMULOSW): New enum "unspec" values.
23648 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23649 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23650 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23651 altivec_vmulosw): New patterns.
23652 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23653 VMULOSW): Add definitions.
23655 2017-06-20 Julia Koval <julia.koval@intel.com>
23657 * config/i386/i386.c: Fix rounding expand for new pattern.
23658 * config/i386/subst.md: Fix pattern (parallel -> unspec).
23660 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23662 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
23663 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
23665 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23667 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
23670 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23672 * config/aarch64/aarch64-cores.def: Rearrange to sort by
23673 architecture, then by implementer ID.
23674 * config/aarch64/aarch64-tune.md: Regenerate.
23676 2017-06-20 Richard Biener <rguenther@suse.de>
23678 PR middle-end/81097
23679 * fold-const.c (split_tree): Fold to type before negating.
23681 2017-06-20 David Malcolm <dmalcolm@redhat.com>
23683 * diagnostic-show-locus.c
23684 (selftest::test_fixit_deletion_affecting_newline): New function.
23685 (selftest::diagnostic_show_locus_c_tests): Call it.
23687 2017-06-20 Andreas Schwab <schwab@suse.de>
23690 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
23693 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23695 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
23696 __ARM_FEATURE_COPROC according to support.
23698 2017-06-20 Jakub Jelinek <jakub@redhat.com>
23700 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
23701 Rewritten to avoid overflow for > 32-bit pointers.
23704 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
23705 by removing enum keyword.
23706 (ubsan_type_descriptor): Likewise. Formatting fix.
23709 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
23710 splitter): Require TARGET_SSE2 in the condition.
23712 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
23715 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
23716 for doing vector set of SFmode on ISA 3.0.
23717 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
23718 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
23720 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
23721 SFmode value into a V4SF variable that was extracted from another
23722 V4SF variable without converting the element to double precision
23723 and back to single precision vector format.
23724 (vsx_insert_extract_v4sf_p9_2): Likewise.
23726 2017-06-19 Jakub Jelinek <jakub@redhat.com>
23728 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
23729 in UWHI to avoid undefined overflow.
23732 * ubsan.h (enum ubsan_encode_value_phase): New.
23733 (ubsan_encode_value): Change second argument to
23734 enum ubsan_encode_value_phase with default value of
23735 UBSAN_ENCODE_VALUE_GENERIC.
23736 * ubsan.c (ubsan_encode_value): Change second argument to
23737 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
23738 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
23739 create_tmp_var_raw instead of create_tmp_var and use a
23741 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
23742 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
23743 ubsan_encode_value callers.
23746 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
23747 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
23748 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
23750 2017-06-19 Richard Biener <rguenther@suse.de>
23752 PR middle-end/81118
23753 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
23754 estimates if we changed anything.
23756 2017-06-19 Richard Biener <rguenther@suse.de>
23758 PR tree-optimization/80887
23759 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
23760 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
23761 simplified lookups, then reset mprts_hook.
23762 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
23764 (try_to_simplify): Likewise.
23766 2017-06-19 Martin Liska <mliska@suse.cz>
23769 * gimplify.c (gimplify_switch_expr):
23770 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
23772 2017-06-19 Martin Liska <mliska@suse.cz>
23774 * doc/install.texi: Document that PGO runs in 4 stages.
23776 2017-06-19 Martin Liska <mliska@suse.cz>
23779 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
23780 to dispatcher function name.
23781 * multiple_target.c (replace_function_decl): New function.
23782 (create_dispatcher_calls): Redirect both edges and references.
23784 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
23786 * profile-count.c (profile_count::dump): Dump quality.
23787 (profile_count::differs_from_p): Update for unsigned val.
23788 * profile-count.h (profile_count_quality): New enum.
23789 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
23791 2017-06-19 Richard Biener <rguenther@suse.de>
23793 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
23794 struct function as arg.
23795 (estimate_numbers_of_iterations): Export overload with loop arg.
23796 (free_numbers_of_iterations_estimates_loop): Use an overload of
23797 free_numbers_of_iterations_estimates instead.
23798 * tree-cfg.c (remove_bb): Adjust.
23799 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
23800 * tree-parloops.c (gen_parallel_loop): Likewise.
23801 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
23803 (tree_unroll_loops_completely): Likewise.
23804 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
23805 Use an overload instead and export.
23806 (estimated_loop_iterations): Adjust.
23807 (max_loop_iterations): Likewise.
23808 (likely_max_loop_iterations): Likewise.
23809 (estimate_numbers_of_iterations): Take struct function as arg
23811 (loop_exits_before_overflow): Adjust.
23812 (free_numbers_of_iterations_estimates_loop): Use an overload.
23813 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
23814 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
23816 2017-06-19 Richard Biener <rguenther@suse.de>
23819 * ipa-prop.c (find_constructor_constant_at_offset): Handle
23820 RANGE_EXPR conservatively.
23822 2017-06-16 Carl Love <cel@us.ibm.com>
23824 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23825 definitions for vec_float, vec_float2, vec_floato,
23826 vec_floate built-ins.
23827 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
23828 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
23830 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
23831 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
23832 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
23833 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
23834 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
23835 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
23836 vec_floato): Add builtin defines.
23837 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
23838 Update the built-in documentation file for the new built-in
23841 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23843 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
23844 (mthumb): Mark as the negative of -marm.
23846 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23848 * doc/invoke.texi (ARM Options, -mcpu): Document supported
23850 (ARM Options, -mtune): Document that this accepts the same
23851 extension options as -mcpu.
23852 (ARM Options, -mfpu): Document addition of -mfpu=auto.
23854 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23856 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
23857 permitted extensions.
23859 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23861 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
23862 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
23863 (armv8-m.main): Add option +nodsp.
23864 * config/arm/arm-cpu-cdata.h: Regenerated.
23866 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23868 * config/arm/t-fuchsia: New file.
23869 * config.gcc (arm*-*-fuchsia*): Use it.
23871 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23873 * config/arm/t-symbian: Rewrite for new option infrastructure.
23875 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23877 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
23878 (MULTILIB_REQUIRED): Likewise.
23880 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23882 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
23883 (MULTILIB_RESUE): Likewise.
23884 (MULTILIB_MATCHES): Likewise.
23885 (MULTLIB_REQUIRED): Likewise.
23887 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23889 * config/arm/t-rtems: Rewrite for new option framework.
23891 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23893 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
23894 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
23895 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
23896 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
23897 * config/arm/t-multilib: ... here.
23898 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
23899 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
23900 armv7-a and armv8*-a when A-profile libraries have not been built.
23901 * config/arm/t-rmprofile: Rewrite.
23903 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23905 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
23906 with a backslash. Remove the backslash after substituting unescaped
23908 * doc/fragments.texi (MULTILIB_REUSE): Document it.
23910 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23912 * config.gcc: (arm*-*-*): When building a-profile libraries, force
23913 the driver to pass through the default setting of -mfloat-abi.
23914 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
23916 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
23917 (all_feat_combs): New rule.
23918 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
23920 * config/arm/t-aprofile: Rewrite.
23922 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23924 * config/arm/arm.h (FPUTYPE_AUTO): Define.
23925 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
23926 fpu is not specified by the user/command-line.
23927 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
23928 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
23929 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
23930 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
23931 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
23932 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
23934 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23936 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
23937 * config/arm/t-arm-elf: Rewritten.
23939 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23941 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
23942 have some floating-point instructions.
23943 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
23944 (TARGET_MAYBE_HARD_FLOAT): New macro.
23945 * config/arm/arm-builtins.c (arm_init_builtins): Use
23946 TARGET_MAYBE_HARD_FLOAT.
23947 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
23949 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23951 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
23952 (configargs.h): Include it.
23953 (arm_print_hint_for_fpu_option): New function.
23954 (arm_parse_fpu_option): New function.
23955 (candidate_extension): New class.
23956 (arm_canon_for_multilib): New function.
23957 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
23958 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
23959 (ARCH_CANONICAL_SPECS): New macro.
23960 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
23962 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23964 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
23965 are set after handling multilib fragments. Set target_cpu_default2
23968 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23970 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
23972 (arm*-*-*): Set target_cpu_default2 to a quoted string.
23973 * config/arm/parsecpu.awk (check_cpu): Validate any extension
23975 (check_arch): Likewise.
23976 * config/arm/arm.c (arm_configure_build_target): Handle
23977 TARGET_CPU_DEFAULT being a string constant. Scan any feature
23978 options in the default.
23980 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23982 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
23983 when an option is an alias of another.
23984 * config/arm/parsecpu.awk (optalias): New parser token.
23985 (gen_comm_data): Mark non-alias options as such. Emit entries
23986 for extension aliases.
23987 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
23988 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
23989 (armv6kz, armv6zk, armv6t2): Likewise.
23990 (armv7): Make vfpv3-d16 an alias.
23991 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
23993 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
23994 Sort in canonical order.
23995 (armv8-a): Sort in canonical order.
23996 (armv8.1-a, armv8.2-a): Likewise.
23997 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
23999 (cortex-a9): Sort in canonical order.
24000 * config/arm/arm.c (selftests.h): Include it.
24001 (arm_test_cpu_arch_data): New function.
24002 (arm_run_self_tests): New function.
24003 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
24004 (targetm): Move declaration to the end of the file.
24005 * arm-cpu-cdata.h: Regenerated.
24007 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24009 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
24010 call to target_mode_check describing the type of option passed.
24011 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
24012 (arm_target_thumb_only): Use arm_parse_arch_option_name or
24013 arm_parse_cpu_option_name to match parameters against list of
24015 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
24016 arm_arch_core_flags data structure.
24017 * config/arm/arm-cpu_cdata.h: Regenerated.
24019 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24021 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
24023 (arm_print_hint_for_cpu_option): Likewise.
24024 (arm_print_hint_for_arch_option): Likewise.
24025 (arm_parse_cpu_option_name): Likewise.
24026 (arm_parse_arch_option_name): Likewise.
24027 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
24028 of entries in the all_fpus list.
24029 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
24030 (arm_parse_cpu_option_name): Declare.
24031 (arm_parse_arch_option_name): Declare.
24032 (arm_parse_option_features): Declare.
24033 (arm_intialize_isa): Declare.
24034 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
24036 (gen_comm_data): ... here. Make definitions non-static.
24037 * config/arm/arm-cpu-data.h: Regenerated.
24038 * config/arm/arm-cpu-cdata.h: Regenerated.
24040 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24042 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
24043 (cpu_arch_extension): New structure.
24044 (cpu_arch_option, arch_option, cpu_option): New structures.
24045 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
24046 architecture types.
24047 (gen_data): Generate new format data tables.
24048 * config/arm/arm.c (cpu_tune): New structure.
24049 (cpu_option, processors): Delete.
24050 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
24051 (arm_print_hint_for_cpu_option): ... this and ...
24052 (arm_print_hint_for_arch_option): ... this.
24053 (arm_parse_arch_cpu_name): Delete. Replace with ...
24054 (arm_parse_cpu_option_name): ... this and ...
24055 (arm_parse_arch_option_name): ... this.
24056 (arm_unrecognized_feature): Change type of target parameter to
24058 (arm_parse_arch_cpu_features): Delete. Replace with ...
24059 (arm_parse_option_features): ... this.
24060 (arm_configure_build_target): Rework to use new configuration data
24062 (arm_print_tune_info): Rework for new configuration data tables.
24063 * config/arm/arm-cpu-data.h: Regenerated.
24064 * config/arm/arm-cpu.h: Regenerated.
24066 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24068 * Makefile.in (OBJS): Move sbitmap.o from here ...
24069 (OBJS-libcommon): ... to here.
24071 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24073 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
24074 (ISA_ALL_CRYPTO): New macro.
24075 (ISA_ALL_SIMD): New macro
24076 (ISA_ALL_FP): New macro.
24077 * config/arm/arm.c (fpu_bitlist): Update initializer.
24078 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
24080 (arm9e): Add fpu. Add option for nofp
24081 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
24082 (arm926ej-s, arm1026ej-s): Likewise.
24083 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
24084 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
24085 neon-fp16, neon-vfpv4, nofp and nosimd.
24086 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
24087 (cortex-a8): Add fpu. Add option for nofp.
24088 (cortex-a9): Add fpu. Add options for nosimd and nofp.
24089 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
24090 (cortex-r4f): Add fpu.
24091 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
24092 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
24094 (cortex-r8): Likewise.
24095 (cortex-m4): Add fpu. Add option for nofp.
24096 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
24097 (cortex-a17.cortex-a7): Likewise.
24098 (cortex-a32): Add fpu. Add options for crypto and nofp.
24099 (cortex-a35, cortex-a53): Likewise.
24100 (cortex-a57): Add fpu. Add option for crypto.
24101 (cortex-a72, cortex-a73): Likewise.
24102 (exynos-m1): Likewise.
24103 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
24104 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
24105 (cortex-m33): Add fpu. Add option for nofp.
24106 * config/arm/arm-cpu-cdata.h: Regenerated
24107 * config/arm/arm-cpu-data.h: Regenerated.
24109 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24111 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
24112 (armv5te, armv5tej): Likewise.
24113 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
24114 (armv7): Add options fp and vfpv3-d16.
24115 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
24116 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
24118 (armv7ve): Likewise.
24119 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
24120 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
24121 (armv8-a): Add nocrypto option.
24122 (armv8.1-a, armv8.2-a): Likewise.
24123 (armv8-m.main): add options fp, fp.dp and nofp.
24125 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24127 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
24129 (armv8-a+crc): Delete.
24130 (armv8.1-a): Add options simd, crypto and nofp.
24131 (armv8.2-a): Add options fp16, simd, crypto and nofp.
24132 (armv8.2-a+fp16): Delete.
24133 (armv8-m.main): Add option dsp.
24134 (armv8-m.main+dsp): Delete.
24135 (cortex-a8): Add fpu. Add nofp option.
24136 (cortex-a9): Add fpu. Add nofp and nosimd options.
24137 * config/arm/parsecpu.awk (gen_data): Generate option tables and
24138 link to main cpu and architecture data structures.
24139 (gen_comm_data): Only put isa attributes from the main architecture
24141 (option): New statement for architecture and CPU entries.
24142 * arm.c (struct cpu_option): New structure.
24143 (struct processors): Add entry for options.
24144 (arm_unrecognized_feature): New function.
24145 (arm_parse_arch_cpu_name): Ignore any characters after the first
24147 (arm_parse_arch_cpu_feature): New function.
24148 (arm_configure_build_target): Separate out any CPU and architecture
24149 features and parse separately. Don't error out if -mfpu=auto is
24150 used with only an architecture string.
24151 (arm_print_asm_arch_directives): New function.
24152 (arm_file_start): Call it.
24153 * config/arm/arm-cpu-cdata.h: Regenerated.
24154 * config/arm/arm-cpu-data.h: Likewise.
24155 * config/arm/arm-tables.opt: Likewise.
24157 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24159 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
24160 assembler when it is not -mfpu=auto.
24162 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24164 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
24165 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
24166 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
24167 (ASM_CPU_SPEC): Rewrite.
24168 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
24169 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
24170 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
24171 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
24172 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
24173 copied string is NUL-terminated. Also strip any characters prefixed
24175 (arm_rewrite_selected_arch): New function.
24176 (arm_rewrite_march): New function.
24178 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24180 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
24181 (x_arm_cpu_string, x_arm_tune_string): Likewise.
24182 (march, mcpu, mtune): Convert to string-based options.
24183 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
24184 (arm_parse_arch_cpu_name): New function.
24185 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
24186 identify selected architecture or CPU.
24187 (arm_option_save): New function.
24188 (TARGET_OPTION_SAVE): Redefine.
24189 (arm_option_restore): Restore string options.
24190 (arm_option_print): Print string options.
24192 2017-06-16 Martin Sebor <msebor@redhat.com>
24194 PR tree-optimization/80933
24195 PR tree-optimization/80934
24196 * builtins.c (fold_builtin_3): Do not handle bcmp here.
24197 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
24198 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
24199 (gimple_fold_builtin): Call them.
24201 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24203 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
24204 as unlikely; update profile.
24206 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24208 * predict.c (force_edge_cold): Handle declaring edges impossible
24211 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24213 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
24215 (try_unroll_loop_completely): Fix reporting.
24217 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24219 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
24221 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
24224 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
24225 if given a non-constant argument for an intrinsic which requires a
24228 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24230 * profile.c (compare_freqs): New function.
24231 (branch_prob): Sort edge list.
24232 (find_spanning_tree): Assume that the list is priority sorted.
24234 2017-06-16 Richard Biener <rguenther@suse.de>
24236 PR tree-optimization/81090
24237 * passes.def (pass_record_bounds): Remove.
24238 * tree-pass.h (make_pass_record_bounds): Likewise.
24239 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
24240 make_pass_record_bounds): Likewise.
24241 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
24242 not free niter estimates at the beginning but at the end.
24243 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
24245 2017-06-16 Richard Biener <rguenther@suse.de>
24247 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
24248 initializer to workaround ICE in host GCC 4.8.
24250 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24252 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
24254 (clone_inlined_nodes): Update.
24256 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24258 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
24259 prefetch settings, and enable prefetching by default at -O3.
24261 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24263 * config/aarch64/aarch64.c (aarch64_override_options_internal):
24264 Set flag_prefetch_loop_arrays according to tuning data.
24266 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24268 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
24269 New tune structure.
24270 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
24271 [Unrelated to main purpose of the patch] Place the pointer field last
24272 to enable type checking errors when tune structure are wrongly merged.
24273 * config/aarch64/aarch64.c (generic_prefetch_tune,)
24274 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
24275 (thunderx2t99_prefetch_tune): New tune constants.
24276 (tune_params *_tunings): Update all tunings (no functional change).
24277 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
24278 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
24279 from tunings structures.
24281 2017-06-16 Jakub Jelinek <jakub@redhat.com>
24284 * ubsan.c (instrument_null): Add T argument, use it instead
24285 of computing it based on IS_LHS.
24286 (instrument_object_size): Likewise.
24287 (pass_ubsan::execute): Adjust instrument_null and
24288 instrument_object_size callers to pass gimple_get_lhs or
24289 gimple_assign_rhs1 result to it. Use instrument_null instead of
24290 calling get_base_address and instrument_mem_ref. Handle
24291 aggregate call arguments for object-size sanitization.
24293 2017-06-16 Yury Gribov <tetra2005@gmail.com>
24295 PR tree-optimization/81089
24296 * tree-vrp.c (is_masked_range_test): Validate operands of
24299 2017-06-15 Martin Sebor <msebor@redhat.com>
24302 * dumpfile.c (dump_register): Avoid calling memset to initialize
24303 a class with a default ctor.
24304 * gcc.c (struct compiler): Remove const qualification.
24305 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
24306 * hash-table.h: Ditto.
24307 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
24309 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
24310 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
24312 * params.h (struct param_info): Make struct members non-const.
24313 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
24314 with default initialization.
24315 * vec.h (vec_copy_construct, vec_default_construct): New helper
24317 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
24318 with vec_copy_construct.
24319 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
24320 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
24321 * doc/invoke.texi (-Wclass-memaccess): Document.
24323 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24325 * emit-rtl.h (is_leaf): Update comment about local
24326 register allocator.
24328 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
24331 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
24332 for a variable to have a section before checking if the section has a
24334 Set section to.persistent if persistent attribute is set.
24335 Warn if .persistent attribute is used on an automatic variable.
24337 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
24339 PR rtl-optimization/80474
24340 * reorg.c (update_block): Do not ignore instructions in a delay slot.
24342 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
24344 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
24347 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
24349 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
24350 (casesi): Emit bounds checking as RTL.
24351 (casesi_internal_mips16_<mode>): Remove bounds checking.
24353 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
24355 * config/xtensa/xtensa.c (xtensa_option_override): Append
24356 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
24357 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
24358 xtensa_doloop_hooks): Define unconditionally.
24359 (xtensa_reorg_loops): Only call reorg_loops in the presence of
24361 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
24362 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
24363 for it in xtensa_option_override.
24364 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
24365 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
24367 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
24369 * doc/cppopts.texi: Document '-' special value to -MF.
24371 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
24373 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
24374 (cortex_a53_fconst): Likewise.
24375 (cortex_a53_fpmul): Likewise.
24376 (cortex_a53_f_load_64): Likewise.
24377 (cortex_a53_f_load_many): Likewise.
24378 (cortex_a53_advsimd_alu): Likewise.
24379 (cortex_a53_advsimd_alu_q): Likewise.
24380 (cortex_a53_advsimd_mul): Likewise.
24381 (cortex_a53_advsimd_mul_q): Likewise.
24382 (fpmac bypass): Add new bypass for fpmac-fpmac case.
24383 Add missing fmul, r2f_cvt and fconst cases.
24385 2017-06-14 Richard Biener <rguenther@suse.de>
24387 PR middle-end/81088
24388 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
24390 (fold_binary_loc): When associating do not treat pre-existing
24391 TREE_OVERFLOW on literal constants as a reason to allow
24392 TREE_OVERFLOW on associated literal constants.
24394 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24396 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
24397 (MASK_FEATURES): New macro.
24398 * config/sparc/sparc.c (sparc_option_override): Remove the special
24399 handling of -mfpu and generalize it to all MASK_FEATURES switches.
24401 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24403 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
24404 a division of 0 if non-call exceptions are enabled.
24406 2017-06-14 Andrew Pinski <apinski@cavium.com>
24407 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24410 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
24411 Improve vector initialization code gen for only variable case.
24413 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24415 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
24417 2017-06-14 Richard Biener <rguenther@suse.de>
24419 PR tree-optimization/81083
24420 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
24423 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24425 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
24426 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
24427 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
24428 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
24429 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
24430 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
24432 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24434 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
24435 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
24437 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24439 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
24441 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24443 * config/rs6000/t-rtems: Don't handle SPE.
24445 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24447 * config/rs6000/t-linux: Don't handle SPE.
24449 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24451 * config/rs6000/eabispe.h: Delete file.
24453 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24455 * config/rs6000/t-spe: Delete file.
24457 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24459 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
24460 (rs6000_legitimate_offset_address_p): Return false for anything in
24461 V2SImode or V2SFmode.
24463 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24465 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
24466 except V2SF and V2SI. Rearrange the vector modes, and add comments.
24467 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
24469 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
24470 (rs6000_legitimate_offset_address_p): Ditto.
24471 (rs6000_emit_move): Ditto.
24472 (rs6000_init_builtins): Remove V4HI_type_node.
24474 2017-06-13 Martin Liska <mliska@suse.cz>
24477 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
24478 (gate_asan): Likewise.
24479 * asan.h (asan_no_sanitize_address_p): Remove the function.
24480 (sanitize_flags_p): New function.
24481 * builtins.def: Fix coding style.
24482 * common.opt: Use renamed enum value.
24483 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
24484 * doc/extend.texi: Document no_sanitize attribute.
24485 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
24486 to SANITIZE_UNDEFINED_NONDEFAULT.
24487 * gcc.c (sanitize_spec_function): Use the renamed enum value.
24488 * gimple-fold.c (optimize_atomic_compare_exchange_p):
24489 Use sanitize_flags_p.
24490 * gimplify.c (gimplify_function_tree): Likewise.
24491 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
24492 * opts.c (parse_no_sanitize_attribute): New function.
24493 (common_handle_option): Use renamed enum value.
24494 * opts.h (parse_no_sanitize_attribute): Declare.
24495 * tree.c (sanitize_flags_p): New function.
24496 * tree.h: Declared here.
24497 * tsan.c: Use sanitize_flags_p.
24498 * ubsan.c (ubsan_expand_null_ifn): Likewise.
24499 (instrument_mem_ref): Likewise.
24500 (instrument_bool_enum_load): Likewise.
24501 (do_ubsan_in_current_function): Remove the function.
24502 (pass_ubsan::execute): Use sanitize_flags_p.
24503 * ubsan.h: Remove do_ubsan_in_current_function
24504 * tree-cfg.c (print_no_sanitize_attr_value): New function.
24505 (dump_function_to_file): Use it here.
24507 2017-06-13 Martin Jambor <mjambor@suse.cz>
24509 PR tree-optimization/80803
24510 PR tree-optimization/81063
24511 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
24512 (propagate_subaccesses_across_link): Enqueue subtree whenever
24513 necessary instead of relying on the caller.
24515 2017-06-13 Martin Jambor <mjambor@suse.cz>
24517 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
24518 that have a first_link.
24519 (sort_and_splice_var_accesses): Do not check first_link before
24521 (subtree_mark_written_and_enqueue): Likewise.
24522 (propagate_all_subaccesses): Likewise and do not stop at first
24523 parent with a first_link.
24525 2017-06-13 Martin Jambor <mjambor@suse.cz>
24527 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
24530 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24532 * match.pd: New pattern.
24534 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24536 * tree-vrp.c (is_masked_range_test): New function.
24537 (register_edge_assert_for): Determine ranges for
24540 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24542 PR tree-optimization/67328
24543 * fold-const.c (maskable_range_p): New function.
24544 (build_range_check): Generate bittests if possible.
24546 2017-06-13 Martin Liska <mliska@suse.cz>
24548 * gimple-pretty-print.c (dump_probability): Add new argument.
24549 (dump_edge_probability): Dump both probability and count.
24550 (dump_gimple_label): Likewise.
24551 (dump_gimple_bb_header): Likewise.
24553 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
24556 * config/avr/avr-devices.c: Fix indentation.
24557 * config/avr/gen-avr-mmcu-specs.c: Dito.
24559 2017-06-13 Richard Biener <rguenther@suse.de>
24561 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
24562 instead get vector type from stmt_info.
24563 (vectorizable_reduction): Adjust. Remove dead code.
24565 2017-06-13 Richard Biener <rguenther@suse.de>
24567 PR middle-end/81065
24568 * fold-const.c (extract_muldiv_1): Remove bogus distribution
24569 case of C * (x * C2 + C3).
24570 (fold_addr_of_array_ref_difference): Properly fold index difference.
24572 2017-06-12 David S. Miller <davem@davemloft.net>
24575 * config/sparc/sparc.md (return expander): Emit frame blockage if
24576 function uses alloca.
24578 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
24580 * combine.c (make_field_assignment): Check len rather than the mode
24581 precision when calling force_to_mode.
24583 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
24585 Support multilibs and devices that see flash in RAM address range.
24588 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
24589 (avr_mcu_t) <flash_pm_offset>: New field.
24590 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
24591 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
24592 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
24593 (AVR_TINY_PM_OFFSET): Remove macro.
24594 * config/avr/avr.opt (-mshort-calls): New option.
24595 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24596 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
24597 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
24598 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
24599 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
24600 instead of avr_arch->have_jmp_call.
24601 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
24602 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
24603 avr_arch->flash_pm_offset to define.
24604 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
24605 new field flash_pm_offset. Add entry for avrxmega3.
24606 (avr_texinfo): Add entry for avrxmega3.
24607 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
24608 attiny212, attiny214,
24609 attiny412, attiny414, attiny416, attiny417,
24610 attiny814, attiny816, attiny817,
24611 attiny1614, attiny1616, attiny1617,
24612 attiny3214, attiny3216, attiny3217.
24613 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
24614 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
24615 (avr_print_operand_address) [AVR_TINY]: Same.
24616 (avr_asm_init_sections) <readonly_data_section>: Only patch
24617 callback if avr_arch->flash_pm_offset = 0.
24618 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
24619 for rodata if avr_arch->flash_pm_offset != 0.
24620 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
24621 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
24622 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
24623 (m_options): Append opt_rcall.
24624 (m_dirnames): Append dir_rcall.
24625 * config/avr/t-multilib: Regenerate.
24627 * configure.ac [target=avr]: Check whether avrxmega3 default
24628 linker description file works as needed.
24629 * configure: Regenerate.
24630 * doc/avr-mmcu.texi: Regenerate.
24631 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
24632 <__AVR_ARCH__>: Document avrxmega3 and 103.
24633 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
24634 <__AVR_SHORT_CALLS__>: Document it.
24635 <__AVR_PM_BASE_ADDRESS__>: Document it.
24636 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
24637 (AVR Variable Attributes) <progmem>: Document this is
24638 not needed for avrxmega3.
24639 (AVR Named Address Spaces) <__flash>: Dito.
24641 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
24643 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
24645 2017-06-12 Doug Rupp <rupp@adacore.com>
24647 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
24648 Append vxworks-stdint.h to the tm_file list.
24649 * config/vxworks-stdint.h: New file.
24651 2017-06-12 Martin Liska <mliska@suse.cz>
24653 PR tree-optimization/81041
24654 * tree-profile.c (gimple_gen_ic_func_profiler):
24655 Create an extra BB in profile-generate
24656 (gimple_gen_time_profiler): Likewise.
24658 2017-06-12 Jakub Jelinek <jakub@redhat.com>
24660 PR tree-optimization/81003
24661 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
24662 (update_range_test): Use it instead of force_gimple_operand_gsi.
24664 2017-06-12 Richard Biener <rguenther@suse.de>
24666 PR tree-optimization/81053
24667 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
24668 with backedge value not defined in loop. Simplify def stmt
24671 2017-06-11 Tom de Vries <tom@codesourcery.com>
24674 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
24676 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
24677 nvptx_cannot_force_const_mem.
24679 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24681 * opts.c (finish_options): Move test for flag_split_stack after
24682 it has been initialized.
24684 2017-06-11 Jason Merrill <jason@redhat.com>
24686 * tree.h (id_equal): New.
24687 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
24688 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
24689 instead of strcmp of IDENTIFIER_POINTER.
24691 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24693 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
24694 (mark_all_inlined_calls_cdtor): Fix formating.
24695 (inline_transform): Rescale profile before inlining.
24697 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24699 * cgraph.h (cgraph_edge::clone): Update prototype.
24700 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
24701 (cgraph_node::create_clone): Update.
24702 (cgraph_node::create_version_clone): Update.
24703 * tree-inline.c (copy_bb): Update.
24704 (expand_call_inline): Update.
24706 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
24708 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
24709 factored out from ...
24710 (rs6000_emit_prologue): ... here.
24712 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
24714 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
24715 factored out from ...
24716 (rs6000_emit_prologue): ... here.
24718 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24720 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
24722 (handle_missing_profiles): Fix computation of tp_first_run.
24723 (counts_to_freqs): Do not touch freqs when count is 0.
24725 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24727 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
24730 2017-06-10 Tom de Vries <tom@codesourcery.com>
24732 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
24733 attributes): Document signal effective target.
24735 2017-06-10 Tom de Vries <tom@codesourcery.com>
24737 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
24738 Document effective target stack_size.
24740 2017-06-09 David Malcolm <dmalcolm@redhat.com>
24742 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
24743 to the edit_context if they can be auto-applied.
24745 2017-06-9 Ian Lance Taylor <iant@golang.org>
24747 * opts.c (finish_options): If -fsplit-stack, disable implicit
24748 -forder-blocks-and-partition.
24749 * doc/invoke.texi (Optimize Options): Document that when using
24750 -fsplit-stack -forder-blocks-and-partition is not implicitly
24753 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24755 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
24756 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
24757 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
24758 * builtins.def (abort, trap, unreachable): Declare cold.
24759 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
24760 * tree-core.h (ECF_COLD): New.
24761 * tree.c (set_call_expr_flags): Handle ECF_COLD.
24762 (build_common_builtin_nodes): Mark unreachable and abort as cold.
24764 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24766 * predict.c (unlikely_executed_stmt_p): Cleanup.
24768 2017-06-09 Richard Biener <rguenther@suse.de>
24770 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
24771 model if the ref is always written to.
24773 2017-06-09 Tamar Christina <tamar.christina@arm.com>
24775 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
24777 2017-06-09 Tamar Christina <tamar.christina@arm.com>
24779 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
24782 2017-06-09 Tom de Vries <tom@codesourcery.com>
24785 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
24786 "target cannot support label values" when encountering LABEL_REF.
24788 2017-06-09 Martin Liska <mliska@suse.cz>
24790 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
24791 (gimple_gen_ic_func_profiler): Emit direct comparison
24792 of __gcov_indirect_call_callee with NULL.
24793 (gimple_gen_time_profiler): Change probability from
24794 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
24796 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24798 * profile.c (edge_gcov_counts): Turn to pointer.
24799 (compute_branch_probabilities, compute_branch_probabilities): Update.
24800 (branch_prob): Do not clear edge_gcov_count.
24801 * profile.h (edge_gcov_counts): Turn to pointer.
24802 (edge_gcov_count): Update.
24804 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24806 * gimple.h (gimple_check_failed): Mark cold.
24808 2017-06-09 Richard Biener <rguenther@suse.de>
24810 PR tree-optimization/66623
24811 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
24812 refactor check_reduction into two parts, properly computing
24813 whether we have to check reduction validity for outer loop
24816 2017-06-09 Richard Biener <rguenther@suse.de>
24818 PR tree-optimization/79483
24819 * graphite-scop-detection.c (order): New global.
24820 (get_order): Compute bb to order mapping that satisfies code
24821 generation constraints.
24822 (cmp_pbbs): New helper.
24823 (build_scops): Start domwalk at entry block, sort generated
24826 2017-06-09 Richard Biener <rguenther@suse.de>
24828 PR middle-end/81007
24829 * ipa-polymorphic-call.c
24830 (ipa_polymorphic_call_context::restrict_to_inner_class):
24831 Skip FIELD_DECLs with error_mark_node type.
24832 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
24835 2017-06-09 Martin Liska <mliska@suse.cz>
24837 * predict.c (struct branch_predictor): New struct.
24838 (test_prediction_value_range): New test.
24839 (predict_c_tests): New function.
24840 * selftest-run-tests.c (selftest::run_tests): Run the function.
24841 * selftest.h: Declare new tests.
24843 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
24846 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
24847 gen_add3_insn did not fail.
24848 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
24849 r0, construct that number in a temporary reg and add that reg to r0.
24850 If asked to put the result in r0 as well, fail.
24852 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
24854 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
24855 for early expansion of vec_eqv.
24857 2017-06-08 Jakub Jelinek <jakub@redhat.com>
24859 PR middle-end/81005
24860 * ubsan.c (instrument_null): Avoid pointless code temporary.
24861 (pass_ubsan::execute): Instrument aggregate arguments of calls.
24863 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
24867 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
24870 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
24873 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24875 * predict.c (unlikely_executed_edge_p): Move ahead.
24876 (probably_never_executed_edge_p): Use it.
24878 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
24880 PR middle-end/79988
24881 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
24882 gimple_call_builtin_p call.
24884 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24886 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
24887 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
24888 rtl_check_failed_type2, rtl_check_failed_code1,
24889 rtl_check_failed_code2, rtl_check_failed_code_mode,
24890 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
24891 rtvec_check_failed_bounds, rtl_check_failed_flag,
24892 _fatal_insn_not_found, _fatal_insn): Likewise.
24893 * tree.h (tree_contains_struct_check_failed,
24894 tree_check_failed, tree_not_check_failed,
24895 tree_class_check_failed, tree_range_check_failed,
24896 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
24897 tree_vec_elt_check_failed, phi_node_elt_check_failed,
24898 tree_operand_check_failed, omp_clause_check_failed,
24899 omp_clause_operand_check_failed, omp_clause_range_check_failed):
24902 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24904 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
24905 flag_branch_probabilities.
24906 * ipa-inline.c (edge_badness): Likewise.
24907 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
24908 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
24909 * predict.c (maybe_hot_frequency_p): Likewise.
24910 (probably_never_executed): Likewise.
24911 * sched-ebb.c (schedule_ebbs): Likewise.
24912 * sched-rgn.c (find_single_block_region): Likewise.
24913 * tracer.c (tail_duplicate): Likewise.
24915 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24917 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
24918 longer requires x_flag_profile_use.
24920 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24922 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
24923 instead of flag_reorder_blocks_and_partition.
24924 * dbxout.c (dbxout_function_end): Likewise.
24925 * dwarf2out.c (gen_subprogram_die): Likewise.
24926 * haifa-sched.c (sched_create_recovery_edges): Likewise.
24927 * hw-doloop.c (reorg_loops): Likewise.
24928 * varasm.c (assemble_start_function,
24929 assemble_end_function): Likewise.
24930 (decide_function_section): Do not check for
24931 flag_reorder_blocks_and_partition.
24933 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
24935 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
24937 (chkp_get_hard_register_fake_addr_expr): Ditto.
24938 (chkp_build_addr_expr): Add check for hard reg case.
24939 (chkp_parse_array_and_component_ref): Ditto.
24940 (chkp_find_bounds_1): Ditto.
24941 (chkp_process_stmt): Don't generate bounds store for
24944 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24946 * predict.c (maybe_hot_bb_p): Do not check profile status.
24947 (maybe_hot_edge_p): Likewise.
24948 (probably_never_executed): Check for zero counts even if profile
24950 (unlikely_executed_edge_p): New function.
24951 (unlikely_executed_stmt_p): New function.
24952 (unlikely_executed_bb_p): New function.
24953 (set_even_probabilities): Use unlikely predicates.
24954 (combine_predictions_for_bb): Likewise.
24955 (predict_paths_for_bb): Likewise.
24956 (predict_paths_leading_to_edge): Likewise.
24957 (determine_unlikely_bbs): New function.
24958 (estimate_bb_frequencies): Use it.
24959 (compute_function_frequency): Use zero counts even if profile is
24961 * profile-count.h: Fix typo.
24963 2017-08-08 Julia Koval <julia.koval@intel.com>
24965 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
24966 _mm512_mask_cvtsepi16_storeu_epi8,
24967 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
24968 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
24969 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
24970 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
24971 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
24972 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
24973 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
24974 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
24975 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
24976 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
24977 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
24978 __builtin_ia32_pmovuswb256mem_mask,
24979 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
24980 __builtin_ia32_pmovwb512mem_mask): New builtins.
24982 2017-08-08 Julia Koval <julia.koval@intel.com>
24984 PR target/73350,80862
24985 * config/i386/subst.md (round): Fix round pattern.
24986 * config/i386/i386.c (ix86_erase_embedded_rounding):
24987 Fix erasing rounding for the fixed pattern.
24989 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24991 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
24993 2017-06-08 Martin Liska <mliska@suse.cz>
24995 PR gcov-profile/80911
24996 * gcov.c (block_info::block_info): New constructor.
24998 2017-06-07 Carl Love <cel@us.ibm.com>
25000 * config/rs6000/rs6000-c: The return type of the following
25001 built-in functions was implemented as int not long long. Fix sign
25002 of return value for the unsigned version of vec_mulo and vec_mule.
25003 vector unsigned long long vec_bperm (vector unsigned long long,
25004 vector unsigned char)
25005 vector signed long long vec_mule (vector signed int,
25007 vector unsigned long long vec_mule (vector unsigned int,
25008 vector unsigned int)
25009 vector signed long long vec_mulo (vector signed int,
25011 vector unsigned long long vec_mulo (vector unsigned int,
25012 vector unsigned int)
25013 * doc/extend.texi: Fix the documentation for the built-in
25016 2017-06-07 Carl Love <cel@us.ibm.com>
25019 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
25022 2017-06-07 Carl Love <cel@us.ibm.com>
25024 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
25025 support, Generate doublehv for signed int/float for BE case only.
25027 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
25029 * doc/invoke.texi (mcx16): Rewrite.
25031 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25033 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
25034 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
25035 *mov<mode>_softfloat, and an anonymous splitter): Use
25036 nonimmediate_operand instead of rs6000_nonimmediate_operand.
25038 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25040 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
25042 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25043 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
25044 (rs6000_debug_reg_global): Adjust.
25045 (rs6000_init_hard_regno_mode_ok): Adjust.
25046 (rs6000_dbx_register_number): Adjust.
25047 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
25048 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25049 Remove SPE_ACC and SPEFSCR.
25050 (REG_ALLOC_ORDER): Ditto.
25051 (FRAME_POINTER_REGNUM): Change to 111.
25052 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
25053 (REG_CLASS_NAMES): Ditto.
25054 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
25055 (REGISTER_NAMES): Ditto.
25056 (ADDITIONAL_REG_NAMES): Ditto.
25057 (rs6000_reg_names): Ditto.
25058 * config/rs6000/rs6000.md: Renumber some register number
25061 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25063 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
25065 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25066 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
25068 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
25069 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25070 Delete the SPE high registers.
25071 (REG_ALLOC_ORDER): Ditto.
25072 (enum reg_class): Remove SPE_HIGH_REGS.
25073 (REG_CLASS_NAMES): Ditto.
25074 (REG_CLASS_CONTENTS): Delete the SPE high registers.
25075 (REGISTER_NAMES): Ditto.
25076 (rs6000_reg_names): Ditto.
25077 * doc/tm.texi.in: Remove SPE as example.
25078 * doc/tm.texi: Regenerate.
25080 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25082 * config/rs6000/8540.md (ppc8540_brinc): Delete.
25083 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
25084 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
25085 * config/rs6000/rs6000.md (type): Remove "brinc".
25087 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25089 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
25090 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
25091 * config/rs6000/linuxspe.h: Delete file.
25092 * config/rs6000/rs6000.md: Don't include spe.md.
25093 * config/rs6000/spe.h: Delete file.
25094 * config/rs6000/spe.md: Delete file.
25095 * config/rs6000/t-rs6000: Remove spe.md.
25097 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25099 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
25100 (reg_or_none500mem_operand): Delete.
25101 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
25102 instead of reg_or_none500mem_operand.
25104 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25106 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
25107 handling of SPE flags.
25108 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
25110 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25112 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
25114 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
25115 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
25116 paired_divv2sf3): Similar.
25117 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
25118 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
25119 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
25121 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
25122 Rename the paired_* instruction patterns.
25123 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
25125 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
25126 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
25127 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
25128 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
25129 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
25130 PAIRED_VECTOR_MODE.
25131 (struct machine_function): Delete field spe_insn_chain_scanned_p.
25132 (spe_func_has_64bit_regs_p): Delete.
25133 (spe_expand_predicate_builtin): Delete.
25134 (spe_expand_evsel_builtin): Delete.
25135 (TARGET_DWARF_REGISTER_SPAN): Do not define.
25136 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
25137 (invalid_e500_subreg): Delete.
25138 (rs6000_legitimize_address): Always force_reg op2 as well, for
25139 paired single memory accesses.
25140 (rs6000_member_type_forces_blk): Delete.
25141 (rs6000_spe_function_arg): Delete.
25142 (rs6000_expand_unop_builtin): Delete SPE handling.
25143 (rs6000_expand_binop_builtin): Ditto.
25144 (spe_expand_stv_builtin): Delete.
25145 (bdesc_2arg_spe): Delete.
25146 (spe_expand_builtin): Delete.
25147 (spe_expand_predicate_builtin): Delete.
25148 (spe_expand_evsel_builtin): Delete.
25149 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
25150 (spe_init_builtins): Delete.
25151 (spe_func_has_64bit_regs_p): Delete.
25152 (savres_routine_name): Delete "info" parameter. Adjust callers.
25153 (rs6000_emit_stack_reset): Ditto.
25154 (rs6000_dwarf_register_span): Delete.
25155 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
25156 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
25157 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
25159 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
25161 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
25162 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
25163 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
25164 mulv2sf3, divv2sf3): Delete expanders.
25166 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25168 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
25170 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25172 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
25173 * config/rs6000/rs6000.c: Ditto.
25175 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25177 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
25178 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
25179 comparison_operator.
25181 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25183 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
25184 * config/rs6000/rs6000.opt: Ditto.
25185 * config/rs6000/t-rtems: Ditto.
25187 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25189 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
25190 TARGET_E500_SINGLE by 0, simplify.
25191 * config/rs6000/rs6000.c: Ditto.
25192 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
25193 (spe_build_register_parallel): Delete.
25194 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
25195 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
25196 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
25197 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
25198 (E500_CONVERT): Delete.
25199 * config/rs6000/spe.md: Remove many patterns and all define_constants.
25201 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25203 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
25204 * config/rs6000/dfp.md: Ditto.
25205 (negdd2, *negdd2_fpr): Merge.
25206 (absdd2, *absdd2_fpr): Merge.
25207 (negtd2, *negtd2_fpr): Merge.
25208 (abstd2, *abstd2_fpr): Merge.
25209 * config/rs6000/e500.h: Delete file.
25210 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
25211 TARGET_FPRS by 1 and simplify.
25212 * config/rs6000/rs6000-c.c: Ditto.
25213 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
25214 TARGET_DF_SPE by 0.
25215 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
25217 * config/rs6000/rs6000.md: Ditto.
25218 (floatdidf2, *floatdidf2_fpr): Merge.
25219 (move_from_CR_gt_bit): Delete.
25220 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
25221 (E500_CR_IOR_COMPARE): Delete.
25222 (All patterns that require !TARGET_FPRS): Delete.
25223 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
25225 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25227 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
25229 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25231 * graphds.c (add_edge): Intitialize edge's attached data.
25232 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
25233 pointer parameter. Call pointed function on each edge during
25234 graph traversing. Skip traversing the edge when the function
25236 (graphds_dfs, graphds_scc): Ditto.
25237 (for_each_edge): New parameter. Pass the new parameter to callback
25239 * graphds.h (skip_edge_callback): New function pointer type.
25240 (graphds_dfs, graphds_scc): New function pointer parameter.
25241 (graphds_edge_callback, for_each_edge): New parameter.
25243 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25245 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
25246 out code checking if runtime alias check is possible to below ...
25247 Call the new function.
25248 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
25249 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
25251 2017-06-07 Marek Polacek <polacek@redhat.com>
25254 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
25255 TYPE_OVERFLOW_WRAPS check.
25257 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25259 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
25260 if versioning is required.
25261 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
25262 peeling with the check for versioning.
25264 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25266 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
25267 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
25268 Set true to new parameter if new ssa variable is defined.
25269 (vect_gen_vector_loop_niters): Refactor. Set range information
25270 for the new vector loop bound variable.
25271 (vect_do_peeling): Ditto.
25273 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25275 * tree-affine.c (ssa.h): Include header file.
25276 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
25277 has wrapping overflow behavior.
25279 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25281 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
25283 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25285 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
25286 (tree_to_aff_combination): ... here.
25288 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25290 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
25291 reg_pressure model function.
25292 (ivopts_global_cost_for_size): Delete.
25293 (determine_set_costs, iv_ca_recount_cost): Call new model function
25294 ivopts_estimate_reg_pressure.
25296 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25298 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
25299 expensive than udiv. Remove floating point cases from mod.
25301 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25303 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
25304 Increase idiv cost.
25306 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25308 * config/aarch64/aarch64.md
25309 (copysignsf3): Fix mask generation.
25311 2017-06-07 Jakub Jelinek <jakub@redhat.com>
25313 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
25315 (class dump_manager): Add register_dumps method.
25316 * dumpfile.c: Include langhooks.h.
25317 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
25318 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
25319 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
25320 (dump_manager::dump_register): Start with 512 entries instead of 32.
25321 (dump_manager::register_dumps): New method.
25322 * toplev.c (general_init): Instead of invoking register_dumps
25323 langhook, invoke register_dumps method on the dump manager.
25324 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
25327 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
25329 * doc/md.texi: Clarify the restrictions on a define_insn condition.
25330 Say that # requires an associated define_split to exist, and that
25331 the define_split must be suitable for use after register allocation.
25333 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25335 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
25336 (compute_outgoing_frequencies): Also initialize zero counts.
25337 (find_many_sub_basic_blocks): Do not produce uninitialized profile
25338 around loops; preserve more of profile when nothing changes.
25340 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
25342 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
25344 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
25345 * config/arm/arm-cpu-cdata.h: Regenerate.
25346 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
25347 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
25348 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
25349 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
25350 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
25352 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
25354 * config/arm/t-rmprofile: Likewise.
25355 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
25357 2017-06-06 David S. Miller <davem@davemloft.net>
25360 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
25361 blockage if function uses alloca.
25363 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25365 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
25366 New "uid" fields to hold pretty-print IDs of group and ref.
25367 Memory references are now identified as <group_id>:<ref_id>
25368 instead of using [random] addresses.
25369 (dump_mem_details): Simplify, no functional change.
25370 (dump_mem_ref): Simplify and make output more concise.
25371 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
25372 (find_or_create_group): Initialize group uid.
25373 (record_ref): Initialize ref uid. Improve debug output.
25374 (prune_group_by_reuse, should_issue_prefetch_p,)
25375 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
25376 (mark_nontemporal_store, determine_loop_nest_reuse):
25377 Improve debug output.
25379 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25381 * dbgcnt.def (prefetch): New debug counter.
25382 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
25383 (schedule_prefetches): Stop issueing prefetches if debug counter
25386 2017-06-06 Tom de Vries <tom@codesourcery.com>
25388 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
25389 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
25391 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25393 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
25394 Use aarch64_reg_or_zero predicate for operand 4.
25395 (aarch64_compare_and_swap<mode> define_insn_and_split):
25396 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
25397 (aarch64_store_exclusive<mode>): Likewise for operand 2.
25399 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25401 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
25402 (arm_compute_save_core_reg_mask): This.
25403 (thumb1_compute_save_reg_mask): Rename into ...
25404 (thumb1_compute_save_core_reg_mask): This.
25405 (arm_compute_save_reg0_reg12_mask): Adapt comment.
25406 (arm_compute_frame_layout): Likewise.
25408 2017-06-06 Richard Biener <rguenther@suse.de>
25410 PR tree-optimization/80974
25411 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
25412 keep or clear leaders SSA info.
25414 2017-06-06 Tom de Vries <tom@codesourcery.com>
25416 * config/nvptx/nvptx.c (split_mode_p): New function.
25417 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
25419 2017-06-06 Tom de Vries <tom@codesourcery.com>
25421 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
25423 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25426 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
25429 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25431 * shrink-wrap.c (handle_simple_exit): Update profile.
25432 (try_shrink_wrapping): Upate profile.
25434 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25436 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
25437 (tree_guess_outgoing_edge_probabilities): New.
25438 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
25439 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
25441 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25443 * ipa-split.c (split_function): Initialize return bb profile.
25445 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25447 * profile.c (compute_branch_probabilities): Also initialize
25448 EXIT_BLOCK profile.
25450 2017-06-06 Richard Biener <rguenther@suse.de>
25452 PR tree-optimization/80928
25453 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
25454 (vect_analyze_loop_operations): Properly guard analysis for
25456 (vect_transform_loop): Likewise.
25457 (vect_analyze_loop_2): Also reset SLP type on PHIs.
25458 (vect_model_induction_cost): Do not cost for pure SLP.
25459 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
25460 of induction in inner loop vectorization.
25461 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
25462 (vect_get_and_check_slp_defs): Handle vect_induction_def.
25463 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
25465 (vect_analyze_slp_cost_1): Cost induction.
25466 (vect_detect_hybrid_slp_stmts): Handle PHIs.
25467 (vect_get_slp_vect_defs): Likewise.
25468 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
25469 (vect_transform_stmt): Handle SLP reductions.
25470 * tree-vectorizer.h (vectorizable_induction): Adjust.
25472 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25474 * config/rs6000/rs6000.c (make_resolver_func): Update
25475 init_lowered_empty_function call.
25477 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
25479 * doc/invoke.texi: Document the -fprofile-abs-path option.
25480 * common.opt (fprofile-abs-path): New option.
25481 * gcov-io.h (gcov_write_filename): Declare.
25482 * gcov-io.c (gcov_write_filename): New function.
25483 * coverage.c (coverage_begin_function): Use gcov_write_filename.
25484 * profile.c (output_location): Likewise.
25486 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25488 * shring-wrap.c: Revert accidental commit.
25490 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
25492 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
25494 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25496 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
25498 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
25499 profile in callgraph edge.
25500 * profile-count.h (apply_probability): If THIS is 0, then result is 0
25501 (apply_scale): Likewise.
25502 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
25503 Also scale profile when inlining function with zero profile.
25504 (initialize_cfun): Update exit block profile even when it is zero.
25505 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
25506 when profile is read.
25508 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25510 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
25511 (CLONE_*): New constants to define the processors we can generate
25512 code for with the target_clone attribute.
25513 (rs6000_clone_map): New array to identify which clone processors
25514 the current program is running on.
25515 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
25516 target_clone attribute.
25517 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
25518 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
25519 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
25520 (cpu_expand_builtin): Add support for target_clone attribute.
25521 (rs6000_valid_attribute_p): Allow "default" attribute.
25522 (get_decl_name): New debug function to simplify printing the
25523 current function name in debugging statements.
25524 (rs6000_clone_priority): New functions to support the target_clone
25525 attribute, and be able to generate code to switch between ISA 2.05
25526 through ISA 3.0 (power6 through power9).
25527 (rs6000_compare_version_priority): Likewise.
25528 (rs6000_get_function_versions_dispatcher): Likewise.
25529 (make_resolver_func): Likewise.
25530 (add_condition_to_bb): Likewise.
25531 (dispatch_function_versions): Likewise.
25532 (rs6000_generate_version_dispatcher_body): Likewise.
25533 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
25534 (fusion_gpr_load_p): Fix a spacing issue.
25535 * doc/extend.texi (Common Function Attributes): Document that the
25536 PowerPC supports the target_clone attribute.
25538 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25540 * config/arm/arm.h: explain F symbol found in description of ARM
25541 register allocation in its legend.
25543 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25545 * config/mips/frame-header-opt.c: Include profile-count.h.
25546 * config/riscv/riscv.c: Include profile-count.h
25548 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25550 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
25552 (sm_set_flag_if_changed): Add bbs field.
25553 (execute_sm_if_changed_flag_set): Pass BBS.
25554 (execute_sm): Update.
25556 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25558 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
25561 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25563 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
25564 (peephole2): New peephole2 to emit the above.
25565 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
25567 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25569 * config/aarch64/aarch64.c (define_peephole2 above
25570 *sub_<shift>_<mode>): New peephole.
25572 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25574 * config/i386/i386.c (make_resolver_func): Update.
25575 * Makefile.in: Add profile-count.h and profile-count.o
25576 * auto-profile.c (afdo_indirect_call): Update to new API.
25577 (afdo_set_bb_count): Update.
25578 (afdo_propagate_edge): Update.
25579 (afdo_propagate_circuit): Update.
25580 (afdo_calculate_branch_prob): Update.
25581 (afdo_annotate_cfg): Update.
25582 * basic-block.h: Include profile-count.h
25583 (struct edge_def): Turn count to profile_count.
25584 (struct basic_block_def): Likewie.
25585 (REG_BR_PROB_BASE): Move to profile-count.h
25586 (RDIV): Move to profile-count.h
25587 * bb-reorder.c (max_entry_count): Turn to profile_count.
25588 (find_traces): Update.
25589 (rotate_loop):Update.
25590 (connect_traces):Update.
25591 (sanitize_hot_paths):Update.
25592 * bt-load.c (migrate_btr_defs): Update.
25593 * cfg.c (RDIV): Remove.
25594 (init_flow): Use alloc_block.
25595 (alloc_block): Uninitialize count.
25596 (unchecked_make_edge): Uninitialize count.
25597 (check_bb_profile): Update.
25598 (dump_edge_info): Update.
25599 (dump_bb_info): Update.
25600 (update_bb_profile_for_threading): Update.
25601 (scale_bbs_frequencies_int): Update.
25602 (scale_bbs_frequencies_gcov_type): Update.
25603 (scale_bbs_frequencies_profile_count): New.
25604 * cfg.h (update_bb_profile_for_threading): Update.
25605 (scale_bbs_frequencies_profile_count): Declare.
25606 * cfgbuild.c (compute_outgoing_frequencies): Update.
25607 (find_many_sub_basic_blocks): Update.
25608 * cfgcleanup.c (try_forward_edges): Update.
25609 (try_crossjump_to_edge): Update.
25610 * cfgexpand.c (expand_gimple_tailcall): Update.
25611 (construct_exit_block): Update.
25612 * cfghooks.c (verify_flow_info): Update.
25613 (dump_bb_for_graph): Update.
25614 (split_edge): Update.
25615 (make_forwarder_block): Update.
25616 (duplicate_block): Update.
25617 (account_profile_record): Update.
25618 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
25619 (get_estimated_loop_iterations): Update.
25620 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
25621 (single_likely_exit): Update.
25622 * cfgloopmanip.c (scale_loop_profile): Update.
25624 (set_zero_probability): Update.
25625 (lv_adjust_loop_entry_edge): Update.
25626 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
25627 (purge_dead_edges): Update.
25628 (rtl_account_profile_record): Update.
25629 * cgraph.c (cgraph_node::create): Uninitialize count.
25630 (symbol_table::create_edge): Uninitialize count.
25631 (cgraph_update_edges_for_call_stmt_node): Update.
25632 (cgraph_edge::dump_edge_flags): Update.
25633 (cgraph_node::dump): Update.
25634 (cgraph_edge::maybe_hot_p): Update.
25635 * cgraph.h: Include profile-count.h
25636 (create_clone), create_edge, create_indirect_edge): Update.
25637 (cgraph_node): Turn count to profile_count.
25638 (cgraph_edge0: Likewise.
25639 (make_speculative, clone): Update.
25640 (create_edge): Update.
25641 (init_lowered_empty_function): Update.
25642 * cgraphclones.c (cgraph_edge::clone): Update.
25643 (duplicate_thunk_for_node): Update.
25644 (cgraph_node::create_clone): Update.
25645 * cgraphunit.c (cgraph_node::analyze): Update.
25646 (cgraph_node::expand_thunk): Update.
25647 * final.c (dump_basic_block_info): Update.
25648 * gimple-streamer-in.c (input_bb): Update.
25649 * gimple-streamer-out.c (output_bb): Update.
25650 * graphite.c (print_global_statistics): Update.
25651 (print_graphite_scop_statistics): Update.
25652 * hsa-brig.c: Include basic-block.h.
25653 * hsa-dump.c: Include basic-block.h.
25654 * hsa-gen.c (T sum_slice): Update.
25655 (convert_switch_statements):Update.
25656 * hsa-regalloc.c: Include basic-block.h.
25657 * ipa-chkp.c (chkp_produce_thunks): Update.
25658 * ipa-cp.c (struct caller_statistics): Update.
25659 (init_caller_stats): Update.
25660 (gather_caller_stats): Update.
25661 (ipcp_cloning_candidate_p): Update.
25662 (good_cloning_opportunity_p): Update.
25663 (get_info_about_necessary_edges): Update.
25664 (dump_profile_updates): Update.
25665 (update_profiling_info): Update.
25666 (update_specialized_profile): Update.
25667 (perhaps_add_new_callers): Update.
25668 (decide_about_value): Update.
25669 (ipa_cp_c_finalize): Update.
25670 * ipa-devirt.c (struct odr_type_warn_count): Update.
25671 (struct decl_warn_count): Update.
25672 (struct final_warning_record): Update.
25673 (possible_polymorphic_call_targets): Update.
25674 (ipa_devirt): Update.
25675 * ipa-fnsummary.c (redirect_to_unreachable): Update.
25676 * ipa-icf.c (sem_function::merge): Update.
25677 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
25678 * ipa-inline.c (compute_uninlined_call_time): Update.
25679 (compute_inlined_call_time): Update.
25680 (want_inline_small_function_p): Update.
25681 (want_inline_self_recursive_call_p): Update.
25682 (edge_badness): Update.
25683 (lookup_recursive_calls): Update.
25684 (recursive_inlining): Update.
25685 (inline_small_functions): Update.
25686 (dump_overall_stats): Update.
25687 (dump_inline_stats): Update.
25688 * ipa-profile.c (ipa_profile_generate_summary): Update.
25689 (ipa_propagate_frequency): Update.
25690 (ipa_profile): Update.
25691 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
25692 * ipa-utils.c (ipa_merge_profiles): Update.
25693 * loop-doloop.c (doloop_modify): Update.
25694 * loop-unroll.c (report_unroll): Update.
25695 (unroll_loop_runtime_iterations): Update.
25696 * lto-cgraph.c (lto_output_edge): Update.
25697 (lto_output_node): Update.
25698 (input_node): Update.
25699 (input_edge): Update.
25700 (merge_profile_summaries): Update.
25701 * lto-streamer-in.c (input_cfg): Update.
25702 * lto-streamer-out.c (output_cfg): Update.
25703 * mcf.c (create_fixup_graph): Update.
25704 (adjust_cfg_counts): Update.
25705 (sum_edge_counts): Update.
25706 * modulo-sched.c (sms_schedule): Update.
25707 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
25708 * predict.c (maybe_hot_count_p): Update.
25709 (probably_never_executed): Update.
25710 (dump_prediction): Update.
25711 (combine_predictions_for_bb): Update.
25712 (propagate_freq): Update.
25713 (handle_missing_profiles): Update.
25714 (counts_to_freqs): Update.
25715 (rebuild_frequencies): Update.
25716 (force_edge_cold): Update.
25717 * predict.h: Include profile-count.h
25718 (maybe_hot_count_p, counts_to_freqs): UPdate.
25719 * print-rtl-function.c: Do not include cfg.h
25720 * print-rtl.c: Include basic-block.h
25721 * profile-count.c: New file.
25722 * profile-count.h: New file.
25723 * profile.c (is_edge_inconsistent): Update.
25724 (correct_negative_edge_counts): Update.
25725 (is_inconsistent): Update.
25726 (set_bb_counts): Update.
25727 (read_profile_edge_counts): Update.
25728 (compute_frequency_overlap): Update.
25729 (compute_branch_probabilities): Update; Initialize and deinitialize
25731 (branch_prob): Update.
25732 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
25733 (edge_gcov_count): New.
25734 (bb_gcov_count): New.
25735 * shrink-wrap.c (try_shrink_wrapping): Update.
25736 * tracer.c (better_p): Update.
25737 * trans-mem.c (expand_transaction): Update.
25738 (ipa_tm_insert_irr_call): Update.
25739 (ipa_tm_insert_gettmclone_call): Update.
25740 * tree-call-cdce.c: Update.
25741 * tree-cfg.c (gimple_duplicate_sese_region): Update.
25742 (gimple_duplicate_sese_tail): Update.
25743 (gimple_account_profile_record): Update.
25744 (execute_fixup_cfg): Update.
25745 * tree-inline.c (copy_bb): Update.
25746 (copy_edges_for_bb): Update.
25747 (initialize_cfun): Update.
25748 (freqs_to_counts): Update.
25749 (copy_cfg_body): Update.
25750 (expand_call_inline): Update.
25751 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25752 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
25753 (try_unroll_loop_completely): Update.
25754 (try_peel_loop): Update.
25755 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25756 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
25757 * tree-ssa-loop-split.c (connect_loops): Update.
25758 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
25759 * tree-ssa-reassoc.c (branch_fixup): Update.
25760 * tree-ssa-tail-merge.c (replace_block_by): Update.
25761 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
25762 (compute_path_counts): Update.
25763 (update_profile): Update.
25764 (recompute_probabilities): Update.
25765 (update_joiner_offpath_counts): Update.
25766 (estimated_freqs_path): Update.
25767 (freqs_to_counts_path): Update.
25768 (clear_counts_path): Update.
25769 (ssa_fix_duplicate_block_edges): Update.
25770 (duplicate_thread_path): Update.
25771 * tree-switch-conversion.c (case_bit_test_cmp): Update.
25772 (struct switch_conv_info): Update.
25773 * tree-tailcall.c (decrease_profile): Update.
25774 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
25775 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25776 * value-prof.c (check_counter): Update.
25777 (gimple_divmod_fixed_value): Update.
25778 (gimple_mod_pow2): Update.
25779 (gimple_mod_subtract): Update.
25780 (gimple_ic_transform): Update.
25781 (gimple_stringop_fixed_value): Update.
25782 * value-prof.h (gimple_ic): Update.
25784 2017-06-02 Carl Love <cel@us.ibm.com>
25786 * config/rs6000/rs6000-c: Add support for built-in functions
25787 vector double vec_doublee (vector signed int);
25788 vector double vec_doublee (vector unsigned int);
25789 vector double vec_doublee (vector float);
25790 vector double vec_doubleh (vector signed int);
25791 vector double vec_doubleh (vector unsigned int);
25792 vector double vec_doubleh (vector float);
25793 vector double vec_doublel (vector signed int);
25794 vector double vec_doublel (vector unsigned int);
25795 vector double vec_doublel (vector float);
25796 vector double vec_doubleo (vector signed int);
25797 vector double vec_doubleo (vector unsigned int);
25798 vector double vec_doubleo (vector float);.
25799 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
25800 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
25802 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
25803 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
25804 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
25806 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
25807 vec_doublel, vec_doubleh.
25808 * doc/extend.texi: Update the built-in documentation file for the
25809 new built-in functions.
25811 2017-06-02 David Malcolm <dmalcolm@redhat.com>
25814 * ipa-inline-analysis.c (free_growth_caches): Set
25815 edge_removal_hook_holder to NULL after removing it.
25817 2017-06-02 Sudakshina Das <sudi.das@arm.com>
25819 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
25820 comparision with zero.
25822 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
25823 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25824 for early expansion of vec_min and vec_max builtins.
25825 (builtin_function_type): Add min/max unsigned variants to those
25826 identified as having unsigned arguments.
25828 2017-06-02 Olivier Hainque <hainque@adacore.com>
25830 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
25832 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25834 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
25835 Use VALL_F16 iterator rather than VALL.
25837 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25839 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
25840 Emit CBNZ inside loop when doing a strong exchange and comparing
25841 against zero. Generate the CC flags after the loop.
25843 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
25845 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
25846 (dl_section_ref): New.
25847 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
25848 On AIX, append an expression to subtract the size of the
25849 section length to dl_section_ref.
25851 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
25853 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25854 for early expansion of vector absolute builtins.
25856 2017-06-02 Richard Biener <rguenther@suse.de>
25858 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
25859 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
25861 2017-06-02 Richard Biener <rguenther@suse.de>
25863 PR tree-optimization/80948
25864 * tree-tailcall.c (find_tail_calls): Track stmts to move in
25865 stmt order as well.
25867 2017-06-02 Richard Biener <rguenther@suse.de>
25869 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
25871 * tree-vect-stmts.c (process_use): Do not mark backedge defs
25872 for inductions as relevant.
25874 2017-06-02 Richard Biener <rguenther@suse.de>
25876 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
25877 (vectorizable_induction): ... this. Remove dead code.
25879 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
25881 * builtins. (expand_builtin_alloca): Remove second parameter and
25882 infer its value from the first parameter instead.
25883 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
25885 2017-06-02 Jakub Jelinek <jakub@redhat.com>
25887 PR rtl-optimization/80903
25888 * loop-doloop.c (add_test): Unshare sequence.
25890 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
25892 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
25894 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
25896 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
25898 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
25899 xlogue_layout::get_instance, logue_layout::xlogue_layout,
25900 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
25901 (xlogue_layout::get_stub_rtx): Make static.
25902 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
25903 (xlogue_layout::compute_stub_managed_regs): Rename to...
25904 (xlogue_layout::count_stub_managed_regs): ...this.
25905 (xlogue_layout::is_stub_managed_reg): New function.
25906 (xlogue_layout::m_stub_names): Rename to...
25907 (xlogue_layout::s_stub_names): ...this, make static.
25908 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
25909 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
25910 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
25911 xlogue_layout::s_stub_names): Instantiate statics.
25912 (stub_managed_regs): Remove.
25913 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
25914 (disable_call_ms2sysv_xlogues): Rename to...
25915 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
25916 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
25918 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
25919 change after reload_completed.
25920 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
25922 (ix86_expand_prologue): Likewise.
25923 (ix86_expand_epilogue): Likewise.
25924 (ix86_expand_split_stack_prologue): Likewise.
25925 (ix86_compute_frame_layout): Remove frame parameter ...
25926 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
25927 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
25929 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
25930 (ix86_frame): Move from here ...
25931 * config/i386/i386.h (ix86_frame): ... to here.
25932 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
25933 complete ix86_frame data structure instead. Remove some_ld_name.
25935 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
25937 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
25938 symbols that hold a DECL_VALUE_EXPR.
25940 2017-06-01 Martin Jambor <mjambor@suse.cz>
25942 PR tree-optimization/80898
25943 * tree-sra.c (process_subtree_disqualification): Removed.
25944 (disqualify_candidate): Do not acll
25945 process_subtree_disqualification.
25946 (subtree_mark_written_and_enqueue): New function.
25947 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
25948 RHS has been disqualified and re-queue LHS if necessary. Apart
25949 from that, ignore disqualified RHS.
25951 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25953 * config/s390/s390.c (s390_emit_epilogue): Disable early return
25954 address fetch for z10 or later.
25956 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25958 * config/arc/arc.md (tst_movb): Add guard when splitting.
25960 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25962 * config/arc/arc.c (arc_can_eliminate): Test against
25963 arc_frame_pointer_needed.
25965 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25967 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
25968 to prevent store reordering.
25969 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
25970 (type): Add block type.
25971 (stack_tie): Define special instruction to be used in
25974 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25976 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
25977 constraint. It is not valid for the pattern.
25978 (noncommutative_binary_comparison): Likewise.
25980 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25982 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
25985 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25987 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
25988 be used by the reg-alloc.
25990 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25992 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
25993 reg-alloc when having mul64 or mul32x16 instructions.
25994 (mulsidi3): Likewise.
25995 (umulsidi3): Likewise.
25996 (mulsi32x16): New pattern.
25997 (mulsi64): Likewise.
25998 (mulsidi64): Likewise.
25999 (umulsidi64): Likewise.
26000 (MUL32x16_REG): Define.
26001 (mul64_600): Use MUL32x16_REG.
26002 (mac64_600): Likewise.
26003 (umul64_600): Likewise.
26004 (umac64_600): Likewise.
26006 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26008 * config/arc/arc.md (mulsi3_700): Make it commutative.
26010 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
26012 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
26014 (*sign_extendsidi2_insn): Likewise for movstosw.
26016 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26018 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
26019 the type of the input discriminant value. Convert the
26020 discriminant value of signedness vary.
26022 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
26024 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
26025 Add to -Wall section.
26027 2017-06-01 Richard Biener <rguenther@suse.de>
26029 PR middle-end/66313
26030 * fold-const.c (fold_plusminus_mult_expr): If the factored
26031 factor may be zero use a wrapping type for the inner operation.
26032 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
26033 and handle moved defs.
26034 (process_assignment): Properly guard the unary op case. Return a
26035 tri-state indicating that moving the stmt before the call may allow
26036 to continue. Pass through to_move.
26037 (find_tail_calls): Handle moving unrelated defs before
26040 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
26043 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
26044 splitter result in the canonical way.
26046 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
26048 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
26049 also for 32bit target. Update insn attributes.
26050 (zero-extendsidi2 splitter): Allow all registers for operand 1.
26052 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
26054 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
26055 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
26056 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
26057 (_mm_maskz_min_ss): New intrinsics.
26059 2017-05-31 Martin Liska <mliska@suse.cz>
26061 * tree-vect-loop.c (vect_create_epilog_for_reduction):
26062 Change comment style to one we normally use.
26063 (vectorizable_reduction): Likewise.
26064 (vectorizable_induction): Likewise.
26065 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
26066 (vectorizable_call): Likewise.
26067 (vectorizable_simd_clone_call): Likewise.
26068 (vectorizable_conversion): Likewise.
26069 (vectorizable_assignment): Likewise.
26070 (vectorizable_shift): Likewise.
26071 (vectorizable_operation): Likewise.
26072 (vectorizable_store): Likewise.
26073 (vectorizable_load): Likewise.
26074 * tree-vectorizer.h: Likewise.
26076 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
26078 * passes.c (emergency_dump_function): New.
26079 * tree-pass.h (emergency_dump_function): Declare.
26080 * plugin.c (plugins_internal_error_function): Remove.
26081 * plugin.h (plugins_internal_error_function): Remove declaration.
26082 * toplev.c (internal_error_function): New static function. Use it...
26083 (general_init): ...here.
26085 2017-05-31 Graham Markall <graham.markall@embecosm.com>
26087 * config/arc/arc.c (arc_print_operand): Handle constant operands.
26088 (arc_rtx_costs): Add costs for new patterns.
26089 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
26090 * config/arc/predicates.md: Add _1_2_3_operand predicate.
26092 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26094 * tree-ssa-strlen.c (get_next_strinfo): New function.
26095 (get_stridx_plus_constant): Use it.
26096 (zero_length_string): Likewise.
26097 (adjust_related_strinfos): Likewise.
26098 (adjust_last_stmt): Likewise.
26100 2017-05-31 Richard Biener <rguenther@suse.de>
26103 * config/i386/i386.c (ix86_expand_builtin): Remove assert
26104 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
26106 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26108 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
26109 loop_vinfo argument and use of dependence distance vectors.
26110 Check instead whether the two references differ only in their
26111 initial value and assume that they have the same alignment if the
26112 difference is a multiple of the vector alignment.
26113 (vect_analyze_data_refs_alignment): Update call accordingly.
26115 2017-05-31 Martin Liska <mliska@suse.cz>
26118 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
26120 2017-05-31 Bin Cheng <bin.cheng@arm.com>
26122 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
26123 (create_intersect_range_checks): Move from ...
26124 * tree-data-ref.c (create_intersect_range_checks_index)
26125 (create_intersect_range_checks): ... to here.
26126 (create_runtime_alias_checks): New function factored from ...
26127 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
26128 here. Call above function.
26129 * tree-data-ref.h (create_runtime_alias_checks): New function.
26131 2017-05-31 Bin Cheng <bin.cheng@arm.com>
26133 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
26134 segment length for dr_b and compute it in wide_int.
26136 2017-05-31 Richard Biener <rguenther@suse.de>
26138 PR tree-optimization/80906
26139 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
26140 and pass through iv_map.
26141 (copy_bb_and_scalar_dependences): Adjust.
26142 (translate_pending_phi_nodes): Likewise.
26143 (copy_loop_close_phi_args): Handle code-generating IVs instead
26146 2017-05-30 David Malcolm <dmalcolm@redhat.com>
26148 * diagnostic-color.c (color_dict): Add "type-diff".
26149 (parse_gcc_colors): Update comment.
26150 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
26151 -fdiagnostics-show-template-tree and -fno-elide-type.
26152 (GCC_COLORS): Add type-diff to example.
26154 (-fdiagnostics-show-template-tree): New.
26155 (-fno-elide-type): New.
26156 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
26157 the pp_format_decoder callback. Call any m_format_postprocessor's
26159 (pretty_printer::pretty_printer): Initialize
26160 m_format_postprocessor.
26161 (pretty_printer::~pretty_printer): Delete any
26162 m_format_postprocessor.
26163 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
26164 (class format_postprocessor): New class.
26165 (struct pretty_printer::format_decoder): Document the new parameters.
26166 (struct pretty_printer::m_format_postprocessor): New field.
26167 * tree-diagnostic.c (default_tree_printer): Update for new
26168 bool and const char ** params.
26169 * tree-diagnostic.h (default_tree_printer): Likewise.
26171 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
26173 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
26174 (lwa_operand): Delete rs6000_gen_cell_microcode test.
26175 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
26176 rs6000_gen_cell_microcode code.
26177 (rs6000_final_prescan_insn): Delete.
26178 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
26179 "warn-cell-microcode" entries.
26180 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
26181 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
26182 throughout. Change cc_reg_not_micro_cr0_operand to
26183 cc_reg_not_cr0_operand throughout.
26184 (*extendhi<mode>2_noload): Delete.
26185 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
26186 (mwarn-cell-microcode): Delete.
26187 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
26188 -mgen-cell-microcode and -mwarn-cell-microcode.
26190 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
26193 * config/i386/constraints.md (Yd): New constraint.
26195 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
26196 and (?Yd, r) alternatives. Update insn attributes.
26197 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
26198 and (?*Yd, r) alternatives. Update insn attributes.
26199 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
26201 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
26203 * gimplify.c (gimplify_modify_expr): Don't create a
26204 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
26207 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
26209 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
26211 2017-05-30 Richard Biener <rguenther@suse.de>
26213 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
26214 and reduc_def fields.
26215 (STMT_VINFO_REDUC_TYPE): New define.
26216 (STMT_VINFO_REDUC_DEF): Likewise.
26217 (vect_force_simple_reduction): Adjust prototype.
26218 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
26219 (vect_is_simple_reduction): Remove check_reduction argument.
26220 (vect_force_simple_reduction): Adjust and set
26221 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
26222 (vectorizable_reduction): Do not re-do reduction analysis
26223 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
26224 * tree-parloops.c (gather_scalar_reductions): Adjust.
26226 2017-05-30 Richard Biener <rguenther@suse.de>
26228 PR middle-end/80901
26229 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
26232 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
26234 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
26235 Introduce unknown_misalignment parameter and remove vf.
26236 (vect_peeling_hash_get_lowest_cost):
26237 Pass unknown_misalignment parameter.
26238 (vect_enhance_data_refs_alignment):
26239 Fix unsupportable data ref treatment.
26241 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26243 * tree-vect-data-refs.c (vect_get_data_access_cost):
26244 Workaround for SLP handling.
26245 (vect_enhance_data_refs_alignment):
26246 Compute costs for doing no peeling at all, compare to the best
26247 peeling costs so far and avoid peeling if cheaper.
26249 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26251 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
26252 Return peeling info and set costs to zero for unlimited cost
26254 (vect_enhance_data_refs_alignment): Also inspect all datarefs
26255 with unknown misalignment. Compute and costs for unknown
26256 misalignment, compare them to the costs for known misalignment
26257 and choose the cheapest for peeling.
26259 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26261 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
26262 (vect_get_peeling_costs_all_drs): Create function.
26263 (vect_peeling_hash_get_lowest_cost):
26264 Use vect_get_peeling_costs_all_drs.
26265 (vect_peeling_supportable): Create function.
26266 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
26268 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26270 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
26271 DR_HAS_NEGATIVE_STEP.
26272 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
26273 (vect_enhance_data_refs_alignment): Use.
26274 (vect_duplicate_ssa_name_ptr_info): Use.
26275 * tree-vectorizer.h (dr_misalignment): Use.
26276 (known_alignment_for_access_p): Use.
26278 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
26281 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
26283 (has_section_name): New function.
26285 2017-05-30 Martin Liska <mliska@suse.cz>
26288 * auto-profile.c (get_function_decl_from_block): Fix
26291 2017-05-30 Richard Biener <rguenther@suse.de>
26293 PR middle-end/80876
26294 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
26296 2017-05-30 Martin Liska <mliska@suse.cz>
26298 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
26299 * dumpfile.h (struct dump_file_info): Remove ctors.
26301 2017-05-30 Martin Liska <mliska@suse.cz>
26303 * predict.def: Fix GNU coding style.
26305 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
26307 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
26308 Mark 'to' argument with ATTRIBUTE_UNUSED.
26310 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
26312 * config/xtensa/xtensa.c (xtensa_emit_call): Use
26313 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
26314 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
26317 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
26319 * doc/install.texi (Options specification): Restore entry of
26320 --enable-sjlj-exceptions.
26322 2017-05-27 Michael Eager <eager@eagercon.com>
26325 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
26327 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
26329 * config/microblaze/microblaze.h
26330 (FIXED_REGISTERS): Update in macro.
26331 (CALL_USED_REGISTERS): Update in macro.
26333 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
26335 * doc/install.texi: Add links to macOS binary distributions.
26337 2017-05-27 Jakub Jelinek <jakub@redhat.com>
26341 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26343 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
26345 2017-05-26 Martin Liska <mliska@suse.cz>
26347 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
26349 2017-05-26 Martin Liska <mliska@suse.cz>
26351 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
26352 always leading ';; '.
26353 (dump_bb_info): Likewise.
26354 (brief_dump_cfg): Likewise.
26355 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
26356 * dumpfile.c: Remove usage of TDF_VERBOSE.
26357 * dumpfile.h (enum dump_kind): Likewise.
26358 (dump_gimple_bb_header): Do not use TDF_COMMENT.
26359 * print-tree.c (debug_verbose): Remove.
26360 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
26361 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
26362 * tree-diagnostic.c (default_tree_printer): Replace
26363 TDF_DIAGNOSTIC with TDF_SLIM.
26365 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26367 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
26368 in parameter loop, rather than loop_vinfo.
26369 (create_intersect_range_checks): Ditto.
26370 (vect_create_cond_for_alias_checks): Update call to above functions.
26372 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26374 PR tree-optimization/80815
26375 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
26376 for merging runtime alias checks. Handle negative DR_STEPs.
26378 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26380 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
26382 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
26383 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
26384 out code pruning runtime alias checks.
26385 * tree-data-ref.c (prune_runtime_alias_test_list): New function
26386 factored out from above.
26387 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
26389 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
26391 (prune_runtime_alias_test_list): New decalaration.
26393 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26395 * tree-vect-data-refs.c (compare_tree): Rename and move ...
26396 * tree-data-ref.c (data_ref_compare_tree): ... to here.
26397 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
26398 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
26399 (operator==, comp_dr_with_seg_len_pair): Ditto.
26400 (vect_prune_runtime_alias_test_list): Ditto.
26402 2017-05-26 Martin Liska <mliska@suse.cz>
26405 * params.def: Bound partial-inlining-entry-probability param.
26407 2017-05-26 Marek Polacek <polacek@redhat.com>
26410 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
26413 2017-05-26 Richard Biener <rguenther@suse.de>
26415 PR tree-optimization/80842
26416 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
26419 2017-05-26 Richard Biener <rguenther@suse.de>
26421 PR tree-optimization/80844
26422 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
26424 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
26426 * doc/md.texi (Machine Constraints): Update x86 family
26427 machine constraints section to match 'config/i386/constraints.md'.
26429 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
26431 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
26433 2017-05-25 Nathan Sidwell <nathan@acm.org>
26435 * doc/invoke.texi (--enable-languages): Update documentation.
26437 2017-05-25 Martin Liska <mliska@suse.cz>
26439 * dumpfile.c: Add TDF_FOLDING.
26440 * dumpfile.h (enum dump_kind): Likewise.
26441 * genmatch.c (dt_simplify::gen_1): Use it.
26443 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26445 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
26447 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26449 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
26450 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
26452 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26454 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
26455 * match.pd (X == C): Rewrite it here.
26456 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
26457 with_certain_nonzero_bits2): New predicates.
26458 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
26460 2017-05-24 Nathan Sidwell <nathan@acm.org>
26462 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
26465 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
26468 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26470 * config/powerpcspe: New port. Files are copied from the rs6000
26471 port, with "rs6000" in filenames replaced by "powerpcspe".
26473 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
26475 PR rtl-optimization/80754
26476 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
26478 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
26480 * config/sparc/sparc.md (length): Return the correct value for -mflat
26481 sibcalls to match output_sibcall.
26483 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26487 * config/rs6000/rs6000.c (struct machine_function): Add new field
26489 (rs6000_get_separate_components): Init that field, use it.
26490 (rs6000_components_for_bb): Use the field.
26492 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
26494 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
26496 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
26498 PR middle-end/80823
26499 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
26501 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26504 * config/s390/s390.c (s390_check_qrst_address): Check incoming
26505 address against address_operand predicate.
26506 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
26508 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
26510 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
26511 parameters passed indirectly.
26513 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
26515 * config/i386/i386.md (*movdi_internal): Remove SSE4
26516 alternative 18 (?r, *v). Update insn attributes.
26517 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
26518 Update insn attributes.
26519 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
26520 Update insn attributes.
26521 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
26522 alternative 1 (r, v). Remove isa attribute.
26523 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
26524 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
26525 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
26527 2017-05-23 Tom de Vries <tom@codesourcery.com>
26529 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
26532 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26534 * cgraphunit.c (symbol_table::process_new_functions): Update.
26535 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
26536 (inline_generate_summary): Rename to ...
26537 (ipa_fn_summary_generate): ... this one.
26538 (inline_read_summary): Rename to ...
26539 (ipa_fn_summary_read): ... this one.
26540 (inline_write_summary): Rename to ...
26541 (ipa_fn_summary_write): ... this one.
26542 (inline_free_summary): Rename to ...
26543 (ipa_free_fn_summary): ... this one.
26544 (pass_data_local_fn_summary, pass_local_fn_summary,
26545 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
26546 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
26547 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
26548 make_pass_ipa_fn_summary): New.
26549 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
26550 inline_write_summary, inline_free_summary): Remove.
26551 (ipa_free_fn_summary) : New.
26552 * ipa-inline.c (ipa_inline): Update.
26553 (pass_ipa_inline): Do not generate summaries.
26554 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
26556 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
26557 and add pass_ipa_fn_summary.
26558 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
26560 (make_pass_inline_parameters): Remove.
26562 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
26564 * omp-low.c (struct omp_context): Remove "default_kind" member.
26567 * omp-offload.c (execute_oacc_device_lower): Remove the
26568 parallelism dimensions function attributes for unparallelized
26569 OpenACC kernels constructs.
26571 2017-05-23 Martin Liska <mliska@suse.cz>
26573 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
26575 (cgraph_edge::make_speculative): Likewise.
26576 (cgraph_edge::resolve_speculation): Likewise.
26577 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
26578 (cgraph_node::dump): Likewise.
26579 * cgraph.h: Likewise.
26580 * cgraphunit.c (analyze_functions): Likewise.
26581 (symbol_table::compile): Likewise.
26582 * ipa-cp.c (print_all_lattices): Likewise.
26583 (determine_versionability): Likewise.
26584 (initialize_node_lattices): Likewise.
26585 (ipcp_verify_propagated_values): Likewise.
26586 (estimate_local_effects): Likewise.
26587 (update_profiling_info): Likewise.
26588 (create_specialized_node): Likewise.
26589 (perhaps_add_new_callers): Likewise.
26590 (decide_about_value): Likewise.
26591 (decide_whether_version_node): Likewise.
26592 (identify_dead_nodes): Likewise.
26593 (ipcp_store_bits_results): Likewise.
26594 * ipa-devirt.c (dump_targets): Likewise.
26595 (ipa_devirt): Likewise.
26596 * ipa-icf.c (sem_item::dump): Likewise.
26597 (sem_function::equals): Likewise.
26598 (sem_variable::equals): Likewise.
26599 (sem_item_optimizer::read_section): Likewise.
26600 (sem_item_optimizer::execute): Likewise.
26601 (congruence_class::dump): Likewise.
26602 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
26603 (dump_inline_summary): Likewise.
26604 (estimate_node_size_and_time): Likewise.
26605 (inline_analyze_function): Likewise.
26606 * ipa-inline-transform.c (inline_call): Likewise.
26607 * ipa-inline.c (report_inline_failed_reason): Likewise.
26608 (want_early_inline_function_p): Likewise.
26609 (edge_badness): Likewise.
26610 (update_edge_key): Likewise.
26611 (inline_small_functions): Likewise.
26612 * ipa-profile.c (ipa_profile): Likewise.
26613 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
26614 (ipa_make_edge_direct_to_target): Likewise.
26615 (remove_described_reference): Likewise.
26616 (ipa_impossible_devirt_target): Likewise.
26617 (propagate_controlled_uses): Likewise.
26618 (ipa_print_node_params): Likewise.
26619 (ipcp_transform_function): Likewise.
26620 * ipa-pure-const.c (pure_const_read_summary): Likewise.
26621 (propagate_pure_const): Likewise.
26622 * ipa-reference.c (generate_summary): Likewise.
26623 (read_write_all_from_decl): Likewise.
26624 (propagate): Likewise.
26625 (ipa_reference_read_optimization_summary): Likewise.
26626 * ipa-utils.c (ipa_merge_profiles): Likewise.
26627 * ipa.c (walk_polymorphic_call_targets): Likewise.
26628 (symbol_table::remove_unreachable_nodes): Likewise.
26629 (ipa_single_use): Likewise.
26630 * passes.c (execute_todo): Likewise.
26631 * predict.c (drop_profile): Likewise.
26632 * symtab.c (symtab_node::get_dump_name): New function.
26633 (symtab_node::dump_name): Likewise.
26634 (symtab_node::dump_asm_name): Likewise.
26635 (symtab_node::dump_references): Likewise.
26636 (symtab_node::dump_referring): Likewise.
26637 (symtab_node::dump_base): Likewise.
26638 (symtab_node::debug_symtab): Likewise.
26639 * tree-sra.c (convert_callers_for_node): Likewise.
26640 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
26641 * value-prof.c (init_node_map): Likewise.
26643 2017-05-23 Martin Liska <mliska@suse.cz>
26645 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
26646 and symtab_node::debug_symtab to symbol_table::debug.
26647 * cgraphunit.c (analyze_functions): Use the renamed function.
26648 (symbol_table::compile): Likewise.
26649 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
26650 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
26651 * passes.c (execute_todo): Likewise.
26652 * symtab.c (symbol_table::dump): New function.
26653 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
26655 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26657 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
26658 that nonconst implies exec.
26660 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26662 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
26663 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
26664 (inline_edge_summary_vec): Turn into ...
26665 (ipa_call_summaries): ... this one.
26666 (redirect_to_unreachable, edge_set_predicate,
26667 evaluate_properties_for_edge, inline_summary_alloc,
26668 reset_ipa_call_summary, reset_inline_summary,
26669 inline_summary_t::duplicate): Update.
26670 (inline_edge_duplication_hook): Turn to ...
26671 (ipa_call_summary_t::duplicate): ... this one.
26672 (inline_edge_removal_hook): Turn to ...
26673 (ipa_call_summary_t::remove): ... this one.
26674 (dump_inline_edge_summary): Turn to ...
26675 (dump_ipa_call_summary): ... this one.
26676 (estimate_function_body_sizes): Update.
26677 (inline_update_callee_summaries): Update.
26678 (remap_edge_change_prob): Update.
26679 (remap_edge_summaries): Update.
26680 (inline_merge_summary): Update.
26681 (do_estimate_edge_time): Update.
26682 (inline_generate_summary): Update.
26683 (inline_read_section): Update.
26684 (inline_read_summary): Update.
26685 (inline_free_summary): Update.
26686 * ipa-inline.c (can_inline_edge_p): Update.
26687 (compute_inlined_call_time): Update.
26688 (want_inline_small_function_p): Update.
26689 (edge_badness): Update.
26690 (early_inliner): Update.
26691 * ipa-inline.h (inline_edge_summary): Turn to ...
26692 (ipa_call_summary): ... this one.
26693 (ipa_call_summary_t): New class.
26694 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
26695 (ipa_call_summaries): New.
26696 (inline_edge_summary): Remove.
26697 (estimate_edge_growth): Update.
26698 * ipa-profile.c (ipa_propagate_frequency_1): Update.
26699 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26700 * ipa-split.c (execute_split_functions): Update.
26701 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
26703 2017-05-23 Tom de Vries <tom@codesourcery.com>
26705 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
26706 attributes): Document rdrand effective target.
26708 2017-05-23 Tom de Vries <tom@codesourcery.com>
26710 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
26711 attributes): Sort alphabetically.
26713 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
26715 * config/avr/genmultilib.awk: Use gsub instead of gensub.
26717 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
26720 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
26721 V2DF/V2DI splat into two separate patterns, one that handles
26722 registers, and the other that only handles memory. Drop support
26723 for splatting from a GPR on ISA 2.07 and then splitting the
26724 splat into direct move and splat.
26725 (vsx_splat_<mode>_reg): Likewise.
26726 (vsx_splat_<mode>_mem): Likewise.
26728 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
26730 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
26732 2017-05-22 Jakub Jelinek <jakub@redhat.com>
26734 PR middle-end/80809
26735 * omp-low.c (finish_taskreg_remap): New function.
26736 (finish_taskreg_scan): If unit size of ctx->record_type
26737 is non-constant, unshare the size expression and replace
26738 decls in it with possible outer var refs.
26740 PR middle-end/80809
26741 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
26742 GOVD_SHARED rather than GOVD_PRIVATE with it.
26743 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
26744 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
26746 PR middle-end/80853
26747 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
26748 as last argument to build_outer_var_ref for pointer bases of array
26749 section reductions.
26751 2017-05-19 Martin Sebor <msebor@redhat.com>
26753 * print-tree.c (print_node): Print DECL_READ_P flag.
26755 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26757 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
26758 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
26759 * cgraph.c: Likewise.
26760 * cgraphunit.c: Likewise.
26761 * gengtype.c: Likewise.
26762 * ipa-cp.c: Likewise.
26763 * ipa-devirt.c: Likewise.
26764 * ipa-icf.c: Likewise.
26765 * ipa-predicate.c: Likewise.
26766 * ipa-profile.c: Likewise.
26767 * ipa-prop.c: Likewise.
26768 * ipa-split.c: Likewise.
26770 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
26771 edge_predicate_pool, dump_inline_hints,
26772 inline_summary::account_size_time, redirect_to_unreachable,
26773 edge_set_predicate, set_hint_predicate,
26774 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
26775 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
26776 inline_summary_t::remove, remap_hint_predicate_after_duplication,
26777 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
26778 ipa_call_summary_t::remove, initialize_growth_caches,
26779 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
26780 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
26781 mark_modified, unmodified_parm_1, unmodified_parm,
26782 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
26783 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
26784 compute_bb_predicates, will_be_nonconstant_expr_predicate,
26785 will_be_nonconstant_predicate, record_modified_bb_info,
26786 get_minimal_bb, record_modified, param_change_prob,
26787 phi_result_unknown_predicate, predicate_for_phi_result,
26788 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
26789 estimate_function_body_sizes, compute_inline_parameters,
26790 compute_inline_parameters_for_curren, pass_data_inline_parameters,
26791 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
26792 inline_update_callee_summaries, remap_edge_change_prob,
26793 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
26794 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
26795 inline_analyze_function, inline_summary_t::insert,
26796 inline_generate_summary, read_ipa_call_summary, inline_read_section,
26797 inline_read_summary, write_ipa_call_summary, inline_write_summary,
26798 inline_free_summary): Move to ipa-fnsummary.h
26799 (predicate_t): Remove.
26800 * ipa-fnsummary.c: New file.
26801 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
26802 (enum inline_hints_vals, inline_hints, agg_position_info,
26803 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
26804 inline_summaries, ipa_call_summary, ipa_call_summary_t,
26805 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
26806 dump_inline_summary, dump_inline_hints, inline_generate_summary,
26807 inline_read_summary, inline_write_summary, inline_free_summary,
26808 inline_analyze_function, initialize_inline_failed,
26809 inline_merge_summary, inline_update_overall_summary,
26810 compute_inline_parameters): Move to ipa-fnsummary.h
26811 * ipa-fnsummary.h: New file.
26812 * ipa-inline-transform.h: Include ipa-inline.h.
26813 * ipa-inline.c: LIkewise.
26815 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26817 * ipa-inline.c (edge_badness): Use inlined_time instead of
26818 inline_summaries->get.
26820 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26822 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
26824 2017-05-22 Nathan Sidwell <nathan@acm.org>
26826 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
26827 (fdump-lang): Document 'raw' option.
26828 * dumpfile.h (TDI_tu): Delete.
26829 * dumpfile.c (dump_files): Remove translation-unit.
26830 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
26832 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
26834 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
26835 command option from $(AWK) call.
26836 * config/avr/genmultilib.awk: Simplify and rewrite so that it
26837 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
26838 [FORMAT]: Remove handling of variable.
26839 * config/avr/t-multilib: Regenerate.
26841 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26843 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
26845 (dump_inline_summary): Do not print self_time.
26846 (estimate_function_body_sizes): Do not set self_time.
26847 (compute_inline_parameters): Likewise.
26848 (inline_read_section, inline_write_summary): Do not stream self_time.
26849 * ipa-inline.h (inline_summary): Drop self_time.
26851 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26853 * ipa-inline-analysis.c (account_size_time): Rename to ...
26854 (inline_summary::account_size_time): ... this one.
26855 (reset_ipa_call_summary): Turn to ...
26856 (ipa_call_summary::reset): ... this one.
26857 (reset_inline_summary): Turn to ...
26858 (inline_summary::reset): ... this one.
26859 (inline_summary_t::remove): Update.
26860 (inline_summary_t::duplicate): Update.
26861 (ipa_call_summary_t::remove): Update.
26862 (dump_inline_summary): Update.
26863 (estimate_function_body_sizes): Update.
26864 (compute_inline_parameters): Update.
26865 (estimate_node_size_and_time): Update.
26866 (inline_merge_summary): Update.
26867 (inline_update_overall_summary): Update.
26868 (inline_read_section): Update.
26869 (inline_write_summary): Update.
26870 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
26871 add account_size_time and reset member functions.
26872 (ipa_call_summary): Add reset function.
26873 * ipa-predicate.h (predicate::operator &): Constify.
26875 2017-05-22 Richard Biener <rguenther@suse.de>
26877 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
26879 2017-05-19 Jason Merrill <jason@redhat.com>
26881 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
26883 2017-05-19 Marek Polacek <polacek@redhat.com>
26886 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
26887 TYPE_OVERFLOW_WRAPS checks.
26889 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
26891 * tree-core.h (enum omp_clause_default_kind): Add
26892 "OMP_CLAUSE_DEFAULT_PRESENT".
26893 * tree-pretty-print.c (dump_omp_clause): Handle it.
26894 * gimplify.c (enum gimplify_omp_var_data): Add
26895 "GOVD_MAP_FORCE_PRESENT".
26896 (gimplify_adjust_omp_clauses_1): Map it to
26897 "GOMP_MAP_FORCE_PRESENT".
26898 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
26900 * gimplify.c (oacc_default_clause): Clarify.
26902 2017-05-19 Nathan Sidwell <nathan@acm.org>
26904 LANG_HOOK_REGISTER_DUMPS
26905 * toplev.c (general_init): Call register dump lang hook.
26906 * doc/invoke.texi: Document -fdump-lang option family.
26907 * dumpfile.c (dump_files): Remove class dump here.
26908 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
26909 * dumpfile.h (tree_dump_index): Remove TDI_class.
26910 * langhooks-def.h (lhd_register_dumps): Declare.
26911 (LANG_HOOKS_REGISTER_DUMPS): Define.
26912 (LANG_HOOKS_INITIALIZER): Add it.
26913 * langhooks.c (lhd_register_dumps): Define.
26914 * langhooks.h (struct lang_hooks): Add register_dumps.
26916 2017-05-19 Nathan Sidwell <nathan@acm.org>
26918 * context.h (context::set_passes): New.
26919 * context.c (context::context): Do not create pass manager.
26920 * toplev.c (general_init): Create pass manager here.
26922 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
26924 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
26925 use this splitter if two add or or instructions would also work for
26926 the constant we want to generate.
26928 2017-05-19 Richard Biener <rguenther@suse.de>
26931 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
26932 predicate evaluation.
26934 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
26936 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
26938 * ipa-inline.c (want_inline_small_function_p): Do not cast to
26941 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
26943 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
26944 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
26945 (inline_edge_summary_vec): Turn into ...
26946 (ipa_call_summaries): ... this one.
26947 (redirect_to_unreachable, edge_set_predicate,
26948 evaluate_properties_for_edge, inline_summary_alloc,
26949 reset_ipa_call_summary, reset_inline_summary,
26950 inline_summary_t::duplicate): Update.
26951 (inline_edge_duplication_hook): Turn to ...
26952 (ipa_call_summary_t::duplicate): ... this one.
26953 (inline_edge_removal_hook): Turn to ...
26954 (ipa_call_summary_t::remove): ... this one.
26955 (dump_inline_edge_summary): Turn to ...
26956 (dump_ipa_call_summary): ... this one.
26957 (estimate_function_body_sizes): Update.
26958 (inline_update_callee_summaries): Update.
26959 (remap_edge_change_prob): Update.
26960 (remap_edge_summaries): Update.
26961 (inline_merge_summary): Update.
26962 (do_estimate_edge_time): Update.
26963 (inline_generate_summary): Update.
26964 (inline_read_section): Update.
26965 (inline_read_summary): Update.
26966 (inline_free_summary): Update.
26967 * ipa-inline.c (can_inline_edge_p): Update.
26968 (compute_inlined_call_time): Update.
26969 (want_inline_small_function_p): Update.
26970 (edge_badness): Update.
26971 (early_inliner): Update.
26972 * ipa-inline.h (inline_edge_summary): Turn to ...
26973 (ipa_call_summary): ... this one.
26974 (ipa_call_summary_t): New class.
26975 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
26976 (ipa_call_summaries): New.
26977 (inline_edge_summary): Remove.
26978 (estimate_edge_growth): Update.
26979 * ipa-profile.c (ipa_propagate_frequency_1): Update.
26980 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26981 * ipa-split.c (execute_split_functions): Update.
26982 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
26984 2017-05-19 Richard Biener <rguenther@suse.de>
26986 PR middle-end/80764
26987 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
26989 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
26991 * config/rs6000/rs6000.c (struct machine_function): Add field
26992 fpr_is_wrapped_separately.
26993 (rs6000_get_separate_components): Use 64 components. Handle the
26994 new FPR components.
26995 (rs6000_components_for_bb): Handle the FPR components.
26996 (rs6000_emit_prologue_components): Handle the FPR components.
26997 (rs6000_emit_epilogue_components): Handle the FPR components.
26998 (rs6000_set_handled_components): Handle the FPR components.
26999 (rs6000_emit_prologue): Don't output prologue code for those FPRs
27000 that are already separately shrink-wrapped.
27001 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
27002 that are already separately shrink-wrapped.
27004 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27007 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
27010 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
27011 (define_peephole2 for Altivec d-form load): Add peepholes to catch
27012 cases where the register allocator uses a move and an offsettable
27013 memory operation to/from a FPR register on ISA 2.06/2.07.
27014 (define_peephole2 for Altivec d-form store): Likewise.
27016 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
27019 * config/i386/mmx.md (*mov<mode>_internal): Enable
27020 alternatives 11, 12, 13 and 14 also for 32bit targets.
27021 Remove alternatives 15, 16, 17 and 18.
27022 * config/i386/sse.md (vec_concatv2di): Change
27023 alternative (!x, *y) to (x, ?!*Yn).
27025 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
27027 * dumpfile.h (enum dump_kind): Remove stray comma.
27029 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27031 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
27032 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
27033 predicate::num_conditions
27034 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
27035 (CHANGED): turn into predicate::changed.
27036 (agg_position_info): Move to ipa-predicate.h
27037 (add_condition, predicate::add_clause, predicate::operator &=,
27038 predicate::or_with, predicate::evaluate, predicate::probability,
27039 dump_condition, dump_clause, predicate::dump,
27040 predicate::remap_after_duplication, predicate::remap_after_inlining,
27041 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
27042 (evaluate_conditions_for_known_args): Update.
27043 (set_cond_stmt_execution_predicate): Update.
27044 * ipa-inline.h: Include ipa-predicate.h
27045 (condition, inline_param_summary, conditions, agg_position_info,
27046 predicate): Move to ipa-predicate.h
27047 * ipa-predicate.c: New file.
27048 * ipa-predicate.h: New file.
27050 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
27052 * final.c (leaf_function_p): Check we are not in a sequence.
27054 2017-05-18 Martin Liska <mliska@suse.cz>
27056 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
27057 * dumpfile.c (dump_register): Use new enum dump_kind.
27058 (get_dump_file_name): Likewise.
27059 (dump_enable_all): Likewise.
27060 (dump_switch_p_1): Likewise.
27061 (enable_rtl_dump_file): Remove usage of TDF_RTL.
27062 * dumpfile.h (enum dump_kind): New enum type.
27063 (struct dump_file_info): Create constructor and
27064 format fields and comments.
27065 * passes.c (pass_manager::register_one_dump_file):
27067 * statistics.c (statistics_early_init): Likewise.
27068 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
27069 TDF_TREE with TDF_SLIM.
27070 (gather_memory_references_ref): Likewise.
27072 2017-05-18 Martin Liska <mliska@suse.cz>
27074 * vec.h (struct vnull): Use it.
27076 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27078 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
27079 (true_predicate, false_predicate, true_predicate_p,
27080 false_predicate_p): Remove.
27081 (single_cond_predicate, not_inlined_predicate): Turn to member function
27083 (add_condition): Update.
27084 (add_clause): Turn to...
27085 (predicate::add_clause): ... this one; update; allow passing NULL
27087 (and_predicates): Turn to ...
27088 (predicate::operator &=): ... this one.
27089 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
27090 (or_predicates): Turn to ...
27091 (predicate::or_with): ... this one.
27092 (evaluate_predicate): Turn to ...
27093 (predicate::evaluate): ... this one.
27094 (predicate_probability): Turn to ...
27095 (predicate::probability): ... this one.
27096 (dump_condition): Update.
27097 (dump_predicate): Turn to ...
27098 (predicate::dump): ... this one.
27099 (account_size_time): Update.
27100 (edge_set_predicate): Update.
27101 (set_hint_predicate): UPdate.
27102 (evaluate_conditions_for_known_args): Update.
27103 (evaluate_properties_for_edge): Update.
27104 (remap_predicate_after_duplication): Turn to...
27105 (predicate::remap_after_duplication): ... this one.
27106 (remap_hint_predicate_after_duplication): Update.
27107 (inline_summary_t::duplicate): UPdate.
27108 (dump_inline_edge_summary): Update.
27109 (dump_inline_summary): Update.
27110 (set_cond_stmt_execution_predicate): Update.
27111 (set_switch_stmt_execution_predicate): Update.
27112 (compute_bb_predicates): Update.
27113 (will_be_nonconstant_expr_predicate): Update.
27114 (will_be_nonconstant_predicate): Update.
27115 (phi_result_unknown_predicate): Update.
27116 (predicate_for_phi_result): Update.
27117 (array_index_predicate): Update.
27118 (estimate_function_body_sizes): Update.
27119 (estimate_node_size_and_time): Update.
27120 (estimate_ipcp_clone_size_and_time): Update.
27121 (remap_predicate): Rename to ...
27122 (predicate::remap_after_inlining): ... this one.
27123 (remap_hint_predicate): Update.
27124 (inline_merge_summary): Update.
27125 (inline_update_overall_summary): Update.
27126 (estimate_size_after_inlining): Update.
27127 (read_predicate): Rename to ...
27128 (predicate::stream_in): ... this one.
27129 (read_inline_edge_summary): Update.
27130 (write_predicate): Rename to ...
27131 (predicate::stream_out): ... this one.
27132 (write_inline_edge_summary): Update.
27133 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
27134 (clause_t): Turn to uint32_t
27135 (predicate): Turn to class; implement constructor and operators
27137 (size_time_entry): Update.
27138 (inline_summary): Update.
27139 (inline_edge_summary): Update.
27141 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
27143 * fold-const.c (fold_binary_loc): Move transformation...
27144 * match.pd (C - X CMP X): ... here.
27146 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
27148 * config/sparc/sparc.c (sparc_option_override): Set function
27149 alignment for -mcpu=niagara7 to 64 to match the I$ line.
27150 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
27152 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
27154 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
27156 2017-05-18 Marek Polacek <polacek@redhat.com>
27159 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
27160 (pass_ubsan::execute): Call gimple_assign_single_p instead of
27161 gimple_assign_load_p.
27163 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
27165 PR middle-end/80692
27166 * real.c (do_compare): Give decimal_do_compare preference over
27167 comparing just the signs.
27169 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
27171 * doc/md.texi (Canonicalization of Instructions): Describe the
27172 canonical form of instructions that inherently set a condition
27175 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
27177 PR middle-end/80775
27178 * tree-cfg.c: Move deletion of unreachable case statements to after
27179 the merging of consecutive case labels.
27181 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
27183 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
27184 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
27185 restoring of callee-saved registers.
27187 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
27189 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
27190 * config/visium/visium.c (single_set_and_flags): Likewise.
27191 * config/visium/visium.md (Substitutions): Likewise.
27193 2017-05-17 Martin Liska <mliska@suse.cz>
27195 * cfg.c: Introduce dump_flags_t type and
27196 use it instead of int type.
27198 * cfghooks.c: Likewise.
27199 * cfghooks.h (struct cfg_hooks): Likewise.
27200 * cfgrtl.c: Likewise.
27201 * cfgrtl.h: Likewise.
27202 * cgraph.c (cgraph_node::get_body): Likewise.
27203 * coretypes.h: Likewise.
27204 * domwalk.c: Likewise.
27205 * domwalk.h: Likewise.
27206 * dumpfile.c (struct dump_option_value_info): Likewise.
27207 (dump_enable_all): Likewise.
27208 (dump_switch_p_1): Likewise.
27209 (opt_info_switch_p): Likewise.
27210 * dumpfile.h (enum tree_dump_index): Likewise.
27211 (struct dump_file_info): Likewise.
27212 * genemit.c: Likewise.
27213 * generic-match-head.c: Likewise.
27214 * gengtype.c (open_base_files): Likewise.
27215 * gimple-pretty-print.c: Likewise.
27216 * gimple-pretty-print.h: Likewise.
27217 * graph.c (print_graph_cfg): Likewise.
27218 * graphite-scop-detection.c (dot_all_sese): Likewise.
27219 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
27220 * loop-unroll.c (report_unroll): Likewise.
27221 * passes.c (pass_manager::register_one_dump_file): Likewise.
27222 * print-tree.c: Likewise.
27223 * statistics.c: Likewise.
27224 * tree-cfg.c: Likewise.
27225 * tree-cfg.h: Likewise.
27226 * tree-dfa.c: Likewise.
27227 * tree-dfa.h: Likewise.
27228 * tree-dump.c (dump_function): Likewise.
27229 * tree-dump.h (struct dump_info): Likewise.
27230 * tree-pretty-print.c: Likewise.
27231 * tree-pretty-print.h: Likewise.
27232 * tree-ssa-live.c: Likewise.
27233 * tree-ssa-live.h: Likewise.
27234 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
27235 * tree-vect-loop.c: Likewise.
27236 * tree-vect-slp.c: Likewise.
27238 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
27239 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27241 PR tree-optimization/80457
27242 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
27243 of all arguments to a statement as scalar_to_vec operations.
27244 (vectorizable_call): Adjust call to vect_model_simple_cost for
27246 (vectorizable_conversion): Likewise.
27247 (vectorizable_assignment): Likewise.
27248 (vectorizable_shift): Likewise.
27249 (vectorizable_operation): Likewise.
27250 (vectorizable_comparison): Likewise.
27251 (vect_is_simple_cond): Record the def types for operands.
27252 (vectorizable_condition): Likewise, call vect_model_simple_cost.
27253 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
27254 for statement argument count.
27256 2017-05-16 Carl Love <cel@us.ibm.com>
27258 * config/rs6000/rs6000-c: Add support for built-in functions
27259 vector unsigned long long vec_bperm (vector unsigned long long,
27260 vector unsigned char)
27261 vector signed long long vec_mule (vector signed int,
27263 vector unsigned long long vec_mule (vector unsigned int,
27264 vector unsigned int)
27265 vector signed long long vec_mulo (vector signed int,
27267 vector unsigned long long vec_mulo (vector unsigned int,
27268 vector unsigned int)
27269 vector signed char vec_sldw (vector signed char,
27270 vector signed char,
27272 vector unsigned char vec_sldw (vector unsigned char,
27273 vector unsigned char,
27275 vector signed short vec_sldw (vector signed short,
27276 vector signed short,
27278 vector unsigned short vec_sldw (vector unsigned short,
27279 vector unsigned short,
27281 vector signed int vec_sldw (vector signed int,
27284 vector unsigned int vec_sldw (vector unsigned int,
27285 vector unsigned int,
27287 vector signed long long vec_sldw (vector signed long long,
27288 vector signed long long,
27290 vector unsigned long long vec_sldw (vector unsigned long long,
27291 vector unsigned long long,
27293 * config/rs6000/rs6000-c: Add support for built-in functions
27294 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
27295 * config/rs6000/altivec.h: Add defintion for vec_sldw.
27296 * doc/extend.texi: Update the built-in documentation for the
27297 new built-in functions.
27299 2017-05-16 Marek Polacek <polacek@redhat.com>
27303 * tree.c (save_expr): Don't fold the expression.
27305 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
27307 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
27308 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
27309 and (?*y,m). Update insn attributes.
27311 2017-05-16 Martin Liska <mliska@suse.cz>
27313 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
27314 flags argument of print_gimple_stmt, print_gimple_expr,
27315 print_generic_stmt and print_generic_expr.
27316 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
27317 * coretypes.h: Likewise.
27318 * except.c (dump_eh_tree): Likewise.
27319 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
27320 * gimple-pretty-print.h: Likewise.
27321 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
27322 (backprop::push_to_worklist): Likewise.
27323 (backprop::pop_from_worklist): Likewise.
27324 (backprop::process_use): Likewise.
27325 (backprop::intersect_uses): Likewise.
27326 (note_replacement): Likewise.
27327 * gimple-ssa-store-merging.c
27328 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
27329 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
27330 (pass_store_merging::execute): Likewise.
27331 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
27332 (ssa_base_cand_dump_callback): Likewise.
27333 (dump_incr_vec): Likewise.
27334 (replace_refs): Likewise.
27335 (replace_mult_candidate): Likewise.
27336 (create_add_on_incoming_edge): Likewise.
27337 (create_phi_basis): Likewise.
27338 (insert_initializers): Likewise.
27339 (all_phi_incrs_profitable): Likewise.
27340 (introduce_cast_before_cand): Likewise.
27341 (replace_one_candidate): Likewise.
27342 * gimplify.c (gimplify_expr): Likewise.
27343 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
27344 (set_rename): Likewise.
27345 (rename_uses): Likewise.
27346 (copy_loop_phi_nodes): Likewise.
27347 (add_close_phis_to_merge_points): Likewise.
27348 (copy_loop_close_phi_args): Likewise.
27349 (copy_cond_phi_args): Likewise.
27350 (graphite_copy_stmts_from_block): Likewise.
27351 (translate_pending_phi_nodes): Likewise.
27352 * graphite-poly.c (print_pdr): Likewise.
27353 (dump_gbb_cases): Likewise.
27354 (dump_gbb_conditions): Likewise.
27355 (print_scop_params): Likewise.
27356 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
27357 (build_cross_bb_scalars_use): Likewise.
27358 (gather_bbs::before_dom_children): Likewise.
27359 * hsa-dump.c (dump_hsa_immed): Likewise.
27360 * ipa-cp.c (print_ipcp_constant_value): Likewise.
27361 (get_replacement_map): Likewise.
27362 * ipa-inline-analysis.c (dump_condition): Likewise.
27363 (estimate_function_body_sizes): Likewise.
27364 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
27365 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
27366 * ipa-prop.c (ipa_dump_param): Likewise.
27367 (ipa_print_node_jump_functions_for_edge): Likewise.
27368 (ipa_modify_call_arguments): Likewise.
27369 (ipa_modify_expr): Likewise.
27370 (ipa_dump_param_adjustments): Likewise.
27371 (ipa_dump_agg_replacement_values): Likewise.
27372 (ipcp_modif_dom_walker::before_dom_children): Likewise.
27373 * ipa-pure-const.c (check_stmt): Likewise.
27374 (pass_nothrow::execute): Likewise.
27375 * ipa-split.c (execute_split_functions): Likewise.
27376 * omp-offload.c (dump_oacc_loop_part): Likewise.
27377 (dump_oacc_loop): Likewise.
27378 * trans-mem.c (tm_log_emit): Likewise.
27379 (tm_memopt_accumulate_memops): Likewise.
27380 (dump_tm_memopt_set): Likewise.
27381 (dump_tm_memopt_transform): Likewise.
27382 * tree-cfg.c (gimple_verify_flow_info): Likewise.
27383 (print_loop): Likewise.
27384 * tree-chkp-opt.c (chkp_print_addr): Likewise.
27385 (chkp_gather_checks_info): Likewise.
27386 (chkp_get_check_result): Likewise.
27387 (chkp_remove_check_if_pass): Likewise.
27388 (chkp_use_outer_bounds_if_possible): Likewise.
27389 (chkp_reduce_bounds_lifetime): Likewise.
27390 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
27391 (chkp_mark_completed_bounds): Likewise.
27392 (chkp_register_incomplete_bounds): Likewise.
27393 (chkp_mark_invalid_bounds): Likewise.
27394 (chkp_maybe_copy_and_register_bounds): Likewise.
27395 (chkp_build_returned_bound): Likewise.
27396 (chkp_get_bound_for_parm): Likewise.
27397 (chkp_build_bndldx): Likewise.
27398 (chkp_get_bounds_by_definition): Likewise.
27399 (chkp_generate_extern_var_bounds): Likewise.
27400 (chkp_get_bounds_for_decl_addr): Likewise.
27401 * tree-chrec.c (chrec_apply): Likewise.
27402 * tree-data-ref.c (dump_data_reference): Likewise.
27403 (dump_subscript): Likewise.
27404 (dump_data_dependence_relation): Likewise.
27405 (analyze_overlapping_iterations): Likewise.
27406 * tree-inline.c (expand_call_inline): Likewise.
27407 (tree_function_versioning): Likewise.
27408 * tree-into-ssa.c (dump_defs_stack): Likewise.
27409 (dump_currdefs): Likewise.
27410 (dump_names_replaced_by): Likewise.
27411 (dump_update_ssa): Likewise.
27412 (update_ssa): Likewise.
27413 * tree-object-size.c (pass_object_sizes::execute): Likewise.
27414 * tree-parloops.c (build_new_reduction): Likewise.
27415 (try_create_reduction_list): Likewise.
27416 (ref_conflicts_with_region): Likewise.
27417 (oacc_entry_exit_ok_1): Likewise.
27418 (oacc_entry_exit_single_gang): Likewise.
27419 * tree-pretty-print.h: Likewise.
27420 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
27421 (get_scalar_evolution): Likewise.
27422 (add_to_evolution): Likewise.
27423 (get_loop_exit_condition): Likewise.
27424 (analyze_evolution_in_loop): Likewise.
27425 (analyze_initial_condition): Likewise.
27426 (analyze_scalar_evolution): Likewise.
27427 (instantiate_scev): Likewise.
27428 (number_of_latch_executions): Likewise.
27429 (gather_chrec_stats): Likewise.
27430 (final_value_replacement_loop): Likewise.
27431 (scev_const_prop): Likewise.
27432 * tree-sra.c (dump_access): Likewise.
27433 (disqualify_candidate): Likewise.
27434 (create_access): Likewise.
27435 (reject): Likewise.
27436 (maybe_add_sra_candidate): Likewise.
27437 (create_access_replacement): Likewise.
27438 (analyze_access_subtree): Likewise.
27439 (analyze_all_variable_accesses): Likewise.
27440 (sra_modify_assign): Likewise.
27441 (initialize_constant_pool_replacements): Likewise.
27442 (find_param_candidates): Likewise.
27443 (decide_one_param_reduction): Likewise.
27444 (replace_removed_params_ssa_names): Likewise.
27445 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
27446 * tree-ssa-copy.c (dump_copy_of): Likewise.
27447 (copy_prop_visit_cond_stmt): Likewise.
27448 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
27449 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
27450 (record_equivalences_from_stmt): Likewise.
27451 * tree-ssa-dse.c (compute_trims): Likewise.
27452 (delete_dead_call): Likewise.
27453 (delete_dead_assignment): Likewise.
27454 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
27455 (forward_propagate_into_cond): Likewise.
27456 (pass_forwprop::execute): Likewise.
27457 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
27458 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
27460 (move_computations_worker): Likewise.
27461 (execute_sm): Likewise.
27462 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
27463 (remove_exits_and_undefined_stmts): Likewise.
27464 (remove_redundant_iv_tests): Likewise.
27465 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
27466 (adjust_iv_update_pos): Likewise.
27467 * tree-ssa-math-opts.c (bswap_replace): Likewise.
27468 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
27469 (value_replacement): Likewise.
27470 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
27471 * tree-ssa-pre.c (print_pre_expr): Likewise.
27472 (get_representative_for): Likewise.
27473 (create_expression_by_pieces): Likewise.
27474 (insert_into_preds_of_block): Likewise.
27475 (eliminate_insert): Likewise.
27476 (eliminate_dom_walker::before_dom_children): Likewise.
27477 (eliminate): Likewise.
27478 (remove_dead_inserted_code): Likewise.
27479 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
27480 * tree-ssa-reassoc.c (get_rank): Likewise.
27481 (eliminate_duplicate_pair): Likewise.
27482 (eliminate_plus_minus_pair): Likewise.
27483 (eliminate_not_pairs): Likewise.
27484 (undistribute_ops_list): Likewise.
27485 (eliminate_redundant_comparison): Likewise.
27486 (update_range_test): Likewise.
27487 (optimize_range_tests_var_bound): Likewise.
27488 (optimize_vec_cond_expr): Likewise.
27489 (rewrite_expr_tree): Likewise.
27490 (rewrite_expr_tree_parallel): Likewise.
27491 (linearize_expr): Likewise.
27492 (break_up_subtract): Likewise.
27493 (linearize_expr_tree): Likewise.
27494 (attempt_builtin_powi): Likewise.
27495 (attempt_builtin_copysign): Likewise.
27496 (transform_stmt_to_copy): Likewise.
27497 (transform_stmt_to_multiply): Likewise.
27498 (dump_ops_vector): Likewise.
27499 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
27500 (print_scc): Likewise.
27501 (set_ssa_val_to): Likewise.
27502 (visit_reference_op_store): Likewise.
27503 (visit_use): Likewise.
27504 (sccvn_dom_walker::before_dom_children): Likewise.
27505 (run_scc_vn): Likewise.
27506 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
27508 (expr_hash_elt::print): Likewise.
27509 (const_and_copies::pop_to_marker): Likewise.
27510 (const_and_copies::record_const_or_copy_raw): Likewise.
27511 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
27512 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
27513 (dump_predicates): Likewise.
27514 (find_uninit_use): Likewise.
27515 (warn_uninitialized_phi): Likewise.
27516 (pass_late_warn_uninitialized::execute): Likewise.
27517 * tree-ssa.c (verify_vssa): Likewise.
27518 (verify_ssa): Likewise.
27519 (maybe_optimize_var): Likewise.
27520 * tree-vrp.c (dump_value_range): Likewise.
27521 (dump_all_value_ranges): Likewise.
27522 (dump_asserts_for): Likewise.
27523 (register_edge_assert_for_2): Likewise.
27524 (vrp_visit_cond_stmt): Likewise.
27525 (vrp_visit_switch_stmt): Likewise.
27526 (vrp_visit_stmt): Likewise.
27527 (vrp_visit_phi_node): Likewise.
27528 (simplify_cond_using_ranges_1): Likewise.
27529 (fold_predicate_in): Likewise.
27530 (evrp_dom_walker::before_dom_children): Likewise.
27531 (evrp_dom_walker::push_value_range): Likewise.
27532 (evrp_dom_walker::pop_value_range): Likewise.
27533 (execute_early_vrp): Likewise.
27535 2017-05-16 Richard Biener <rguenther@suse.de>
27537 * dwarf2out.c (loc_list_from_tree_1): Do not create
27538 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
27540 2017-05-16 Richard Biener <rguenther@suse.de>
27542 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
27544 (note_variable_value_in_expr): If we resolved the decl ref
27545 do not push to the stack.
27547 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
27549 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
27550 operations in fast-math mode.
27551 (vaddq_f16): Likewise.
27552 (vmul_f16): Likewise.
27553 (vmulq_f16): Likewise.
27554 (vsub_f16): Likewise.
27555 (vsubq_f16): Likewise.
27556 * config/arm/neon.md (add<mode>3): New.
27558 (fma:<VH:mode>3): New. Also remove outdated comment.
27561 2017-05-16 Martin Liska <mliska@suse.cz>
27565 * ipa-devirt.c (warn_types_mismatch): Fix typo.
27566 (odr_types_equivalent_p): Likewise.
27568 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
27570 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
27572 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
27575 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
27576 non-interunit SSE move alternatives with '?'.
27577 (zero-extendsidi peephole2): New peephole to skip intermediate
27578 general register in SSE zero-extend sequence.
27580 2017-05-15 Jeff Law <law@redhat.com>
27582 * reorg.c (relax_delay_slots): Create a new variable to hold
27583 the temporary target rather than clobbering TARGET_LABEL.
27585 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
27586 missing argument to extract_bit_field call.
27587 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
27589 2017-05-15 Martin Liska <mliska@suse.cz>
27592 * gcc.c (process_command): Do not allow empty argument of -o option.
27594 2017-05-15 Renlin Li <renlin.li@arm.com>
27596 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
27597 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
27598 * config/aarch64/constraints.md (Usf): Add long call check.
27599 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
27600 (call_value): Likewise.
27601 (sibcall): Likewise.
27602 (sibcall_value): Likewise.
27604 (call_value_insn): New.
27605 (sibcall_insn): Update rtx pattern.
27606 (sibcall_value_insn): Likewise.
27607 (call_internal): Remove.
27608 (call_value_internal): Likewise.
27609 (sibcall_internal): Likewise.
27610 (sibcall_value_internal): Likewise.
27611 (call_reg): Likewise.
27612 (call_symbol): Likewise.
27613 (call_value_reg): Likewise.
27614 (call_value_symbol): Likewise.
27616 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
27619 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
27621 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
27623 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
27624 compatible with CCGOCmode and with CCZmode.
27626 2017-05-14 Martin Sebor <msebor@redhat.com>
27628 PR middle-end/77671
27629 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
27630 (gimple_fold_builtin_snprintf): Same.
27631 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
27632 (gimple_fold_builtin_snprintf): Same.
27633 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
27634 of character types.
27635 (is_call_safe): New function.
27636 (try_substitute_return_value): Call it.
27637 (try_simplify_call): New function.
27638 (pass_sprintf_length::handle_gimple_call): Call it.
27640 2017-05-14 Martin Sebor <msebor@redhat.com>
27642 PR middle-end/80669
27643 * builtins.c (expand_builtin_stpncpy): Simplify.
27645 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
27647 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
27648 * config/i386/i386.h
27649 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
27650 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
27651 (struct machine_function): Add new members call_ms2sysv,
27652 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
27653 (struct machine_frame_state): New fields sp_realigned and
27654 sp_realigned_offset.
27655 * config/i386/i386.c
27656 (enum xlogue_stub): New enum.
27657 (enum xlogue_stub_sets): New enum.
27658 (class xlogue_layout): New class.
27659 (struct ix86_frame): New fields stack_realign_allocate_offset,
27660 stack_realign_offset and outlined_save_offset. Modify comments to
27661 detail stack layout when using out-of-line stubs.
27662 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
27663 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
27664 -mcall-ms2sysv-xlogues.
27665 (stub_managed_regs): New static variable.
27666 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
27667 registers managed by out-of-line stub.
27668 (disable_call_ms2sysv_xlogues): New function.
27669 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
27670 m->call_ms2sysv when appropriate and compute frame layout for
27672 (sp_valid_at, fp_valid_at): New inline functions.
27673 (choose_basereg): New function.
27674 (choose_baseaddr): Add align parameter, use choose_basereg and modify
27676 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
27677 Use align parameter of choose_baseaddr to generated aligned SSE movs
27679 (pro_epilogue_adjust_stack): Modify to track
27680 machine_frame_state::sp_realigned.
27681 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
27682 (ix86_nsaved_sseregs): Likewise.
27683 (ix86_emit_save_regs): Likewise.
27684 (ix86_emit_save_regs_using_mov): Likewise.
27685 (ix86_emit_save_sse_regs_using_mov): Likewise.
27686 (get_scratch_register_on_entry): Likewise.
27687 (gen_frame_set): New function.
27688 (gen_frame_load): Likewise.
27689 (gen_frame_store): Likewise.
27690 (emit_outlined_ms2sysv_save): Likewise.
27691 (emit_outlined_ms2sysv_restore): Likewise.
27692 (ix86_expand_prologue): Modify stack re-alignment code and call
27693 emit_outlined_ms2sysv_save when appropriate.
27694 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
27695 parameter rtx_insn *insn, which allows the function to be used to only
27696 generate the notes.
27697 (ix86_expand_epilogue): Modify validity checks of frame and stack
27698 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
27699 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
27700 * config/i386/predicates.md
27701 (save_multiple): New predicate.
27702 (restore_multiple): Likewise.
27703 * config/i386/sse.md
27704 (save_multiple<mode>): New pattern.
27705 (save_multiple_realign<mode>): Likewise.
27706 (restore_multiple<mode>): Likewise.
27707 (restore_multiple_and_return<mode>): Likewise.
27708 (restore_multiple_leave_return<mode>): Likewise.
27709 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
27711 2017-05-14 Julia Koval <julia.koval@intel.com>
27713 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
27714 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
27715 (__builtin_ia32_xsetbv): New builtins.
27716 * config/i386/i386.c (ix86_expand_special_args_builtin):
27718 (ix86_expand_builtin): Special expand for new intrinsics.
27719 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
27720 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
27721 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
27723 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27725 * cfganal.c (inverted_post_order_compute): Change argument type
27727 * cfganal.h (inverted_post_order_compute): Adjust prototype.
27728 * df-core.c (rest_of_handle_df_initialize): Adjust.
27729 (rest_of_handle_df_finish): Likewise.
27730 (df_analyze_1): Likewise.
27731 (df_analyze): Likewise.
27732 (loop_inverted_post_order_compute): Change argument to be a vec *.
27733 (df_analyze_loop): Adjust.
27734 (df_get_n_blocks): Likewise.
27735 (df_get_postorder): Likewise.
27736 * df.h (struct df_d): Change field to be a vec.
27737 * lcm.c (compute_laterin): Adjust.
27738 (compute_available): Likewise.
27739 * lra-lives.c (lra_create_live_ranges_1): Likewise.
27740 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
27741 * tree-ssa-pre.c (compute_antic): Likewise.
27743 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27745 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
27746 (depth_first_search::depth_first_search): Change structure init
27747 function to this constructor.
27748 (depth_first_search::add_bb): Rename function to this member.
27749 (depth_first_search::execute): Likewise.
27750 (flow_dfs_compute_reverse_finish): Adjust.
27752 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27754 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
27755 (longest_simple_path): Likewise.
27756 * shrink-wrap.c (spread_components): Likewise.
27757 (disqualify_problematic_components): Likewise.
27758 (emit_common_heads_for_components): Likewise.
27759 (emit_common_tails_for_components): Likewise.
27760 (insert_prologue_epilogue_for_components): Likewise.
27762 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27764 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
27767 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27769 * df-core.c (df_set_blocks): Start using auto_bitmap.
27770 (df_compact_blocks): Likewise.
27771 * df-problems.c (df_rd_confluence_n): Likewise.
27772 * df-scan.c (df_insn_rescan_all): Likewise.
27773 (df_process_deferred_rescans): Likewise.
27774 (df_update_entry_block_defs): Likewise.
27775 (df_update_exit_block_uses): Likewise.
27776 (df_entry_block_bitmap_verify): Likewise.
27777 (df_exit_block_bitmap_verify): Likewise.
27778 (df_scan_verify): Likewise.
27779 * lra-constraints.c (lra_constraints): Likewise.
27780 (undo_optional_reloads): Likewise.
27781 (lra_undo_inheritance): Likewise.
27782 * lra-remat.c (calculate_gen_cands): Likewise.
27783 (do_remat): Likewise.
27784 * lra-spills.c (assign_spill_hard_regs): Likewise.
27785 (spill_pseudos): Likewise.
27786 * tree-ssa-pre.c (bitmap_set_and): Likewise.
27787 (bitmap_set_subtract_values): Likewise.
27789 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27791 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
27792 management with auto_bitmap.
27793 (fix_inter_tick): Likewise.
27794 (fix_recovery_deps): Likewise.
27795 * ira.c (add_store_equivs): Likewise.
27796 (find_moveable_pseudos): Likewise.
27797 (split_live_ranges_for_shrink_wrap): Likewise.
27798 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
27799 (rtx_reuse_manager::seen_def_p): Likewise.
27800 (rtx_reuse_manager::set_seen_def): Likewise.
27801 * print-rtl.h (class rtx_reuse_manager): Likewise.
27803 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27805 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
27807 (migrate_btr_def): Likewise.
27808 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
27809 * df-core.c (loop_post_order_compute): Likewise.
27810 (loop_inverted_post_order_compute): Likewise.
27811 * hsa-common.h: Likewise.
27812 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
27813 * init-regs.c (initialize_uninitialized_regs): Likewise.
27814 * ipa-inline.c (resolve_noninline_speculation): Likewise.
27815 (inline_small_functions): Likewise.
27816 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
27817 * ira.c (combine_and_move_insns): Likewise.
27818 (build_insn_chain): Likewise.
27819 * loop-invariant.c (find_invariants): Likewise.
27820 * lower-subreg.c (propagate_pseudo_copies): Likewise.
27821 * predict.c (tree_predict_by_opcode): Likewise.
27822 (predict_paths_leading_to): Likewise.
27823 (predict_paths_leading_to_edge): Likewise.
27824 (estimate_loops_at_level): Likewise.
27825 (estimate_loops): Likewise.
27826 * shrink-wrap.c (try_shrink_wrapping): Likewise.
27827 (spread_components): Likewise.
27828 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
27829 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
27830 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
27831 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
27832 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
27833 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
27834 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
27835 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
27836 (mark_threaded_blocks): Likewise.
27837 (thread_through_all_blocks): Likewise.
27838 * tree-ssa.c (verify_ssa): Likewise.
27839 (execute_update_addresses_taken): Likewise.
27840 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
27842 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27844 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
27846 (post_order_compute): Likewise.
27847 (inverted_post_order_compute): Likewise.
27848 (pre_and_rev_post_order_compute_fn): Likewise.
27850 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27852 * genrecog.c (int_set::int_set): Explicitly construct our
27853 auto_vec base class.
27854 * vec.h (auto_vec::auto_vec): New constructor.
27856 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27858 * bitmap.h (class auto_bitmap): New constructor taking
27859 bitmap_obstack * argument.
27861 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27863 * bitmap.h (class auto_bitmap): Change type of m_bits to
27864 bitmap_head, and adjust ctor / dtor and member operators.
27866 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
27868 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
27869 when returned register mode doesn't match original mode.
27871 2017-05-12 Jeff Law <law@redhat.com>
27872 Jakub Jelinek <jakub@redhat.com>
27874 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
27875 we look for cc setter after the compare-elim changes.
27876 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
27877 within the vector to match what compare-elim now expects.
27878 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
27879 (xorsi3_flags, one_cmplsi2_flags): Likewise.
27881 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
27882 after the compare-elim changes.
27883 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
27884 the vector to match what compare-elim now expects.
27885 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
27886 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
27887 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
27888 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
27889 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
27891 * config/visium/visium.c (single_set_and_flags): Fix where
27892 we look for cc setter after the compare-elim changes.
27893 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
27894 with the vector to match what compare-elim now expects.
27895 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
27896 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
27897 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
27898 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
27899 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
27900 (neg<mode>2_insn_set_overflow): Likewise.
27902 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
27904 PR middle-end/79794
27905 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
27906 maybe_expand_insn call, set ops[0].target. If still set after call,
27907 set alt_rtl. Add extra arg to recursive calls.
27908 (extract_bit_field): Add alt_rtl argument. Pass to
27910 * expmed.h (extract_bit_field): Fix prototype.
27911 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
27912 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
27913 to extract_bit_field_calls.
27914 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
27915 Pass alt_rtl to extract_bit_field calls.
27916 * calls.c (store_unaligned_arguments_into_psuedos)
27917 load_register_parameters): Pass extra NULL to extract_bit_field calls.
27918 * optabs.c (maybe_legitimize_operand): Clear op->target when call
27920 * optabs.h (struct expand_operand): Add target bitfield.
27922 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
27924 * compare-elim.c (try_eliminate_compare): Canonicalize
27925 operation with embedded compare to
27926 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
27927 (set (reg) (operation)].
27929 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
27931 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
27934 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
27935 cost of adding a carry flag for ADC instruction.
27936 [case MINUS]: Ignore the cost of subtracting a carry flag
27937 for SBB instruction.
27939 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
27941 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
27943 * config/rs6000/bmiintrin.h: New file.
27944 * config/rs6000/bmi2intrin.h: New file.
27945 * config/rs6000/x86intrin.h: New file.
27947 2017-05-12 Jeff Law <law@redhat.com>
27949 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
27952 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
27954 PR middle-end/80707
27955 * tree-cfg.c: Remove cfg edges of unreachable case statements.
27957 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27959 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
27960 early expansion of vector divide builtins.
27961 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
27962 builtins identified as having unsigned arguments.
27964 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27966 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
27967 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
27968 expansion of vector logical operations (and, andc, or, xor,
27971 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27973 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
27974 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
27976 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27978 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
27979 early GIMPLE expansion of vector multiplies.
27981 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27983 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
27984 TARGET_HAVE_MOVT conditional.
27985 (movt splitter): Likewise.
27987 2017-05-12 Richard Biener <rguenther@suse.de>
27989 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
27990 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
27991 Fold all stmts not inplace.
27993 2017-05-12 Richard Biener <rguenther@suse.de>
27995 PR tree-optimization/80713
27996 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
27997 inserted_exprs bit for not removed stmts.
27999 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
28001 PR middle-end/69921
28002 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
28003 parallelized" attribute for parallelized OpenACC kernels.
28004 * omp-offload.c (execute_oacc_device_lower): Use it.
28006 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
28007 Set "oacc kernels" attribute.
28008 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
28009 parameter. Adjust all users.
28010 (oacc_fn_attrib_kernels_p): Remove function.
28011 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
28012 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
28013 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
28014 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
28015 assert "oacc kernels" attribute is set.
28017 2017-05-11 Carl Love <cel@us.ibm.com>
28019 * config/rs6000/rs6000-c: Add support for built-in functions
28020 vector unsigned char vec_popcnt (vector signed char)
28021 vector unsigned char vec_popcnt (vector unsigned char)
28022 vector unsigned short vec_popcnt (vector signed short)
28023 vector unsigned short vec_popcnt (vector unsigned short)
28024 vector unsigned int vec_popcnt (vector signed int)
28025 vector unsigned int vec_popcnt (vector unsigned int)
28026 vector unsigned long long vec_popcnt (vector signed long long)
28027 vector unsigned long long vec_popcnt (vector unsigned long long)
28028 vector signed long long vec_slo (vector signed long long,
28029 vector signed char)
28030 vector signed long long vec_slo (vector signed long long,
28031 vector unsigned char)
28032 vector unsigned long long vec_slo (vector unsigned long long,
28033 vector signed char)
28034 vector unsigned long long vec_slo (vector unsigned long long,
28035 vector unsigned char)
28036 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
28037 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
28038 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
28039 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
28040 * doc/extend.texi: Update the built-in documentation file for the
28041 new built-in functions.
28043 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
28045 * attribs.h (sorted_attr_string): Move machine independent
28046 functions for target clone support from the i386 port to common
28047 code. Rename ix86_function_versions to common_function_versions.
28048 Rename make_name to make_unique_name.
28049 (common_function_versions): Likewise.
28050 (make_unique_name): Likewise.
28051 (make_dispatcher_decl): Likewise.
28052 (is_function_default_version): Likewise.
28053 * attribs.c (attr_strcmp): Likewise.
28054 (sorted_attr_string): Likewise.
28055 (common_function_versions): Likewise.
28056 (make_unique_name): Likewise.
28057 (make_dispatcher_decl): Likewise.
28058 (is_function_default_version): Likewise.
28059 * config/i386/i386.c (attr_strcmp): Likewise.
28060 (sorted_attr_string): Likewise.
28061 (ix86_function_versions): Likewise.
28062 (make_name): Likewise.
28063 (make_dispatcher_decl): Likewise.
28064 (is_function_default_version): Likewise.
28065 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
28067 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28070 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
28071 Account for direct move costs for vec_construct of integer
28074 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
28077 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
28078 (UNSPEC_STX_ATOMIC): Ditto.
28079 (loaddi_via_sse): New insn.
28080 (storedi_via_sse): Ditto.
28081 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
28082 Update corresponding peephole2 patterns.
28083 (atomic_storedi_fpu): Ditto.
28085 2017-05-11 Julia Koval <julia.koval@intel.com>
28087 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
28088 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
28090 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
28091 (__builtin_ia32_rsqrt14ss_mask): New builtins.
28092 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
28094 2017-05-11 Nathan Sidwell <nathan@acm.org>
28096 * graphite-poly.c: Include dumpfile.h.
28098 * dumpfle.h (dump_function): Declare here ...
28099 * tree-dump.h (dump_function): ... not here.
28100 * dumpfile.c: #include tree-cfg.h.
28101 (dump_function): Move here from ...
28102 * tree-dump.c (dump_function): ... here.
28103 * gimplify.c: #include splay-tree.h, not tree-dump.h.
28104 * graphite-poly.c: Don't include tree-dump.h.
28105 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
28106 * print-tree.c: Likewise.
28107 * stor-layout.c: Likewise.
28108 * tree-nested.c: Likewise.
28110 * dumpfile.c (dump_start): Use TDF_FLAGS.
28111 (dump_enable_all): Fix TDF_KIND check thinko.
28113 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
28115 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
28116 array entries to represent two legal parameterizations of the
28117 overloaded __builtin_cmpb function, as represented by the
28118 P6_OV_BUILTIN_CMPB constant.
28119 (altivec_resolve_overloaded_builtin): Add special case handling
28120 for the __builtin_cmpb function, as represented by the
28121 P6_OV_BUILTIN_CMPB constant.
28122 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
28123 (BU_P6_64BIT_2): New macro.
28124 (BU_P6_OVERLOAD_2): New macro
28125 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
28126 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
28127 (CMPB): Add overload support to represent both 32-bit and 64-bit
28128 compare-bytes function.
28129 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
28130 support for TARGET_CMPB.
28131 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
28132 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28133 documentation of the __builtin_cmpb overloaded built-in function.
28135 2017-05-11 Richard Biener <rguenther@suse.de>
28137 PR tree-optimization/80705
28138 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
28139 bases are not vectorizable.
28141 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28143 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
28144 when counting register pressure.
28146 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28148 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
28149 (iv_ca_more_deps): Renamed to ...
28150 (iv_ca_compare_deps): ... this.
28151 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
28153 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28155 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
28157 (determine_group_iv_costs): ... here.
28158 (find_inv_vars_cb): Record inv var if it's not recorded before.
28160 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28162 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
28163 (get_shiftadd_cost): Ditto.
28165 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28167 * tree-ssa-address.c: Include header file.
28168 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
28170 (add_to_parts): Refactor.
28171 (addr_to_parts): New parameter. Update use of move_hint_to_base.
28172 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
28175 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28177 PR tree-optimization/53090
28178 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
28180 (extract_cond_operands): Detect condition with IV on both sides
28181 and return COMP_IV_EXPR_2.
28182 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
28183 (rewrite_use_compare): Simplify by removing call to function
28184 extract_cond_operands.
28186 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28188 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
28189 (extract_cond_operands): Detect condition comparing against non-
28190 invariant bound and return appropriate enum value.
28191 (find_interesting_uses_cond): Update use of extract_cond_operands.
28192 Handle its return value accordingly.
28193 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
28195 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28197 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
28198 nonlinear iv_use computation in loop invariant sensitive way.
28200 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28202 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
28203 (find_iv_candidates): Call relate_compare_use_with_all_cands.
28205 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28207 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
28208 (dump_cand): Support iv_cand.inv_exprs.
28209 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
28211 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
28214 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28216 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
28218 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
28219 as local function. Include necessary header files.
28220 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
28222 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28224 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
28226 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28228 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
28229 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
28230 RSHIFT_EXPR and BIT_NOT_EXPR.
28232 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28234 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
28235 (adjust_setup_cost): New parameter supporting round up adjustment.
28236 (struct address_cost_data): Delete.
28237 (force_expr_to_var_cost): Don't bound cost with spill_cost.
28238 (split_address_cost, ptr_difference_cost): Delete.
28239 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
28240 (struct ainc_cost_data): New struct.
28241 (get_address_cost_ainc): New function.
28242 (get_address_cost, get_computation_cost): Reimplement.
28243 (determine_group_iv_cost_address): Record inv_expr for all uses of
28245 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
28246 (iv_ca_has_deps): Reimplemented to ...
28247 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
28249 (iv_ca_extend): Call iv_ca_more_deps.
28251 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28253 * tree-ssa-address.c (struct mem_address): Move to header file.
28254 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
28255 * tree-ssa-address.h (struct mem_address): Move from C file.
28256 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
28258 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28260 * tree-affine.h (aff_combination_type): New interface.
28261 (aff_combination_zero_p): Remove static.
28262 (aff_combination_const_p): New interface.
28263 (aff_combination_singleton_var_p): New interfaces.
28265 2017-05-11 Richard Biener <rguenther@suse.de>
28267 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28268 Skip unreachable blocks and destinations.
28269 (eliminate): Move stmt removal and fixup ...
28270 (fini_eliminate): ... here. Skip inserted exprs.
28271 (pass_pre::execute): Move fini_pre after fini_eliminate.
28272 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
28273 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
28274 PRE to get rid of dead code that has invalid SSA form and
28275 split critical edges again.
28277 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28279 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28281 2017-05-11 Richard Biener <rguenther@suse.de>
28283 * passes.c (execute_function_todo): Verify loops if they are
28284 said to be up-to-date.
28285 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
28286 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
28288 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
28291 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
28292 handle calling assemble_external ourself.
28295 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
28296 modes with zero size. Enhance comment.
28298 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28300 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
28301 built-ins for vec_xl and vec_xst with short and char pointer
28304 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
28306 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
28307 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
28308 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
28309 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
28310 (_mm_maskz_min_round_ss): New intrinsics.
28311 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
28312 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
28313 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
28314 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
28315 (__builtin_ia32_minss_mask_round): New builtins.
28316 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28317 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28318 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
28320 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
28321 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
28323 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
28326 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
28328 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
28329 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
28330 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
28331 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
28332 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
28333 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
28334 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
28335 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28336 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
28337 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
28338 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
28339 (__builtin_ia32_mulss_mask_round): New builtins.
28340 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28341 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28342 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
28344 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
28345 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
28347 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
28350 2017-05-10 Julia Koval <julia.koval@intel.com>
28352 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
28353 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
28354 (_mm256_setr_m128i): New intrinsics.
28356 2017-05-10 Julia Koval <julia.koval@intel.com>
28358 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
28359 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
28360 (_mm_maskz_rcp14_ss): New intrinsics.
28361 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
28362 (__builtin_ia32_rcp14ss_mask): New builtins.
28363 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
28365 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
28367 PR tree-optimization/51513
28368 * tree-cfg.c (gimple_seq_unreachable_p): New function.
28369 (assert_unreachable_fallthru_edge_p): Use it.
28370 (group_case_labels_stmt): Likewise.
28371 * tree-cfg.h: Prototype it.
28372 * stmt.c: Include cfghooks.h and tree-cfg.h.
28373 (emit_case_dispatch_table) <gap_label>: New local variable.
28374 Use it to fill dispatch table gaps.
28375 Test for default_label before updating probabilities.
28376 (expand_case) <default_label>: Remove unneeded initialization.
28377 Test for unreachable default case statement and remove its edge.
28378 Set default_label accordingly.
28379 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
28381 2017-05-10 Carl Love <cel@us.ibm.com>
28383 * config/rs6000/rs6000-c: Add support for built-in functions
28384 vector signed char vec_neg (vector signed char)
28385 vector signed short int vec_neg (vector short int)
28386 vector signed int vec_neg (vector signed int)
28387 vector signed long long vec_neg (vector signed long long)
28388 vector float vec_neg (vector float)
28389 vector double vec_neg (vector double)
28390 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
28392 * config/rs6000/altivec.h: Add define for vec_neg
28393 * doc/extend.texi: Update the built-in documentation for the
28394 new built-in functions.
28396 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28398 PR tree-optimization/77644
28399 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
28401 2017-05-10 Nathan Sidwell <nathan@acm.org>
28403 * dumpfile.h (TDI_lang_all): New.
28404 (TDF_KIND): New. Renumber others
28405 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
28407 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
28409 (get_dump_file_name): Adjust suffix generation.
28410 (dump_enable_all): Use TDF_KIND.
28411 * doc/invoke.texi (-fdump-lang-all): Document.
28413 * dumpfile.h: Tabify.
28415 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
28418 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
28419 Move member access before delete.
28421 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
28423 * tree-inline.c (expand_call_inline): Split block at stmt
28426 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
28429 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
28430 are now unused after splitting mov{sf,sd}_hardfloat.
28431 (f32_lr2): Likewise.
28432 (f32_lm): Likewise.
28433 (f32_lm2): Likewise.
28434 (f32_li): Likewise.
28435 (f32_li2): Likewise.
28436 (f32_lv): Likewise.
28437 (f32_sr): Likewise.
28438 (f32_sr2): Likewise.
28439 (f32_sm): Likewise.
28440 (f32_sm2): Likewise.
28441 (f32_si): Likewise.
28442 (f32_si2): Likewise.
28443 (f32_sv): Likewise.
28444 (f32_dm): Likewise.
28445 (f32_vsx): Likewise.
28446 (f32_av): Likewise.
28447 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
28448 For movsf, order stores so the VSX stores occur before the GPR
28449 store which encourages the register allocator to use a traditional
28450 FPR instead of a GPR. For movsd, order the stores so that the GPR
28451 store comes before the VSX stores to allow the power6 to work.
28452 This is due to the power6 not having a 32-bit integer store
28453 instruction from a FPR.
28454 (movsf_hardfloat): Likewise.
28455 (movsd_hardfloat): Likewise.
28457 2017-05-09 Martin Sebor <msebor@redhat.com>
28459 PR translation/80280
28460 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
28463 PR translation/80280
28464 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
28465 data member added in r247778.
28466 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
28468 2017-05-09 Nathan Sidwell <nathan@acm.org>
28470 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
28472 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
28475 2017-05-09 Marek Polacek <polacek@redhat.com>
28477 * doc/invoke.texi: Fix typo.
28479 2017-05-09 Richard Biener <rguenther@suse.de>
28481 * tree-vrp.c (vrp_val_is_max): Adjust comment.
28482 (vrp_val_is_min): Likewise.
28483 (set_value_range_to_value): Likewise.
28484 (set_value_range_to_nonnegative): Likewise.
28485 (gimple_assign_nonzero_p): Likewise.
28486 (gimple_stmt_nonzero_p): Likewise.
28487 (vrp_int_const_binop): Likewise. Remove unreachable case.
28488 (adjust_range_with_scev): Adjust comments.
28489 (compare_range_with_value): Likewise.
28490 (extract_range_from_phi_node): Likewise.
28491 (test_for_singularity): Likewise.
28493 2017-05-09 Richard Biener <rguenther@suse.de>
28495 * tree-vrp.c (get_single_symbol): Add assert that we don't
28496 get overflowed constants as invariant part.
28497 (compare_values_warnv): Add comment before the TREE_NO_WARNING
28498 checks. Use wi::cmp instead of recursing for integer constants.
28499 (compare_values): Just ignore whether we assumed undefined
28500 overflow instead of failing the compare.
28501 (extract_range_for_var_from_comparison_expr): Add comment before the
28502 TREE_NO_WARNING sets.
28503 (test_for_singularity): Likewise.
28504 (extract_range_from_comparison): Do not disable optimization
28505 when we assumed undefined overflow.
28506 (extract_range_basic): Remove init of unused var.
28508 2017-05-09 Richard Biener <rguenther@suse.de>
28510 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
28511 (extract_range_from_multiplicative_op_1): Adjust.
28512 (extract_range_from_binary_expr_1): Use int_const_binop.
28514 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
28517 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
28518 rs6000_store_data_bypass_p in seven define_bypass directives and
28519 in several comments.
28520 * config/rs6000/rs6000-protos.h: Add prototype for
28521 rs6000_store_data_bypass_p function.
28522 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
28523 function implements slightly different (rs6000-specific) semantics
28524 than store_data_bypass_p, returning false rather than aborting
28525 with assertion error when arguments do not satisfy the
28526 requirements of store data bypass.
28527 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
28528 rs6000_store_data_bypass_p.
28530 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
28532 * config/xtensa/xtensa-protos.h
28533 (xtensa_initial_elimination_offset): New declaration.
28534 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28535 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
28536 macro definition, add case for FRAME_POINTER_REGNUM when
28537 FRAME_GROWS_DOWNWARD.
28538 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
28539 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
28540 xtensa_initial_elimination_offset.
28542 2017-05-08 Nathan Sidwell <nathan@acm.org>
28544 * doc/invoke.texi: Alphabetize -fdump options.
28546 2017-05-08 Martin Sebor <msebor@redhat.com>
28548 PR translation/80280
28549 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
28551 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
28553 * target.def (compute_frame_layout): New optional target hook.
28554 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
28555 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
28556 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
28558 * reload1.c (verify_initial_elim_offsets): Likewise.
28559 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
28560 (use_simple_return_p): Call arm_compute_frame_layout if needed.
28561 (arm_get_frame_offsets): Split up into this ...
28562 (arm_compute_frame_layout): ... and this function.
28564 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
28566 * config/aarch64/constraints.md (Usa): New constraint.
28567 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
28569 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
28571 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
28572 with_multilib_list after it has been checked.
28574 2017-05-08 Richard Biener <rguenther@suse.de>
28576 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
28577 (bitmap_set_subtract_values): Likewise.
28579 2017-05-08 Richard Biener <rguenther@suse.de>
28581 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
28582 (gimple_assign_nonzero): ... this and remove strict_overflow_p
28584 (gimple_stmt_nonzero_warnv_p): Rename to ...
28585 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
28587 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
28588 (extract_range_basic): Adjust, do not disable propagation on
28589 strict overflow sensitive simplification.
28590 (vrp_visit_cond_stmt): Likewise.
28592 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
28594 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
28595 body size unconditionally.
28597 2017-05-07 Jeff Law <law@redhat.com>
28600 2017-05-06 Jeff Law <law@redhat.com>
28601 PR tree-optimization/78496
28602 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
28605 PR tree-optimization/78496
28606 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
28607 (simplify_stmt_using_ranges): Call it.
28608 (vrp_dom_walker::before_dom_children): Extract equivalences
28609 from an ASSERT_EXPR with an equality comparison against a
28612 2017-05-06 Jeff Law <law@redhat.com>
28614 PR tree-optimization/78496
28615 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
28618 PR tree-optimization/78496
28619 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
28620 (simplify_stmt_using_ranges): Call it.
28621 (vrp_dom_walker::before_dom_children): Extract equivalences
28622 from an ASSERT_EXPR with an equality comparison against a
28625 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
28627 * lra-constraints.c (lra_copy_reg_equiv): New function.
28628 (split_reg): Use it to copy equivalence information from the
28629 original register to the spill register.
28631 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
28633 PR rtl-optimization/75964
28634 * simplify-rtx.c (simplify_const_relational_operation): Remove
28635 invalid handling of comparisons of integer ABS.
28637 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
28639 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
28640 initialize to zero.
28641 (init_regs): Remove declaration.
28642 (function_arg_advance_32): Initialize error_p as boolean variable.
28644 2017-05-05 Nathan Sidwell <nathan@acm.org>
28646 * store-motion.c (remove_reachable_equiv_notes): Reformat long
28647 lines. Use for (;;).
28649 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28651 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
28652 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
28653 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
28654 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
28655 VF=2 that require versioning.
28657 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28659 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
28662 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28664 * diagnostic.h (diagnostic_override_option_index): Convert from
28665 macro to inline function.
28667 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28669 * diagnostic.c (last_module_changed_p): New function.
28670 (set_last_module): New function.
28671 (diagnostic_report_current_module): Convert macro usage to
28672 the above functions.
28673 * diagnostic.h (diagnostic_context::last_module): Strengthen
28674 from const line_map * to const line_map_ordinary *.
28675 (diagnostic_last_module_changed): Delete macro.
28676 (diagnostic_set_last_module): Delete macro.
28678 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28680 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
28681 with diagnostic_report_diagnostic.
28682 (diagnostic_n_impl_richloc): Likewise.
28683 * diagnostic.h (report_diagnostic): Delete macro.
28684 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
28685 with diagnostic_report_diagnostic.
28686 * substring-locations.c (format_warning_va): Likewise.
28688 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28690 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
28691 save/restor of format_spec. Move option-printing code to...
28692 (print_option_information): ...this new function, and
28693 reimplement by simply printing to the pretty_printer,
28694 rather than appending to the format string.
28696 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28698 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
28699 handling logic into...
28700 (update_effective_level_from_pragmas): ...this new function.
28702 2017-05-04 Andrew Waterman <andrew@sifive.com>
28704 * config/riscv/riscv.opt (mstrict-align): New option.
28705 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
28706 (SLOW_UNALIGNED_ACCESS): Define.
28707 (riscv_slow_unaligned_access): Declare.
28708 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
28710 (riscv_slow_unaligned_access): New variable.
28711 (rocket_tune_info): Set slow_unaligned_access to true.
28712 (optimize_size_tune_info): Set slow_unaligned_access to false.
28713 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
28714 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
28715 (riscv_option_override): Set riscv_slow_unaligned_access.
28716 * doc/invoke.texi: Add -mstrict-align to RISC-V.
28718 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
28720 * config/riscv/riscv.md: Unify indentation.
28722 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28727 * config/rs6000/rs6000.md (u code attribute): Add FIX and
28729 (extendsi<mode>2): Add support for doing sign extension via
28730 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
28731 don't have ISA 3.0 instructions.
28732 (extendsi<mode>2 splitter): Likewise.
28733 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
28734 generate the normal insns since SImode can now go in vector
28735 registers. Disallow the special UNSPECs needed for previous
28736 machines to hide SImode being used. Add new insns
28737 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
28738 (fix_trunc<mode>si2_stfiwx): Likewise.
28739 (fix_trunc<mode>si2_internal): Likewise.
28740 (fixuns_trunc<mode>si2): Likewise.
28741 (fixuns_trunc<mode>si2_stfiwx): Likewise.
28742 (fctiw<u>z_<mode>_smallint): Likewise.
28743 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
28744 of floating point to 32-bit integer from doing a direct move to
28745 the GPR registers to do a store.
28746 (fctiwz_<mode>): Break long line.
28748 2017-05-05 Bin Cheng <bin.cheng@arm.com>
28750 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
28751 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
28752 (addr_list, addr_offset_valid_p): New.
28753 (split_address_groups): Check offset validity with above function.
28754 (gt-tree-ssa-loop-ivopts.h): Include header file.
28756 2017-05-05 Nathan Sidwell <nathan@acm.org>
28758 * config.gcc (arm*-*-*): Add missing 'fi'.
28760 2017-05-05 Steve Ellcey <sellcey@cavium.com>
28762 * doc/invoke.texi (-fopt-info): Explicitly say order of options
28763 included in -fopt-info does not matter.
28764 * doc/optinfo.texi (-fopt-info): Fix description of default
28765 behavour. Explicitly say order of options included in -fopt-info
28768 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28770 * config.gcc: Allow combinations of aprofile and rmprofile values for
28771 --with-multilib-list.
28772 * config/arm/t-multilib: New file.
28773 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
28774 variables. Remove setting of ISA and floating-point ABI in
28775 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
28776 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
28777 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
28778 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
28780 * config/arm/t-rmprofile: Likewise except for the matches changes.
28781 * doc/install.texi (--with-multilib-list): Document the combination of
28782 aprofile and rmprofile values and warn about pitfalls in doing that.
28784 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
28786 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
28787 (movdi_aarch64): Likewise.
28789 2017-05-05 Jakub Jelinek <jakub@redhat.com>
28791 PR tree-optimization/80632
28792 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
28794 (build_arrays): Initialize it for virtual phis.
28795 (fix_phi_nodes): Use it for virtual phis.
28797 PR tree-optimization/80558
28798 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
28799 [x, y] op z into [x op, y op z] for op & or | if conditions
28802 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
28803 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28806 * config/arm/arm.md (use_literal_pool): Remove.
28807 (64-bit immediate split): No longer takes cost into consideration
28808 if arm_disable_literal_pool is enabled.
28809 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
28810 used when arm_disable_literal_pool is enabled.
28811 (arm_max_const_double_inline_cost): Remove use of
28812 arm_disable_literal_pool.
28813 (push_minipool_fix): Add assert.
28814 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
28815 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
28816 (no_literal_pool_sf_immediate): New.
28818 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28820 PR tree-optimization/80613
28821 * tree-ssa-dce.c (propagate_necessity): Remove cases for
28822 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
28824 2017-05-05 Richard Biener <rguenther@suse.de>
28826 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
28828 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
28830 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
28831 of this flag from insn conditions due to removal from r247495.
28833 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
28835 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
28837 (arm_early_store_addr_dep_ptr): Likewise.
28838 * config/arm/aarch-common-protos.h
28839 (arm_early_load_addr_dep_ptr): Add prototype.
28840 (arm_early_store_addr_dep_ptr): Likewise.
28841 * config/arm/cortex-a53.md: Add new bypasses.
28843 2017-05-05 Jakub Jelinek <jakub@redhat.com>
28845 * tree.c (next_type_uid): Change type to unsigned.
28846 (type_hash_canon): Decrement back next_type_uid if
28847 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
28848 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
28851 2017-05-04 Martin Sebor <msebor@redhat.com>
28853 * builtins.c: Fix a trivial typo in a comment.
28855 PR middle-end/79234
28856 * builtins.c (check_sizes): Adjust to handle reading past the end.
28857 Avoid printing excessive upper bound of ranges. Use %E to print
28858 tree nodes instead of converting them to %wu.
28859 (expand_builtin_memchr): New function.
28860 (compute_dest_size): Rename...
28861 (compute_objsize): ...to this.
28862 (expand_builtin_memcpy): Adjust.
28863 (expand_builtin_mempcpy): Adjust.
28864 (expand_builtin_strcat): Adjust.
28865 (expand_builtin_strcpy): Adjust.
28866 (check_strncat_sizes): Adjust.
28867 (expand_builtin_strncat): Adjust.
28868 (expand_builtin_strncpy): Adjust and simplify.
28869 (expand_builtin_memset): Adjust.
28870 (expand_builtin_bzero): Adjust.
28871 (expand_builtin_memcmp): Adjust.
28872 (expand_builtin): Handle memcmp.
28873 (maybe_emit_chk_warning): Check strncat just once.
28875 2017-05-04 Martin Sebor <msebor@redhat.com>
28877 PR preprocessor/79214
28878 PR middle-end/79222
28879 PR middle-end/79223
28880 * builtins.c (check_sizes): Add inlining context and issue
28881 warnings even when -Wno-system-headers is set.
28882 (check_strncat_sizes): Same.
28883 (expand_builtin_strncat): Same.
28884 (expand_builtin_memmove): New function.
28885 (expand_builtin_stpncpy): Same.
28886 (expand_builtin): Handle memmove and stpncpy.
28888 2017-05-04 Bin Cheng <bin.cheng@arm.com>
28890 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
28891 which is not used any more.
28893 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28895 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
28897 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28899 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
28900 (cortexa53_tunings): Likewise.
28901 (cortexa57_tunings): Likewise.
28902 (cortexa72_tunings): Likewise.
28903 (cortexa73_tunings): Likewise.
28905 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28907 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
28908 Set loop alignment to 8.
28910 2017-05-04 Martin Sebor <msebor@redhat.com>
28912 PR translation/80280
28913 * builtins.c (expand_builtin_object_size): Add missing quoting to
28914 %D and like directives.
28915 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
28916 (hsa_type_for_tree_type): Same.
28917 (verify_function_arguments): Same.
28918 * symtab.c (symbol_table::change_decl_assembler_name): Same.
28919 * varasm.c (get_section): Same.
28922 2017-05-04 Martin Sebor <msebor@redhat.com>
28924 PR translation/80280
28925 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
28927 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28929 * config/aarch64/aarch64.c (generic_addrcost_table):
28930 Change HI/TI mode setting.
28932 2017-05-04 Martin Jambor <mjambor@suse.cz>
28934 PR tree-optimization/80622
28935 * tree-sra.c (comes_initialized_p): New function.
28936 (build_accesses_from_assign): Only set write lazily when
28937 comes_initialized_p is false.
28938 (analyze_access_subtree): Use comes_initialized_p.
28939 (propagate_subaccesses_across_link): Assert !comes_initialized_p
28940 instead of testing for PARM_DECL.
28942 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28944 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
28945 constraint on operand 0 to allow more general addressing modes.
28946 Adjust output template.
28947 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
28949 * config/aarch64/aarch64-protos.h
28950 (aarch64_address_valid_for_prefetch_p): Declare prototype.
28951 * config/aarch64/constraints.md (Dp): New address constraint.
28952 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
28955 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
28957 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
28958 update use of estimate_ipcp_clone_size_and_time.
28959 (estimate_local_effects): Update use of
28960 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
28961 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
28962 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
28963 Return nonspecialized time.
28965 2017-05-04 Richard Biener <rguenther@suse.de>
28967 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
28968 for the last VUSE which def dominates the PHI. Directly call
28970 (get_continuation_for_phi_1): Remove.
28972 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
28974 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
28975 to explain the use of truncating division. Cap the number of
28976 iterations to the maximum given by nb_iterations_upper_bound,
28979 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28981 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
28982 * configure: Regenerate.
28983 * config.in: Regenerate.
28984 * config/i386/driver-mingw32.c: new file.
28985 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
28986 * config.host: Link driver-mingw32.o on MinGW host.
28987 * doc/install.texi: Document new --enable-mingw-wildcard configure
28990 2017-05-04 Marek Polacek <polacek@redhat.com>
28992 PR tree-optimization/80612
28993 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
28995 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28996 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
28998 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
28999 (movt splitter): Likewise.
29000 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
29001 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
29002 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
29003 block for Thumb-1 with MOVT.
29004 (thumb2_legitimate_address_p): Move code block ...
29005 (can_avoid_literal_pool_for_label_p): ... into this new function.
29006 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
29008 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
29009 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
29010 "M-profile targets with the MOVT instruction".
29012 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29014 * config/arm/arm-builtins.c (arm_init_builtins): Rename
29015 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
29016 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
29018 2017-05-04 Martin Liska <mliska@suse.cz>
29020 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
29021 variable cond_code.
29023 2017-05-04 Richard Biener <rguenther@suse.de>
29025 * tree.c (array_at_struct_end_p): Handle arrays at struct
29026 end with flexarrays more conservatively. Refactor and treat
29027 arrays of arrays or aggregates more strict. Fix
29028 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
29029 * tree.c (array_at_struct_end_p): Adjust prototype.
29030 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
29031 * gimple-fold.c (get_range_strlen): Likewise.
29032 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
29034 2017-05-04 Richard Biener <rguenther@suse.de>
29036 PR tree-optimization/31130
29037 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
29039 (supports_overflow_infinity): Likewise.
29040 (is_negative_overflow_infinity): Likewise.
29041 (is_positive_overflow_infinity): Likewise.
29042 (is_overflow_infinity): Likewise.
29043 (stmt_overflow_infinity): Likewise.
29044 (overflow_infinity_range_p): Likewise.
29045 (usable_range_p): Remove as always returning true.
29046 (make_overflow_infinity): Remove.
29047 (negative_overflow_infinity): Likewise.
29048 (positive_overflow_infinity): Likewise.
29049 (avoid_overflow_infinity): Likewise.
29050 (set_value_range): Adjust accordingly.
29051 (set_value_range_to_nonnegative): Likewise, remove now unused
29052 overflow_infinity arg.
29053 (vrp_operand_equal_p): Adjust.
29054 (update_value_range): Likewise.
29055 (range_int_cst_singleton_p): Likewise.
29056 (operand_less_p): Likewise.
29057 (compare_values_warnv): Likewise.
29058 (extract_range_for_var_from_comparison_expr): Likewise.
29059 (vrp_int_const_binop): Likewise.
29060 (zero_nonzero_bits_from_vr): Likewise.
29061 (extract_range_from_multiplicative_op_1): Likewise.
29062 (extract_range_from_binary_expr_1): Likewise.
29063 (extract_range_from_unary_expr): Likewise.
29064 (extract_range_from_comparison): Likewise.
29065 (extract_range_basic): Likewise.
29066 (adjust_range_with_scev): Likewise.
29067 (compare_ranges): Likewise.
29068 (compare_range_with_value): Likewise.
29069 (dump_value_range): Likewise.
29070 (test_for_singularity): Likewise, remove strict_overflow_p parameter
29072 (simplify_cond_using_ranges): Adjust.
29074 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
29076 * brig-builtins.def: Added a builtin for class_f64.
29077 * builtin-types.def: Added a builtin type needed by class_f64.
29079 2017-05-03 Jason Merrill <jason@redhat.com>
29081 * timevar.def: Add TV_CONSTEXPR.
29083 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29085 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
29087 2017-05-03 Martin Jambor <mjambor@suse.cz>
29089 * ipa-prop.c (ipa_update_after_lto_read): Removed.
29090 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
29091 * ipa-cp.c (ipcp_propagate_stage): Do not call
29092 ipa_update_after_lto_read.
29093 * ipa-inline.c (ipa_inline): Likewise.
29095 2017-05-03 Martin Jambor <mjambor@suse.cz>
29097 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
29098 tag. Added a default constructor and a destructor.
29099 (ipa_edge_args_sum_t): New class;
29100 (ipa_edge_args_sum): Declare.
29101 (ipa_edge_args_vector): Remove declaration.
29102 (IPA_EDGE_REF): Use ipa_edge_args_sum.
29103 (ipa_free_edge_args_substructures): Remove declaration.
29104 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
29105 (ipa_edge_args_info_available_for_edge_p): Likewise.
29106 * ipa-prop.c (ipa_edge_args_vector): Removed.
29107 (edge_removal_hook_holder): Likewise.
29108 (edge_duplication_hook_holder): Likewise.
29109 (ipa_edge_args_sum): New variable.
29110 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
29111 ipa_edge_args_vector.
29112 (ipa_free_edge_args_substructures): Likewise.
29113 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
29114 ipa_edge_args_vector.
29115 (ipa_edge_removal_hook): Turned into method
29116 ipa_edge_args_sum_t::remove.
29117 (ipa_edge_duplication_hook): Turned into method
29118 ipa_edge_args_sum_t::duplicate.
29119 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
29120 registering edge hooks.
29121 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
29122 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
29123 ipa_edge_args_sum instead of ipa_edge_args_vector.
29124 * ipa-profile.c (ipa_profile): Likewise.
29126 2017-05-03 Martin Jambor <mjambor@suse.cz>
29128 * symbol-summary.h (function_summary): New method exists.
29129 (function_summary::symtab_removal): Deallocate through release.
29130 (call_summary): New class.
29131 (gt_ggc_mx): New overload.
29132 (gt_pch_nx): Likewise.
29133 (gt_pch_nx): Likewise.
29135 2017-05-03 Jeff Law <law@redhat.com>
29137 PR tree-optimization/78496
29138 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
29139 from simplify_cond_using_ranges. Split off code to walk
29140 backwards through casts into ...
29141 (simplify_cond_using_ranges_2): New function.
29142 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
29143 (execute_vrp): After identifying jump threads, call
29144 simplify_cond_using_ranges_2.
29146 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
29149 * ipa-inline.h (inline_summary): Add ctor.
29150 (create_ggc): Do not use ggc_cleared_alloc.
29152 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
29153 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29155 * gcc.c (handle_braces): Support escaping in switch matching
29157 * doc/invoke.texi (Spec Files): Document it.
29158 Remove superfluous @code markup in items.
29160 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29162 * diagnostic-show-locus.c (struct column_range): New struct.
29163 (get_affected_columns): New function.
29164 (get_printed_columns): New function.
29165 (struct correction): New struct.
29166 (correction::ensure_capacity): New function.
29167 (correction::ensure_terminated): New function.
29168 (struct line_corrections): New struct.
29169 (line_corrections::~line_corrections): New dtor.
29170 (line_corrections::add_hint): New function.
29171 (layout::print_trailing_fixits): Reimplement in terms of the new
29173 (selftest::test_overlapped_fixit_printing): New function.
29174 (selftest::diagnostic_show_locus_c_tests): Call it.
29176 2017-05-03 Nathan Sidwell <nathan@acm.org>
29178 Canonicalize canonical type hashing
29179 * tree.h (type_hash_canon_hash): Declare.
29180 * tree.c (type_hash_list, attribute_hash_list): Move into
29181 type_hash_canon_hash.
29182 (build_type_attribute_qual_variant): Break out hash code calc into
29183 type_hash_canon_hash.
29184 (type_hash_canon_hash): New. Generic type hash computation.
29185 (build_range_type_1, build_array_type_1, build_function_type,
29186 build_method_type_directly, build_offset_type, build_complex_type,
29187 make_vector_type): Call it.
29189 2017-05-03 Richard Biener <rguenther@suse.de>
29191 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
29192 When all DRs have unknown misaligned do not always peel
29193 when there is a store but apply the same costing model as if
29194 there were only loads.
29196 2017-05-03 Richard Biener <rguenther@suse.de>
29199 PR tree-optimization/80492
29200 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
29201 compare_base_decls returning dont-know properly.
29203 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
29205 * config/arm/iterators.md (CCSI): New mode iterator.
29206 (arch): New mode attribute.
29207 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
29208 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
29209 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
29210 code iterator for success result mode.
29211 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
29212 the corresponding new insn generators.
29214 2017-05-03 Bin Cheng <bin.cheng@arm.com>
29217 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29218 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29220 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
29222 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
29223 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
29224 (DDR_A): Wrap DDR argument in brackets.
29225 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
29226 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
29227 (DDR_REVERSED_P): Likewise.
29229 2017-05-03 Jakub Jelinek <jakub@redhat.com>
29231 PR tree-optimization/79472
29232 * tree-switch-conversion.c (struct switch_conv_info): Add
29233 contiguous_range and default_case_nonstandard fields.
29234 (collect_switch_conv_info): Compute contiguous_range and
29235 default_case_nonstandard fields, don't clear final_bb if
29236 contiguous_range and only the default case doesn't have the required
29238 (check_all_empty_except_final): Set default_case_nonstandard instead
29239 of failing if contiguous_range and the default case doesn't have empty
29241 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
29242 and only the default case doesn't have the required constants. Skip
29244 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
29245 if default_case_nonstandard.
29246 (build_constructors): Build constant 1 just once. Assert that default
29247 values aren't inserted in between cases if contiguous_range. Skip
29249 (build_arrays): Skip virtual phis.
29250 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
29251 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
29252 Handle virtual phis.
29253 (gen_inbound_check): Handle default_case_nonstandard case.
29254 (process_switch): Adjust check_final_bb caller. Call
29255 gather_default_values with the first non-default case instead of
29256 default case if default_case_nonstandard.
29258 2017-05-02 Nathan Sidwell <nathan@acm.org>
29260 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
29261 check. Fix formatting.
29263 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
29265 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
29266 errors when comparing specialized and unspecialized times.
29268 2017-05-02 David Malcolm <dmalcolm@redhat.com>
29270 * diagnostic-show-locus.c
29271 (layout::should_print_annotation_line_p): Make private.
29272 (layout::print_annotation_line): Make private.
29273 (layout::annotation_line_showed_range_p): Make private.
29274 (layout::show_ruler): Make private.
29275 (layout::print_source_line): Make private. Pass in line and
29276 line_width, rather than calling location_get_source_line. Drop
29278 (layout::print_leading_fixits): New method.
29279 (layout::print_any_fixits): Rename to...
29280 (layout::print_trailing_fixits): ...this, and make private.
29281 Don't print newline fixits.
29282 (diagnostic_show_locus): Move logic for printing one row into...
29283 (layout::print_line): ...this new function. Move the
29284 location_get_source_line call and error-handling from
29285 print_source_line to here. Call print_leading_fixits, and rename
29286 print_any_fixits to print_trailing_fixits.
29287 (selftest::test_fixit_insert_containing_newline): Update now that
29288 newlines are partially supported.
29289 (selftest::test_fixit_insert_containing_newline_2): New test.
29290 (selftest::test_fixit_replace_containing_newline): Update comments.
29291 (selftest::diagnostic_show_locus_c_tests): Call the new test.
29292 * edit-context.c (class added_line): New class.
29293 (class edited_line): Describe newline handling in comment.
29294 (edited_line::actually_edited_p): New method.
29295 (edited_line::print_content): Delete redundant decl.
29296 (edited_line::m_predecessors): New field.
29297 (edited_file::print_content): Call edited_line::print_content.
29298 (edited_file::print_diff): Update to support newlines.
29299 (edited_file::print_diff_hunk): Likewise.
29300 (edited_file::print_run_of_changed_lines): New function.
29301 (edited_file::print_diff_line): Convert to...
29302 (print_diff_line): ...this.
29303 (edited_file::get_effective_line_count): New function.
29304 (edited_line::edited_line): Initialize new field m_predecessors.
29305 (edited_line::~edited_line): Clean up m_predecessors.
29306 (edited_line::apply_fixit): Handle newlines.
29307 (edited_line::get_effective_line_count): New function.
29308 (edited_line::print_content): New function.
29309 (edited_line::print_diff_lines): New function.
29310 (selftest::test_applying_fixits_insert_containing_newline): New
29312 (selftest::test_applying_fixits_replace_containing_newline): New
29314 (selftest::insert_line): New function.
29315 (selftest::test_applying_fixits_multiple_lines): Add example of
29317 (selftest::edit_context_c_tests): Call the new tests.
29319 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29321 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
29322 parameter cand. Update dump information.
29323 (get_computation_cost): Update uses.
29325 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29327 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
29328 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
29329 (get_computation_at, rewrite_use_address): Update use of
29330 get_computation_aff.
29332 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29334 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
29335 (get_computation): Delete.
29336 (get_computation_cost): Implement like get_computation_cost_at.
29337 Use get_computation_at.
29338 (get_computation_cost_at): Delete.
29339 (rewrite_use_nonlinear_expr): Use get_computation_at.
29340 (rewrite_use_compare, remove_unused_ivs): Ditto.
29342 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29344 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
29346 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29348 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
29349 (ivopts_global_cost_for_size): Rename parameter and update uses.
29350 (iv_ca_recount_cost): Update uses.
29351 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
29352 candidates seperately in n_invs and n_cands.
29353 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
29355 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29357 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
29358 (find_inv_vars_cb): New.
29359 (find_depends): Renamed to ...
29360 (find_inv_vars): ... this.
29361 (add_candidate_1, force_var_cost): Call find_inv_vars.
29362 (split_address_cost, determine_group_iv_cost_cond): Ditto.
29364 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29366 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
29367 inv_vars. Add inv_exprs.
29368 (struct iv_cand): Rename depends_on to inv_vars.
29369 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
29370 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
29371 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
29372 (dump_cand): Dump inv_vars.
29373 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
29374 (record_invariant, find_depends, add_candidate_1): Ditto.
29375 (set_group_iv_cost, force_var_cost): Ditto.
29376 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
29377 (get_computation_cost_at, get_computation_cost): Ditto.
29378 (determine_group_iv_cost_generic): Ditto.
29379 (determine_group_iv_cost_address): Ditto.
29380 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
29381 (determine_group_iv_costs): Ditto.
29382 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
29383 (iv_ca_set_remove_invariants): Renamed to ...
29384 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
29385 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
29386 (iv_ca_set_add_invariants): Renamed to ...
29387 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
29388 (iv_ca_set_cp): Use iv_ca_set_add_invs.
29389 (iv_ca_has_deps): Support inv_vars and inv_exprs.
29390 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
29391 (create_new_ivs): Remove useless dump.
29393 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29395 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
29397 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
29398 (iv_ca_set_no_cp, create_new_iv): Ditto.
29400 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29402 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29404 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29406 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
29407 function tree_check2.
29409 2017-05-02 Martin Liska <mliska@suse.cz>
29411 * doc/gcov.texi: Add missing preposition.
29412 * gcov.c (function_info::function_info): Properly fill up
29413 all member variables.
29415 2017-05-02 Tamar Christina <tamar.christina@arm.com>
29417 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
29419 2017-05-02 Tamar Christina <tamar.christina@arm.com>
29421 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
29423 2017-05-02 Martin Liska <mliska@suse.cz>
29426 * lto-streamer-in.c (lto_read_tree_1): Remove
29427 LTO_STREAMER_DEBUG.
29428 * lto-streamer.c (struct tree_hash_entry): Likewise.
29429 (struct tree_entry_hasher): Likewise.
29430 (tree_entry_hasher::hash): Likewise.
29431 (tree_entry_hasher::equal): Likewise.
29432 (lto_streamer_init): Likewise.
29433 (lto_orig_address_map): Likewise.
29434 (lto_orig_address_get): Likewise.
29435 (lto_orig_address_remove): Likewise.
29436 * lto-streamer.h: Likewise.
29437 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
29438 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
29440 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
29442 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
29443 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
29444 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
29445 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
29446 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
29447 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
29448 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
29449 (mm_maskz_sub_ss): New intrinsics.
29450 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29451 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
29452 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
29453 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
29454 (__builtin_ia32_subss_mask_round): New builtins.
29455 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29456 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29457 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
29459 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
29460 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
29462 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
29465 2017-05-02 Martin Jambor <mjambor@suse.cz>
29467 PR tree-optimization/78687
29468 * tree-sra.c (access): New field parent.
29469 (process_subtree_disqualification): New function.
29470 (disqualify_candidate): Call it.
29471 (build_accesses_from_assign): Reset write flag if creating an
29473 (build_access_subtree): Fill in parent field and also prpagate
29474 down grp_write flag.
29475 (create_artificial_child_access): New parameter set_grp_write, set
29476 grp_write to its value.
29477 (propagate_subaccesses_across_link): Also propagate grp_write flag
29479 (propagate_all_subaccesses): Push the closest parent back to work
29480 queue if add_access_to_work_queue returned true.
29482 2017-05-02 Richard Biener <rguenther@suse.de>
29484 * common.opt (fstrict-overflow): Alias negative to fwrapv.
29485 * doc/invoke.texi (fstrict-overflow): Remove all traces of
29486 -fstrict-overflow documentation.
29487 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
29488 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
29489 flag_strict_overflow.
29490 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
29491 * lto-opts.c (lto_write_options): Do not stream it.
29492 * lto-wrapper.c (merge_and_complain): Do not handle it.
29493 * opts.c (default_options_table): Do not set -fstrict-overflow.
29494 (finish_options): Likewise do not clear it when sanitizing.
29495 * simplify-rtx.c (simplify_const_relational_operation): Do not
29496 test flag_strict_overflow.
29498 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29500 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
29501 using enabled attribute.
29502 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
29503 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
29504 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
29505 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
29506 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
29507 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
29508 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
29509 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
29510 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
29511 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
29513 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29515 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
29517 2017-05-02 Richard Biener <rguenther@suse.de>
29519 PR tree-optimization/80591
29521 2017-04-10 Richard Biener <rguenther@suse.de>
29523 * tree-ssa-structalias.c (find_func_aliases): Properly handle
29526 2017-05-02 Richard Biener <rguenther@suse.de>
29528 PR tree-optimization/80549
29529 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
29530 (cleanup_tree_cfg_noloop): Create forwarders to known loop
29531 headers if they do not have a preheader.
29533 2017-05-02 Martin Liska <mliska@suse.cz>
29536 * common.opt: Fix typo.
29537 * doc/invoke.texi: Likewise.
29539 2017-05-01 Jan Beulich <jbeulich@suse.com>
29541 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
29542 swapping, add (x,x,m,x,n) alternative.
29544 2017-05-01 Nathan Sidwell <nathan@acm.org>
29546 * calls.c (combine_pending_stack_adjustment_and_call): Remove
29547 unnecessary unadjusted_alignment check.
29549 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29552 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
29554 * gimplify.c (gimplify_cilk_detach): New function.
29555 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
29556 * tree-core.h: Document EXPR_CILK_SPAWN.
29557 * tree.h (EXPR_CILK_SPAWN): Define.
29559 2017-05-01 David Malcolm <dmalcolm@redhat.com>
29561 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
29562 to use new fixit_hint representation, using the "replace" logic.
29563 (get_line_span_for_fixit_hint): Likewise.
29564 (layout::print_any_fixits): Likewise.
29565 (selftest::test_one_liner_many_fixits): Rename to...
29566 (selftest::test_one_liner_many_fixits_1): ...this, and update
29567 comment and expected output to reflect that the multiple fix-it
29568 hints are now consolidated into one insertion.
29569 (selftest::test_one_liner_many_fixits_2): New test.
29570 (selftest::test_diagnostic_show_locus_one_liner): Update for
29572 (selftest::test_fixit_consolidation): Update for fix-it API
29574 * diagnostic.c (print_parseable_fixits): Likewise.
29575 * edit-context.c (edited_line::m_line_events): Convert from
29576 auto_vec <line_event *> to auto_vec <line_event>.
29577 (class line_event): Convert from abstract base class to a concrete
29578 class, taking over the role of replace_event.
29579 (class insert_event): Delete.
29580 (class replace_event): Rename to class line_event. Convert to
29582 (edit_context::add_fixits): Reimplement.
29583 (edit_context::apply_insert): Delete.
29584 (edit_context::apply_replace): Rename to...
29585 (edit_context::apply_fixit): ...this. Convert to half-open range.
29586 (edited_file::apply_insert): Delete.
29587 (edited_file::apply_replace): Rename to...
29588 (edited_file::apply_fixit): ...this.
29589 (edited_line::~edited_line): Drop deletion of events.
29590 (edited_line::apply_insert): Delete.
29591 (edited_line::apply_replace): Rename to...
29592 (edited_line::apply_fixit): ...this. Convert to half-open range.
29593 Update for change to type of m_line_events.
29594 * edit-context.h (edit_context::apply_insert): Delete.
29595 (edit_context::apply_replace): Rename to...
29596 (edit_context::apply_fixit): ...this.
29598 2017-05-01 Martin Sebor <msebor@redhat.com>
29600 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
29603 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
29606 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
29607 __get_cpuid_max returns 0.
29608 (__get_cpuid_count): Ditto.
29610 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
29612 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
29613 replacement expression is another instance of one of its arguments.
29615 2017-05-01 Jakub Jelinek <jakub@redhat.com>
29618 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
29619 check for stack push/pop autoinc.
29620 * config/i386/i386.c (ix86_agi_dependent): Return false
29621 if the only reason why modified_in_p returned true is that
29622 addr is SP based and set_insn is a push or pop.
29624 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
29626 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
29629 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
29632 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
29633 (account_size_time): Use two predicates - exec_pred and
29635 (evaluate_conditions_for_known_args): Compute both clause and
29637 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
29638 (inline_summary_t::duplicate): Update.
29639 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
29641 (compute_inline_parameters): Likewise.
29642 (estimate_edge_size_and_time): Update caluclation of time.
29643 (estimate_node_size_and_time): Compute both time and nonspecialized
29645 (estimate_ipcp_clone_size_and_time): Update.
29646 (inline_merge_summary): Update.
29647 (do_estimate_edge_time): Update.
29648 (do_estimate_edge_size): Update.
29649 (do_estimate_edge_hints): Update.
29650 (inline_read_section, inline_write_summary): Stream both new predicates.
29651 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
29653 (compute_inlined_call_time): Cleanup.
29654 (big_speedup_p): Update.
29655 (edge_badness): Update.
29656 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
29657 (size_time_entry): Replace predicate by exec_predicate and
29658 nonconst_predicate.
29659 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
29660 (estimate_edge_time): Return also nonspec_time.
29661 (reset_edge_growth_cache): Update.
29663 2017-04-29 Jakub Jelinek <jakub@redhat.com>
29665 PR rtl-optimization/80491
29666 * ifcvt.c (noce_process_if_block): When looking for x setter
29667 with missing else_bb, don't check only the insn right before
29668 cond_earliest, but look for the last insn that x is modified in
29669 within the same bb.
29671 PR rtl-optimization/80491
29672 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
29674 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
29676 PR tree-optimization/80487
29677 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
29679 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29681 PR tree-optimization/79697
29682 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
29683 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
29684 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
29686 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
29687 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
29689 2017-04-28 Martin Sebor <msebor@redhat.com>
29691 PR tree-optimization/80523
29692 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
29693 (init_target_to_host_charmap, target_to_host, target_strtol10): New
29695 (maybe_warn, format_directive, parse_directive): Use new functions.
29696 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
29698 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
29700 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
29702 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
29704 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
29705 target_header_dir): Set correctly.
29706 * configure: Regenerated.
29707 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29708 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29709 instead of SYSTEM_HEADER_DIR.
29711 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
29713 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
29714 (estimate_local_effects): Likewise.
29715 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
29716 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
29717 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
29718 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
29719 do_estimate_edge_time, estimate_edge_time): Likewise.
29720 * ipa-inline-analysis.c (estimate_node_size_and_time,
29721 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
29722 (estimate_time_after_inlining): Remove.
29724 2017-04-28 Martin Liska <mliska@suse.cz>
29726 * doc/gcov.texi: Enhance documentation of gcov.
29728 2017-04-28 Martin Liska <mliska@suse.cz>
29730 * doc/gcov.texi: Sort options in alphabetic order.
29731 * doc/gcov-dump.texi: Likewise.
29732 * doc/gcov-tool.texi: Likewise.
29733 * gcov.c (print_usage): Likewise.
29734 * gcov-dump.c (print_usage): Likewise.
29735 * gcov-tool.c (print_merge_usage_message): Likewise.
29736 (print_rewrite_usage_message): Likewise.
29737 (print_overlap_usage_message): Likewise.
29739 2017-04-28 Martin Liska <mliska@suse.cz>
29741 PR gcov-profile/53915
29742 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
29744 2017-04-28 Martin Liska <mliska@suse.cz>
29746 PR gcov-profile/79891
29747 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
29748 is marked by compiler as living on a line.
29749 (get_cycles_count): Remove usage of the union.
29750 (output_intermediate_file): Likewise.
29751 (find_source): Fix GNU coding style.
29752 (accumulate_line_counts): Remove old non-all block mode.
29753 (output_lines): Remove usage of the union.
29754 * profile.c (output_location): Include all BBs, even if
29755 belonging to a same line (and file) as a previous BB.
29757 2017-04-28 Martin Liska <mliska@suse.cz>
29759 * gcov.c (process_args): Handle new argument 'w'.
29760 (read_graph_file): Assign ID to BBs.
29761 (output_branch_count): Display BB # if verbose flag is set.
29762 (output_lines): Likewise for arcs.
29763 (print_usage): Add '--verbose' option help.
29764 * doc/gcov.texi: Document --verbose (-w) option.
29766 2017-04-28 Martin Liska <mliska@suse.cz>
29768 * gcov.c (struct block_location_info): New struct.
29769 (process_file): Fill up the new structure.
29770 (read_graph_file): Replace usage of encoding by the newly added
29772 (add_line_counts): Likewise.
29773 (accumulate_line_counts): Remove usage of the union.
29774 (function_info::function_info): New function.
29775 (function_info::~function_info): Likewise.
29776 (process_file): Call delete instead of release_function.
29777 (release_function): Release the function.
29778 (release_structures): Call delete instead of release_function.
29779 (solve_flow_graph): Replace usage of num_blocks.
29780 (find_exception_blocks): Likewise.
29781 (output_lines): Fix GNU coding style.
29783 2017-04-28 Martin Liska <mliska@suse.cz>
29786 * coverage.c (coverage_remove_note_file): New function.
29787 * coverage.h: Declare the function.
29788 * toplev.c (finalize): Clean if an error has been seen.
29790 2017-04-28 Martin Liska <mliska@suse.cz>
29792 PR gcov-profile/80031
29793 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
29794 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
29795 * gcov.c (read_graph_file): Read just number of blocks.
29796 * profile.c (branch_prob): Do not stream 0 flags per a basic
29799 2017-04-28 Martin Liska <mliska@suse.cz>
29801 * gcov-dump.c (tag_*): Add new argument to declarations.
29802 (dump_gcov_file): Likewise.
29803 (tag_blocks): Add and use new argument depth.
29804 (tag_arcs): Likewise.
29805 (tag_lines): Likewise.
29806 (tag_counters): Likewise.
29807 (tag_summary): Likewise.
29808 (dump_working_sets): Use depth to do a proper indentation.
29810 2017-04-28 Jakub Jelinek <jakub@redhat.com>
29813 * cgraph.h (symtab_node::debug_symtab): No longer inline.
29814 * symtab.c (symtab_node::debug_symtab): Move definition here.
29816 2017-04-28 Richard Biener <rguenther@suse.de>
29818 * lto-streamer.h (LTO_major_version): Bump to 7.
29820 2017-04-28 Richard Biener <rguenther@suse.de>
29822 * tree-vrp.c (assert_info): New struct.
29823 (add_assert_info): New helper.
29824 (register_edge_assert_for_2): Refactor to add asserts to a vector
29826 (register_edge_assert_for_1): Likewise.
29827 (register_edge_assert_for): Likewise.
29828 (finish_register_edge_assert_for): New helper actually registering
29829 asserts where live on edge.
29830 (find_conditional_asserts): Adjust.
29831 (find_switch_asserts): Likewise.
29832 (evrp_dom_walker::try_find_new_range): Generalize.
29833 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
29835 2017-04-27 Marek Polacek <polacek@redhat.com>
29838 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
29839 arg10 and arg11 to itype.
29841 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
29843 * doc/extend.texi (Object Size Checking): Improve grammar.
29845 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
29848 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
29849 that the logic for permitting reciprocal estimates matches that
29852 2017-04-27 Jakub Jelinek <jakub@redhat.com>
29855 * tree.c (type_cache_hasher::equal): Only compare
29856 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
29857 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
29858 non-aggregate element types.
29859 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
29860 about the flag on ARRAY_TYPEs in the comment, formatting fix.
29862 2017-04-27 Richard Biener <rguenther@suse.de>
29864 PR middle-end/80533
29865 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
29866 stripping ARRAY_REFs from MEM_EXPR make sure we're not
29867 keeping a reference to a trailing array.
29869 2017-04-27 Richard Biener <rguenther@suse.de>
29871 PR middle-end/80539
29872 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
29873 being in loop-closed SSA form conservatively.
29874 (chrec_fold_multiply_poly_poly): Likewise.
29876 2017-04-27 Tamar Christina <tamar.christina@arm.com>
29878 PR middle-end/79665
29879 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
29880 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
29882 2017-04-27 Jakub Jelinek <jakub@redhat.com>
29885 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
29886 (aarch64_function_arg_alignment): Return unsigned int again, but still
29887 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
29888 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
29889 Don't emit -Wpsabi note.
29890 (aarch64_function_arg_boundary): Likewise.
29891 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
29894 2017-04-26 Nathan Sidwell <nathan@acm.org>
29896 * tree.h (crc32_unsigned_n): Declare.
29897 (crc32_unsigned, crc32_unsigned): Make inline.
29898 * tree.c (crc32_unsigned_bits): Replace with ...
29899 (crc32_unsigned_n): ... this.
29900 (crc32_unsigned, crc32_byte): Remove.
29901 (crc32_string): Remove unnecessary braces.
29903 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
29905 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
29906 * ipa-inline-analysis.c (MAX_TIME): Remove.
29907 (account_size_time): Use sreal for time.
29908 (dump_inline_summary): Update.
29909 (estimate_function_body_sizes): Update.
29910 (estimate_edge_size_and_time): Update.
29911 (estimate_calls_size_and_time): Update.
29912 (estimate_node_size_and_time): Update.
29913 (inline_merge_summary): Update.
29914 (inline_update_overall_summary): Update.
29915 (estimate_time_after_inlining): Update.
29916 (inline_read_section): Update.
29917 (inline_write_summary): Update.
29918 * ipa-inline.c (compute_uninlined_call_time): Update.
29919 (compute_inlined_call_time): Update.
29920 (recursive_inlining): Update.
29921 (inline_small_functions): Update.
29922 (dump_overall_stats): Update.
29923 * ipa-inline.h: Include sreal.h.
29924 (size_time_entry): Turn time to sreal.
29925 (inline_summary): Turn self_time nad time to sreal.
29927 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
29929 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
29931 (sreal::stream_out, sreal::stream_in): New.
29932 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
29934 2017-04-25 Jakub Jelinek <jakub@redhat.com>
29936 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
29939 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
29942 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
29944 (dimode_scalar_chain::compute_convert_gain): Ditto.
29945 (dimode_scalar_chain::make_vector_copies): Ditto.
29946 (dimode_scalar_chain::convert_reg): Ditto.
29947 (dimode_scalar_chain::convert_insn): Ditto.
29948 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
29949 (VI248_AVX512BW_1): New mode iterator.
29950 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
29951 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
29954 2017-04-25 Martin Sebor <msebor@redhat.com>
29956 PR tree-optimization/80497
29957 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
29958 constants are representable in HOST_WIDE_INT.
29959 (parse_directive): Ditto.
29961 2017-04-25 Martin Sebor <msebor@redhat.com>
29964 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
29965 (new_zero_array): Adjust signature.
29966 (dom_info::dom_init): Used unsigned rather that size_t.
29967 (dom_info::dom_info): Same.
29969 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
29970 Jakub Jelinek <jakub@redhat.com>
29973 * config/arm/arm.c: Include gimple.h.
29974 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
29975 returns negative, increment ncrn only if it returned positive.
29976 (arm_needs_doubleword_align): Return int instead of bool,
29977 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
29978 members, but if there is any such non-FIELD_DECL
29979 > PARM_BOUNDARY aligned decl, return -1 instead of false.
29980 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
29981 returns negative, increment nregs only if it returned positive.
29982 (arm_setup_incoming_varargs): Likewise.
29983 (arm_function_arg_boundary): Emit -Wpsabi note if
29984 arm_needs_doubleword_align returns negative, return
29985 DOUBLEWORD_ALIGNMENT only if it returned positive.
29987 2017-04-25 Marek Polacek <polacek@redhat.com>
29990 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
29991 first argument to type.
29993 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
29996 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
29997 type checks to test for compatibility instead of equality.
29999 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30000 Jakub Jelinek <jakub@redhat.com>
30003 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
30005 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
30006 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
30007 the alignment computation, but return their maximum in warn_alignment.
30008 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30009 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
30011 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
30012 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30015 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30017 * config/arc/simdext.md (dmpyh): Fix typo.
30019 2017-04-25 Richard Biener <rguenther@suse.de>
30021 PR tree-optimization/80492
30022 * alias.c (compare_base_decls): Handle registers with asm
30023 specification conservatively.
30024 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
30025 compare_base_decls returning dont-know properly.
30027 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30029 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
30030 (legitimate_offset_address_p): New function.
30031 (arc_legitimate_address_p): Use above function.
30033 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30035 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
30037 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30039 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
30040 ACCH registers whenever they are available.
30042 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30044 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
30045 double regs fix when not used.
30047 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30049 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
30051 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
30052 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
30054 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30056 * config/arc/arc.c (arc_output_addsi): Check for h-register class
30057 when emitting short ADD instructions.
30059 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30061 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
30063 (cmpsi_cc_c_insn): Likewise.
30064 (cbranchsi4_scratch): Compute proper instruction length using
30065 compact_hreg_operand.
30066 * config/arc/predicates.md (compact_hreg_operand): New predicate.
30068 2017-04-25 Richard Biener <rguenther@suse.de>
30070 PR middle-end/80509
30071 * passes.c (pass_manager::pass_manager): Initialize
30072 m_name_to_pass_map.
30074 2017-04-25 Richard Biener <rguenther@suse.de>
30076 PR tree-optimization/79201
30077 * tree-ssa-sink.c (statement_sink_location): Handle calls.
30079 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30082 * config/s390/vector.md: Split MEM->GPR vector moves for
30083 non-s_operand addresses.
30085 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30088 * config/s390/predicates.md (reload_const_wide_int_operand): New
30090 * config/s390/s390.md ("movti"): Remove d/P alternative.
30091 ("movti_bigconst"): New pattern definition.
30093 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30096 * s390-protos.h (s390_expand_cs_hqi): Removed.
30097 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
30098 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
30099 modes as well as CCZ1mode and CCZmode.
30100 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
30101 signature of s390_emit_compare_and_swap.
30102 (s390_expand_cs_hqi): Likewise, make static.
30103 (s390_expand_cs_tdsi): Generate an explicit compare before trying
30104 compare-and-swap, in some cases.
30105 (s390_expand_cs): Wrapper function.
30106 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
30108 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
30109 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
30110 patterns for small and large integers. Forbid symref memory operands.
30111 Move expander to s390.c. Require cc register.
30112 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
30113 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
30114 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
30115 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
30116 symref memory operands. Remove CC mode and call s390_match_ccmode
30118 ("atomic_exchange<mode>"): Allow and implement all integer modes.
30120 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30122 * config/s390/s390.md (define_peephole2): New peephole to help
30123 combining the load-and-test pattern with volatile memory.
30125 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30127 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
30128 with CCZmode for TARGET_Z196.
30130 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30132 PR rtl-optimization/80501
30133 * combine.c (make_compound_operation_int): Set subreg_code to SET
30134 even for AND with mask of the sign bit of mode.
30136 PR rtl-optimization/80500
30137 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
30138 sum's initial value.
30140 2017-04-25 Julian Brown <julian@codesourcery.com>
30141 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30143 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
30145 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
30147 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
30149 2017-04-25 Julian Brown <julian@codesourcery.com>
30150 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30152 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
30153 (thunderx2t99_sha): New Reservation.
30155 2017-04-25 Julian Brown <julian@codesourcery.com>
30156 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30158 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
30159 type for 1-element load.
30161 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
30163 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
30165 2017-04-24 Martin Jambor <mjambor@suse.cz>
30167 PR tree-optimization/80293
30168 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
30169 char arrays not totally scalarizable if it is false.
30170 (analyze_all_variable_accesses): Pass correct value in the new
30171 parameter. Add a statistics counter.
30173 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
30175 PR middle-end/79931
30176 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
30178 2017-04-24 Richard Biener <rguenther@suse.de>
30180 PR tree-optimization/80494
30181 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
30182 out for complex types.
30184 2017-04-24 Richard Biener <rguenther@suse.de>
30186 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
30187 * tree-ssa-sccvn.c (print_scc): Print SCC size.
30188 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
30189 (DFS): Adjust and never fail.
30190 (sccvn_dom_walker::fail): Remove.
30191 (sccvn_dom_walker::before_dom_children): Adjust.
30192 (run_scc_vn): Likewise and never fail.
30193 * tree-ssa-pre.c (pass_pre::execute): Adjust.
30194 (pass_fre::execute): Likewise.
30196 2017-04-24 Richard Biener <rguenther@suse.de>
30198 PR tree-optimization/79725
30199 * tree-ssa-sink.c (statement_sink_location): Return whether
30200 failure reason was zero uses. Move that check later.
30201 (sink_code_in_bb): Deal with zero uses by removing the stmt
30204 2017-04-24 Richard Biener <rguenther@suse.de>
30207 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
30208 pointer-based references.
30210 2017-04-24 Richard Biener <rguenther@suse.de>
30213 * pass_manager.h (pass_manager::operator new): Remove.
30214 (pass_manager::operator delete): Likewise.
30215 * passes.c (pass_manager::operator new): Remove.
30216 (pass_manager::operator delete): Likewise.
30217 (pass_manager::pass_manager): Zero individual pass members.
30219 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
30222 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
30223 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
30224 Check "XEXP (src, 1)" operand here.
30225 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
30226 Check "XEXP (src, 1)" operand here.
30227 (dimode_scalar_chain::make_vector_copies): Detect count register
30228 of a shift instruction. Zero extend count register from QImode
30229 to DImode to satisfy vector shift pattern count operand predicate.
30230 Substitute vector shift count operand with a DImode copy.
30231 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
30234 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
30236 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
30237 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
30238 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
30239 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
30240 (UNSPEC_NOREX_MEM): Remove definition.
30242 2017-04-21 Richard Biener <rguenther@suse.de>
30244 PR tree-optimization/79547
30245 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
30246 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
30247 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
30248 without any constraints.
30250 2017-04-21 Richard Biener <rguenther@suse.de>
30252 PR tree-optimization/78847
30253 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
30255 2017-04-21 Richard Biener <rguenther@suse.de>
30257 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
30258 (build_distinct_type_copy): Likewise.
30259 (build_variant_type_copy): Likewise.
30260 * tree.c (build_qualified_type): Pass down mem-stat info.
30261 (build_distinct_type_copy): Likewise.
30262 (build_variant_type_copy): Likewise.
30264 2017-04-21 Richard Biener <rguenther@suse.de>
30266 PR tree-optimization/80237
30267 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
30269 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
30270 for a simplified result.
30272 2016-04-21 Richard Biener <rguenther@suse.de>
30274 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
30275 sth as strict as a simple_iv but a chrec without symbols and an
30276 operand defined in the loop we are peeling (and not some subloop).
30277 (propagate_constants_for_unrolling): Propagate all constants.
30279 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
30282 * config/i386/i386.c (print_reg): Remove assert for disalowed
30283 regno values, call output_operand_lossage instead.
30285 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
30288 * config/i386/constraints.md (Yc): New register constraint.
30289 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
30290 Use Yc constraint for alternative 2 of operand 0. Remove
30291 preferred_for_speed attribute.
30293 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
30295 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
30296 lastprivate clauses in SIMT case.
30298 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
30300 * doc/invoke.texi (-Wextra-semi): Document new warning option.
30302 2017-04-20 Richard Biener <rguenther@suse.de>
30304 PR tree-optimization/57796
30305 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
30306 as N scalar stores.
30307 (vect_model_load_cost): Cost gathers as N scalar loads.
30309 2017-04-20 Richard Biener <rguenther@suse.de>
30311 * ggc-page.c (ggc_allocated_p): Rename to ...
30312 (safe_lookup_page_table_entry): ... this and return the lookup
30314 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
30316 2017-04-20 Richard Biener <rguenther@suse.de>
30318 PR tree-optimization/80453
30319 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
30320 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
30321 from the conditions.
30322 (vn_phi_eq): Pass them down.
30323 (vn_phi_lookup): Record them.
30324 (vn_phi_insert): Likewise.
30326 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
30328 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
30329 uninitialized variable warning to avoid buffer overrun.
30331 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
30334 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
30335 is suppressed for '{ 0 }' in C.
30337 2017-04-20 Jakub Jelinek <jakub@redhat.com>
30339 * BASE-VER: Set to 8.0.0.
30341 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
30343 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
30344 priority .init_array and .fini_array section with SECTION_NOTYPE
30347 2017-04-20 Jakub Jelinek <jakub@redhat.com>
30349 PR middle-end/80423
30350 * tree.h (build_array_type): Add typeless_storage default argument.
30351 * tree.c (type_cache_hasher::equal): Also compare
30352 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
30353 (build_array_type): Add typeless_storage argument, set
30354 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
30356 (build_nonshared_array_type): Adjust build_array_type_1 caller.
30357 (build_array_type): Likewise. Add typeless_storage argument.
30359 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
30360 Jakub Jelinek <jakub@redhat.com>
30362 PR tree-optimization/80426
30363 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
30364 operation on symbolic operands, also compute the overflow for the
30365 invariant part when the operation degenerates into a negation.
30367 2017-04-19 Jakub Jelinek <jakub@redhat.com>
30370 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
30371 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
30374 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
30376 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
30379 * config/avr/avr.c (tree.h): Include it.
30380 (cgraph.h): Include it.
30381 (avr_encode_section_info): Don't warn for uninitialized progmem
30382 variable if it's just an alias.
30384 2017-04-19 Richard Biener <rguenther@suse.de>
30387 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
30388 when needed by AutoPGO.
30390 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
30393 * doc/lto.texi: Remove an extra 'that'.
30395 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
30397 PR rtl-optimization/80429
30398 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
30399 are only used in debug insns.
30401 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
30402 Vladimir Makarov <vmakarov@redhat.com>
30404 * config/sparc/predicates.md (input_operand): Add comment. Return
30405 true for any memory operand when LRA is in progress.
30406 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
30408 2017-04-18 Jeff Law <law@redhat.com>
30411 * mips.md ({return,simple_return}_internal): Do not overwrite
30414 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30416 PR tree-optimization/80443
30417 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
30418 instead of adding 1, subtract -1 and similarly instead of subtracting
30421 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
30423 PR rtl-optimization/80357
30424 * haifa-sched.c (tmp_bitmap): New variable.
30425 (model_recompute): Handle duplicate use records.
30426 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
30427 (free_global_sched_pressure_data): Free it.
30429 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
30432 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
30433 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30434 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30435 instead of SYSTEM_HEADER_DIR.
30437 2017-04-18 Jeff Law <law@redhat.com>
30439 PR middle-end/80422
30440 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
30441 predecessors after walking up the insn chain.
30443 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30446 * dwarf2out.c (modified_type_die): Try harder not to emit internal
30447 sizetype type into debug info.
30449 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30452 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
30453 unneeded test for TARGET_UPPER_REGS_SF.
30454 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
30456 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30459 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
30460 instead of gsi_after_labels.
30462 2017-04-18 Jeff Law <law@redhat.com>
30464 * regcprop.c (maybe_mode_change): Avoid creating copies of the
30468 2017-04-13 Jeff Law <law@redhat.com>
30469 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30470 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30472 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
30475 * config/avr/avr.c (intl.h): Include it.
30476 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
30478 2017-04-18 Martin Liska <mliska@suse.cz>
30480 PR gcov-profile/78783
30481 * gcov-tool.c (gcov_output_files): Validate that destination
30482 file is either removed by the tool or by a user.
30484 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30485 Guy Benyei <guybe@mellanox.com>
30487 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
30488 block, and do not negate it, the stored id is already negative.
30490 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30492 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
30494 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
30497 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
30498 masks of options that should be turned off if the VSX vector
30499 options are turned off.
30500 (OTHER_P8_VECTOR_MASKS): Likewise.
30501 (OTHER_VSX_VECTOR_MASKS): Likewise.
30502 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
30503 rs6000_disable_incompatible_switches to validate no type switches
30505 (rs6000_incompatible_switch): New function to disallow turning on
30506 other vector options if -mno-vsx, -mno-power8-vector, or
30507 -mno-power9-vector are specified.
30509 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30511 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
30513 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30515 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
30516 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
30517 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
30518 (ARG_POINTER_CFA_OFFSET): Likewise.
30520 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30522 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
30523 conditions to take advantage of various optimizations.
30525 2017-04-13 Jeff Law <law@redhat.com>
30527 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30528 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30529 (zero_extendsidi2_dext): Likewise.
30531 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30534 * fold-const.c (fold_ternary_loc): Revert
30535 use op0 instead of fold_convert_loc (loc, type, arg0) part of
30538 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
30540 PR rtl-optimization/80343
30541 * lra-remat.c (update_scratch_ops): Assign original hard reg to
30542 new scratch pseudo.
30544 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
30547 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
30548 to ubsan_encode_value.
30550 2017-04-13 Jeff Law <law@redhat.com>
30552 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
30553 appearing in DEBUG_INSNs.
30555 2017-04-13 Martin Liska <mliska@suse.cz>
30557 PR gcov-profile/80413
30558 * gcov-io.c (gcov_write_string): Copy to buffer just when
30559 allocated size is greater than zero.
30561 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30564 * dwarf2out.c (decls_for_scope): Ignore declarations of
30565 current_function_decl in BLOCK_NONLOCALIZED_VARS.
30567 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
30570 * ipa-visibility.c (non_local_p): Fix typos.
30571 (localize_node): When localizing symbol in same comdat group,
30572 dissolve the group only when we know external symbols are going
30574 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
30576 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30578 PR tree-optimization/79390
30579 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
30580 order does not result in usable sequence, retry with reversed operand
30586 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
30587 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
30588 op0 instead of fold_convert_loc (loc, type, arg0).
30590 2017-04-12 Jeff Law <law@redhat.com>
30592 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
30593 has a delay slot in the generated code.
30595 * config/cris/cris.md (cris_preferred_reload_class): Return
30596 GENNONACR_REGS rather than GENERAL_REGS.
30598 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30601 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
30602 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
30603 signedness of the result type.
30605 2017-04-12 Richard Biener <rguenther@suse.de>
30606 Jeff Law <law@redhat.com>
30608 PR tree-optimization/80359
30609 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
30610 trim stores to TARGET_MEM_REFs.
30612 2017-04-12 Richard Biener <rguenther@suse.de>
30614 PR tree-optimization/79390
30615 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
30616 threading case even more.
30618 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
30621 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
30622 for quad_address_p for TImode, instead of just not indexed_address.
30624 2017-04-12 Richard Biener <rguenther@suse.de>
30625 Bernd Edlinger <bernd.edlinger@hotmail.de>
30627 PR middle-end/79671
30628 * alias.c (component_uses_parent_alias_set_from): Handle
30629 TYPE_TYPELESS_STORAGE.
30630 (get_alias_set): Likewise.
30631 * tree-core.h (tree_type_common): Add typeless_storage flag.
30632 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
30633 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
30634 for types containing members with TYPE_TYPELESS_STORAGE.
30635 (place_field): Likewise.
30636 (layout_type): Likewise for ARRAY_TYPE.
30637 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
30638 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
30639 TYPE_TYPELESS_STORAGE.
30640 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
30642 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30645 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30646 first argument to type.
30648 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30652 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
30653 CONST0_RTX (mode) rather than const0_rtx where appropriate.
30654 (rs6000_expand_binop_builtin): Likewise.
30655 (rs6000_expand_ternop_builtin): Likewise; also add missing
30656 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
30657 vshasigma built-ins.
30658 * doc/extend.texi: Document that vec_xxpermdi's third argument
30659 must be a constant.
30661 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
30663 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
30664 Use shift_const cost parameter when calculating gain of STV shifts.
30666 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
30668 PR rtl-optimization/70478
30669 * lra-constraints.c (process_alt_operands): Check memory for
30670 disfavoring memory insn operand.
30672 2017-04-11 Jakub Jelinek <jakub@redhat.com>
30674 PR middle-end/80100
30675 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
30676 left shift in unsigned HOST_WIDE_INT type.
30678 PR rtl-optimization/80385
30679 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
30680 (not (neg X)) into (plus X -1) for complex or non-integral modes.
30683 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
30684 if they have any depend clauses.
30686 2017-04-11 Martin Liska <mliska@suse.cz>
30689 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
30690 * ipa-split.c (split_function): Create a local comdat symbol
30691 if caller is in a comdat group.
30693 2017-04-11 Martin Liska <mliska@suse.cz>
30696 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
30699 2017-04-11 Martin Sebor <msebor@redhat.com>
30701 PR middle-end/80364
30702 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
30703 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
30705 (directive::set_width, directive::set_precision, format_character):
30707 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
30710 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
30713 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
30714 conflict, set target->arch_name instead of target->cpu_name.
30716 2017-04-11 Richard Biener <rguenther@suse.de>
30718 PR tree-optimization/80374
30719 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
30720 build_zero_cst, remove fold_convertible_p check again.
30722 2017-04-11 Martin Liska <mliska@suse.cz>
30725 * ubsan.c (instrument_object_size): Do not instrument register
30728 2017-04-11 Jakub Jelinek <jakub@redhat.com>
30731 * config/i386/i386-builtin-types.def
30732 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
30733 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
30734 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
30735 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
30736 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
30737 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
30738 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
30739 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
30740 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
30741 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
30742 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
30743 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
30744 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
30745 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
30746 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
30747 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
30748 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
30749 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
30750 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
30751 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
30752 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
30753 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
30754 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
30755 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
30756 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
30757 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
30758 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
30759 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
30760 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
30761 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
30762 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
30763 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
30764 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
30765 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
30766 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
30767 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
30768 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
30769 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
30770 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
30771 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
30772 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
30773 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
30774 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
30775 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
30776 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
30777 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
30779 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
30780 flag to second_arg_count, handle 4 argument function type _COUNT
30781 aliases, handle second_arg_count on second argument rather than last.
30783 2017-04-10 Jeff Law <law@redhat.com>
30785 PR tree-optimization/80374
30786 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
30787 record anything if we can not convert integer_zero_node to the
30790 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
30793 * config/rs6000/rs6000.c (rs6000_option_override_internal):
30794 Enhance special handling given to the TARGET_P9_MINMAX option in
30795 relation to certain other options.
30797 2017-04-10 Bin Cheng <bin.cheng@arm.com>
30799 PR tree-optimization/80153
30800 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
30801 remove POINTER_PLUS_EXPR's base part directly, rather than through
30804 2017-04-10 Richard Biener <rguenther@suse.de>
30805 Bin Cheng <bin.cheng@arm.com>
30807 PR tree-optimization/80153
30808 * tree-affine.c (aff_combination_to_tree): Get base pointer from
30809 the first element of pointer type aff_tree. Build result expr in
30811 (add_elt_to_tree): Convert to type unconditionally. Remove other
30812 fold_convert calls.
30813 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
30814 (rewrite_use_nonlinear_expr): Check invariant using iv information.
30816 2017-04-10 Richard Biener <rguenther@suse.de>
30818 * tree-ssa-structalias.c (find_func_aliases): Properly handle
30821 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
30823 PR rtl-optimization/70478
30824 * lra-constraints.c (curr_small_class_check): New.
30825 (update_and_check_small_class_inputs): New.
30826 (process_alt_operands): Update curr_small_class_check. Disfavor
30827 alternative insn memory operands. Check available regs for small
30830 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
30833 * config/mips/mips.opt (-mvirt): Update description.
30834 * doc/invoke.texi (-mvirt): Likewise.
30836 2017-04-10 Richard Biener <rguenther@suse.de>
30838 PR middle-end/80362
30839 * fold-const.c (fold_binary_loc): Look at unstripped ops when
30840 looking for NEGATE_EXPR in -A / -B to A / B folding.
30842 2017-04-10 Martin Liska <mliska@suse.cz>
30844 PR gcov-profile/80224
30845 * gcov.c (print_usage): Fix usage string.
30846 (get_gcov_intermediate_filename): Remove.
30847 (output_gcov_file): Use both for normal and intermediate format.
30848 (generate_results): Do not initialize special file for
30849 intermediate format.
30851 2017-04-10 Richard Biener <rguenther@suse.de>
30853 PR tree-optimization/80304
30854 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
30857 2017-04-10 Nathan Sidwell <nathan@acm.org>
30860 * config/rs6000/rs6000.c (rs6000_vector_type): New.
30861 (rs6000_init_builtins): Use it.
30863 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30865 * config/arm/arm.md (<mrc>): Add mode to SET source.
30866 (<mrrc>): Likewise.
30868 2017-04-10 Richard Biener <rguenther@suse.de>
30870 PR middle-end/80344
30871 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
30873 2017-04-10 Jakub Jelinek <jakub@redhat.com>
30876 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
30877 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
30878 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
30879 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
30880 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
30881 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
30882 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
30883 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
30884 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
30885 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
30886 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
30887 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
30888 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
30889 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
30890 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
30891 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
30892 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
30893 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
30894 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
30895 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
30896 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
30897 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
30898 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
30900 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
30902 PR rtl-optimization/70478
30903 * lra-constraints.c: Reverse the last patch.
30905 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
30907 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
30908 Add comment for WCHAR_T.
30910 2017-04-08 Martin Liska <mliska@suse.cz>
30913 2017-04-07 Martin Liska <mliska@suse.cz>
30916 * ipa-split.c (split_function): Add function part to a same comdat
30919 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30922 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
30924 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
30926 * rs6000/rs6000.c (vec_load_pendulum): Rename...
30927 (vec_pairing): ...to this.
30928 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
30929 (rs6000_sched_init): Adjust for name change.
30930 (struct rs6000_sched_context): Likewise.
30931 (rs6000_init_sched_context): Likewise.
30932 (rs6000_set_sched_context): Likewise.
30934 2017-04-07 Jakub Jelinek <jakub@redhat.com>
30940 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
30942 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
30943 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
30944 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
30946 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
30948 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
30950 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
30952 PR rtl-optimization/70703
30953 * ira-color.c (update_conflict_hard_regno_costs): Use
30954 int64_t instead of HOST_WIDE_INT.
30956 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
30958 PR rtl-optimization/70478
30959 * lra-constraints.c (process_alt_operands): Disfavor alternative
30960 insn memory operands.
30962 2017-04-07 Jeff Law <law@redhat.com>
30964 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
30965 CALL and NOTE_INSN_CALL_ARG_LOCATION.
30967 2017-04-07 Martin Liska <mliska@suse.cz>
30970 * config/aarch64/aarch64.c (aarch64_process_target_attr):
30971 Show error message instead of an ICE.
30973 2017-04-07 Martin Liska <mliska@suse.cz>
30976 * ipa-split.c (split_function): Add function part to a same comdat
30979 2017-04-07 Richard Biener <rguenther@suse.de>
30981 PR middle-end/80341
30982 * tree.c (get_unwidened): Also handle ! for_type case for
30984 * convert.c (do_narrow): Split out from ...
30985 (convert_to_integer_1): ... here. Do not pass final truncation
30986 type to get_unwidened for TRUNC_DIV_EXPR.
30988 2017-04-07 Richard Biener <rguenther@suse.de>
30990 * tree-affine.c (wide_int_ext_for_comb): Take type rather
30992 (aff_combination_const): Adjust.
30993 (aff_combination_scale): Likewise.
30994 (aff_combination_add_elt): Likewise.
30995 (aff_combination_add_cst): Likewise.
30996 (aff_combination_convert): Likewise.
30997 (add_elt_to_tree): Likewise. Remove unused argument.
30998 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
31000 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
31002 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
31004 * config/arm/arm.c (arm_default_short_enums): Use
31005 ARM_DEFAULT_SHORT_ENUMS.
31006 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
31008 2017-04-06 Jakub Jelinek <jakub@redhat.com>
31011 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
31012 members with redundant out-of-class redeclaration.
31014 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31017 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
31018 * config/i386/i386.md (*zero_extendsidi2):
31019 Add (?*x,*x) and (?*v,*v) alternatives.
31021 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31024 * config/i386/i386.c (ix86_expand_builtin)
31025 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
31026 mode from insn data. Convert operands to insn operand mode.
31027 Copy operands that don't satisfy insn predicate to a register.
31029 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
31031 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
31034 2017-04-06 Richard Biener <rguenther@suse.de>
31036 PR tree-optimization/80334
31037 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
31038 preserve alignment of accesses.
31040 2017-04-06 Richard Biener <rguenther@suse.de>
31042 PR tree-optimization/80262
31043 * tree-sra.c (build_ref_for_offset): Preserve address-space
31045 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
31046 Drop useless address-space information on MEM_REF offsets.
31048 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
31050 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
31052 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31054 PR rtl-optimization/70703
31055 * ira-color.c (update_conflict_hard_regno_costs): Use
31056 HOST_WIDE_INT instead of long.
31058 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
31061 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
31062 not defined for x86_64 target. Add -mmmx target option when __SSE2__
31064 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
31065 for x86_64 target. Handle -m3dnowa option.
31067 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31069 PR rtl-optimization/70703
31070 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
31071 (update_conflict_hard_regno_costs): Use long instead of unsigned
31072 arithmetic for cost calculation.
31074 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31075 Bernd Edlinger <bernd.edlinger@hotmail.de>
31078 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
31081 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
31084 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
31085 ptr_mode with Pmode throughout.
31086 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
31087 into probe_stack_range and use DImode.
31089 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
31092 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
31093 call_eh_return is true.
31095 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31097 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
31098 Initialize last_match_fntype_index.
31100 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31103 * tree-nvr.c: Include internal-fn.h.
31104 (pass_return_slot::execute): Ignore internal calls without
31107 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31108 Richard Biener <rguenther@suse.de>
31111 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
31112 captures used multiple times, except for the last use.
31113 * generic-match-head.c: Include gimplify.h.
31115 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31117 PR tree-optimization/79390
31118 * target.h (struct noce_if_info): Declare.
31119 * targhooks.h (default_noce_conversion_profitable_p): Declare.
31120 * target.def (noce_conversion_profitable_p): New target hook.
31121 * ifcvt.h (struct noce_if_info): New type, moved from ...
31122 * ifcvt.c (struct noce_if_info): ... here.
31123 (noce_conversion_profitable_p): Renamed to ...
31124 (default_noce_conversion_profitable_p): ... this. No longer
31126 (noce_try_store_flag_constants, noce_try_addcc,
31127 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
31128 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
31129 instead of noce_conversion_profitable_p.
31130 * config/i386/i386.c: Include ifcvt.h.
31131 (ix86_option_override_internal): Don't override
31132 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
31133 (ix86_noce_conversion_profitable_p): New function.
31134 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
31135 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
31136 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
31137 * doc/tm.texi: Regenerated.
31139 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31141 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
31144 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31147 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
31148 instructions for small multiply cores.
31150 2017-04-04 Jeff Law <law@redhat.com>
31152 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
31154 (mips_expand_vec_perm_const): Initialize elements in orig_perm
31155 that are not set by the loop over the elements.
31157 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31160 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
31161 int mode, convert_modes it to mode as unsigned, otherwise use
31162 lowpart_subreg to mode rather than SImode.
31163 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
31164 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
31165 Use DImode instead of SImode for the shift count operand.
31166 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
31169 2017-04-04 Richard Biener <rguenther@suse.de>
31171 PR middle-end/80281
31172 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
31173 arithmetic done for the negate or the plus. Simplify.
31174 (A - (-B) -> A + B): Likewise.
31175 * fold-const.c (split_tree): Make sure to not negate pointers.
31177 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
31179 PR rtl-optimization/60818
31180 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
31181 a compare of comparisons with the thing compared if this results
31182 in a different machine mode.
31184 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
31186 * alias.c (base_alias_check): Fix typo in comment.
31187 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
31188 * cgraphunit.c (symbol_table::compile): Likewise.
31189 * collect2.c (maybe_run_lto_and_relink): Likewise.
31190 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
31191 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
31192 * config/avr/avr.c (avr_map_op_t): Likewise.
31193 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
31194 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
31195 * config/epiphany/epiphany.md (movcc): Likewise.
31196 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
31197 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
31199 * config/mips/mips.c (mips_save_restore_reg): Likewise.
31200 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
31201 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
31202 * config/sh/sh.c (sh_rtx_costs): Likewise.
31203 * fold-const.c (fold_truth_andor): Likewise.
31204 * genautomata.c (collapse_flag): Likewise.
31205 * gengtype.h (struct type::u::s): Likewise.
31206 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
31207 * input.c (FORMAT_AMOUNT): Likewise.
31208 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
31209 (known_aggs_to_agg_replacement_list): Likewise.
31210 * ipa-inline-analysis.c: Likewise.
31211 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
31212 * ipa-polymorphic-call.c
31213 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
31214 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
31215 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
31217 * modulo-sched.c (apply_reg_moves): Likewise.
31218 * omp-expand.c (build_omp_regions_1): Likewise.
31219 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
31220 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
31221 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
31222 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
31223 * value-prof.c: Likewise.
31224 * var-tracking.c (val_reset): Likewise.
31226 2017-04-03 Richard Biener <rguenther@suse.de>
31228 PR tree-optimization/80275
31229 * fold-const.c (split_address_to_core_and_offset): Handle
31232 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
31234 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
31235 descriptors is at least equal to that of functions.
31237 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
31239 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
31241 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
31244 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
31245 (mov<IMOD4:mode>): New expander.
31246 (*mov<IMOD4:mode>_internal): New insn and split pattern.
31248 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
31250 PR rtl-optimization/79405
31251 * fwprop.c (propagations_left): New variable.
31252 (forward_propagate_into): Decrement it.
31253 (fwprop_init): Initialize it.
31254 (fw_prop): If the variable has reached zero, stop propagating.
31255 (fwprop_addr): Ditto.
31257 2017-03-31 Jakub Jelinek <jakub@redhat.com>
31260 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
31261 a FUNCTION_DECL, pass it as decl instead of origin to
31264 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
31266 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
31269 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
31272 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
31273 TARGET_VSX_SMALL_INTEGER.
31275 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31277 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
31278 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
31280 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
31282 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
31283 extraction from odd-numbered MSA register.
31285 2017-03-31 Jakub Jelinek <jakub@redhat.com>
31287 PR middle-end/80173
31288 * expmed.c (store_bit_field_1): Don't attempt to create
31289 a word subreg out of hard registers wider than word if they
31290 have HARD_REGNO_NREGS of 1 for their mode.
31292 PR middle-end/80163
31293 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
31294 conversions to integer types wider than word and pointer.
31297 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
31298 (rtx_equal_for_cselib_p): Pass 0 to it.
31299 * cselib.c (cselib_hasher::equal): Likewise.
31300 (rtx_equal_for_cselib_1): Add depth argument. If depth
31301 is 128, don't look up VALUE locs and punt. Increment
31302 depth in recursive calls when walking VALUE locs.
31304 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
31306 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
31307 (make_gcov_file_name): Use the canonical path name for generating
31309 (read_line): Fix handling of files with ascii null bytes.
31311 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
31313 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
31314 to initialise a vector register instead
31315 of using a const_int.
31317 2017-03-30 Jakub Jelinek <jakub@redhat.com>
31319 PR translation/80189
31320 * gimplify.c (omp_default_clause): Use %qs instead of %s in
31321 diagnostic messages.
31323 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
31326 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
31327 (dfp_diex_<mode>): Update mode of operand 1.
31328 * doc/extend.texi (dxex, dxexq): Document change to return type.
31329 (diex, diexq): Document change to argument type.
31331 2017-03-30 Martin Jambor <mjambor@suse.cz>
31334 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
31335 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
31336 it reflects the signature changes performed at the callee side.
31337 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
31338 to cgraph_build_function_type_skip_args.
31339 (build_function_decl_skip_args): Adjust call to the above function.
31341 2017-03-30 Jakub Jelinek <jakub@redhat.com>
31344 * config/i386/sse.md
31345 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
31346 register as dest whenever it is a MEM not rtx_equal_p to the
31347 corresponding dup operand, and when forcing into reg move the
31348 reg into the memory afterwards.
31349 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
31350 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
31351 for the force_reg mode.
31352 (avx512vl_vextractf128<mode>): Use register as dest either
31353 always when a MEM, or when it is a MEM not rtx_equal_p to the
31354 corresponding dup operand, or even not when it is a CONST_VECTOR
31355 depending on the mode and lo vs. hi.
31356 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
31358 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
31359 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
31360 Likewise. Require that operands[2] is even.
31361 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
31362 Remove extraneous parens. Require that operands[2] is a multiple
31364 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
31365 operands[0] is a MEM if <mask_applied>, the predicates/constraints
31366 disallow memory then.
31368 2017-03-30 Richard Biener <rguenther@suse.de>
31370 PR tree-optimization/77498
31371 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
31372 to non-constants over backedges.
31374 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
31376 PR rtl-optimization/80233
31377 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
31378 as last_combined_insn. Do not test for BARRIER_P separately.
31380 2017-03-29 Andreas Schwab <schwab@suse.de>
31383 * calls.c (prepare_call_address): Convert funexp to Pmode before
31384 copying to temp reg.
31386 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31388 PR tree-optimization/80158
31389 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
31390 Handle possible future case of more than one alternate
31392 (replace_rhs_if_not_dup): Likewise.
31393 (replace_one_candidate): Likewise.
31395 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
31397 PR rtl-optimization/80193
31398 * ira.c (ira): Do not check allocation for LRA.
31400 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
31402 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
31403 (nvptx_output_simt_exit): Declare.
31404 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
31405 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
31406 (init_softstack_frame): Move initialization of crtl->is_leaf to...
31407 (nvptx_declare_function_name): ...here. Emit declaration of local
31408 memory space buffer for omp_simt_enter insn.
31409 (nvptx_output_unisimt_switch): New.
31410 (nvptx_output_softstack_switch): New.
31411 (nvptx_output_simt_enter): New.
31412 (nvptx_output_simt_exit): New.
31413 * config/nvptx/nvptx.h (struct machine_function): New fields
31414 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
31415 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
31416 (UNSPECV_SIMT_EXIT): Ditto.
31417 (omp_simt_enter_insn): New insn.
31418 (omp_simt_enter): New expansion.
31419 (omp_simt_exit): New insn.
31420 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
31422 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
31423 (expand_GOMP_SIMT_ENTER_ALLOC): New.
31424 (expand_GOMP_SIMT_EXIT): New.
31425 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
31426 (GOMP_SIMT_ENTER_ALLOC): Ditto.
31427 (GOMP_SIMT_EXIT): Ditto.
31428 * target-insns.def (omp_simt_enter): New insn.
31429 (omp_simt_exit): Ditto.
31430 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
31432 (lower_rec_simd_input_clauses): Implement SIMT privatization.
31433 (lower_rec_input_clauses): Likewise.
31434 (lower_lastprivate_clauses): Handle SIMT privatization.
31436 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
31437 (ompdevlow_adjust_simt_enter): New.
31438 (find_simtpriv_var_op): New.
31439 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
31440 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
31442 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
31443 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
31444 (copy_decl_for_dup_finish): Ditto.
31446 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
31448 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
31451 * config/i386/i386.c (ix86_option_override_internal): Always
31452 allow -mpreferred-stack-boundary=3 for 64-bit targets.
31454 2017-03-28 Bin Cheng <bin.cheng@arm.com>
31456 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
31458 2017-03-28 Bin Cheng <bin.cheng@arm.com>
31460 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
31461 mark new edge's irreducible flag accordign to it.
31462 (vect_do_peeling): Check loop preheader edge's irreducible flag
31463 and pass it to function slpeel_add_loop_guard.
31465 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
31467 PR tree-optimization/80218
31468 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
31469 Update block frequencies and counts.
31471 2017-03-28 Richard Biener <rguenther@suse.de>
31473 PR tree-optimization/78644
31474 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
31475 of a simplification result we may not use it at all.
31477 2017-03-28 Richard Biener <rguenther@suse.de>
31480 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
31481 without arguments, generate default definition of a SSA name.
31483 2017-03-28 Richard Biener <rguenther@suse.de>
31485 PR middle-end/80222
31486 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
31487 TYPE_REF_CAN_ALIAS_ALL references.
31488 * fold-const.c (fold_indirect_ref_1): Likewise.
31490 2017-03-28 Martin Liska <mliska@suse.cz>
31493 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
31494 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
31496 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31497 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
31499 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
31500 (EXTRA_SPECS): Define.
31501 (SUBTARGET_EXTRA_SPECS): Likewise.
31502 (SUBTARGET_CPP_SPEC): Likewise.
31503 * config/arc/elf.h (EXTRA_SPECS): Renamed to
31504 SUBTARGET_EXTRA_SPECS.
31505 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
31507 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31509 * config/arc/simdext.md (vst64_insn): Update pattern.
31510 (vld32wh_insn): Likewise.
31511 (vld32wl_insn): Likewise.
31512 (vld64_insn): Likewise.
31513 (vld32_insn): Likewise.
31515 2017-03-28 Marek Polacek <polacek@redhat.com>
31518 * fold-const.c (fold_comparison): Use protected_set_expr_location
31519 instead of SET_EXPR_LOCATION.
31521 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
31523 * tree.c (add_expr): Avoid name lookup warning.
31525 2017-03-27 Jeff Law <law@redhat.com>
31527 PR tree-optimization/80216
31528 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
31529 function name. Limit recursion depth.
31530 (record_temporary_equivalences): Corresponding changes.
31532 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
31534 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
31537 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31540 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
31542 * cfgcleanup.c (reg_note_cfa_p): New array.
31543 (insns_have_identical_cfa_notes): New function.
31544 (old_insns_match_p): Don't cross-jump in between /f
31545 and non-/f instructions. If both i1 and i2 are frame related,
31546 verify all CFA notes, their order and content.
31548 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
31551 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
31552 HImode and SImode with zero extend to DImode to one insn.
31553 (bswap<mode>2_extenddi): Likewise.
31554 (bswapsi2_extenddi): Likewise.
31555 (bswaphi2_extendsi): Likewise.
31556 (bswaphi2): Combine bswap HImode and SImode into one insn.
31557 Separate memory insns from swapping register.
31558 (bswapsi2): Likewise.
31559 (bswap<mode>2): Likewise.
31560 (bswaphi2_internal): Delete, no longer used.
31561 (bswapsi2_internal): Likewise.
31562 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
31563 store, and gpr<-gpr swap insns.
31564 (bswap<mode>2_store): Likewise.
31565 (bswaphi2_reg): Register only splitter, combine with the splitter.
31566 (bswaphi2 splitter): Likewise.
31567 (bswapsi2_reg): Likewise.
31568 (bswapsi2 splitter): Likewise.
31569 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
31570 the insns into load, store, and register/register insns.
31571 (bswapdi2_ldbrx): Likewise.
31572 (bswapdi2_load): Likewise.
31573 (bswapdi2_store): Likewise.
31574 (bswapdi2_reg): Likewise.
31576 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
31578 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
31579 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
31581 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
31584 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
31586 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31587 special handling for target option conflicts between dform
31588 options (-mpower9-dform, -mpower9-dform-vector,
31589 -mpower9-dform-scalar) and -mno-direct-move.
31591 2017-03-27 Richard Biener <rguenther@suse.de>
31593 PR tree-optimization/80181
31594 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
31596 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31598 * config/arc/predicates.md (move_double_src_operand): Replace the
31599 call to move_double_src_operand with a call to address_operand.
31601 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31603 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
31604 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
31605 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
31607 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31609 * config/arc/predicates.md (long_immediate_loadstore_operand):
31610 Consider scaled addresses cases.
31612 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31614 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
31615 restored when in interrupt.
31616 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
31617 doesn't have delay slot.
31619 2017-03-27 Richard Biener <rguenther@suse.de>
31622 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
31623 inlined thunk clones.
31625 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31628 * asan.c (instrument_derefs): Copy over last operand from
31629 original COMPONENT_REF to the new COMPONENT_REF with
31630 DECL_BIT_FIELD_REPRESENTATIVE.
31631 * ubsan.c (instrument_object_size): Likewise.
31633 2017-03-27 Richard Biener <rguenther@suse.de>
31635 PR tree-optimization/80170
31636 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
31637 sure DR/SCEV didnt fold in constants we do not see when looking
31638 at the reference base alignment.
31640 2017-03-27 Richard Biener <rguenther@suse.de>
31642 PR middle-end/80171
31643 * gimple-fold.c (fold_ctor_reference): Properly guard against
31644 NULL return value from canonicalize_constructor_val.
31646 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
31649 * config/i386/i386.c (ix86_expand_builtin)
31650 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
31651 flags reg setting and flags reg using instructions.
31652 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
31653 clobbering instructions to zero extend op2.
31655 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
31657 * doc/install.texi (Configuration) <--with-aix-soname>:
31658 Update link to AIX ld.
31660 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
31662 PR rtl-optimization/80160
31663 PR rtl-optimization/80159
31664 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
31665 reg_alternate_class into account.
31667 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
31670 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
31671 to consider in curr_insn_transform.
31673 2017-03-24 Jakub Jelinek <jakub@redhat.com>
31675 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
31676 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
31677 and emit_mode_inner.
31679 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31681 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
31682 argument to the overloaded builtin variants. Use the new flag to
31683 deprecate certain builtin variants.
31684 * config/s390/s390-builtin-types.def: Add new builtin types.
31685 * config/s390/s390-builtins.h: Support new flags field for
31686 overloaded builtins.
31687 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
31688 (s390_macro_to_expand): Enable vector float data type.
31689 (s390_cpu_cpp_builtins_internal): Indicate support of the new
31690 builtins by incrementing the __VEC__ version number.
31691 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
31693 (s390_resolve_overloaded_builtin): Emit error messages depending
31694 on the builtin flags.
31695 * config/s390/s390.c (s390_expand_builtin): Support additional
31696 flags argument. Change error message to match the messages
31697 emitted in s390-c.c.
31698 * config/s390/s390.md: New UNSPEC_* constants.
31699 (op_type): Add new instruction types.
31700 * config/s390/vecintrin.h: Add new builtins and test data class
31702 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
31703 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
31704 (VEC_INEXACT, VEC_NOINEXACT): New constants.
31705 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
31706 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
31707 ("vec_mergel<mode>"): V_HW -> VEC_HW.
31709 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
31710 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
31711 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
31712 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
31714 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
31715 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
31716 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
31717 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
31719 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
31720 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
31721 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
31722 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
31723 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
31724 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
31725 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
31727 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
31728 ("vec_scatter_element<V_HW_4:mode>_DI")
31729 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
31730 ("vec_fpint<mode>", "vflls")
31731 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
31732 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
31733 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
31734 ("*vec_cmphe<mode>_cc"): ... these.
31736 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
31737 mode constant instead of magic value.
31739 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31741 * config/s390/s390.c (s390_expand_vec_compare): Support other
31742 vector floating point modes than just V2DF.
31743 (s390_expand_vcond): Likewise.
31744 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
31745 (s390_cannot_change_mode_class): Prevent mode changes between TF
31746 and V1TF in vector registers.
31747 * config/s390/s390.md (DF, SF): New mode attributes.
31748 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
31749 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
31750 SFmode support for VRs.
31751 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
31753 (VFT, VF_HW): New mode iterators.
31754 (vw, sdx): New mode attributes.
31755 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
31756 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
31757 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
31758 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
31759 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
31760 also the new vector floating point modes. Renaming to ...
31762 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
31763 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
31764 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
31765 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
31766 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
31767 ("vec_unordered<mode>"): ... these.
31769 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
31770 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
31771 ("*vec_extendv2df"): New insn definitions.
31773 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31775 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
31776 ("mulditi3_2", "*muldi3_sign"): New patterns.
31777 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
31778 rename the pattern definition.
31780 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31782 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
31784 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
31786 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31788 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
31789 instruction if possible.
31790 * config/s390/vector.md (vec_halfnumelts): New mode
31792 ("*vec_vllezlf<mode>"): New pattern.
31794 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31796 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
31797 ("popcountv4si2", "popcountv2di2"): Rename to ...
31798 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
31799 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
31801 ("popcount<mode>2_vxe"): New pattern.
31803 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31805 * common/config/s390/s390-common.c (processor_flags_table): Add
31807 * config.gcc: Add arch12.
31808 * config/s390/driver-native.c (s390_host_detect_local_cpu):
31809 Default to arch12 for unknown CPU model numbers.
31810 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
31811 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
31812 PROCESSOR_max sanity check.
31813 * config/s390/s390-opts.h (enum processor_type): Add
31815 * config/s390/s390.c (processor_table): Add arch12.
31816 (s390_expand_builtin): Add check for B_VXE flag.
31817 (s390_issue_rate): Add PROCESSOR_ARCH12.
31818 (s390_get_sched_attrmask): Likewise.
31819 (s390_get_unit_mask): Likewise.
31820 (s390_sched_score): Enable z13 scheduling for arch12.
31821 (s390_sched_reorder): Likewise.
31822 (s390_sched_variable_issue): Likewise.
31823 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
31825 (s390_tune_attr): Use z13 scheduling also for arch12.
31826 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
31827 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
31828 (TARGET_VXE_P): New macros.
31829 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
31830 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
31831 * config/s390/s390.opt: Add arch12 as processor_type.
31833 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31835 * config/s390/s390.md
31836 ("fixuns_truncdddi2", "fixuns_trunctddi2")
31837 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
31838 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
31840 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
31841 Rename expanders to ...
31843 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
31844 ("fixuns_truncdddi2_emu"): ... these.
31846 ("fixuns_trunc<mode>si2_emu"): New expander.
31848 ("*fixuns_truncdfdi2_z13"): Rename to ...
31849 ("*fixuns_truncdfdi2_vx"): ... this.
31851 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31853 * config/s390/2964.md: Remove the single element vector compare
31854 instructions which are no longer used.
31855 * config/s390/s390.c (s390_select_ccmode): Remove handling of
31857 (s390_canonicalize_comparison): Remove handling of DFmode
31859 (s390_expand_vec_compare_scalar): Remove function.
31860 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
31861 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
31863 ("*cmp<mode>_ccs"): Add wfcdb instruction.
31865 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31867 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
31869 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
31870 will anyway by matched by mov<mode>_64dfp.
31872 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31874 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
31875 vlef/vstef. Add missing operand to vleif.
31877 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31879 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
31880 pair for all vector types with 64 bit elements.
31881 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
31882 * config/s390/vector.md (V_HW_64): ... here.
31883 (V_128_NOSINGLE): New mode iterator.
31884 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
31885 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
31886 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
31887 ("*vec_load_pairv2di"): Change to ...
31888 ("*vec_load_pair<mode>"): ... this one.
31890 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31892 * config/s390/constraints.md: Add comments.
31893 (jKK): Reject element sizes > 8 bytes.
31894 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
31896 * config/s390/s390.md: Add the s_operand checks formerly in
31897 s390_split_ok_p to various splitters where they are still
31899 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
31900 for 128 bit vectors. Plus two splitters.
31902 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31904 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
31907 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31910 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
31911 error if the boundary argument is not constant.
31913 2017-03-24 Jakub Jelinek <jakub@redhat.com>
31915 PR rtl-optimization/80112
31916 * loop-doloop.c (doloop_condition_get): Don't check condition
31917 if cmp isn't SET with IF_THEN_ELSE src.
31919 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31921 PR tree-optimization/80158
31922 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
31923 replacing a candidate statement, also replace it for the
31924 candidate's alternate interpretation.
31925 (replace_rhs_if_not_dup): Likewise.
31926 (replace_one_candidate): Likewise.
31928 2017-03-24 Richard Biener <rguenther@suse.de>
31930 PR tree-optimization/80167
31931 * graphite-isl-ast-to-gimple.c
31932 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
31934 (translate_isl_ast_to_gimple::get_rename): Likewise.
31936 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
31938 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
31939 handling of certain combinations of target options, including the
31940 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
31941 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
31943 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31946 * config/arm/arm.md (*load_multiple): Add reload_completed to
31947 matching condition.
31949 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31950 Richard Biener <rguenth@suse.de>
31952 PR tree-optimization/79908
31953 PR tree-optimization/80136
31954 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
31955 been cast away, gimplify_and_add suffices.
31957 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
31959 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
31961 2017-03-23 Richard Biener <rguenther@suse.de>
31963 PR tree-optimization/80032
31964 * gimplify.c (gimple_push_cleanup): Forced unconditional
31965 cleanups still have to go to the conditional_cleanups
31968 2017-03-22 Jakub Jelinek <jakub@redhat.com>
31970 PR tree-optimization/80072
31971 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
31973 (next_operand_entry_id): Change type to unsigned int.
31974 (sort_by_operand_rank): Make sure to return the right return value
31975 even if unsigned fields are bigger than INT_MAX.
31976 (struct oecount): Change cnt and id type to unsigned int.
31977 (oecount_hasher::equal): Formatting fix.
31978 (oecount_cmp): Make sure to return the right return value
31979 even if unsigned fields are bigger than INT_MAX.
31980 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
31983 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
31984 TREE_READONLY on result if writing it more than once.
31987 * doc/invoke.texi (-fsanitize=thread): Document that with
31988 -fnon-call-exceptions atomics are not able to throw
31992 * tsan.c: Include tree-eh.h.
31993 (instrument_builtin_call): Call maybe_clean_eh_stmt or
31994 maybe_clean_or_replace_eh_stmt where needed.
31995 (instrument_memory_accesses): Add cfg_changed argument.
31996 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
31997 if it returned true.
31998 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
32000 PR rtl-optimization/63191
32001 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
32002 wrapper function, moved the whole old content into ...
32003 (ix86_delegitimize_address_1): ... this. New inline function.
32004 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
32005 true as last argument instead of ix86_delegitimize_address.
32007 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32009 * config/aarch64/aarch64.c (generic_branch_cost): Copy
32010 cortexa57_branch_cost.
32012 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32014 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
32016 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32019 * doc/md.texi (Constraints): Document wA constraint.
32020 * config/rs6000/constraints.md (wA): New.
32021 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
32022 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
32023 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
32024 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
32026 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
32029 * gimplify.c (is_oacc_declared): New function.
32030 (oacc_default_clause): Use it to set default flags for acc declared
32031 variables inside parallel regions.
32032 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
32033 declared variables.
32034 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
32035 declare attribute to any decl as necessary.
32037 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
32040 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
32041 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
32042 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
32043 (arm_arch_lpae): This.
32044 * config/arm/arm.c (arm_arch7ve): Rename into ...
32045 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
32046 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
32049 2017-03-22 Martin Liska <mliska@suse.cz>
32052 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
32053 error message instead of an ICE.
32055 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32057 * doc/extend.texi (6.11 Additional Floating Types): Revise.
32059 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
32061 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
32063 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32066 2017-03-21 Martin Sebor <msebor@redhat.com>
32068 * doc/extend.texi: Use "cannot" instead of "can't."
32069 * doc/hostconfig.texi: Same.
32070 * doc/install.texi: Same.
32071 * doc/invoke.texi: Same.
32072 * doc/loop.texi: Same.
32073 * doc/md.texi: Same.
32074 * doc/objc.texi: Same.
32075 * doc/rtl.texi: Same.
32076 * doc/tm.texi: Same.
32077 * doc/tm.texi.in: Same.
32078 * doc/trouble.texi: Same.
32080 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
32083 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
32084 (collect_checksum_attributes): Set it.
32085 (die_checksum_ordered): Use it.
32087 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32089 PR tree-optimization/79908
32090 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
32091 change: For a VA_ARG whose LHS has been cast away, use
32092 force_gimple_operand to construct the side effects.
32094 2017-03-21 David Malcolm <dmalcolm@redhat.com>
32096 PR translation/80001
32097 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
32098 more amenable to translation.
32099 (oacc_loop_auto_partitions): Likewise.
32101 2017-03-21 Marek Polacek <polacek@redhat.com>
32102 Martin Sebor <msebor@redhat.com>
32104 PR tree-optimization/80109
32105 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
32106 on INTEGRAL_TYPE_P.
32108 2017-03-21 Jakub Jelinek <jakub@redhat.com>
32109 Segher Boessenkool <segher@kernel.crashing.org>
32112 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
32113 check reg_used_between_p between insn and one of succ or succ2
32114 depending on if succ is artificial insn not inserted into insn
32117 2017-03-21 Martin Liska <mliska@suse.cz>
32119 PR gcov-profile/80081
32120 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
32121 * doc/gcc.texi: Include gcov-dump stuff.
32122 * doc/gcov-dump.texi: New file.
32124 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
32126 PR rtl-optimization/79150
32127 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
32128 conditional jump, if the jump is the last insn of the loop.
32130 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32131 Richard Biener <rguenth@suse.de>
32133 PR tree-optimization/79908
32134 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32135 been cast away, use force_gimple_operand to construct the side
32138 2017-03-21 Martin Liska <mliska@suse.cz>
32140 PR libfortran/79956
32141 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
32144 2017-03-21 Brad Spengler <spender@grsecurity.net>
32147 * plugin.c (htab_hash_plugin): New function.
32148 (add_new_plugin): Use it and adjust.
32149 (parse_plugin_arg_opt): Adjust.
32150 (init_one_plugin): Likewise.
32152 2017-03-21 Richard Biener <rguenther@suse.de>
32154 PR tree-optimization/80032
32155 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
32156 if set force the cleanup to happen unconditionally.
32157 (gimplify_target_expr): Push inserted clobbers with force_uncond
32158 to avoid them being removed by control-dependent DCE.
32160 2017-03-21 Richard Biener <rguenther@suse.de>
32162 PR tree-optimization/80122
32163 * tree-inline.c (copy_bb): Do not expans va-arg packs or
32164 va_arg_pack_len when the inlined call stmt requires pack
32166 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
32168 2017-03-21 Jakub Jelinek <jakub@redhat.com>
32171 * tsan.c (instrument_builtin_call): If the memory model argument
32172 is not a constant, assume it is valid.
32175 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
32178 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
32180 PR rtl-optimization/79910
32181 * combine.c (can_combine_p): Do not allow combining an I0 or I1
32182 if its dest is used by an insn before I2 (other than the combined
32183 insns themselves, which are properly handled already).
32185 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
32188 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
32190 * combine.c (record_used_regs): New static function.
32191 (try_combine): Handle situations where there is an additional
32192 instruction between I2 and I3 which needs to have a LOG_LINK
32196 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
32198 * combine.c (try_combine): Delete redundant i1 test. Call
32199 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
32201 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32204 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
32205 alternatives 13/14.
32207 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32209 PR tree-optimization/80054
32210 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
32211 the optimization if a PHI or any of its arguments is not dominated
32212 by the candidate's basis. Use gphi* rather than gimple* as
32214 (replace_profitable_candidates): Clean up a gimple* variable that
32215 should be a gphi* variable.
32217 2017-03-20 Martin Sebor <msebor@redhat.com>
32220 * doc/extend.texi (attribute constructor): Document present limitation.
32222 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
32225 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
32226 __POWER9_VECTOR__ #ifdef control, change template definition to
32227 use Power9-specific built-in function.
32228 (vec_any_eq): Likewise.
32229 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
32230 to control outcomes from this test.
32231 (vector_ae_<mode>p): For VEC_F modes, likewise.
32233 2017-03-20 Ian Lance Taylor <iant@google.com>
32235 * config/i386/i386.c (ix86_function_regparm): Save an extra
32236 register for -fsplit-stack with DECL_STATIC_CHAIN.
32238 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
32241 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
32242 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
32244 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
32246 * config/riscv/riscv.c (riscv_print_operand): Use "fence
32248 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
32251 2017-03-20 Marek Polacek <polacek@redhat.com>
32254 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
32256 2017-03-20 Richard Biener <rguenther@suse.de>
32258 PR tree-optimization/80113
32259 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
32260 allocate extra SSA name for PHI def.
32261 (add_close_phis_to_outer_loops): Likewise.
32262 (add_close_phis_to_merge_points): Likewise.
32263 (copy_loop_close_phi_args): Likewise.
32264 (copy_cond_phi_nodes): Likewise.
32266 2017-03-20 Martin Liska <mliska@suse.cz>
32268 PR middle-end/79753
32269 * tree-chkp.c (chkp_build_returned_bound): Do not build
32270 returned bounds for a LHS that's not a BOUNDED_P type.
32272 2017-03-20 Martin Liska <mliska@suse.cz>
32276 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
32277 COMPLEX_CST and VECTOR_CST.
32279 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32282 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
32283 target operand. A new splitter adds the clobber statement in case
32284 the target operand is dead anyway.
32286 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
32288 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
32289 to age-old versions of binutils and glibc.
32291 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
32293 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
32295 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
32297 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
32299 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
32301 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
32302 requirement for binutils 2.13.
32304 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
32306 * combine.c (try_combine): Delete redundant i1 test. Call
32307 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
32309 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
32311 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
32312 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
32314 <riscv64-*-elf>: Re-arrange section
32315 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
32316 <riscv32-*-linux>: Likewise.
32317 <riscv64-*-elf>: Likewise
32318 <riscv64-*-linux>: Likewise.
32320 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
32323 * aarch64.opt(verbose-cost-dump): Fix typo.
32325 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
32328 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
32329 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
32331 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
32333 * reload.c (find_reloads): When reloading a nonoffsettable address,
32334 use RELOAD_OTHER for it and its address reloads.
32336 PR rtl-optimization/79910
32337 * combine.c (record_used_regs): New static function.
32338 (try_combine): Handle situations where there is an additional
32339 instruction between I2 and I3 which needs to have a LOG_LINK
32342 2017-03-17 Jeff Law <law@redhat.com>
32344 PR tree-optimization/71437
32345 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
32346 conditional in the hash table first.
32347 (vrp_dom_walker::before_dom_children): Extract condition from
32348 ASSERT_EXPR. Record condition, its inverion and any implied
32349 conditions as well.
32351 2017-03-17 Marek Polacek <polacek@redhat.com>
32352 Markus Trippelsdorf <markus@trippelsdorf.de>
32354 PR tree-optimization/80079
32355 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
32358 2017-03-17 Richard Biener <rguenther@suse.de>
32360 PR middle-end/80075
32361 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
32362 Properly verify the LHS before the RHS possibly claims to be
32364 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
32367 2017-03-17 Martin Jambor <mjambor@suse.cz>
32369 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
32370 (List of -O2 options): Likewise.
32371 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
32374 2017-03-17 Tom de Vries <tom@codesourcery.com>
32376 * gcov-dump.c (print_usage): Print bug_report_url.
32378 2017-03-17 Richard Biener <rguenther@suse.de>
32380 PR middle-end/80050
32381 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
32382 (parser::peek): Likewise.
32384 2017-03-17 Richard Biener <rguenther@suse.de>
32386 PR tree-optimization/80048
32387 * sese.c (free_sese_info): Properly release rename_map and
32388 copied_bb_map elements.
32390 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
32392 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
32393 Add linked-list forward and backlinks. Insert on
32394 construction, remove on destruction.
32395 (class pass_store_merging): Add m_stores_head field.
32396 (pass_store_merging::terminate_and_process_all_chains):
32397 Iterate over m_stores_head list.
32398 (pass_store_merging::terminate_all_aliasing_chains):
32400 (pass_store_merging::execute): Check for debug stmts first.
32401 Push new chains onto the m_stores_head stack.
32403 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
32406 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
32407 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
32408 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
32410 2017-03-16 Jeff Law <law@redhat.com>
32412 PR tree-optimization/71437
32413 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
32414 member function. Implementation moved into after_dom_children
32415 member function and into the threader's thread_outgoing_edges
32417 (dom_opt_dom_walker::after_dom_children): Simplify by moving
32418 some code into new thread_outgoing_edges.
32419 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
32420 definition. Simplify marker handling (do it here). Assume we always
32421 have the available expression and the const/copies tables.
32422 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
32424 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
32425 * tree-vrp.c (equiv_stack): No longer file scoped.
32426 (vrp_dom_walker): New class.
32427 (vrp_dom_walker::before_dom_children): New member function.
32428 (vrp_dom_walker::after_dom_children): Likewise.
32429 (identify_jump_threads): Setup domwalker. Use it rather than
32430 walking edges in a random order by hand. Simplify setup/finalization.
32431 (finalize_jump_threads): Remove.
32432 (vrp_finalize): Do not call identify_jump_threads here.
32433 (execute_vrp): Do it here instead and call thread_through_all_blocks
32436 PR tree-optimization/71437
32437 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
32439 (simplify_stmt_for_jump_threading): Add basic_block argument. All
32441 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
32442 (dom_opt_dom_walker::thread_across_edge): Remove
32443 handle_dominating_asserts argument. All callers changed.
32444 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
32445 changes. Remove calls to lhs_of_dominating_assert. Other
32446 uses of handle_dominating_asserts turn into unconditional code
32447 (simplify_control_stmt_condition_1): Likewise.
32448 (simplify_control_stmt_condition): Likewise.
32449 (thread_through_normal_block, thread_across_edge): Likewise.
32450 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
32451 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
32452 object if it is not an SSA_NAME.
32453 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
32454 before calling into the VRP specific simplifiers.
32455 (identify_jump_threads): Remove handle_dominating_asserts
32458 2017-03-16 Jakub Jelinek <jakub@redhat.com>
32461 * tree-diagnostic.c (default_tree_printer): No longer static.
32462 * tree-diagnostic.h (default_tree_printer): New prototype.
32464 2017-03-16 Tamar Christina <tamar.christina@arm.com>
32466 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
32467 Change ins into fmov.
32469 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32471 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
32472 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
32473 Use h_con constraint for operand 1.
32474 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
32475 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
32477 2017-03-15 Jeff Law <law@redhat.com>
32479 PR tree-optimization/71437
32480 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
32481 (record_temporary_equivalences): Use it.
32483 PR tree-optimization/71437
32484 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
32485 tree-ssa-scopedtables.
32486 (lookup_avail_expr, build_and_record_new_cond): Likewise.
32487 (record_conditions, record_cond, vuse_eq): Likewise.
32488 (record_edge_info): Adjust to API tweak of record_conditions.
32489 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
32490 (record_temporary_equivalences, optimize_stmt): Likewise.
32491 (eliminate_redundant_computations): Likewise.
32492 (record_equivalences_from_stmt): Likewise.
32493 * tree-ssa-scopedtables.c: Include options.h and params.h.
32494 (vuse_eq): New function, moved from tree-ssa-dom.c
32495 (build_and_record_new_cond): Likewise.
32496 (record_conditions): Likewise. Accept vector of conditions rather
32497 than edge_equivalence structure for first argument.
32498 for the first argument.
32499 (avail_exprs_stack::lookup_avail_expr): New member function, moved
32500 from tree-ssa-dom.c.
32501 (avail_exprs_stack::record_cond): Likewise.
32502 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
32503 from tree-ssa-dom.c.
32504 (avail_exprs_stack): Add new member functions lookup_avail_expr
32506 (record_conditions): Declare.
32508 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
32511 * lra-constraints.c (process_alt_operands): Increase reject for
32512 reloading an input/output operand.
32514 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
32517 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
32518 insns to convert from signed/unsigned char/short to IEEE 128-bit
32520 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
32522 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
32525 * config/i386/i386.c (ix86_vector_duplicate_value): Create
32526 subreg of inner mode for values already in registers.
32528 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
32530 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
32531 iteration reg is used after the loop.
32533 2017-03-14 Martin Sebor <msebor@redhat.com>
32535 PR tree-optimization/79800
32536 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
32537 precision in negative-positive range.
32538 (format_floating): Call non-const overload with adjusted precision.
32540 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
32543 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
32546 2017-03-14 Martin Sebor <msebor@redhat.com>
32548 PR middle-end/80020
32549 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
32550 * builtins.def (aligned_alloc): Use it.
32553 * Makefile.in (GTFILES): Add calls.c.
32554 * calls.c: Include "gt-calls.h".
32556 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
32558 PR rtl-optimization/79728
32559 * regs.h (struct target_regs): New field
32560 x_contains_allocatable_regs_of_mode.
32561 (contains_allocatable_regs_of_mode): New macro.
32562 * reginfo.c (init_reg_sets_1): Initialize it, and change
32563 contains_reg_of_mode so it includes global regs as well.
32564 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
32565 rather than contains_regs_of_mode.
32567 2017-03-14 Martin Liska <mliska@suse.cz>
32569 * doc/invoke.texi: Document options that can't be combined with
32570 -fcheck-pointer-bounds.
32572 2017-03-14 Martin Liska <mliska@suse.cz>
32574 PR middle-end/79831
32575 * doc/invoke.texi (-Wchkp): Document the option.
32577 2017-03-14 Martin Liska <mliska@suse.cz>
32579 * Makefile.in: Install gcov-dump.
32581 2017-03-14 Martin Liska <mliska@suse.cz>
32583 * multiple_target.c (expand_target_clones): Bail out for
32584 an invalid attribute.
32586 2017-03-14 Richard Biener <rguenther@suse.de>
32588 * alias.c (struct alias_set_entry): Pack properly.
32589 * cfgloop.h (struct loop): Likewise.
32590 * cse.c (struct set): Likewise.
32591 * ipa-utils.c (struct searchc_env): Likewise.
32592 * loop-invariant.c (struct invariant): Likewise.
32593 * lra-remat.c (struct cand): Likewise.
32594 * recog.c (struct change_t): Likewise.
32595 * rtl.h (struct address_info): Likewise.
32596 * symbol-summary.h (function_summary): Likewise.
32597 * tree-loop-distribution.c (struct partition): Likewise.
32598 * tree-object-size.c (struct object_size_info): Likewise.
32599 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
32600 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
32601 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
32602 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
32603 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
32604 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
32605 (struct _stmt_vec_info): Likewise.
32607 2017-03-14 Martin Liska <mliska@suse.cz>
32610 * multiple_target.c (create_dispatcher_calls): Check that
32611 a target can create a function dispatcher.
32613 2017-03-14 Martin Liska <mliska@suse.cz>
32616 * multiple_target.c (expand_target_clones): Drop local.local
32617 flag for default implementation.
32619 2017-03-14 Richard Biener <rguenther@suse.de>
32621 PR tree-optimization/80030
32622 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
32624 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
32626 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
32627 gcc_fallthrough() instead of __attribute__((fallthrough));
32629 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
32631 * doc/gcc.texi: Remove "up" link to (DIR).
32632 * doc/gccint.texi: Ditto.
32634 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
32636 * doc/install.texi (Specific) <avr>: Remove reference to
32639 2017-03-13 Jeff Law <law@redhat.com>
32641 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
32642 attribute rather than comments.
32644 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
32645 match_scratch operand is highest.
32647 2017-03-13 Martin Liska <mliska@suse.cz>
32649 PR middle-end/78339
32650 * ipa-pure-const.c (warn_function_noreturn): If the declarations
32651 is a CHKP clone, use original declaration.
32653 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32655 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
32656 (arc_conditional_register_usage): Use a different allocation order
32657 when optimizing for size.
32658 * common/config/arc/arc-common.c (arc_option_optimization_table):
32659 Section anchors default on when optimizing for size.
32661 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32663 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
32665 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32667 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
32668 * config/arc/arc.md (cpu_facility): Add cd variant.
32669 (*movqi_insn): Add code density variant.
32670 (*movhi_insn): Likewise.
32671 (*movqi_insn): Likewise.
32672 (*addsi3_mixed): Likewise.
32673 (subsi3_insn): Likewise.
32675 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32677 * config/arc/arc.md (movsi_cond_exec): Update constraint.
32679 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32681 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
32682 expressions with MINUS and UNARY ops.
32684 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32687 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
32689 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
32690 between vec_select and vector argument.
32691 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
32692 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
32693 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
32694 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
32695 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
32696 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
32698 2017-03-13 Richard Biener <rguenther@suse.de>
32701 * params.def (vect-max-peeling-for-alignment): Fix typo.
32703 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32705 * doc/install.texi (Specific) <mips-*-*>: Remove description of
32706 issue that only occurred with binutils below 2.18.
32708 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32710 * doc/install.texi (Specific) <cris-axis-elf>: No longer
32711 refer to binutils 2.11/2.12 minimum.
32713 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32715 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
32716 ftp.kernel.org and simplify binutils requirement.
32718 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
32720 * doc/invoke.texi (Warning Options): Fix spelling of link-time
32722 (Optimize Options): Ditto. Also remove redundancy.
32724 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32726 PR translation/79848
32727 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
32729 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
32730 to G_ to avoid double translation.
32732 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32734 PR translation/79923
32735 * auto-profile.c (get_combined_location): Convert leading
32736 character of diagnostics to lower case and remove trailing period.
32737 (read_profile): Likewise for various diagnostics.
32738 * config/arm/arm.c (arm_option_override): Remove trailing period
32739 from various diagnostics.
32740 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
32741 (msp430_expand_delay_cycles): Likewise.
32743 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32746 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
32747 full command-line argument, rather than just "str".
32748 (aarch64_validate_march): Likewise.
32749 (aarch64_validate_mtune): Likewise.
32751 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
32753 PR rtl-optimization/78911
32754 * lra-assigns.c (must_not_spill_p): New function.
32755 (spill_for): Use it.
32757 2017-03-10 Jakub Jelinek <jakub@redhat.com>
32759 PR tree-optimization/79981
32760 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
32761 ATOMIC_COMPARE_EXCHANGE ifn result.
32762 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
32763 IFN_ATOMIC_COMPARE_EXCHANGE.
32765 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32768 * opts.c (parse_sanitizer_options): Add missing question mark to
32769 "did you mean" message.
32771 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32773 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
32775 (VMULEUH_UNS): Likewise.
32776 (VMULOUB_UNS): Likewise.
32777 (VMULOUH_UNS): Likewise.
32778 * config/rs6000/rs6000.c (builtin_function_type): Remove
32779 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
32781 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32784 * read-rtl-function.c (function_reader::read_rtx_operand): Update
32785 x with result of extra_parsing_for_operand_code_0.
32786 (function_reader::extra_parsing_for_operand_code_0): Convert
32787 return type from void to rtx, returning x. When reading
32788 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
32789 larger size containing struct block_symbol.
32791 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
32793 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
32794 -mfloat128-hardware without -m64.
32796 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
32799 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
32800 entries to the case statement that marks unsigned arguments to
32801 overloaded functions.
32803 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
32805 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
32806 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
32808 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
32811 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
32812 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
32814 2017-03-10 Martin Liska <mliska@suse.cz>
32818 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
32819 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
32822 2017-03-10 Olivier Hainque <hainque@adacore.com>
32824 * tree-switch-conversion (array_value_type): Start by resetting
32825 candidate type to it's main variant.
32827 2017-03-10 Jakub Jelinek <jakub@redhat.com>
32829 PR rtl-optimization/79909
32830 * combine.c (try_combine): Use simplify_replace_rtx on individual
32831 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
32832 of the whole CALL_INSN_FUNCTION_USAGE.
32834 PR tree-optimization/79972
32835 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
32836 get_range_info on SSA_NAMEs. Formatting fixes.
32838 2017-03-10 Richard Biener <rguenther@suse.de>
32839 Jakub Jelinek <jakub@redhat.com>
32841 PR tree-optimization/77975
32842 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
32843 edge to be constant.
32844 (get_val_for): For constant x return it. Formatting fix.
32845 (loop_niter_by_eval): Avoid pointless looping if the next iteration
32846 would use the same bases as the current one.
32848 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32850 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
32851 instead of vec_select for V1TImode.
32852 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
32854 (VSX_LE_128): Add V1TI to this mode iterator.
32855 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
32856 (*vsx_le_perm_store_<mode>): Likewise.
32857 (pre-reload splitter for VSX stores): Likewise.
32858 (post-reload splitter for VSX stores): Likewise.
32859 (*vsx_xxpermdi2_le_<mode>): Likewise.
32860 (*vsx_lxvd2x2_le_<mode>): Likewise.
32861 (*vsx_stxvd2x2_le_<mode>): Likewise.
32863 2017-03-09 Michael Eager <eager@eagercon.com>
32865 Correct failures with --enable-checking=yes,rtl.
32867 * config/microblaze/microblaze.c (microblaze_expand_shift):
32868 Replace GET_CODE test with CONST_INT_P and INTVAL test with
32869 test for const0_rtx.
32870 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
32871 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
32873 2017-03-09 Richard Biener <rguenther@suse.de>
32875 PR tree-optimization/79977
32876 * graphite-scop-detection.c (scop_detection::merge_sese):
32877 Handle the case of extra exits to blocks dominating the entry.
32879 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
32881 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
32884 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
32886 PR rtl-optimization/79949
32887 * lra-constraints.c (process_alt_operands): Check memory when
32888 trying to predict a cycle. Print about the overall increase.
32890 2017-03-09 Richard Biener <rguenther@suse.de>
32892 PR middle-end/79971
32893 * gimple-expr.c (useless_type_conversion_p): Preserve
32894 TYPE_SATURATING for fixed-point types.
32896 2017-03-09 Richard Biener <rguenther@suse.de>
32899 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
32900 alignment of BLKmode params.
32902 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32905 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
32906 (VALL_NO_V2Q): Likewise.
32908 * config/aarch64/aarch64-simd.md
32909 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
32911 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
32912 VALL_NO_V2Q mode iterator.
32913 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
32915 2017-03-09 Martin Liska <mliska@suse.cz>
32917 PR tree-optimization/79631
32918 * tree-chkp-opt.c (chkp_is_constant_addr): Call
32919 tree_int_cst_sign_bit just for INTEGER constants.
32921 2017-03-09 Martin Liska <mliska@suse.cz>
32925 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
32928 2017-03-09 Marek Polacek <polacek@redhat.com>
32931 * tree.c (inchash::add_expr): Handle TREE_VEC.
32933 2017-03-09 Martin Liska <mliska@suse.cz>
32936 (chkp_narrow_size_and_offset): New function.
32937 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
32938 (void chkp_parse_bit_field_ref): New function.
32939 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
32940 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
32942 2017-03-09 Martin Liska <mliska@suse.cz>
32945 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
32946 (chkp_find_bounds_1): Remove gcc_unreachable.
32948 2017-03-09 Jakub Jelinek <jakub@redhat.com>
32951 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
32952 BUILT_IN_SYNC*, determine the access type from the size suffix and
32953 always build a MEM_REF with that type. Handle forgotten
32954 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
32957 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
32958 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
32959 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
32960 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
32961 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
32962 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
32963 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
32964 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
32965 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
32966 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
32967 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
32968 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
32969 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
32970 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
32971 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
32972 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
32973 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
32974 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
32975 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
32976 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
32977 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
32978 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
32979 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
32980 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
32981 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
32982 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
32983 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
32984 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
32985 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
32986 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
32987 definitions outside of __OPTIMIZE__ guarded section.
32990 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
32991 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
32992 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
32993 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
32996 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32998 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
32999 ("vfenez<mode>"): Add missing constraints.
33001 2017-03-08 Martin Sebor <msebor@redhat.com>
33004 * config/nds32/nds32.c (nds32_option_override):
33005 Fix misspelled diagnostic.
33007 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33010 * gimplify.c (gimplify_omp_for): Replace index var in outer
33011 taskloop statement with an artificial variable and add
33012 OMP_CLAUSE_PRIVATE clause for it.
33014 2017-03-08 Richard Biener <rguenther@suse.de>
33016 PR tree-optimization/79955
33017 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
33018 for accesses that are completely outside of the variable.
33020 2017-03-08 Andrew Haley <aph@redhat.com>
33022 PR tree-optimization/79943
33023 * tree-ssa-loop-split.c (compute_new_first_bound): When
33024 calculating the new upper bound, (END-BEG) should be added, not
33027 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33029 * config/avr/avr.md (setmemhi): Make sure match_dup
33030 operand number comes before match_scratch.
33032 2017-03-08 Richard Biener <rguenther@suse.de>
33034 PR tree-optimization/79920
33035 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
33036 with ncopies == 1 to ...
33037 (vect_transform_slp_perm_load): ... here. Properly compute
33038 all element loads by iterating VF times over the group. Do
33039 not handle ncopies (computed in a broken way) in
33040 vect_create_mask_and_perm.
33042 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33045 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
33046 is a uniform vector, use uniform_vector_p return value instead of
33047 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
33049 2017-03-07 Marek Polacek <polacek@redhat.com>
33051 PR middle-end/79809
33052 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
33053 (alloca_call_type): Likewise.
33055 2017-03-07 Martin Liska <mliska@suse.cz>
33057 * gcov.c (process_args): Put comment to correct location.
33059 2017-03-07 Martin Liska <mliska@suse.cz>
33061 PR middle-end/68270
33062 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
33063 Use array_at_struct_end_p instead of DECL_CHAIN (field).
33064 (chkp_narrow_bounds_for_field): Likewise.
33065 (chkp_parse_array_and_component_ref): Pass one more argument to
33068 2017-03-07 Richard Biener <rguenther@suse.de>
33070 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
33073 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
33075 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
33076 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
33078 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33081 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
33082 to end of description.
33083 (PARAM_MAX_STORES_TO_MERGE): Likewise.
33085 2017-03-07 Jakub Jelinek <jakub@redhat.com>
33087 PR rtl-optimization/79901
33088 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
33090 (*avx512f_<code><mode>3<mask_name>): ... this.
33091 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
33092 iterator instead of VI8_AVX2_AVX512BW.
33094 PR rtl-optimization/79901
33095 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
33096 min/max expander, expand it using expand_vec_cond_expr.
33099 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
33102 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33105 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
33106 to void * for PCH reasons.
33107 * dwarf2out.c (output_loc_operands, output_die): Cast
33108 v.val_vec.array to unsigned char *.
33110 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
33113 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
33116 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
33118 PR rtl-optimization/79571
33119 * lra-constraints.c (process_alt_operands): Calculate static
33120 reject and subtract it from overall when only addresses will be
33123 2017-03-06 Julia Koval <julia.koval@intel.com>
33126 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
33127 incoming stack boundary to 128 for 64-bit targets.
33129 2017-03-06 Richard Biener <rguenther@suse.de>
33131 PR tree-optimization/79894
33132 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
33133 to NULL after folding it.
33135 2017-03-06 Richard Biener <rguenther@suse.de>
33137 PR tree-optimization/79824
33138 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
33139 check disabling peeling for gaps.
33141 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
33143 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
33144 attributes): Document gettimeofday.
33146 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
33148 * config/s390/s390.c (s390_option_override_internal): Set
33149 PARAM_MIN_VECT_LOOP_BOUND
33151 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
33153 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
33154 * config/s390/s390.md: Likewise.
33156 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33159 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
33160 (<avx2_avx512>_perm<mode>): Rename to ...
33161 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
33163 (<avx512>_perm<mode>_mask): Rename to ...
33164 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
33166 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
33167 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
33168 instead of VI8F_256_512.
33169 (avx512f_perm<mode>): New define_expand.
33170 (avx512f_perm<mode>_mask): Likewise.
33171 (avx512f_perm<mode>_1<mask_name>): New define_insn.
33172 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
33174 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33176 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
33177 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
33179 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
33181 2017-03-06 Martin Liska <mliska@suse.cz>
33184 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
33185 when having a SSA NAME w/o VAR_DECL assigned to it.
33187 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33189 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
33190 msa_dpsub_<su>_d): Fix MODE for vec_select.
33192 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33194 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
33196 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
33198 2017-03-06 Richard Biener <rguenther@suse.de>
33200 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
33201 * plugin.c (register_plugin_info): Likewise.
33202 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
33204 2017-03-05 Jakub Jelinek <jakub@redhat.com>
33206 * config/i386/sse.md (sse_storehps, sse_storelps,
33207 avx_<castmode><avxsizesuffix>_<castmode>,
33208 avx512f_<castmode><avxsizesuffix>_<castmode>,
33209 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
33210 in condition that at least one operand is not a MEM.
33212 2017-03-03 Jakub Jelinek <jakub@redhat.com>
33214 PR middle-end/79805
33215 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
33216 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
33218 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
33219 gimple_call_nothrow_p flag based on whether original builtin can throw.
33220 If it can, emit following stmts on the fallthrough edge.
33221 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
33222 don't create new bb if inserting just debug stmts on the edge, try to
33223 insert them on the fallthru bb or just reset debug stmts.
33225 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
33228 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
33229 restore recog_data (including the operand rtxes inside it) around
33230 the call to get_insn_template.
33232 2017-03-03 Martin Sebor <msebor@redhat.com>
33234 PR tree-optimization/79699
33235 * context.c (context::~context): Free MPFR caches to avoid
33236 a memory leak on program exit.
33238 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33240 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
33241 Use wide_int::ulow () instead of .elt (0).
33243 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
33245 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
33246 (*pushxf): Limit oF constraint to 32bit targets and add oC
33247 constraint for 64bit targets.
33248 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
33249 (*pushdf): Change rmF constraint to rmC.
33251 2017-03-03 Martin Liska <mliska@suse.cz>
33253 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
33254 Remove unused variable.
33256 2017-03-03 Jakub Jelinek <jakub@redhat.com>
33259 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
33260 is a memory operand, increase num_memory.
33261 (ix86_expand_args_builtin): Likewise.
33263 2017-03-03 Jan Hubicka <jh@suse.cz>
33266 * ipa-devirt.c (maybe_record_node): Properly handle
33267 __cxa_pure_virtual visibility.
33269 2017-03-03 Martin Liska <mliska@suse.cz>
33271 PR tree-optimization/79803
33272 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
33274 (pass_loop_prefetch::execute): Disabled optimization if an
33275 assumption about L1 cache size is not met.
33277 2017-03-03 Martin Liska <mliska@suse.cz>
33279 PR rtl-optimization/79574
33280 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
33281 (hash_scan_set): Likewise.
33282 (dump_hash_table): Likewise.
33283 (hoist_code): Likewise.
33285 2017-03-03 Richard Biener <rguenther@suse.de>
33287 * fixed-value.c (fixed_from_string): Restore use of elt (1)
33288 in place of uhigh ().
33289 (fixed_convert_from_real): Likewise.
33291 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
33294 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
33296 2017-03-03 Richard Biener <rguenther@suse.de>
33298 PR middle-end/79818
33299 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
33300 TYPE_OVERFLOW_UNDEFINED check.
33302 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33304 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
33306 (vector_ae_<mode>_p): Likewise.
33307 (vector_nez_<mode>_p): Likewise.
33308 (vector_ne_v2di_p): Likewise.
33309 (vector_ae_v2di_p): Likewise.
33310 (vector_ne_<mode>_p): Likewise.
33311 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
33313 (vsx_tsqrt<mode>2_fe): Likewise.
33315 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
33318 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
33320 2017-03-02 Jakub Jelinek <jakub@redhat.com>
33322 PR rtl-optimization/79780
33323 * cprop.c (one_cprop_pass): When second and further conditional trap
33324 in a single basic block is turned into an unconditional trap, turn it
33325 into a deleted note to avoid RTL verification failures.
33327 2017-03-02 Richard Biener <rguenther@suse.de>
33329 * fold-const.c (const_binop): Use ulow () instead of elt (0).
33331 2017-03-02 Richard Biener <rguenther@suse.de>
33333 PR tree-optimization/79345
33335 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
33336 param and abort the walk, returning -1 if it is hit.
33337 (walk_aliased_vdefs): Take a limit param and pass it on.
33338 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
33339 defaulting to 0 and return a signed int.
33340 * tree-ssa-uninit.c (struct check_defs_data): New struct.
33341 (check_defs): New helper.
33342 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
33343 about uninitialized memory.
33344 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
33345 bogus uninitialized warning.
33346 (fixed_convert_from_real): Likewise.
33348 2017-03-02 Bin Cheng <bin.cheng@arm.com>
33350 PR tree-optimization/66768
33351 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
33352 iv_use if base object can't be determined.
33354 2017-03-02 Jakub Jelinek <jakub@redhat.com>
33356 PR tree-optimization/79345
33357 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
33358 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
33359 (get_pattern_stats): Initialize it.
33360 * genemit.c (gen_expand): Verify match_scratch numbers come after
33361 match_operand/match_dup numbers.
33362 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
33363 match_scratch numbers.
33364 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
33366 * config/s390/s390.md (trunctdsd2): Likewise.
33368 2017-03-02 Richard Biener <rguenther@suse.de>
33370 * wide-int.h (wide_int_storage::operator=): Implement in terms
33373 2017-03-02 Richard Biener <rguenther@suse.de>
33375 PR tree-optimization/79777
33376 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
33377 the to insert expression to sth existing.
33379 2017-03-01 Martin Sebor <msebor@redhat.com>
33381 PR middle-end/79692
33382 * gimple-ssa-sprintf.c
33383 (directive::known_width_and_precision): New function.
33384 (format_integer): Use it.
33385 (get_mpfr_format_length): Consider the full range of precision
33386 when computing %g output with the # flag. Set the likely byte
33387 count to 3 rather than 1 when precision is indeterminate.
33388 (format_floating): Correct the lower bound of precision.
33390 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33392 * doc/invoke.texi: Document default code model for 64-bit Linux.
33394 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33397 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
33398 udiv rather than div since input pattern is unsigned.
33400 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
33402 * config/i386/i386.c (print_reg): Warn for values of
33403 unsupported size in integer register.
33405 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
33408 * config/rs6000/predicates.md (current_file_function_operand): Do
33409 not allow self calls to be local if the function is replaceable.
33411 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
33414 * config/rs6000/altivec.h (vec_ctz and others): Change the
33415 preprocessor macro that controls conditional compilation from
33416 _ARCH_PWR9 to __POWER9_VECTOR__.
33417 (vec_all_ne): Change parameterization of __altivec_scalar_pred
33418 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
33419 control (instead of _ARCH_PWR9 control) so that template
33420 definition uses power9-specific function.
33421 (vec_any_eq): Likewise.
33422 (vec_all_ne): Change macro definition to use a power9-specific
33423 expansion under #ifdef __POWER9_VECTOR__ control (instead of
33424 _ARCH_PWR9 control).
33425 (vec_any_eq) Likewise.
33426 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
33427 expansion for CMPNEF to remove support for xvcmpnesp instruction.
33428 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
33429 support for xvcmpnedp instruction.
33430 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
33431 macro expansion so that Power9 implementation of vec_all_ne does
33432 not use the AltiVec predicate framework.
33433 (VCMPNEH_P): Likewise.
33434 (VCMPNEW_P): Likewise.
33435 (VCMPNED_P): Likewise.
33436 (VCMPNEFP_P): Likewise.
33437 (VCMPNEDP_P): Likewise.
33438 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
33439 implementation of vec_any_eq to not use AltiVec predicate
33441 (VCMPAEH_P): Likewise.
33442 (VCMPAEW_P): Likewise.
33443 (VCMPAED_P): Likewise.
33444 (VCMPAEFP_P): Likewise.
33445 (VCMPAEDP_P): Likewise.
33446 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
33447 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
33448 not use the AltiVec predicate framework.
33449 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
33450 of vec_any_eq to not use AltiVec predicate framework.
33451 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
33452 support for predefined __POWER9_VECTOR__ macro to indicate that
33453 Power9 instruction selection is enabled.
33454 (altivec_overloaded_builtins): Remove extraneous
33455 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
33456 function argument types RS6000_BTI_bool_V16QI and
33457 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
33458 entry for overloaded function argument types RS6000_BTI_bool_V4SI
33459 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
33460 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
33461 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
33462 Power9 for implementations of vec_cmpne. Change the signature for
33463 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
33464 (representing vec_all_ne) to remove the previously described first
33465 argument of type RS6000_BTI_INTSI, as this was an artifact of
33466 reliance on the AltiVec predicate framework, which is no longer
33467 used in the implementation of these functions. Add
33468 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
33469 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
33470 since, unlike the AltiVec predicate framework implementation, we
33471 do not share function descriptors between vec_alle and vec_anyeq.
33472 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
33473 set of modes that receive special treatment even when
33474 TARGET_P9_VECTOR is true. The special treatment emits code that
33475 does not depend on Power9 instructions.
33476 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
33477 define_expand to not rely on AltiVec predicate framework.
33478 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33480 (vector_ne_v2di_p): Change this define_expand to not rely on
33481 AltiVec predicate framework.
33482 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
33484 (vector_ne_<mode>_p): Change this define_expand to not rely on
33485 AltiVec predicate framework.
33486 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33488 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
33489 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
33490 define_insn pattern.
33491 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
33492 define_insn pattern because the xvcmpne<VSs>. instruction is not
33494 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
33495 instruction is not supported.
33497 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33499 * config/nvptx/nvptx.c: Include intl.h.
33501 2017-03-01 Martin Jambor <mjambor@suse.cz>
33504 * ipa-prop.h (ipa_bits): Removed field known.
33505 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
33506 to pointers. Adjusted their comments to warn about their sharing.
33507 (ipcp_transformation_summary): Change bits to a vector of pointers.
33508 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
33509 (ipa_get_ipa_bits_for_value): Declare.
33510 * tree-vrp.h (value_range): Mark as GTY((for_user)).
33511 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
33512 (ipa_bits_hash_table): Likewise.
33513 (ipa_vr_ggc_hash_traits): Likewise.
33514 (ipa_vr_hash_table): Likewise.
33515 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
33516 being pointers and vr_known being removed.
33517 (ipa_set_jf_unknown): Likewise.
33518 (ipa_get_ipa_bits_for_value): New function.
33519 (ipa_set_jfunc_bits): Likewise.
33520 (ipa_get_value_range): New overloaded functions.
33521 (ipa_set_jfunc_vr): Likewise.
33522 (ipa_compute_jump_functions_for_edge): Use the above functions to
33523 construct bits and vr parts of jump functions.
33524 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
33525 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33527 (ipcp_grow_transformations_if_necessary): Also allocate
33528 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33530 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
33531 them. Fix too long lines.
33532 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
33533 vr_known being removed.
33534 (ipa_read_jump_function): Use new setter functions to construct bits
33535 and vr parts of jump functions or set them to NULL.
33536 (write_ipcp_transformation_info): Adjust for bits being pointers.
33537 (read_ipcp_transformation_info): Likewise.
33538 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
33540 Include gt-ipa-prop.h.
33541 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
33543 (ipcp_store_bits_results): Likewise.
33544 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
33545 Do not write to existing jump functions but use a temporary instead.
33547 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33550 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
33551 attempt to use its first operand as BIT_FIELD_REF base.
33553 2017-03-01 Richard Biener <rguenther@suse.de>
33555 PR middle-end/79721
33556 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
33557 interpolating formula in wrapping arithmetic.
33558 (chrec_apply): Convert chrec_evaluate return value to wanted type.
33560 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33562 PR tree-optimization/79734
33563 * tree-vect-generic.c (expand_vector_condition): Optimize
33564 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
33565 Handle VEC_COND_EXPR where comparison has different inner width from
33566 type's inner width.
33568 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
33570 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
33571 markup, and similar issues. Remove @opindex entries for things
33572 that aren't options. Add missing -mmpy-option entries.
33574 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33576 PR tree-optimization/79737
33577 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
33578 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
33579 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
33580 instead of byte_size. Formatting fix.
33581 (shift_bytes_in_array_right): Formatting fix.
33583 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
33586 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
33587 condition on optimize for the leaf function test.
33589 2017-02-28 Martin Liska <mliska@suse.cz>
33592 * read-rtl-function.c (function_reader::handle_unknown_directive):
33593 Bail out when one uses -flto.
33595 2017-02-28 Martin Liska <mliska@suse.cz>
33597 * common.opt: Replace space with tabular for options of <number>
33599 * config/i386/i386.opt: Show <number> value for
33600 -mlarge-data-threshold.
33601 * opts.c (print_filtered_help): Do not display number in hexadecimal
33604 2017-02-28 Martin Liska <mliska@suse.cz>
33606 * common.opt: Fix --help=option -Q for options which are of
33609 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
33611 * config/i386/i386.c (print_reg): Error out for values
33612 of 8-bit size in invalid integer register.
33614 2017-02-28 Martin Sebor <msebor@redhat.com>
33616 PR tree-optimization/79691
33617 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
33619 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33622 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
33623 gcc_unreachable with output_operand_lossage.
33625 2017-02-28 Richard Biener <rguenther@suse.de>
33627 PR tree-optimization/79740
33628 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
33630 (visit_nary_op): Insert the nary into the hashtable if we
33631 pattern-matched sth.
33632 * tree-ssa-pre.c (eliminate_insert): Robustify.
33634 2017-02-28 Richard Biener <rguenther@suse.de>
33636 PR middle-end/79731
33637 * fold-const.c (decode_field_reference): Reject out-of-bound
33640 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33642 * config/i386/i386.c: Include intl.h.
33643 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
33644 instead of just cond ? "..." : "...".
33645 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
33646 * coverage.c (read_counts_file): Likewise.
33647 * omp-offload.c: Include intl.h.
33648 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
33649 of just cond ? "..." : "...".
33650 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
33651 of just cond ? "..." : "...".
33653 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
33656 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
33658 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
33659 'tune for' CPU name.
33660 * config/arm/arm-cpu-data.h: Regenerated.
33662 2017-02-28 Richard Biener <rguenther@suse.de>
33664 PR tree-optimization/79732
33665 * tree-inline.c (expand_call_inline): Do not shadow var.
33667 2017-02-28 Richard Biener <rguenther@suse.de>
33669 PR tree-optimization/79723
33670 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
33671 address-space properly.
33673 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
33675 * doc/optinfo.texi (Optimization groups): Fix option used for
33677 * doc/invoke.texi (-fopt-info): Document "omp".
33678 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
33679 (OPTGROUP_ALL): Add OPTGROUP_OMP.
33680 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
33681 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
33682 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
33684 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
33686 * dumpfile.c (optgroup_options): Instead of "openmp", associate
33687 OPTGROUP_OMP with "omp".
33689 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
33692 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
33693 for arithmetic shift of unsigned V2DI.
33695 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
33697 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
33698 arc/linux.h headers.
33699 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
33700 (LINK_SPEC): Likewise.
33701 (ARC_TLS_EXTRA_START_SPEC): Likewise.
33702 (EXTRA_SPECS): Likewise.
33703 (STARTFILE_SPEC): Likewise.
33704 (ENDFILE_SPEC): Likewise.
33705 (LIB_SPEC): Likewise.
33706 (TARGET_SDATA_DEFAULT): Likewise.
33707 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
33708 (MULTILIB_DEFAULTS): Likewise.
33709 (DWARF2_UNWIND_INFO): Likewise.
33710 * config/arc/big.h: New file.
33711 * config/arc/elf.h: Likewise.
33712 * config/arc/linux.h: Likewise.
33713 * config/arc/t-uClibc: Remove.
33715 2017-02-27 Bin Cheng <bin.cheng@arm.com>
33717 PR tree-optimization/77536
33718 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
33719 (tree_transform_and_unroll_loop): Use above function to compute the
33720 estimated niter of unrolled loop and use it when scaling profile.
33721 Also use count info rather than frequency if it's non-zero.
33722 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
33723 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
33724 (vect_transform_loop): Call above function.
33726 2017-02-27 Richard Biener <rguenther@suse.de>
33728 PR tree-optimization/45397
33729 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
33730 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
33731 (visit_nary_op): Add pattern matching for CSEing sign-changed
33732 or truncated operations with wider ones.
33734 2017-02-27 Richard Biener <rguenther@suse.de>
33736 PR tree-optimization/79690
33737 * tree-vect-stmts.c (vectorizable_store): Use vector type
33738 built from the DR with address-space.
33740 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
33742 * doc/invoke.texi (Optimize Options): Refine the description
33743 of asan-use-after-return.
33745 2017-02-25 Alan Modra <amodra@gmail.com>
33747 PR rtl-optimization/79584
33748 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
33749 base, not ad->base_term, the reg within base. Remove assertion
33750 that ad->base == ad->base_term. Replace gen_int_mode using
33751 bogus mode with const0_rtx.
33753 2017-02-25 Jakub Jelinek <jakub@redhat.com>
33755 PR middle-end/79396
33756 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
33757 FMA_EXPR like tcc_binary or tcc_unary.
33759 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
33762 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
33764 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
33765 (output_loc_operands): Handle DW_OP_call_ref and
33766 DW_OP_GNU_variable_value.
33767 (struct variable_value_struct): New type.
33768 (struct variable_value_hasher): Likewise.
33769 (variable_value_hash): New variable.
33770 (string_types): Remove.
33771 (copy_loc_descr): New function.
33772 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
33773 (prepend_loc_descr_to_each): New function.
33774 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
33775 instead of add_loc_descr_to_each if the first argument is single
33776 location list and the second has multiple.
33777 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
33778 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
33779 when looking for variable value which doesn't have other location info.
33780 (loc_list_from_tree): Formatting fix.
33781 (gen_array_type_die): Simplify DW_AT_string_length handling.
33782 (adjust_string_types): Remove.
33783 (gen_subprogram_die): Don't call adjust_string_types nor test/set
33784 string_types. Call resolve_variable_values.
33785 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
33786 (resolve_addr_in_expr): Likewise. Add A argument.
33787 (copy_deref_exprloc): Remove deref argument. Adjust for the
33788 original expression being DW_OP_GNU_variable_value with optionally
33789 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
33790 optionally after it.
33791 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
33792 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
33793 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
33794 (variable_value_hasher::hash, variable_value_hasher::equal): New
33796 (resolve_variable_value_in_expr, resolve_variable_value,
33797 resolve_variable_values, note_variable_value_in_expr,
33798 note_variable_value): New functions.
33799 (dwarf2out_early_finish): Call note_variable_value on all toplevel
33802 2017-02-24 Jakub Jelinek <jakub@redhat.com>
33805 * opts.h (handle_generated_option): Add GENERATED_P argument.
33806 * opts-common.c (handle_option): Adjust function comment.
33807 (handle_generated_option): Add GENERATED_P argument, pass it to
33809 (control_warning_option): Pass false to handle_generated_option
33811 * opts.c (maybe_default_option): Pass true to handle_generated_option
33813 * optc-gen.awk: Likewise.
33815 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33817 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
33818 a REG, look at the REG it is a SUBREG of.
33819 (splitter for cmpeqsi_t): Ditto.
33821 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33823 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
33824 the special USEs with the pattern of the insn, not the insn itself.
33826 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
33829 * doc/invoke.texi: Document -mload-store-pairs.
33831 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33832 Sandra Loosemore <sandra@codesourcery.com>
33834 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
33835 argument isn't a CONST_INT.
33836 (nios2_alternate_compare_const): Assert op is a CONST_INT.
33837 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
33838 (nios2_validate_compare): Bypass alternate compare logic if *op2
33839 is not a CONST_INT.
33840 (ldstwm_operation_p): Return false if first_base is not a REG or
33841 if first_offset is not a CONST_INT.
33843 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33845 * config/cris/cris.md: Use correct operand in a define_peephole2.
33847 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33849 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
33851 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33853 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
33854 this_insn if it is an INSN or JUMP_INSN.
33855 (force_offsettable): Look at base, not at addr.
33856 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
33857 on things that aren't necessarily CONST_INTs.
33859 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
33861 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
33862 -mfpmath=sse is the default also for x86-32 targets with SSE2
33863 instruction set when @option{-ffast-math} is enabled
33865 2017-02-24 Jeff Law <law@redhat.com>
33867 PR rtl-optimizatoin/79286
33868 * ira.c (update_equiv_regs): Drop may_trap_p exception to
33871 2017-02-24 Richard Biener <rguenther@suse.de>
33873 PR tree-optimization/79389
33874 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
33877 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
33879 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
33880 function comment to reflect reality.
33881 (loop_exits_before_overflow): Fix typo in function description.
33883 2017-02-24 Richard Biener <rguenther@suse.de>
33885 PR tree-optimization/79389
33886 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
33887 properly that a threading opportunity exists. Detect conditional
33888 copy/constant propagation opportunities.
33890 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
33892 * config/visium/visium.md (type): Add trap.
33893 (b): New mode attribute.
33894 (*btst): Rename into...
33895 (*btst<mode>): ...this and adjust.
33896 (*cbranchsi4_btst_insn): Rename into...
33897 (*cbranch<mode>4_btst_insn): ...this and adjust.
33898 (trap): New define_insn.
33900 2017-02-23 Jakub Jelinek <jakub@redhat.com>
33902 PR tree-optimization/79389
33903 * ifcvt.c (struct noce_if_info): Add rev_cond field.
33904 (noce_reversed_cond_code): New function.
33905 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
33906 reversed_comparison_code. Formatting fix.
33907 (noce_try_store_flag): Test rev_cond != NULL in addition to
33908 reversed_comparison_code.
33909 (noce_try_store_flag_constants): Likewise.
33910 (noce_try_store_flag_mask): Likewise.
33911 (noce_try_addcc): Use rev_cond if non-NULL instead of
33912 reversed_comparison_code.
33913 (noce_try_cmove_arith): Likewise. Formatting fixes.
33914 (noce_try_minmax, noce_try_abs): Clear rev_cond.
33915 (noce_find_if_block): Initialize rev_cond.
33916 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
33917 instead of false as last argument never attempt to reverse it
33920 2017-02-23 Bin Cheng <bin.cheng@arm.com>
33922 PR tree-optimization/79663
33923 * tree-predcom.c (combine_chains): Process refs in reverse order
33924 only for ZERO length chains, and add explaining comment.
33926 2017-02-23 Jeff Law <law@redhat.com>
33928 PR tree-optimization/79578
33929 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
33930 in call to operand_equal_p.
33932 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
33935 * config/i386/cpuid.h: Fix another undefined behavior.
33937 2017-02-23 Richard Biener <rguenther@suse.de>
33939 PR tree-optimization/79683
33940 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
33941 vector types for data-refs.
33943 2017-02-23 Martin Liska <mliska@suse.cz>
33945 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
33947 2017-02-23 Jakub Jelinek <jakub@redhat.com>
33949 PR middle-end/79665
33950 * internal-fn.c (get_range_pos_neg): Moved to ...
33951 * tree.c (get_range_pos_neg): ... here. No longer static.
33952 * tree.h (get_range_pos_neg): New prototype.
33953 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
33954 are known to be in between 0 and signed maximum inclusive, try to
33955 expand both unsigned and signed divmod and use the cheaper one from
33958 2017-02-22 Jeff Law <law@redhat.com>
33960 PR tree-optimization/79578
33961 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
33962 to compare base operands.
33964 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
33967 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
33968 gpc_reg_operand instead of fpr_reg_operand.
33970 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
33972 * config/mips/mips.c (mips_return_in_memory): Force FP
33973 vector types to be returned in memory for o32 ABI.
33975 2017-02-22 Jakub Jelinek <jakub@redhat.com>
33977 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
33978 instead of DW_TAG_member for static data member declarations and don't
33979 set no_linkage_name for static inline data members.
33980 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
33983 2017-02-22 Martin Liska <mliska@suse.cz>
33985 * doc/invoke.texi: Replace inequality signs with square brackets
33988 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
33991 * lra-constraints.c (simplify_operand_subreg): Handle
33992 WORD_REGISTER_OPERATIONS targets.
33994 2017-02-22 Jakub Jelinek <jakub@redhat.com>
33997 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
33998 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
33999 elimination by swapping fld*.
34001 2017-02-22 Richard Biener <rguenther@suse.de>
34003 PR tree-optimization/79673
34004 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
34005 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
34006 irrelevant address-space qualifiers and avoiding a
34007 ADDR_SPACE_CONVERT_EXPR from fold_convert.
34009 2017-02-22 Richard Biener <rguenther@suse.de>
34011 PR tree-optimization/79666
34012 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
34013 to not symbolically negate if that may introduce undefined
34016 2017-02-22 Martin Liska <mliska@suse.cz>
34019 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
34020 * data-streamer-out.c (streamer_write_gcov_count_stream):
34022 * value-prof.c (stream_out_histogram_value): Make assert more
34023 precise based on type of counter.
34025 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
34028 * config/i386/i386.md (standard_x87sse_constant_load splitter):
34029 Use nonimmediate_operand instead of memory_operand for operand 1.
34030 (float-extend standard_x87sse_constant_load splitter): Ditto.
34032 2017-02-21 Jeff Law <law@redhat.com>
34034 PR tree-optimization/79621
34035 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
34036 blocks with edges to themselves.
34038 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34041 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
34042 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
34043 Use gimple_call_builtin_p.
34046 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
34047 on temporarily removed DEBUG_INSNs.
34049 PR tree-optimization/79649
34050 * tree-loop-distribution.c (classify_partition): Give up on
34051 non-generic address space loads/stores.
34053 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
34055 * doc/loop.texi (Loop manipulation): Remove nonexistent
34056 tree_ssa_loop_version from the documentation.
34057 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
34059 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34062 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
34063 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
34064 * config/rs6000/rs6000.c: Include except.h.
34065 (rs6000_expand_split_stack_prologue): Call
34066 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
34068 2017-02-21 Martin Jambor <mjambor@suse.cz>
34071 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
34072 have been analyzed.
34074 2017-02-21 Martin Jambor <mjambor@suse.cz>
34076 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
34077 for backward compatibility only.
34078 * doc/invoke.texi (Option Summary): Remove all references to
34079 -fipa-cp-alignment.
34081 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
34085 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34087 * lra-constraints.c (curr_insn_transform): Handle
34088 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34090 2017-02-21 Martin Liska <mliska@suse.cz>
34092 * config/i386/i386.opt: Replace -masm-dialect with -masm.
34094 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
34096 PR translation/79638
34097 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
34099 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
34102 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
34103 (arm_function_ok_for_sibcall): Return false for an indirect call by
34104 descriptor if all the argument registers are used.
34105 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
34106 alignment of the function.
34108 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34110 PR tree-optimization/61441
34111 * simplify-rtx.c (simplify_const_unary_operation): For
34112 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
34113 the sNaN unmodified.
34115 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34117 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34118 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34119 instead of SYSTEM_HEADER_DIR.
34121 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
34122 Martin Liška <mliska@suse.cz>
34124 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
34125 Fix typos and grammar, use active voice, and clarify.
34127 2017-02-20 Marek Polacek <polacek@redhat.com>
34129 PR middle-end/79537
34130 * gimplify.c (gimplify_expr): Handle unused *&&L;.
34133 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
34135 2017-02-20 Jakub Jelinek <jakub@redhat.com>
34138 * config/i386/i386.c (ix86_expand_builtin): Handle
34139 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
34140 ix86_builtins_isa[fcode].isa as a requirement of those
34141 flags and any other flag in the bitmask.
34142 (ix86_init_mmx_sse_builtins): Use 0 instead of
34143 ~OPTION_MASK_ISA_64BIT as mask.
34144 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
34145 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
34146 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
34147 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
34149 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34152 * lra-constraints.c (split_reg): Check requested split mode
34153 is supported by the register.
34155 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34157 * lra-constraints.c (simplify_operand_subreg): Remove early
34160 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34163 * lra-constraints.c (curr_insn_transform): Tighten condition
34164 for converting SUBREG reloads from OP_OUT to OP_INOUT.
34166 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34169 * lra-constraints.c (curr_insn_transform): Handle
34170 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34172 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
34175 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
34177 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
34179 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
34182 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
34185 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34187 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
34190 2017-02-18 Jakub Jelinek <jakub@redhat.com>
34192 * final.c (last_columnnum, override_columnnum): New variables.
34193 (final_start_function): Set last_columnnum, pass it to begin_prologue
34194 hook and pass 0 to dwarf2out_begin_prologue.
34195 (final_scan_insn): Update override_columnnum. Pass last_columnnum
34196 to source_line debug hook.
34197 (notice_source_line): Compute last_columnnum and for debug_column_info
34198 return true on column changes.
34199 * debug.h (struct gcc_debug_hooks): Add column argument to
34200 source_line and begin_prologue hooks.
34201 (debug_nothing_int_charstar_int_bool): Remove prototype.
34202 (debug_nothing_int_int_charstar,
34203 debug_nothing_int_int_charstar_int_bool): New prototypes.
34204 (dwarf2out_begin_prologue): Add column argument.
34205 * debug.c (do_nothing_debug_hooks): Adjust source_line and
34206 begin_prologue hooks.
34207 (debug_nothing_int_charstar_int_bool): Remove.
34208 (debug_nothing_int_int_charstar,
34209 debug_nothing_int_int_charstar_int_bool): New functions.
34210 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
34211 through to dwarf2out_source_line.
34212 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
34213 (dwarf2out_source_line): Add column argument, emit it if requested.
34214 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
34216 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
34217 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
34218 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
34219 through to dwarf2out_begin_prologue.
34220 (vmsdbgout_source_line): Add column argument, pass it through to
34221 dwarf2out_source_line.
34222 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
34223 dbxout_source_line caller.
34224 (dbxout_source_line): Add column argument.
34226 * common.opt (gno-column-info, gcolumn-info): New options.
34227 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
34228 (check_die): Also test for multiple DW_AT_decl_column attributes.
34229 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
34230 DW_AT_decl_column if requested.
34231 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
34233 (gen_variable_die): Likewise.
34234 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
34235 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
34238 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
34239 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
34240 (ix86_handle_option): Handle OPT_m3dnowa.
34241 * doc/invoke.texi (-m3dnowa): Document.
34242 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
34243 -m3dnowa instead of -m3dnow -march=athlon.
34246 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
34247 instead of gcc_assert for K, r and R code checks. Formatting fixes.
34249 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34252 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
34253 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
34254 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
34255 generator for vsx_xxpermdi_<mode>_be.
34256 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
34257 force big-endian semantics.
34258 (vsx_xxpermdi_<mode>_be): New define_expand with same
34259 implementation as previous version of vsx_xxpermdi_<mode>.
34261 2017-02-17 Jakub Jelinek <jakub@redhat.com>
34263 PR tree-optimization/79327
34264 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
34265 variable, its initialization and use.
34267 2017-02-17 Julia Koval <julia.koval@intel.com>
34269 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
34270 (OPTION_MASK_ISA_PKU_UNSET): New.
34271 (ix86_handle_option): Handle -mrdpid.
34272 * config/i386/cpuid.h (bit_RDPID): New.
34273 * config/i386/driver-i386.c (host_detect_local_cpu):
34274 Detect RDPID feature.
34275 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
34276 * config/i386/i386-c.c (ix86_target_macros_internal):
34278 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
34279 (ix86_valid_target_attribute_inner_p): Add "rdpid".
34280 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
34281 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
34282 * config/i386/i386.md (define_insn "rdpid"): New.
34283 * config/i386/i386.opt Add -mrdpid.
34284 * config/i386/immintrin.h (_rdpid_u32): New.
34286 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
34288 PR rtl-optimization/79541
34289 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
34290 instead of transforming it into USE.
34292 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
34294 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
34295 If HONOR_SNANS (SFmode) force the input to a register.
34296 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
34297 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
34298 an frsp or similar insn.
34300 2017-02-17 Martin Liska <mliska@suse.cz>
34302 PR rtl-optimization/79577
34303 * params.def (selsched-max-sched-times): Increase minimum to 1.
34305 2017-02-17 Martin Liska <mliska@suse.cz>
34307 PR rtl-optimization/79574
34308 * gcse.c (want_to_gcse_p): Prevent integer overflow.
34310 2017-02-17 Martin Liska <mliska@suse.cz>
34312 PR tree-optimization/79529
34313 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
34314 ssa_defined_default_def_p to handle cases which are implicitly
34316 * tree-ssa.c (ssa_defined_default_def_p): New function.
34317 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
34318 which are implicitly defined.
34319 * tree-ssa.h (ssa_defined_default_def_p): Declare.
34321 2017-02-17 Richard Biener <rguenther@suse.de>
34323 PR middle-end/79576
34324 * params.def (max-ssa-name-query-depth): Limit to 10.
34326 2017-02-17 Richard Biener <rguenther@suse.de>
34328 PR tree-optimization/79552
34329 * tree-ssa-structalias.c (visit_loadstore): Properly verify
34332 2017-02-17 Richard Biener <rguenther@suse.de>
34335 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
34337 2017-02-17 Marek Polacek <polacek@redhat.com>
34339 PR middle-end/79536
34340 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
34341 (fold_negate_expr): New wrapper.
34343 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
34345 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
34346 Correct terminology and de-emphasize pre-standard behavior.
34348 2017-02-16 Alan Modra <amodra@gmail.com>
34350 PR rtl-optimization/79286
34351 * ira.c (def_dominates_uses): New function.
34352 (update_equiv_regs): Don't create an equivalence for insns that
34353 may trap where the register def does not dominate the use.
34355 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
34357 PR rtl-optimization/78127
34358 * lra.c (lra): Call lra_eliminate before finish the loop after
34361 2017-02-16 Richard Biener <rguenther@suse.de>
34363 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
34365 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
34366 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
34367 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
34368 (isl_val_int_from_wi): New function.
34369 (extract_affine_gmp): Rename to ...
34370 (extract_affine_wi): ... this, take a widest_int.
34371 (extract_affine_int): Just wrap extract_affine_wi.
34372 (add_param_constraints): Use isl_val_int_from_wi.
34373 (add_loop_constraints): Likewise, and extract_affine_wi.
34375 2017-02-15 Jeff Law <law@redhat.com>
34377 PR middle-end/79521
34378 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
34379 ira_init_register_move_cost_if_necessary.
34381 2017-02-15 Martin Sebor <msebor@redhat.com>
34383 PR middle-end/32003
34384 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
34385 removed in a prior commit.
34387 2017-02-15 Bin Cheng <bin.cheng@arm.com>
34389 PR tree-optimization/79347
34390 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
34391 counters during peeling.
34393 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
34395 * Makefile.in (site.exp): Remove "set ISLVER".
34397 2017-02-15 Jakub Jelinek <jakub@redhat.com>
34400 * real.c (real_from_integer): Call real_convert even for decimal.
34402 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
34405 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
34407 2017-02-14 Andrew Pinski <apinski@cavium.com>
34409 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
34410 cores and change the partno/implementer to be correct.
34411 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
34412 the 'B" as the implementer.
34413 * config/aarch64/aarch64-tune.md: Regenerate.
34415 2017-02-14 Carl Love <cel@us.ibm.com>
34417 * config/rs6000/rs6000.c: Add case statement entry to make the
34418 xvcvuxdsp built-in argument unsigned.
34419 * config/rs6000/vsx.md: Fix the source and return operand types so they
34420 match the instruction definitions from the ISA document. Fix typo
34421 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
34424 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
34427 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
34428 member early_clobber_alts.
34429 * lra-lives.c (reg_early_clobber_p): New.
34430 (process_bb_lives): Use it.
34431 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
34432 (debug_operand_data): Initialize early_clobber_alts.
34433 (setup_operand_alternative): Set up early_clobber_alts.
34434 (collect_non_operand_hard_regs): Ditto. Pass early clobber
34435 alternatives to new_insn_reg.
34436 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
34438 (lra_update_insn_regno_info): Pass the new arg.
34440 2017-02-14 Jakub Jelinek <jakub@redhat.com>
34442 PR middle-end/79505
34443 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
34444 (new_oacc_loop_raw): Don't clear already cleared fields.
34447 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
34448 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
34449 _mm512_prefetch_i64gather_ps): New inline functions and macros.
34451 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
34454 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
34456 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
34459 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
34460 the extra instruction to the right place to store 128-bit constant
34463 2017-02-14 Martin Sebor <msebor@redhat.com>
34465 PR middle-end/79448
34466 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
34467 warning for strings of unknown length.
34469 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
34471 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
34473 2017-02-14 Jeff Law <law@redhat.com>
34476 * ira-costs.c (scan_one_insn): Initialize register move costs
34477 for pseudos seen in USE/CLOBBER insns.
34479 PR tree-optimization/79095
34480 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
34481 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
34482 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
34483 if the operands are known to be not equal, then the resulting range
34485 (intersect_ranges): If the new range is ~[0,0] and the old range is
34486 wide, then prefer ~[0,0].
34487 * tree-vrp.c (overflow_comparison_p_1): New function.
34488 (overflow_comparison_p): New function.
34489 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
34490 if NAME is used in an overflow test.
34491 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
34492 overflow check that can be expressed as an equality test, then adjust
34493 ops to be that equality test.
34495 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34497 * config/s390/s390-builtin-types.def: Remove flags argument.
34498 * config/s390/s390.c (s390_init_builtins): Likewise.
34500 2017-02-14 Martin Liska <mliska@suse.cz>
34502 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
34503 vector. Fix trailing white spaces.
34505 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
34507 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
34510 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34512 PR rtl-optimization/68664
34513 * config/arm/arm.c (arm_sched_can_speculate_insn):
34514 New function. Declare prototype.
34515 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34517 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34519 PR rtl-optimization/68664
34520 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
34522 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34524 2017-02-14 Amit Pawar <amit.pawar@amd.com>
34526 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
34527 max skip bytes for function, loop and jump.
34529 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34531 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
34532 ABS_EXPR for gimple dump.
34534 2017-02-14 Jakub Jelinek <jakub@redhat.com>
34537 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
34539 PR tree-optimization/79408
34540 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
34541 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
34542 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
34543 also if rhs1 is INTEGER_CST.
34545 2017-02-14 Richard Biener <rguenther@suse.de>
34547 PR middle-end/79432
34548 * tree-into-ssa.c (insert_phi_nodes): When the function can
34549 have abnormal edges rewrite SSA names with broken use-def
34550 dominance out of SSA and register them for PHI insertion.
34552 2017-02-13 Martin Sebor <msebor@redhat.com>
34554 PR middle-end/79496
34555 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
34556 clearing info.nowrite flag when snprintf size argument is a range.
34558 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34560 * cprop.c (cprop_jump): Add missing space in string literal.
34561 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
34562 (get_constraint_for_component_ref): Likewise.
34563 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
34564 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
34565 * lra-constraints.c (process_alt_operands): Likewise.
34566 * ipa-inline.c (inline_small_functions): Likewise.
34567 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
34568 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
34569 * trans-mem.c (diagnose_tm_1_op): Likewise.
34570 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
34571 (grid_parallel_clauses_gridifiable): Likewise.
34573 * config/nvptx/mkoffload.c (process): Add space in between
34576 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
34577 "MOD4_SSE_REGS" and "ALL_REGS".
34579 * spellcheck.c (test_data): Add , in between "foo" and "food".
34581 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34584 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
34585 boundary crossing check and subsequent code generation agree.
34587 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34589 * config/aarch64/aarch64.c (has_memory_op): Delete.
34590 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
34593 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34595 PR rtl-optimization/79388
34596 PR rtl-optimization/79450
34597 * combine.c (distribute_notes): When removing TEM_INSN for which
34598 corresponding dest has last value recorded, invalidate that last
34601 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34603 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
34604 of explicit '@'. Add missing assembly comment marker on branch costs
34607 2017-02-13 Nathan Sidwell <nathan@acm.org>
34609 * gengtype-lex.l (<in_struct>): Add '/'.
34611 2017-02-13 Martin Liska <mliska@suse.cz>
34614 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
34616 2017-02-13 Richard Biener <rguenther@suse.de>
34618 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
34620 * configure: Re-generate.
34621 * config.in: Likewise.
34622 * graphite-dependences.c: Simplify as if
34623 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
34624 * graphite-isl-ast-to-gimple.c: Likewise.
34625 * graphite-optimize-isl.c: Likewise.
34626 * graphite-poly.c: Likewise.
34627 * graphite-sese-to-poly.c: Likewise.
34628 * graphite.h: Likewise.
34629 * toplev.c: Include isl/version.h and use isl_version () for
34630 printing the ISL version.
34631 * doc/install.texi: Update ISL requirement.
34633 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
34635 * doc/standards.texi (Standards): Update reference to
34638 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
34640 * doc/extend.texi (Named Address Spaces): sourceware.org now
34642 * doc/install.texi (Binaries): Ditto.
34645 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
34647 * doc/cpp.texi: Replace "stringify"/"stringification" with C
34648 standard terminology "stringize"/"stringizing" throughout.
34649 * doc/cppinternals.texi: Likewise.
34651 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
34653 * doc/extend.texi: Fix some spelling mistakes and typos.
34654 * doc/invoke.texi: Likewise.
34656 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
34659 * params.def (inline-min-speedup) Change from 10 to 8.
34661 2017-02-11 Jakub Jelinek <jakub@redhat.com>
34663 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
34666 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
34669 * ipa-inline-analysis.c (get_minimal_bb): New function.
34670 (record_modified): Use it.
34671 (remap_edge_change_prob): Handle also ancestor functions.
34673 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
34675 * doc/contrib.texi (Contributors): Remove broken link into
34676 the Mauve CVS repository.
34678 2017-02-11 Jakub Jelinek <jakub@redhat.com>
34680 PR middle-end/79454
34681 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
34682 result computation whenever lhs doesn't have vector mode, not
34683 just when it has BLKmode.
34685 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
34687 * doc/makefile.texi (profiledbootstrap): Refer to the
34688 installation instructions only in textual form.
34690 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34693 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
34695 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
34697 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
34698 (Specific): Update mingw-w64 reference.
34700 (Specific): Remove broken link to Renesas RX processor.
34702 2017-02-10 Richard Biener <rguenther@suse.de>
34704 * toplev.c (process_options): Do not mention obsolete graphite
34705 options when printing sorry message about missing graphite support.
34706 Mention -floop-nest-optimize.
34708 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
34710 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
34711 (vtst_p16): Likewise.
34712 (vtstq_p8): Likewise.
34713 (vtstq_p16): Likewise.
34715 (vtstq_p64): Likewise.
34716 * config/arm/arm_neon.h (vgetq_lane_p64): New.
34717 (vset_lane_p64): New.
34718 (vsetq_lane_p64): New.
34720 2017-02-10 Jakub Jelinek <jakub@redhat.com>
34722 PR tree-optimization/79411
34723 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
34724 stmt operands are SSA_NAMEs used in abnormal phis.
34725 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
34728 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
34731 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
34734 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
34736 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
34738 2017-02-09 Jakub Jelinek <jakub@redhat.com>
34740 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
34744 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
34745 not arbitrary TREE_CONSTANT.
34748 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
34749 "omp declare target link" attribute unless is_global_var.
34750 * omp-offload.c (find_link_var_op): Likewise.
34752 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
34753 Chung-Lin Tang <cltang@codesourcery.com>
34755 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
34757 (gimplify_adjust_omp_clauses): Don't delete TILE.
34758 (gimplify_omp_for): Deal with TILE.
34759 * internal-fn.c (expand_GOACC_TILE): New function.
34760 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
34762 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
34763 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
34765 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
34766 avoid DIV for outermost collapse var.
34767 (expand_oacc_for): Insert tile element loop as needed. Adjust.
34768 Remove out of date comments, fix whitespace.
34769 * omp-general.c (omp_extract_for_data): Deal with tiling.
34770 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
34771 adjust OLF_DIM_BASE value.
34772 (struct omp_for_data): Add tiling field.
34773 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
34774 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
34775 for auto loops. Remove default auto determining, moved to
34776 oacc_loop_fixed_partitions.
34777 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
34778 stmts, add e_mask field.
34779 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
34780 (oacc_thread_numbers): Use oacc_dim_call.
34781 (oacc_xform_tile): New.
34782 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
34783 (finish_oacc_loop): Adjust for ifns vector.
34784 (oacc_loop_discover_walk): Append loop abstraction sites to list,
34785 add case for GOACC_TILE fns.
34786 (oacc_loop_xform_loop): Delete.
34787 (oacc_loop_process): Iterate over call list directly, and add
34788 handling for GOACC_TILE fns.
34789 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
34791 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
34792 vector partitioning to outer loops. Assign 2 partitions to loops
34793 when available. Add TILE handling.
34794 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
34795 (execite_oacc_device_lower): Process GOACC_TILE fns,
34796 ignore unknown specs.
34797 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
34798 * tree.c (omp_clause_num_ops): Adjust TILE ops.
34799 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
34801 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
34803 * configure.ac (ACX_BUGURL): Update.
34804 * configure: Regenerate.
34806 2017-02-09 Richard Biener <rguenther@suse.de>
34808 PR tree-optimization/69823
34809 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
34810 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
34812 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
34814 * config/arc/arc-c.def: Add __NPS400__ definition.
34815 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
34816 (TARGET_NPS400): Define.
34818 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
34820 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
34822 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
34823 pointer, arch_info.
34824 (arc_cpu_types): Fill the arch_info field with a pointer into the
34825 arc_arch_types table.
34826 (arc_selected_cpu): Declare.
34827 * config/arc/arc.c (arc_selected_cpu): Make global.
34828 (arc_selected_arch): Delete.
34829 (arc_base_cpu): Delete.
34830 (arc_override_options): Remove references to deleted variables,
34831 update access to arch information.
34832 (ARC_OPT): Update access to arch information.
34833 (ARC_OPTX): Likewise.
34834 * config/arc/arc.h (arc_base_cpu): Remove declaration.
34835 (TARGET_ARC600): Update access to arch information.
34836 (TARGET_ARC601): Likewise.
34837 (TARGET_ARC700): Likewise.
34838 (TARGET_EM): Likewise.
34839 (TARGET_HS): Likewise.
34840 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
34843 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
34846 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
34847 condition/operands for integer GE/LE/GEU/LEU operations.
34849 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
34851 PR translation/79397
34852 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
34855 2017-02-08 Martin Jambor <mjambor@suse.cz>
34858 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
34859 whether allocation happened.
34860 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
34861 nothing was allocated.
34863 2017-02-08 Jakub Jelinek <jakub@redhat.com>
34865 PR tree-optimization/79408
34866 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
34867 constant, but SSA_NAME with a known integer range, use the minimum
34868 of that range instead of op1 to determine if modulo can be replaced
34869 with its first operand.
34871 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34873 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
34875 2017-02-08 Richard Biener <rguenther@suse.de>
34877 PR tree-optimization/71824
34878 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
34879 Check all loops contained in the merged region.
34881 2017-02-07 Andrew Pinski <apinski@cavium.com>
34883 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
34885 2017-02-07 Andrew Pinski <apinski@cavium.com>
34887 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
34888 (thunderxt88): Likewise.
34889 (thunderxt81): Disable LSE and change v8.1 to v8.
34890 (thunderxt83): Likewise.
34892 2017-02-07 Jakub Jelinek <jakub@redhat.com>
34893 Richard Biener <rguenther@suse.de>
34895 PR middle-end/79399
34896 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
34897 type from int to size_t.
34898 * ira-costs.c (struct_costs_size): Change type from int to size_t.
34900 2017-02-07 Jakub Jelinek <jakub@redhat.com>
34902 PR rtl-optimization/79386
34903 * cprop.c (bypass_conditional_jumps): Initialize
34904 bypass_last_basic_block already before splitting bbs after
34905 unconditional traps...
34906 (bypass_conditional_jumps): ... rather than here.
34909 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
34910 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
34911 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
34912 fix -masm=intel patterns.
34914 2017-02-07 Richard Biener <rguenther@suse.de>
34916 PR tree-optimization/79256
34917 PR middle-end/79278
34918 * builtins.c (get_object_alignment_2): Use min_align_of_type
34919 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
34920 and ADJUST_FIELD_ALIGN.
34922 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
34924 * doc/tm.texi: Regenerate.
34925 * stor-layout.c (layout_decl): Adjust.
34926 (update_alignment_for_field): Likewise.
34927 (place_field): Likewise.
34928 (min_align_of_type): Likewise.
34929 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
34930 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
34931 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
34932 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
34933 * config/frv/frv.c (frv_adjust_field_align): Likewise.
34934 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
34935 * config/i386/i386.c (x86_field_alignment): Likewise.
34936 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
34937 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
34938 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
34939 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
34940 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
34941 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
34945 2017-01-30 Richard Biener <rguenther@suse.de>
34947 PR tree-optimization/79256
34948 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
34949 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
34952 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
34954 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
34955 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
34956 builtins to SImode and emit a zero-extend, if necessary.
34958 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
34960 * docs/invoke.texi (RISC-V Options): Alphabetize.
34962 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
34964 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
34967 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
34969 * config/riscv/riscv.c: New file.
34970 * common/config/riscv/riscv-common.c: Likewise.
34971 * config.gcc: Likewise.
34972 * config/riscv/constraints.md: Likewise.
34973 * config/riscv/elf.h: Likewise.
34974 * config/riscv/generic.md: Likewise.
34975 * config/riscv/linux.h: Likewise.
34976 * config/riscv/multilib-generator: Likewise.
34977 * config/riscv/peephole.md: Likewise.
34978 * config/riscv/pic.md: Likewise.
34979 * config/riscv/predicates.md: Likewise.
34980 * config/riscv/riscv-builtins.c: Likewise.
34981 * config/riscv/riscv-c.c: Likewise.
34982 * config/riscv/riscv-ftypes.def: Likewise.
34983 * config/riscv/riscv-modes.def: Likewise.
34984 * config/riscv/riscv-opts.h: Likewise.
34985 * config/riscv/riscv-protos.h: Likewise.
34986 * config/riscv/riscv.h: Likewise.
34987 * config/riscv/riscv.md: Likewise.
34988 * config/riscv/riscv.opt: Likewise.
34989 * config/riscv/sync.md: Likewise.
34990 * config/riscv/t-elf-multilib: Likewise.
34991 * config/riscv/t-linux: Likewise.
34992 * config/riscv/t-linux-multilib: Likewise.
34993 * config/riscv/t-riscv: Likewise.
34994 * configure.ac: Likewise.
34995 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
34996 Waterman as RISC-V maintainers.
34997 * doc/install.texi: Add RISC-V entries.
34998 * doc/invoke.texi: Add RISC-V options section.
34999 * doc/md.texi: Add RISC-V constraints section.
35000 * configure: Regenerated.
35002 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
35005 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
35006 false values to be constant vectors with all 0 or all 1 bits set.
35007 (vcondu<mode><mode>): Likewise.
35008 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
35010 (fpmask_comparison_operator): Update comment.
35011 (vecint_comparison_operator): New predicate.
35012 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
35013 vector conditionals when the true and false values are constant
35014 vectors with all 0 bits or all 1 bits set.
35016 2017-02-06 Martin Sebor <msebor@redhat.com>
35018 PR tree-optimization/79376
35019 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
35021 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
35023 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
35024 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
35025 to simplify split condition.
35027 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35029 * omp-expand.c (oxpand_omp_atomic_fetch_op,
35030 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
35033 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
35035 PR rtl-optimization/68664
35036 * target.def (can_speculate_insn): New hook.
35037 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
35038 * doc/tm.texi: Regenerate.
35039 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
35040 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
35041 (rs6000_sched_can_speculate_insn): New function.
35043 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35045 PR tree-optimization/79284
35046 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
35047 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
35048 vectorizable_mask_load_store, vectorizable_operation,
35049 vect_is_simple_cond, get_same_sized_vectype): Use it instead
35050 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
35051 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
35052 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
35053 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
35054 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
35055 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
35056 is_gimple_assign (stmt). Replace another such test with
35057 is_gimple_assign (stmt).
35059 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35062 * config/avr/avr.c (rtl-iter.h): Include it.
35063 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
35064 (avr_legitimate_combined_insn): ...and implementation.
35066 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
35068 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
35069 * config/s390/s390.c (s390_const_operand_ok)
35070 (s390_canonicalize_comparison, s390_extract_part)
35071 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
35072 (s390_contiguous_bitmask_p, s390_rtx_costs)
35073 (legitimize_pic_address): Likewise.
35074 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
35075 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
35076 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
35077 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
35078 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
35080 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35082 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
35083 REGNO($0) == REGNO($1).
35085 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35087 * config/s390/linux.h(SIZE_TYPE): Add comment.
35089 2017-02-06 Julian Brown <julian@codesourcery.com>
35090 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35091 Virendra Pathak <virendra.pathak@broadcom.com>
35093 * config/aarch64/aarch64-cores.def: Change the scheduler
35095 * config/aarch64/aarch64.md: Include thunderx2t99.md.
35096 * config/aarch64/thunderx2t99.md: New file.
35098 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35100 * doc/standards.texi (Go Language): Update link to language
35103 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35105 * tree-eh.c (lower_resx): Sanitize profile.
35106 (cleanup_empty_eh_move_lp): Likewise.
35108 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35111 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
35113 * cfgloopmanip.h (loop_version): Update prototype.
35114 * modulo-sched.c (sms_schedule): Update call of loop_version.
35115 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
35116 * tree-parloops.c (gen_parallel_loop): Likewise.
35117 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
35118 * tree-ssa-loop-split.c (split_loop): Likewise.
35119 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
35120 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
35122 2017-02-05 Martin Liska <mliska@suse.cz>
35125 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
35127 (print_operand_address): Initialize a struct to zero.
35129 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35131 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
35132 garbage collector only in textual form.
35134 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35136 * doc/extend.texi (x86 specific memory model extensions for
35137 transactional memory): Simplify a phrase.
35139 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
35142 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
35143 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
35144 (atomic_storedi_1): Likewise.
35146 2017-02-04 Jakub Jelinek <jakub@redhat.com>
35148 PR tree-optimization/79338
35149 * tree-parloops.c (gather_scalar_reductions): Don't call
35150 vect_analyze_loop_form for loop->inner before destroying loop's
35153 2017-02-03 Martin Sebor <msebor@redhat.com>
35155 PR tree-optimization/79327
35156 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
35157 when precision has resulted in leading zeros.
35158 (format_integer): Adjust the likely counter to assume an unknown
35159 argument that may be zero is non-zero.
35161 2017-02-03 Jason Merrill <jason@redhat.com>
35164 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
35165 avoid copying non-taken branch.
35167 2017-02-03 Jakub Jelinek <jakub@redhat.com>
35169 PR tree-optimization/79340
35170 * tree-vect-loop.c (vectorizable_reduction): Release
35171 vec_defs elements after safe_splicing them into other vectors.
35174 PR tree-optimization/79327
35175 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
35176 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
35178 (format_integer): Use wide_int_to_tree instead of build_int_cst
35179 + to_?hwi. If argmin is NULL, just set argmin and argmax to
35180 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
35181 of shortest and longest sequence.
35183 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
35185 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
35186 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
35188 2017-02-03 Walter Lee <walt@tilera.com>
35191 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
35192 after initial stackframe link reg save.
35193 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
35195 2017-02-03 Jakub Jelinek <jakub@redhat.com>
35198 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
35199 wu for stxssp alternative.
35201 2017-02-03 Martin Sebor <msebor@redhat.com>
35203 PR tree-optimization/79352
35204 * gimple-fold.c (get_range_strlen): Add argument.
35205 (get_range_strlen): Change return type to bool.
35206 (get_maxval_strlen): Pass in a dummy argument.
35207 * gimple-fold.h (get_range_strlen): Change return type to bool.
35208 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
35209 * tree.h (array_at_struct_end_p): Add argument.
35210 * tree.c (array_at_struct_end_p): Handle it.
35212 2017-02-03 Martin Liska <mliska@suse.cz>
35215 * multiple_target.c (create_dispatcher_calls): Redirect edge
35216 from a caller of a dispatcher.
35217 (expand_target_clones): Make the clones local.
35218 (ipa_target_clone): Do both target clones and resolvers.
35219 (ipa_dispatcher_calls): Remove the pass.
35220 (pass_dispatcher_calls::gate): Likewise.
35221 (make_pass_dispatcher_calls): Likewise.
35222 * passes.def (pass_target_clone): Put as very first IPA early
35225 2017-02-03 Martin Liska <mliska@suse.cz>
35227 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
35228 in case of a function with ifunc attribute.
35230 2017-02-03 Martin Liska <mliska@suse.cz>
35232 * cgraph.c (cgraph_node::dump): Dump function version info.
35233 * symtab.c (symtab_node::dump_base): Add missing new line.
35235 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
35237 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
35238 (ifcombine_ifandif): Use it.
35240 2017-02-03 Martin Liska <mliska@suse.cz>
35242 * doc/invoke.texi: Document default value for
35243 use-after-scope-direct-emission-threshold.
35245 2017-02-03 Martin Liska <mliska@suse.cz>
35247 PR tree-optimization/79339
35248 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
35249 (format_floating): Likewise.
35251 2017-02-03 Martin Liska <mliska@suse.cz>
35254 * ipa-prop.c (ipa_node_params_t::insert): Remove current
35256 (ipa_node_params_t::remove): Likewise.
35257 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
35258 initialization from removed ipa_node_params_t::insert.
35259 (ipa_node_params::~ipa_node_params): Move from removed
35260 ipa_node_params_t::release.
35261 * symbol-summary.h (symbol_summary::m_released): New member.
35262 Do not release a summary twice. Do not allow to call finalizer
35263 for types of a summary that live in GGC memory.
35265 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35267 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
35270 2017-02-02 Martin Sebor <msebor@redhat.com>
35272 PR middle-end/79275
35273 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
35274 (format_string): Tighten up the range of output for non-constant
35275 strings and correct the expected range for wide non-constant strings.
35277 2017-02-02 Martin Sebor <msebor@redhat.com>
35279 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
35281 PR middle-end/32003
35282 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
35284 (-fdump-tree-@var): Add to index and document how to come up
35285 with pass-specific option and dump file names.
35286 (-fdump-passes): Clarify where to look for output.
35288 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
35290 PR middle-end/77445
35291 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
35292 statistics of the analyzed path; allow threading for speed when
35293 any of BBs along the path are optimized for speed.
35295 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
35297 PR middle-end/78468
35298 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
35299 settings of the virtual registers.
35302 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
35304 * explow.c (get_dynamic_stack_size): Take known alignment of stack
35305 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
35308 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35310 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
35311 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
35313 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35315 * config/s390/s390.md: Add missing comments with the expanded
35317 * config/s390/vector.md: Likewise.
35318 * config/s390/vx-builtins.md: Likewise.
35320 2017-02-02 Jakub Jelinek <jakub@redhat.com>
35323 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
35324 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
35325 conditions on a single line.
35327 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35329 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
35330 __S390_VX__ to __VX__.
35332 2017-02-01 Andrew Pinski <apinski@cavium.com>
35334 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
35335 stmt_info to record_stmt_cost.
35336 (vect_get_known_peeling_cost): Pass stmt_info if known to
35338 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
35339 cpu_vector_cost field into
35340 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
35341 field into vec_int_stmt_cost and vec_fp_stmt_cost.
35342 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
35343 splitting of scalar_stmt_cost and vec_stmt_cost.
35344 (thunderx_vector_cost): Likewise.
35345 (cortexa57_vector_cost): LIkewise.
35346 (exynosm1_vector_cost): Likewise.
35347 (xgene1_vector_cost): Likewise.
35348 (thunderx2t99_vector_cost): Improve after the splitting of the two
35350 (aarch64_builtin_vectorization_cost): Update for the splitting of
35351 scalar_stmt_cost and vec_stmt_cost.
35353 2017-02-01 Torvald Riegel <triegel@redhat.com>
35354 Richard Henderson <rth@redhat.com>
35356 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
35357 conditional on existance of a fast atomic load.
35358 * optabs-query.c (can_atomic_load_p): New function.
35359 * optabs-query.h (can_atomic_load_p): Declare it.
35360 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
35361 no fast atomic load is available for the particular size of access.
35362 (expand_atomic_compare_and_swap): Likewise.
35363 (expand_atomic_load): Likewise.
35364 (expand_atomic_store): Likewise.
35365 (expand_atomic_fetch_op): Likewise.
35366 * testsuite/lib/target-supports.exp
35367 (check_effective_target_sync_int_128): Remove x86 because it provides
35368 no fast atomic load.
35369 (check_effective_target_sync_int_128_runtime): Likewise.
35371 2017-02-01 Richard Biener <rguenther@suse.de>
35373 * graphite.c: Include tree-vectorizer.h for find_loop_location.
35374 (graphite_transform_loops): Provide opt-info for optimized nests.
35375 * tree-parloop.c (parallelize_loops): Provide opt-info for
35376 parallelized loops.
35378 2017-02-01 Richard Biener <rguenther@suse.de>
35380 PR middle-end/79315
35381 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
35382 was not set before.
35384 2017-02-01 Richard Biener <rguenther@suse.de>
35386 PR tree-optimization/71824
35387 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
35388 Verify the loops are valid in the merged SESE region.
35389 (scop_detection::can_represent_loop_1): Check analyzing the
35390 evolution of the number of iterations in the region succeeds.
35392 2017-01-31 Ian Lance Taylor <iant@golang.org>
35394 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
35395 REG_ARGS_SIZE note to 32-bit push insns and call insn.
35397 2017-01-31 David Malcolm <dmalcolm@redhat.com>
35399 PR preprocessor/79210
35400 * input.c (get_substring_ranges_for_loc): Replace line_width
35401 assertion with error-handling.
35403 2017-01-31 Richard Biener <rguenther@suse.de>
35405 PR tree-optimization/77318
35406 * graphite-sese-to-poly.c (extract_affine): Fix assert.
35407 (create_pw_aff_from_tree): Take loop parameter.
35408 (add_condition_to_pbb): Pass loop of the condition to
35409 create_pw_aff_from_tree.
35411 2017-01-31 Jakub Jelinek <jakub@redhat.com>
35413 * config/s390/s390.c (s390_asan_shadow_offset): New function.
35414 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
35416 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
35420 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
35422 (convert_int_to_float128): Likewise.
35423 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
35424 (convert_int_to_float128): Likewise.
35425 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
35426 (UNSPEC_IEEE128_CONVERT): Likewise.
35427 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
35428 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
35429 Use local variables for IBM extended format.
35430 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
35431 (fix_trunc<mode>si2_fprs): Likewise.
35432 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
35433 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
35434 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
35435 to know that we can now have integers of all sizes in vector
35437 (fix<uns>_<mode>di2_hw): Likewise.
35438 (float<uns>_<mode>si2_hw): Likewise.
35439 (fix_<mode>si2_hw): Likewise.
35440 (fixuns_<mode>si2_hw): Likewise.
35441 (float<uns>_<mode>di2_hw): Likewise.
35442 (float_<mode>di2_hw): Likewise.
35443 (float_<mode>si2_hw): Likewise.
35444 (floatuns_<mode>di2_hw): Likewise.
35445 (floatuns_<mode>si2_hw): Likewise.
35446 (xscvqp<su>wz_<mode>): Delete, no longer used.
35447 (xscvqp<su>dz_<mode>): Likewise.
35448 (xscv<su>dqp_<mode>): Likewise.
35449 (ieee128_mfvsrd_64bit): Likewise.
35450 (ieee128_mfvsrd_32bit): Likewise.
35451 (ieee128_mfvsrwz): Likewise.
35452 (ieee128_mtvsrw): Likewise.
35453 (ieee128_mtvsrd_64bit): Likewise.
35454 (ieee128_mtvsrd_32bit): Likewise.
35456 2017-01-31 Martin Liska <mliska@suse.cz>
35459 * ipa-prop.c (ipa_free_all_node_params): Call release method
35460 instead of ~sumbol_summary to not to trigger double times
35463 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
35465 PR tree-optimization/71691
35466 * bitmap.h (class auto_bitmap): New.
35467 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
35468 is_maybe_undefined instead of ssa_undefined_value_p.
35470 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35472 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
35473 __S390_ARCH_LEVEL__ to __ARCH__.
35475 2017-01-31 Jakub Jelinek <jakub@redhat.com>
35477 PR tree-optimization/79267
35478 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
35479 if should_remove_lhs_p is true.
35481 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
35484 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
35485 (add_alignment_attribute): New.
35486 (base_type_die): Add alignment attribute.
35487 (subrange_type_die): Likewise.
35488 (modified_type_die): Likewise.
35489 (gen_array_type_die): Likewise.
35490 (gen_descr_array_type_die: Likewise.
35491 (gen_enumeration_type_die): Likewise.
35492 (gen_subprogram_die): Likewise.
35493 (gen_variable_die): Likewise.
35494 (gen_field_die): Likewise.
35495 (gen_ptr_to_mbr_type_die): Likewise.
35496 (gen_struct_or_union_type_die): Likewise.
35497 (gen_subroutine_type_die): Likewise.
35498 (gen_typedef_die): Likewise.
35499 (base_type_cmp): Compare alignment attribute.
35501 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35504 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
35505 (setb_unsigned) New pattern for setb with CCUNS.
35506 * config/rs6000/rs6000.c (expand_block_compare): Use a different
35507 subfc./subfe sequence to avoid overflow problems. Generate a
35508 shorter sequence with cmpld/setb for power9.
35509 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
35510 for generating subfc. instruction.
35511 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
35512 now uses this instruction.
35514 2017-01-30 Ian Lance Taylor <iant@google.com>
35517 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
35518 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
35520 2017-01-30 Martin Sebor <msebor@redhat.com>
35522 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
35523 Move constant to the right of a relational operator.
35524 (get_mpfr_format_length, format_character, format_string): Ditto.
35525 (should_warn_p, maybe_warn): Same.
35527 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
35529 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
35532 * asan.c (get_translation_unit_decl): Remove function.
35533 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
35535 2017-01-30 Martin Liska <mliska@suse.cz>
35537 PR gcov-profile/79259
35538 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
35539 -fprofile-generate.
35541 2017-01-30 Martin Liska <mliska@suse.cz>
35544 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
35545 Initialize variables with NULL value.
35547 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
35550 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
35552 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
35554 2017-01-30 Richard Biener <rguenther@suse.de>
35556 PR tree-optimization/79276
35557 * tree-vrp.c (process_assert_insertions): Properly adjust common
35558 when removing a duplicate.
35560 2017-01-30 Richard Biener <rguenther@suse.de>
35562 PR tree-optimization/79256
35563 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35564 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35566 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
35568 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
35571 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
35572 ("*r<noxa>sbg_<mode>_sll_bitmask")
35573 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
35574 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
35575 Use contiguous_bitmask_nowrap_operand.
35577 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35580 * config/rs6000/altivec.h (vec_xl): Revise #define.
35581 (vec_xst): Likewise.
35583 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
35585 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
35587 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
35589 PR rtl-optimization/79194
35590 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
35591 traps before call to bypass_conditional_jumps.
35593 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
35595 PR tree-optimization/71374
35596 * lra-constraints.c (check_conflict_input_operands): New.
35597 (match_reload): Use it.
35599 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
35602 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
35603 account to calculate conflict_set.
35605 2017-01-27 Bin Cheng <bin.cheng@arm.com>
35607 PR rtl-optimization/78559
35608 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
35609 other_insn in combine.
35611 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
35613 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
35614 uint16_type_node for BT_UINT16.
35616 2017-01-27 David Malcolm <dmalcolm@redhat.com>
35618 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
35619 "RTL Tests" to menu.
35620 (GIMPLE Tests): New node.
35621 (RTL Tests): New node.
35623 2017-01-27 Richard Biener <rguenther@suse.de>
35625 PR tree-optimization/79245
35626 * tree-loop-distribution.c (distribute_loop): Apply cost
35627 modeling also to detected patterns.
35629 2017-01-27 Richard Biener <rguenther@suse.de>
35631 PR tree-optimization/71433
35632 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
35633 (compare_assert_loc): New function.
35634 (process_assert_insertions): Sort and optimize assert locations
35635 to remove duplicates and push down identical assertions on
35636 edges to their destination block.
35638 2017-01-27 Richard Biener <rguenther@suse.de>
35640 PR tree-optimization/79244
35641 * tree-vrp.c (remove_range_assertions): Forcefully propagate
35642 out SSA names even if abnormal.
35644 2017-01-27 Jakub Jelinek <jakub@redhat.com>
35646 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
35647 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
35648 instead of MPFR_RNDN.
35650 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
35653 * arm.c (arm_option_override): Don't call build_target_option_node
35654 until after doing all option overrides.
35655 (arm_valid_target_attribute_tree): Likewise.
35657 2017-01-27 Martin Liska <mliska@suse.cz>
35659 * doc/invoke.texi (-fprofile-arcs): Document profiling support
35660 for {cd}tors and C++ {cd}tors.
35662 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
35664 * config/s390/s390.md ("*setmem_long_and")
35665 ("*setmem_long_and_31z"): Use zero_extend instead of and.
35667 2017-01-26 Martin Sebor <msebor@redhat.com>
35669 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
35672 2017-01-26 Martin Sebor <msebor@redhat.com>
35674 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
35675 HAVE_DFmode before using XFmode or DFmode.
35676 (parse_directive): Avoid using the z length modifier to avoid
35677 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
35679 PR middle-end/78703
35680 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
35681 to accept adjustment as an array.
35682 (get_int_range): New function.
35683 (struct directive): Make width and prec arrays.
35684 (directive::set_width, directive::set_precision): Call get_int_range.
35685 (format_integer, format_floating): Handle width and precision ranges.
35686 (format_string, parse_directive): Same.
35688 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35691 * dwarf2out.c (generate_skeleton_bottom_up): For children with
35692 comdat_type_p set, just clone them, but keep the children in the
35696 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
35697 which have direct callers with -fvar-tracking-assignments enabled
35699 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
35700 inside of type units.
35702 2017-01-26 Martin Sebor <msebor@redhat.com>
35704 PR middle-end/78703
35705 * gimple-ssa-sprintf.c (struct result_range): Add likely and
35707 (struct format_result): Replace number_chars, number_chars_min,
35708 and number_chars_max with a single member of struct result_range.
35710 (format_result::operator+=): Adjust.
35711 (struct fmtresult): Remove bounded. Handle likely and unlikely
35713 (fmtresult::adjust_for_width_or_precision): New function.
35714 (fmtresult:type_max_digits): New function.
35715 (bytes_remaining): Handle likely and unlikely counters.
35716 (min_bytes_remaining): Remove.
35717 (format_percent): Simplify.
35718 (format_integer, format_floating): Set likely and unlikely counters.
35719 (get_string_length, format_character, format_string): Same.
35720 (format_plain, should_warn_p): New function.
35721 (maybe_warn): Call should_warn_p. Update diagnostic messages
35722 and handle those for all directives, including plain strings.
35723 (format_directive): Handle likely and unlikely counters.
35724 Remove unnecessary quoting from diagnostics. Add an informational
35726 (add_bytes): Remove.
35727 (pass_sprintf_length::compute_format_length): Simplify.
35728 (try_substitute_return_value): Handle likely and unlikely counters.
35730 2017-01-26 Carl Love <cel@us.ibm.com>
35732 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
35733 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
35735 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
35738 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
35739 endianess for subregs into account.
35740 * lra-constraints.c (lra_constraints): Do risky transformations
35741 always on the first iteration.
35742 * lra-lives.c (check_pseudos_live_through_calls): Add arg
35743 last_call_used_reg_set.
35744 (process_bb_lives): Define and use last_call_used_reg_set.
35745 * lra.c (lra): Always continue after lra_constraints on the first
35748 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
35750 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
35752 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
35754 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35756 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
35757 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
35758 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
35759 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
35760 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
35761 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
35762 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
35763 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
35764 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
35766 2017-01-26 Marek Polacek <polacek@redhat.com>
35769 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
35770 for the third operand.
35772 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35774 PR middle-end/79236
35775 * omp-low.c (struct omp_context): Add simt_stmt field.
35776 (scan_omp_for): Return omp_context *.
35777 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
35778 context to the _simt_ SIMD stmt.
35779 (lower_omp_for): For combined SIMD with sibling _simt_
35780 SIMD, make sure to use the same decls in _looptemp_
35781 clauses as in the sibling.
35783 2017-01-26 David Sherwood <david.sherwood@arm.com>
35785 PR middle-end/79212
35786 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
35789 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35792 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
35793 emit fld b; fld a; if possible.
35795 * brig-builtins.def: Update copyright years.
35796 * config/arm/arm_acle_builtins.def: Update copyright years.
35798 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
35801 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
35802 constraint instead of o for the stxsd instruction.
35804 2017-01-25 Carl Love <cel@us.ibm.com>
35806 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
35807 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
35809 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
35811 * doc/invoke.texi (C++ Dialect Options): Fix typo.
35813 2017-01-25 Richard Biener <rguenther@suse.de>
35815 PR tree-optimization/69264
35816 * target.def (vector_alignment_reachable): Improve documentation.
35817 * doc/tm.texi: Regenerate.
35818 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
35820 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
35821 earlier changes with respect to TYPE_USER_ALIGN.
35822 (vector_alignment_reachable_p): Likewise. Improve dumping.
35824 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35827 * config/arm/arm.md (xordi3): Force constant operand into a register
35830 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35832 * doc/invoke.texi (-fstore-merging): Correct default optimization
35833 levels at which it is enabled.
35834 (-O): Move -fstore-merging from list to...
35837 2017-01-25 Richard Biener <rguenther@suse.de>
35840 * omp-expand.c: Include debug.h.
35841 (expand_omp_taskreg): Make sure to generate early debug before
35842 outlining anything from a function.
35843 (expand_omp_target): Likewise.
35844 (grid_expand_target_grid_body): Likewise.
35846 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
35849 * asan.c (get_translation_unit_decl): New function.
35850 (asan_add_global): Extract modules file name from globals
35851 TRANSLATION_UNIT_DECL name.
35853 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
35856 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
35857 for long calls with APCS frame and VFP.
35859 2017-01-24 David Malcolm <dmalcolm@redhat.com>
35861 * cfg.c (original_copy_tables_initialized_p): New function.
35862 * cfg.h (original_copy_tables_initialized_p): New decl.
35863 * cfgrtl.c (relink_block_chain): Guard the call to
35864 free_original_copy_tables with a call to
35865 original_copy_tables_initialized_p.
35866 * cgraph.h (symtab_node::native_rtl_p): New decl.
35867 * cgraphunit.c (symtab_node::native_rtl_p): New function.
35868 (symtab_node::needed_p): Don't assert for early assembly output
35869 for __RTL functions.
35870 (cgraph_node::finalize_function): Set "force_output" for __RTL
35872 (cgraph_node::analyze): Bail out early for __RTL functions.
35873 (analyze_functions): Update assertion to support __RTL functions.
35874 (cgraph_node::expand): Bail out early for __RTL functions.
35875 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
35877 * function.h (struct function): Update comment for field
35879 * gimple-expr.c: Include "tree-pass.h".
35880 (gimple_has_body_p): Return false for __RTL functions.
35881 * Makefile.in (OBJS): Add run-rtl-passes.o.
35882 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
35884 (gcc::pass_manager::get_clean_slate): New accessor.
35885 * passes.c: Include "insn-addr.h".
35886 (should_skip_pass_p): Add logging. Update logic for running
35887 "expand" to be compatible with both __GIMPLE and __RTL. Guard
35888 property-provider override so it is only done for gimple passes.
35890 (skip_pass): New function.
35891 (execute_one_pass): Call skip_pass when skipping passes.
35892 * read-md.c (md_reader::read_char): Support filtering
35893 the input to a subset of line numbers.
35894 (md_reader::md_reader): Initialize fields
35895 m_first_line and m_last_line.
35896 (md_reader::read_file_fragment): New function.
35897 * read-md.h (md_reader::read_file_fragment): New decl.
35898 (md_reader::m_first_line): New field.
35899 (md_reader::m_last_line): New field.
35900 * read-rtl-function.c (function_reader::create_function): Only
35901 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
35902 curr_properties. Set DECL_INITIAL to a dummy block.
35903 (read_rtl_function_body_from_file_range): New function.
35904 * read-rtl-function.h (read_rtl_function_body_from_file_range):
35906 * run-rtl-passes.c: New file.
35907 * run-rtl-passes.h: New file.
35909 2017-01-24 Jeff Law <law@redhat.com>
35911 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
35914 2017-01-24 Bin Cheng <bin.cheng@arm.com>
35916 PR tree-optimization/79159
35917 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
35918 (record_nonwrapping_iv): Improve boundary using above function if no
35919 value range information.
35921 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
35922 Martin Jambor <mjambor@suse.cz>
35924 * brig-builtins.def: New file.
35925 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
35926 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
35927 (DEF_HSAIL_SAT_BUILTIN): Likewise.
35928 (DEF_HSAIL_INTR_BUILTIN): Likewise.
35929 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
35930 * builtin-types.def (BT_INT8): New.
35931 (BT_INT16): Likewise.
35932 (BT_UINT8): Likewise.
35933 (BT_UINT16): Likewise.
35934 (BT_FN_ULONG): Likewise.
35935 (BT_FN_UINT_INT): Likewise.
35936 (BT_FN_UINT_ULONG): Likewise.
35937 (BT_FN_UINT_LONG): Likewise.
35938 (BT_FN_UINT_PTR): Likewise.
35939 (BT_FN_ULONG_PTR): Likewise.
35940 (BT_FN_INT8_FLOAT): Likewise.
35941 (BT_FN_INT16_FLOAT): Likewise.
35942 (BT_FN_UINT32_FLOAT): Likewise.
35943 (BT_FN_UINT16_FLOAT): Likewise.
35944 (BT_FN_UINT8_FLOAT): Likewise.
35945 (BT_FN_UINT64_FLOAT): Likewise.
35946 (BT_FN_UINT16_UINT32): Likewise.
35947 (BT_FN_UINT32_UINT16): Likewise.
35948 (BT_FN_UINT16_UINT16_UINT16): Likewise.
35949 (BT_FN_INT_PTR_INT): Likewise.
35950 (BT_FN_UINT_PTR_UINT): Likewise.
35951 (BT_FN_LONG_PTR_LONG): Likewise.
35952 (BT_FN_ULONG_PTR_ULONG): Likewise.
35953 (BT_FN_VOID_UINT64_UINT64): Likewise.
35954 (BT_FN_UINT8_UINT8_UINT8): Likewise.
35955 (BT_FN_INT8_INT8_INT8): Likewise.
35956 (BT_FN_INT16_INT16_INT16): Likewise.
35957 (BT_FN_INT_INT_INT): Likewise.
35958 (BT_FN_UINT_FLOAT_UINT): Likewise.
35959 (BT_FN_FLOAT_UINT_UINT): Likewise.
35960 (BT_FN_ULONG_UINT_UINT): Likewise.
35961 (BT_FN_ULONG_UINT_PTR): Likewise.
35962 (BT_FN_ULONG_ULONG_ULONG): Likewise.
35963 (BT_FN_UINT_UINT_UINT): Likewise.
35964 (BT_FN_VOID_UINT_PTR): Likewise.
35965 (BT_FN_UINT_UINT_PTR: Likewise.
35966 (BT_FN_UINT32_UINT64_PTR): Likewise.
35967 (BT_FN_INT_INT_UINT_UINT): Likewise.
35968 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
35969 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
35970 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
35971 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
35972 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
35973 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
35974 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
35975 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
35976 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
35977 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
35978 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
35979 * doc/frontends.texi: List BRIG FE.
35980 * doc/install.texi (Testing): Add BRIG tesring requirements.
35981 * doc/invoke.texi (Overall Options): Mention BRIG.
35982 * doc/standards.texi (Standards): Doucment BRIG HSA version.
35984 2017-01-24 Richard Biener <rguenther@suse.de>
35986 PR translation/79208
35987 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
35989 2017-01-24 Martin Jambor <mjambor@suse.cz>
35992 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
35993 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
35994 and known_contexts.
35996 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
35998 PR middle-end/79123
35999 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
36000 casts from signed to unsigned really don't have a range.
36002 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
36004 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
36005 GMP_RNDx for compatiblity.
36007 2017-01-24 Martin Liska <mliska@suse.cz>
36010 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
36011 that would prevent us to call alloca with -1 as argument.
36013 2017-01-24 Jakub Jelinek <jakub@redhat.com>
36015 * dwarf2out.c (output_compilation_unit_header, output_file_names):
36016 Avoid -Wformat-security warning.
36018 2017-01-23 Andrew Pinski <apinski@cavium.com>
36020 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
36023 2017-01-23 Martin Sebor <msebor@redhat.com>
36025 PR middle-end/78703
36026 * gimple-ssa-sprintf.c (warn_level): New global.
36027 (format_integer): Use it here and throughout the rest of the file.
36028 Use the same switch to compute sign as base.
36029 (maybe_warn): New function.
36030 (format_directive): Factor out warnings into maybe_warn.
36031 Add debugging output. Use warn_level.
36032 (add_bytes): Use warn_level.
36033 (pass_sprintf_length::compute_format_length): Add debugging output.
36034 (try_substitute_return_value): Same.
36035 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
36037 PR middle-end/78703
36038 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
36039 (struct fmtresult, format_integer, format_floating): Adjust.
36040 (fmtresult::fmtresult): Set max correctly in two argument ctor.
36041 (get_string_length, format_string,format_directive): Same.
36042 (pass_sprintf_length::compute_format_length): Same.
36043 (try_substitute_return_value): Simplify slightly.
36045 PR middle-end/78703
36046 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
36047 (fmtresult::operator+=): Outlined.
36048 (struct fmtresult): Add ctors.
36049 (struct conversion_spec): Rename...
36050 (struct directive): ...to this. Add and remove data members.
36051 (directive::set_width, directive::set_precision): New functions.
36052 (format_percent): Use fmtresult ctor.
36053 (get_width_and_precision): Remove.
36054 (format_integer): Make naming changes. Avoid computing width and
36056 (format_floating): Same. Adjust indentation.
36057 (format_character, format_none): New functions.
36058 (format_string): Moved character handling to format_character.
36059 (format_directive): Remove arguments, change return type.
36060 (parse_directive): New function.
36061 (pass_sprintf_length::compute_format_length): Move directive
36062 parsing to parse_directive.
36064 2017-01-23 Jakub Jelinek <jakub@redhat.com>
36066 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
36067 (assign_assembler_name_if_needed): ... this.
36068 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
36069 (assign_assembler_name_if_needed): ... this.
36070 (free_lang_data_in_cgraph): Adjust callers.
36071 * cgraphunit.c (cgraph_node::analyze): Likewise.
36072 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
36075 2017-01-23 Richard Biener <rguenther@suse.de>
36077 PR tree-optimization/79088
36078 PR tree-optimization/79188
36079 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
36080 resetting loop bounds after last path deletion. Reset loop
36081 bounds of the target loop, make code match the comments.
36082 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
36083 Make sure loops need no fixups.
36085 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
36087 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
36088 exponent support with double type for first argument.
36089 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
36090 type returned by __builtin_vec_extract_sig,
36091 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
36092 functions from "vector int" to "vector unsigned int" or from
36093 "vector long long int" to "vector unsigned long long int".
36094 Changed type returned by __builtin_vec_extract_exp,
36095 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
36096 functions from "vector int" to "vector unsigned int" or from
36097 "vector long long int" to "vector unsigned long long int".
36098 Changed return type of __builtin_vec_test_data_class,
36099 __builtin_vec_test_data_class_sp, and
36100 __builtin_vec_test_data_class_dp from "vector int" to
36101 "vector bool int" or from "vector long long int" to "vector bool
36102 long long int" and changed second argument type from "unsigned
36103 int" to "int". Added new overloaded function forms "vector float
36104 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
36105 "vector float __builtin_vec_insert_exp_sp (vector float, vector
36106 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
36107 double, vector unsigned long long int)" and "vector double
36108 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
36109 long int)". Changed return type of
36110 __builtin_scalar_test_data_class and
36111 __builtin_scalar_test_data_class_sp and
36112 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
36113 int" and changed second argument from "unsigned int" to "int".
36114 Changed type returned by __builtin_scalar_test_neg,
36115 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
36116 from "int" to "bool int". Added new overloaded function form
36117 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
36118 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
36119 exponent double-precision with floating point first argument.
36120 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
36121 documentation of scalar_test_data_class, scalar_test_neg,
36122 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
36123 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
36124 vec_test_data_class built-in functions to reflect refinements in
36125 their type signatures.
36127 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
36129 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
36131 (aarch64_elf_asm_destructor): Likewise.
36133 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
36135 PR rtl-optimization/78634
36136 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
36137 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
36138 * ifcvt.c (noce_try_cmove): Add missing cost check.
36140 PR rtl-optimization/71724
36141 * combine.c (if_then_else_cond): Look for situations where it is
36142 beneficial to undo the work of one of the recursive calls.
36144 2017-01-23 Bin Cheng <bin.cheng@arm.com>
36146 PR tree-optimization/70754
36147 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
36148 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
36149 combined stmt before it if not NULL.
36150 (combine_chains): Process refs reversely and compute dominance point
36153 2017-01-23 Martin Liska <mliska@suse.cz>
36155 PR tree-optimization/79196
36156 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
36157 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
36159 (strlen_optimize_stmt): Call the renamed function.
36161 2017-01-23 Michael Matz <matz@suse.de>
36163 PR tree-optimization/78384
36164 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
36166 2017-01-23 Richard Biener <rguenther@suse.de>
36168 PR tree-optimization/79186
36169 * tree-vrp.c (register_new_assert_for): Make sure we've seen
36170 both incoming edges before moving an assert.
36172 2017-01-23 Martin Jambor <mjambor@suse.cz>
36174 * ipa-prop.c (load_from_param_1): Removed.
36175 (load_from_unmodified_param): Bits from load_from_param_1 put back
36177 (load_from_param): Removed.
36178 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
36179 with stmt. Reverted back to use of load_from_unmodified_param.
36181 2017-01-23 Martin Jambor <mjambor@suse.cz>
36184 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
36185 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
36186 field a pointer to garbage collected vector, mark lattices and
36187 ipcp_orig_node with GTY((skip)).
36188 (ipa_get_param_count): Adjust to descriptors being a pointer.
36189 (ipa_get_param): Likewise.
36190 (ipa_get_type): Likewise.
36191 (ipa_get_param_move_cost): Likewise.
36192 (ipa_set_param_used): Likewise.
36193 (ipa_get_controlled_uses): Likewise.
36194 (ipa_set_controlled_uses): Likewise.
36195 (ipa_is_param_used): Likewise.
36196 (ipa_node_params_t): Move into garbage collector. New methods insert
36198 (ipa_node_params_sum): Annotate wth GTY(()).
36199 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
36201 (ipa_load_from_parm_agg): Adjust declaration.
36202 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
36203 * ipa-profile.c (ipa_profile): Likewise.
36204 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
36205 (ipa_populate_param_decls): Make descriptors parameter garbage
36207 (ipa_dump_param): Adjust to descriptors being a pointer.
36208 (ipa_alloc_node_params): Likewise.
36209 (ipa_initialize_node_params): Likewise.
36210 (load_from_param_1): Make descriptors parameter garbage collected.
36211 (load_from_unmodified_param): Likewise.
36212 (load_from_param): Likewise.
36213 (ipa_load_from_parm_agg): Likewise.
36214 (ipa_node_params::~ipa_node_params): Removed.
36215 (ipa_free_all_node_params): Remove call to delete operator.
36216 (ipa_node_params_t::insert): New.
36217 (ipa_node_params_t::remove): Likewise.
36218 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
36219 copy known_csts and known_contexts vectors.
36220 (ipa_read_node_info): Adjust to descriptors being a pointer.
36221 (ipcp_modif_dom_walker): Make m_descriptors field garbage
36223 (ipcp_transform_function): Make descriptors variable garbage
36226 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
36228 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
36229 * config/i386/avx512dqintrin.h: Ditto.
36230 * config/i386/avx512fintrin.h: Ditto.
36231 * config/i386/i386.c: Handle new builtins.
36232 * config/i386/i386-builtin.def: Add new builtins.
36233 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
36234 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
36236 2017-01-23 Jakub Jelinek <jakub@redhat.com>
36237 Martin Liska <mliska@suse.cz>
36239 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
36240 * asan.c (asan_expand_poison_ifn): Support stores and use
36241 appropriate ASAN report function.
36242 * internal-fn.c (expand_ASAN_POISON_USE): New function.
36243 * internal-fn.def (ASAN_POISON_USE): Declare.
36244 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
36245 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
36246 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
36247 ASAN_POISON calls w/o LHS.
36248 * tree-ssa.c (execute_update_addresses_taken): Create clobber
36249 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
36250 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
36251 * gimplify.c (asan_poison_variables): Add attribute
36252 use_after_scope_memory to variables that really needs to live
36254 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
36255 having the attribute.
36257 2017-01-23 Martin Liska <mliska@suse.cz>
36259 * asan.c (create_asan_shadow_var): New function.
36260 (asan_expand_poison_ifn): Likewise.
36261 * asan.h (asan_expand_poison_ifn): New declaration.
36262 * internal-fn.c (expand_ASAN_POISON): Likewise.
36263 * internal-fn.def (ASAN_POISON): New builtin.
36264 * sanopt.c (pass_sanopt::execute): Expand
36265 asan_expand_poison_ifn.
36266 * tree-inline.c (copy_decl_for_dup_finish): Make function
36268 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
36269 * tree-ssa.c (is_asan_mark_p): New function.
36270 (execute_update_addresses_taken): Rewrite local variables
36271 (identified just by use-after-scope as addressable) into SSA.
36273 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
36275 * doc/install.texi (Specific): opensource.apple.com uses https
36276 now. Remove trailing slash.
36278 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
36280 * README.Portability: Remove note on an Irix compatibility issue.
36282 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
36284 * gcov.c (INCLUDE_ALGORITHM): Define.
36285 (INCLUDE_VECTOR): Define.
36286 No longer include <vector> and <algorithm> directly.
36288 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
36290 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
36292 * doc/invoke.texi (Code Gen Options): Ditto.
36294 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
36297 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
36299 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
36301 rtl-optimization/79125
36302 * cprop.c (local_cprop_pass): Handle cases where we make an
36303 unconditional trap.
36305 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
36309 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
36310 read from, for big endian.
36312 2017-01-20 Jiong Wang <jiong.wang@arm.com>
36314 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
36315 register pauth builtins for LP64 only.
36317 2017-01-20 Marek Polacek <polacek@redhat.com>
36320 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
36323 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
36325 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
36327 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
36328 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
36329 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
36331 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36334 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
36335 in big-endian mode when they are not a single duplicated value.
36337 2017-01-20 Richard Biener <rguenther@suse.de>
36339 * BASE-VER: Bump to 7.0.1.
36341 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
36343 * omp-low.c (omplow_simd_context): New struct. Use it...
36344 (lower_rec_simd_input_clauses): ...here and...
36345 (lower_rec_input_clauses): ...here to hold common data. Adjust all
36346 references to idx, lane, max_vf, is_simt.
36348 2017-01-20 Graham Markall <graham.markall@embecosm.com>
36350 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
36353 2017-01-20 Martin Jambor <mjambor@suse.cz>
36355 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
36356 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
36358 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
36359 (GTFILES): Rename hsa.c to hsa-common.c.
36360 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
36361 * hsa-dump.c: Likewise.
36362 * hsa-gen.c: Likewise.
36363 * hsa-regalloc.c: Likewise.
36364 * ipa-hsa.c: Likewise.
36365 * omp-expand.c: Likewise.
36366 * omp-low.c: Likewise.
36367 * toplev.c: Likewise.
36369 2017-01-20 Marek Polacek <polacek@redhat.com>
36372 * doc/invoke.texi: Document -Wduplicated-branches.
36373 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
36374 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
36375 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
36376 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
36377 return 0 only when not OEP_LEXICOGRAPHIC.
36378 (fold_build_cleanup_point_expr): Use the expression
36379 location when building CLEANUP_POINT_EXPR.
36380 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
36381 * tree.c (add_expr): Handle error_mark_node.
36383 2017-01-20 Martin Liska <mliska@suse.cz>
36386 * tree-profile.c (init_ic_make_global_vars): Do not call
36388 (gimple_init_gcov_profiler): Likewise.
36390 2017-01-20 Martin Liska <mliska@suse.cz>
36393 * cgraph.h (maybe_create_reference): Remove argument and
36395 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
36397 * ipa-cp.c (create_specialized_node): Likewise.
36398 * symtab.c (symtab_node::maybe_create_reference): Handle
36399 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
36401 2017-01-20 Martin Liska <mliska@suse.cz>
36403 * read-rtl-function.c (function_reader::create_function): Use
36404 build_decl instread of build_decl_stat.
36406 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
36408 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
36409 * config/i386/avx512dqintrin.h: Ditto.
36410 * config/i386/avx512fintrin.h: Ditto.
36411 * config/i386/i386-builtin-types.def: Add new types.
36412 * config/i386/i386.c: Handle new types.
36413 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
36414 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
36415 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
36416 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
36417 (__builtin_ia32_kshiftridi): New.
36418 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
36420 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
36424 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
36425 define to rs6000_init_stack_protect_guard.
36426 (rs6000_init_stack_protect_guard): New function.
36428 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
36429 Yunqiang Su <yunqiang.su@imgtec.com>
36431 * config.gcc (supported_defaults): Add madd4.
36432 (with_madd4): Add validation.
36433 (all_defaults): Add madd4.
36434 * config/mips/mips.opt (mmadd4): New option.
36435 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
36437 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
36439 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
36440 (ISA_HAS_FUSED_MADD4): Likewise.
36441 * doc/invoke.texi (-mmadd4): Document the new option.
36442 * doc/install.texi (--with-madd4): Document the new option.
36444 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36446 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
36447 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
36448 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
36449 (aarch64_init_pauth_hint_builtins): New.
36450 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
36451 (aarch64_expand_builtin): Expand new builtins.
36453 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36455 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
36456 * combine-stack-adj.c (no_unhandled_cfa): Handle
36457 REG_CFA_TOGGLE_RA_MANGLE.
36458 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
36459 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
36460 info for return address signing.
36461 (aarch64_expand_epilogue): Likewise.
36463 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36465 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
36466 * config/aarch64/aarch64-protos.h
36467 (aarch64_return_address_signing_enabled): New declaration.
36468 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
36470 (aarch64_expand_prologue): Sign return address before it's pushed onto
36472 (aarch64_expand_epilogue): Authenticate return address fetched from
36474 (aarch64_override_options): Sanity check for ILP32 and ISA level.
36475 (aarch64_attributes): New function attributes for "sign-return-address".
36476 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
36477 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
36478 ("*do_return"): Generate combined instructions according to key index.
36479 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
36480 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
36482 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
36483 * config/aarch64/aarch64.opt (msign-return-address=): New.
36484 * doc/extend.texi (AArch64 Function Attributes): Documents
36485 "sign-return-address=".
36486 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
36488 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
36490 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
36491 overall option summary.
36493 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36495 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
36496 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
36497 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
36498 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
36500 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
36502 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
36503 -mpower9-minmax by default for -mcpu=power9.
36504 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
36505 128-bit floating point.
36507 2017-01-20 Alan Modra <amodra@gmail.com>
36509 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
36510 optimizing for size.
36512 2017-01-20 Alan Modra <amodra@gmail.com>
36515 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
36516 for strcmp and strncmp from corresponding builtin decl.
36518 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36520 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
36521 instead of i386/rtems-64.h.
36522 * config/i386/rtems-64.h: Remove.
36524 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36528 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
36530 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
36532 2017-01-19 Tamar Christina <tamar.christina@arm.com>
36534 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
36535 Change int to HOST_WIDE_INT.
36536 * config/aarch64/aarch64-protos.h
36537 (aarch64_simd_gen_const_vector_dup): Likewise.
36538 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
36540 2017-01-19 David Malcolm <dmalcolm@redhat.com>
36542 * langhooks-def.h (lhd_type_for_size): New decl.
36543 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
36544 * langhooks.c (lhd_type_for_size): New function, taken from
36547 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
36549 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
36550 define_bypass for CR latency.
36551 (power9-cracked-alu): Update bypass latency and remove power9-branch.
36552 (power9-alu2): Add define_bypass for CR latency.
36554 (power9-mul): Update insn latency.
36555 (power9-mul-compare): Update insn latency, bypass latency and remove
36558 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36560 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
36562 * config/aarch64/aarch64.md
36563 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
36564 aarch64_nopcrelative_literal_loads.
36565 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
36567 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
36569 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
36570 TARGET_LOONGSON_3A.
36571 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
36573 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
36576 * config.gcc (supported_defaults): Add lxc1-sxc1.
36577 (with_lxc1_sxc1): Add validation.
36578 (all_defaults): Add lxc1-sxc1.
36579 * config/mips/mips.opt (mlxc1-sxc1): New option.
36580 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
36582 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
36583 __mips_no_lxc1_sxc1.
36584 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
36585 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
36586 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
36588 2017-01-19 Richard Biener <rguenther@suse.de>
36590 PR tree-optimization/72488
36591 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
36592 sure to restore SSA info.
36593 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
36595 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
36597 PR rtl-optimization/79121
36598 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
36599 of the inner type when shifting an extended value.
36601 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
36604 * symtab.c (symtab_node::equal_address_to): Fix comparing of
36605 interposable aliases.
36607 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
36610 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
36611 Use the evmergelohi instruction.
36612 (mov_si<mode>_e500_subreg4_2_le): Likewise.
36613 (mov_sitf_e500_subreg8_2_be): Likewise.
36614 (mov_sitf_e500_subreg12_2_le): Likewise.
36615 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
36616 (mov_si<mode>_e500_subreg4_2_be): Likewise.
36617 (mov_sitf_e500_subreg8_2_le): Likewise.
36618 (mov_sitf_e500_subreg12_2_be): Likewise.
36620 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36622 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
36623 attribute from vecsimple to vecperm.
36624 (altivec_vbpermq2): Likewise.
36626 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36629 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
36631 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36632 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
36633 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
36634 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
36635 case where N arg is SIZE_MAX.
36636 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
36637 (cmpstrsi): Add pattern.
36639 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
36641 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
36642 __builtin_vec_revb builtins.
36643 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
36644 built-in functions to support generation of the ISA 3.0 XXBR<x>
36645 vector byte reverse instructions.
36646 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
36647 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
36648 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
36649 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
36650 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
36651 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
36652 (P9V_BUILTIN_VEC_REVB): Likewise.
36653 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
36654 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
36655 (p9_xxbrq_v16qi): Likewise.
36656 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
36657 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
36658 (p9_xxbrh_v8hi): Likewise.
36659 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
36660 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
36661 vec_revb built-in functions.
36663 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
36665 PR rtl-optimization/78952
36666 * config/i386/i386.md (any_extract): New code iterator.
36667 (*insvqi_2): Use any_extract for source operand.
36668 (*insvqi_3): Use any_shiftrt for source operand.
36670 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
36672 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
36674 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
36676 2017-01-18 Matthias Klose <doko@ubuntu.com>
36678 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
36680 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36682 * config/rs6000/altivec.h (vec_bperm): Change #define.
36683 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
36684 (altivec_vbpermq2): New define_insn.
36685 (altivec_vbpermd): Likewise.
36686 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
36687 function interface.
36688 (VBPERMD): Likewise.
36689 (VBPERM): New polymorphic function interface.
36690 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
36691 Add entries for P9V_BUILTIN_VEC_VBPERM.
36692 * doc/extend.texi: Add interfaces for vec_bperm.
36694 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36696 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
36697 first letter of error messages.
36698 (s390_resolve_overloaded_builtin): Likewise.
36699 * config/s390/s390.c (s390_expand_builtin): Likewise.
36700 (s390_invalid_arg_for_unprototyped_fn): Likewise.
36701 (s390_valid_target_attribute_inner_p): Likewise.
36702 * config/s390/s390.md ("tabort"): Likewise.
36704 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
36706 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
36707 (ISA_AVOID_DIV_HILO): New macro.
36708 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
36709 (ISA_HAS_DDIV): Likewise.
36711 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36713 * doc/invoke.texi (fabi-version): Correct number of occurrences.
36715 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36717 * doc/invoke.texi (fabi-version): Spelling fix.
36719 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36722 * doc/invoke.texi (fabi-version): Mention mangling fix for
36725 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36728 * doc/invoke.texi (fabi-version): Document discriminator mangling.
36730 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
36733 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
36734 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
36736 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
36738 (stack_protect_test): Ditto.
36739 * config/rs6000/rs6000.opt (mstack-protector-guard=,
36740 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
36742 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
36743 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
36744 -mstack-protector-guard-offset=.
36745 (RS/6000 and PowerPC Options): Ditto.
36747 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
36749 * config/i386/i386.h (MASK_CLASS_P): New define.
36750 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
36751 there are no registers from different register sets also when
36752 mask registers are used. Update function comment.
36753 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
36754 to (*k/*r) and (*k/*km) alternatives.
36756 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
36758 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
36759 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
36760 (EH_RETURN_HANDLER_RTX): New define.
36761 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
36762 Force frame pointer in EH return functions.
36763 (aarch64_expand_epilogue): Add barrier for eh_return.
36764 (aarch64_final_eh_return_addr): Remove.
36765 (aarch64_eh_return_handler_rtx): New function.
36766 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
36768 (aarch64_eh_return_handler_rtx): New prototype.
36770 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36772 * config/rs6000/altivec.h (vec_rlmi): New #define.
36773 (vec_vrlnm): Likewise.
36774 (vec_rlnm): Likewise.
36775 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
36776 (UNSPEC_VRLNM): Likewise.
36777 (VIlong): New mode iterator.
36778 (altivec_vrl<VI_char>mi): New define_insn.
36779 (altivec_vrl<VI_char>nm): Likewise.
36780 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
36782 (VRLDNM): Likewise.
36783 (RLNM): New polymorphic function entry.
36784 (VRLWMI): New monomorphic function entry.
36785 (VRLDMI): Likewise.
36786 (RLMI): New polymorphic function entry.
36787 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
36788 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
36789 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
36792 2017-01-17 Jakub Jelinek <jakub@redhat.com>
36795 * dwarf2out.c (field_byte_offset): Restore the
36796 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
36797 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
36798 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
36801 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
36804 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
36806 2017-01-17 Jakub Jelinek <jakub@redhat.com>
36809 * dwarf2out.c (add_data_member_location_attribute): For constant
36810 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
36811 instead of DW_AT_data_member_location, DW_AT_bit_offset and
36812 DW_AT_byte_size attributes.
36814 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
36816 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
36817 after forcing to constant memory when the code model is medium.
36819 2017-01-17 Julia Koval <julia.koval@intel.com>
36822 * config/i386/avx512fintrin.h
36823 (_mm512_i32gather_ps): Change __addr type to void const*.
36824 (_mm512_mask_i32gather_ps): Ditto.
36825 (_mm512_i32gather_pd): Ditto.
36826 (_mm512_mask_i32gather_pd): Ditto.
36827 (_mm512_i64gather_ps): Ditto.
36828 (_mm512_mask_i64gather_ps): Ditto.
36829 (_mm512_i64gather_pd): Ditto.
36830 (_mm512_mask_i64gather_pd): Ditto.
36831 (_mm512_i32gather_epi32): Ditto.
36832 (_mm512_mask_i32gather_epi32): Ditto.
36833 (_mm512_i32gather_epi64): Ditto.
36834 (_mm512_mask_i32gather_epi64): Ditto.
36835 (_mm512_i64gather_epi32): Ditto.
36836 (_mm512_mask_i64gather_epi32): Ditto.
36837 (_mm512_i64gather_epi64): Ditto.
36838 (_mm512_mask_i64gather_epi64): Ditto.
36839 (_mm512_i32scatter_ps): Change __addr type to void*.
36840 (_mm512_mask_i32scatter_ps): Ditto.
36841 (_mm512_i32scatter_pd): Ditto.
36842 (_mm512_mask_i32scatter_pd): Ditto.
36843 (_mm512_i64scatter_ps): Ditto.
36844 (_mm512_mask_i64scatter_ps): Ditto.
36845 (_mm512_i64scatter_pd): Ditto.
36846 (_mm512_mask_i64scatter_pd): Ditto.
36847 (_mm512_i32scatter_epi32): Ditto.
36848 (_mm512_mask_i32scatter_epi32): Ditto.
36849 (_mm512_i32scatter_epi64): Ditto.
36850 (_mm512_mask_i32scatter_epi64): Ditto.
36851 (_mm512_i64scatter_epi32): Ditto.
36852 (_mm512_mask_i64scatter_epi32): Ditto.
36853 (_mm512_i64scatter_epi64): Ditto.
36854 (_mm512_mask_i64scatter_epi64): Ditto.
36855 * config/i386/avx512pfintrin.h
36856 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
36857 (_mm512_mask_prefetch_i32gather_ps): Ditto.
36858 (_mm512_mask_prefetch_i64gather_pd): Ditto.
36859 (_mm512_mask_prefetch_i64gather_ps): Ditto.
36860 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
36861 (_mm512_prefetch_i32scatter_ps): Ditto.
36862 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
36863 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
36864 (_mm512_prefetch_i64scatter_pd): Ditto.
36865 (_mm512_prefetch_i64scatter_ps): Ditto.
36866 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
36867 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
36868 * config/i386/avx512vlintrin.h
36869 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
36870 (_mm_mmask_i32gather_ps): Ditto.
36871 (_mm256_mmask_i32gather_pd): Ditto.
36872 (_mm_mmask_i32gather_pd): Ditto.
36873 (_mm256_mmask_i64gather_ps): Ditto.
36874 (_mm_mmask_i64gather_ps): Ditto.
36875 (_mm256_mmask_i64gather_pd): Ditto.
36876 (_mm_mmask_i64gather_pd): Ditto.
36877 (_mm256_mmask_i32gather_epi32): Ditto.
36878 (_mm_mmask_i32gather_epi32): Ditto.
36879 (_mm256_mmask_i32gather_epi64): Ditto.
36880 (_mm_mmask_i32gather_epi64): Ditto.
36881 (_mm256_mmask_i64gather_epi32): Ditto.
36882 (_mm_mmask_i64gather_epi32): Ditto.
36883 (_mm256_mmask_i64gather_epi64): Ditto.
36884 (_mm_mmask_i64gather_epi64): Ditto.
36885 (_mm256_i32scatter_ps): Change __addr type to void*.
36886 (_mm256_mask_i32scatter_ps): Ditto.
36887 (_mm_i32scatter_ps): Ditto.
36888 (_mm_mask_i32scatter_ps): Ditto.
36889 (_mm256_i32scatter_pd): Ditto.
36890 (_mm256_mask_i32scatter_pd): Ditto.
36891 (_mm_i32scatter_pd): Ditto.
36892 (_mm_mask_i32scatter_pd): Ditto.
36893 (_mm256_i64scatter_ps): Ditto.
36894 (_mm256_mask_i64scatter_ps): Ditto.
36895 (_mm_i64scatter_ps): Ditto.
36896 (_mm_mask_i64scatter_ps): Ditto.
36897 (_mm256_i64scatter_pd): Ditto.
36898 (_mm256_mask_i64scatter_pd): Ditto.
36899 (_mm_i64scatter_pd): Ditto.
36900 (_mm_mask_i64scatter_pd): Ditto.
36901 (_mm256_i32scatter_epi32): Ditto.
36902 (_mm256_mask_i32scatter_epi32): Ditto.
36903 (_mm_i32scatter_epi32): Ditto.
36904 (_mm_mask_i32scatter_epi32): Ditto.
36905 (_mm256_i32scatter_epi64): Ditto.
36906 (_mm256_mask_i32scatter_epi64): Ditto.
36907 (_mm_i32scatter_epi64): Ditto.
36908 (_mm_mask_i32scatter_epi64): Ditto.
36909 (_mm256_i64scatter_epi32): Ditto.
36910 (_mm256_mask_i64scatter_epi32): Ditto.
36911 (_mm_i64scatter_epi32): Ditto.
36912 (_mm_mask_i64scatter_epi32): Ditto.
36913 (_mm256_i64scatter_epi64): Ditto.
36914 (_mm256_mask_i64scatter_epi64): Ditto.
36915 (_mm_i64scatter_epi64): Ditto.
36916 (_mm_mask_i64scatter_epi64): Ditto.
36917 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
36918 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
36919 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
36920 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
36921 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
36922 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
36923 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
36924 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
36925 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
36926 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
36927 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
36928 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
36929 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
36930 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
36931 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
36932 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
36933 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
36934 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
36935 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
36936 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
36937 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
36938 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
36939 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
36940 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
36941 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
36942 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
36943 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
36944 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
36945 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
36946 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
36947 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
36948 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
36949 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
36950 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
36951 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
36952 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
36953 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
36954 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
36955 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
36956 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
36957 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
36958 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
36959 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
36960 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
36961 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
36962 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
36963 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
36964 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
36965 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
36966 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
36967 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
36968 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
36969 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
36970 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
36971 definitions accordingly.
36973 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
36974 Kuan-Lin Chen <kuanlinchentw@gmail.com>
36977 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
36980 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
36983 * ira-conflicts.c (ira_build_conflicts): Update total conflict
36984 hard regs for inner regno.
36986 2017-01-17 Martin Liska <mliska@suse.cz>
36989 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
36990 assumption and add comment.
36992 2017-01-17 Nathan Sidwell <nathan@acm.org>
36994 * ipa-visibility.c (localize_node): New function, broken out of ...
36995 (function_and_variable_visibility): ... here. Call it.
36997 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
36999 PR middle-end/77445
37000 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
37001 correctly set frequency of oudgoing edge.
37002 (duplicate_thread_path): Fix profile updating.
37004 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37007 * configure.ac: Add GCC_BASE_VER.
37008 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
37009 version from BASE-VER file.
37010 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
37011 (gcc.o): Depend on $(BASEVER).
37012 * common.opt (dumpfullversion): New option.
37013 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
37014 * doc/invoke.texi: Document -dumpfullversion.
37015 * doc/install.texi: Document --with-gcc-major-version-only.
37016 * configure: Regenerated.
37018 2017-01-17 Richard Biener <rguenther@suse.de>
37020 PR tree-optimization/71433
37021 * tree-vrp.c (register_new_assert_for): Merge same asserts
37022 on all incoming edges.
37023 (process_assert_insertions_for): Handle insertions at the
37026 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
37028 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
37029 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
37031 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
37034 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
37037 2017-01-17 Alan Modra <amodra@gmail.com>
37040 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
37041 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
37042 symbolic stack limit when pic.
37044 2017-01-16 Martin Sebor <msebor@redhat.com>
37046 PR tree-optimization/78608
37047 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
37049 2017-01-16 Jeff Law <law@redhat.com>
37052 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
37053 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
37054 for several include directories that may be relative to sysroot.
37055 * config/i386/x-mingw32 (gplus_includedir): Define.
37056 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
37057 (native_system_includedir): Likewise.
37058 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
37059 override if TARGET_SYSTEM_ROOT is defined.
37060 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
37062 PR tree-optimization/79090
37063 PR tree-optimization/33562
37064 PR tree-optimization/61912
37065 PR tree-optimization/77485
37066 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
37067 and computed trims into the dump file.
37069 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
37071 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
37073 2017-01-16 Jakub Jelinek <jakub@redhat.com>
37076 * gimplify.c (gimplify_init_constructor): If want_value and
37077 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
37081 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
37082 sequence. Formatting fixes.
37083 (doloop_optimize): Formatting fixes.
37086 * gcc.c (debug_level_greater_than_spec_func): New function.
37087 (static_spec_functions): Add debug-level-gt spec function.
37088 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
37090 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37091 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37092 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
37093 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
37094 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
37095 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
37097 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
37099 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
37100 QImode fixups to general and mask registers only.
37102 2017-01-16 Carl Love <cel@us.ibm.com>
37104 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
37105 for built-in functions
37106 vector signed char vec_nabs (vector signed char)
37107 vector signed short vec_nabs (vector signed short)
37108 vector signed int vec_nabs (vector signed int)
37109 vector signed long long vec_nabs (vector signed long long)
37110 vector float vec_nabs (vector float)
37111 vector double vec_nabs (vector double)
37112 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
37114 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
37115 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
37116 * doc/extend.texi: Update the documentation file for the new built-in
37119 2017-01-16 Martin Sebor <msebor@redhat.com>
37121 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
37124 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37126 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
37127 UNSPEC_VSX__XXSPLTD to require special splat handling.
37129 2017-01-16 David Malcolm <dmalcolm@redhat.com>
37132 * system.h: Poison strndup.
37134 2017-01-16 Alan Modra <amodra@gmail.com>
37137 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
37140 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
37142 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
37144 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
37146 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
37147 call recog here. Assert that INSN_CODE (insn) is non-negative.
37149 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
37152 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
37154 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
37155 in the currently scheduled RTL fragment.
37157 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
37159 PR rtl-optimization/78751
37160 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
37163 2017-01-14 Jeff Law <law@redhat.com>
37165 PR tree-optimization/79090
37166 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
37167 variable length stores.
37168 (compute_trims): Delete dead assignment to *trim_tail.
37169 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
37172 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
37174 PR rtl-optimization/78626
37175 PR rtl-optimization/78727
37176 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
37177 of a block, and split such blocks after everything else is finished.
37179 2017-01-14 Alan Modra <amodra@gmail.com>
37182 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
37183 target legitimate_combined_insn.
37184 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
37185 (rs6000_legitimate_combined_insn): New function.
37186 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
37188 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
37189 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
37190 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
37192 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
37194 * doc/frontends.texi (G++ and GCC): Remove references to Java.
37196 2017-01-13 Jeff Law <law@redhat.com>
37198 PR tree-optimization/33562
37199 PR tree-optimization/61912
37200 PR tree-optimization/77485
37201 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
37203 (delete_dead_assignment): Likewise.
37204 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
37205 statement to delete_dead_call and delete_dead_assignment.
37207 2017-01-13 David Malcolm <dmalcolm@redhat.com>
37210 * substring-locations.c (format_warning_va): Strengthen case 1 so
37211 that both endpoints of the substring must be within the format
37212 range for just the substring to be printed.
37214 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
37216 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
37217 * config/i386/i386.c (ix86_target_string): Add missing options
37218 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
37219 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
37220 flags_other and ix86_target_other to flags2_other. Display unknown
37222 (ix86_valid_target_attribute_inner_p): Add missing options and
37223 reorder options by implied ISAs, as in ix86_target_string.
37225 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
37227 * hash-table.h (hash_table::too_empty_p): New function.
37228 (hash_table::expand): Use it.
37229 (hash_table::traverse): Likewise.
37230 (hash_table::empty_slot): Use sizeof (value_type) instead of
37231 sizeof (PTR) to convert bytes to elements. Shrink the table
37232 if the current size is excessive for the current number of
37235 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
37237 * ira-costs.c (record_reg_classes): Break from the inner loop
37238 early once alt_fail is known to be true. Update outer loop
37239 handling accordingly.
37241 2017-01-13 Jeff Law <law@redhat.com>
37243 * tree-ssa-dse.c (decrement_count): New function.
37244 (increment_start_addr, maybe_trim_memstar_call): Likewise.
37245 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
37246 when we know the partially dead statement is a mem* function.
37248 PR tree-optimization/61912
37249 PR tree-optimization/77485
37250 * tree-ssa-dse.c: Include expr.h.
37251 (maybe_trim_constructor_store): New function.
37252 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
37254 PR tree-optimization/33562
37255 PR tree-optimization/61912
37256 PR tree-optimization/77485
37257 * doc/invoke.texi: Document new dse-max-object-size param.
37258 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
37259 * tree-ssa-dse.c: Include params.h.
37260 (dse_store_status): New enum.
37261 (initialize_ao_ref_for_dse): New, partially extracted from
37263 (valid_ao_ref_for_dse, normalize_ref): New.
37264 (setup_live_bytes_from_ref, compute_trims): Likewise.
37265 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
37266 (maybe_trim_partially_dead_store): Likewise.
37267 (maybe_trim_complex_store): Likewise.
37268 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
37269 Track what bytes live from the original store. Return tri-state
37270 for dead, partially dead or live.
37271 (dse_dom_walker): Add constructor, destructor and new private members.
37272 (delete_dead_call, delete_dead_assignment): New extracted from
37274 (dse_optimize_stmt): Make a member of dse_dom_walker.
37275 Use initialize_ao_ref_for_dse.
37277 PR tree-optimization/33562
37278 PR tree-optimization/61912
37279 PR tree-optimization/77485
37280 * sbitmap.h (bitmap_count_bits): Prototype.
37281 (bitmap_clear_range, bitmap_set_range): Likewise.
37282 * sbitmap.c (bitmap_clear_range): New function.
37283 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
37285 2017-01-13 Martin Liska <mliska@suse.cz>
37288 * function.c (set_cfun): Add new argument force.
37289 * function.h (set_cfun): Likewise.
37290 * ipa-inline-transform.c (inline_call): Use the function when
37291 strict alising from is dropped for function we inline to.
37293 2017-01-13 Richard Biener <rguenther@suse.de>
37295 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
37296 for dumping GIMPLE INTEGER_CSTs.
37298 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
37300 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
37301 to 201112L since C++17.
37303 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
37306 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
37307 if -fsanitize=kernel-address is present.
37309 2017-01-13 Richard Biener <rguenther@suse.de>
37311 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
37312 as _Literal ( type ) number in case usual suffixes do not
37313 preserve all information.
37315 2017-01-13 Richard Biener <rguenther@suse.de>
37317 PR tree-optimization/77283
37318 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
37319 and ssa-iterators.h.
37320 (is_feasible_trace): Implement a cost model based on joiner
37323 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
37326 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
37327 char or short to __float128/_Float128 directly.
37329 2017-01-12 Martin Sebor <msebor@redhat.com>
37331 to -Wformat-overflow.
37332 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
37333 (min_bytes_remaining): Same.
37334 (get_string_length): Same.
37335 (format_string): Same.
37336 (format_directive): Same.
37338 (pass_sprintf_length::handle_gimple_call): Same.
37340 2017-01-12 Jakub Jelinek <jakub@redhat.com>
37342 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
37343 info.nowrite calls with no lhs that can't throw. Return bool
37344 whether gsi_remove has been called or not.
37345 (pass_sprintf_length::handle_gimple_call): Return bool whether
37346 try_substitute_return_value called gsi_remove. Formatting fix.
37347 (pass_sprintf_length::execute): Don't use gsi_remove if
37348 handle_gimple_call returned true.
37351 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
37352 be removed due to side-effects, don't remove following barrier nor
37353 turn the successor edge into fallthru edge.
37355 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37358 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
37359 element-reversing loads and stores as not swappable.
37361 2017-01-12 Nathan Sidwell <nathan@acm.org>
37362 Nicolai Stange <nicstange@gmail.com>
37364 * combine.c (try_combine): Don't ignore result of overlap checking
37365 loop. Combine overlap & asm check into single loop.
37367 2017-01-12 Richard Biener <rguenther@suse.de>
37369 * tree-pretty-print.c (dump_generic_node): Provide -gimple
37370 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
37372 2017-01-12 Richard Biener <rguenther@suse.de>
37374 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
37375 and TS_TARGET_OPTION directly derive from TS_BASE.
37376 * tree-core.h (tree_optimization_option): Derive from tree_base.
37377 (tree_target_option): Likewise.
37379 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
37381 * config/i386/i386.c (memory_address_length): Increase len
37382 only when rip_relative_addr_p returns false.
37384 2017-01-11 Julia Koval <julia.koval@intel.com>
37386 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
37387 (OPTION_MASK_ISA_SGX_SET): New.
37388 (ix86_handle_option): Handle OPT_msgx.
37389 * config.gcc: Added sgxintrin.h.
37390 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
37391 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
37392 * config/i386/i386.c (ix86_target_string): Add -msgx.
37394 (ix86_option_override_internal): Handle new options.
37395 (ix86_valid_target_attribute_inner_p): Add sgx.
37396 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
37397 * config/i386/i386.opt: Add msgx.
37398 * config/i386/sgxintrin.h: New file.
37399 * config/i386/x86intrin.h: Add sgxintrin.h.
37401 2017-01-11 Jakub Jelinek <jakub@redhat.com>
37404 * fold-const.c (maybe_nonzero_address): Return 1 for function
37406 (tree_single_nonzero_warnv_p): Don't handle function local objects
37410 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
37413 2017-01-11 David Malcolm <dmalcolm@redhat.com>
37416 * opts.c: Include "spellcheck.h"
37417 (struct string_fragment): New struct.
37418 (struct edit_distance_traits<const string_fragment &>): New
37420 (get_closest_sanitizer_option): New function.
37421 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
37423 2017-01-11 Jakub Jelinek <jakub@redhat.com>
37425 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
37427 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
37428 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
37429 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
37430 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
37431 for initial die_offset if dwarf_split_debug_info.
37432 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
37433 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
37435 (output_skeleton_debug_sections): Formatting fix. Use
37436 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
37437 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
37439 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
37441 * config/arm/cortex-a53.md: Add bypasses for
37442 cortex_a53_r2f_cvt.
37443 (cortex_a53_r2f): Only use for transfers.
37444 (cortex_a53_f2r): Likewise.
37445 (cortex_a53_r2f_cvt): Add reservation for conversions.
37446 (cortex_a53_f2r_cvt): Likewise.
37448 2017-01-11 Tamar Christina <tamar.christina@arm.com>
37450 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
37451 to all inlined functions, change static to extern.
37453 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
37456 * config/arm/arm.c (legitimize_pic_address): Handle reference to
37458 (arm_assemble_integer): Likewise.
37460 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
37462 * config.gcc: Use new awk script to check CPU, FPU and architecture
37463 parameters for --with-... options.
37464 * config/arm/parsecpu.awk: New file
37465 * config/arm/arm-cpus.in: New file.
37466 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
37468 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
37470 * config/arm/t-arm: Update dependency rules.
37471 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
37472 of processing .def files.
37473 * config/arm/genopt.sh: Deleted.
37474 * config/arm/gentune.sh: Deleted.
37475 * config/arm/arm-cores.def: Deleted.
37476 * config/arm/arm-arches.def: Deleted.
37477 * config/arm/arm-fpus.def: Deleted.
37478 * config/arm/arm-tune.md: Regenerated.
37479 * config/arm/arm-tables.opt: Regenerated.
37480 * config/arm/arm-cpu.h: New generated file.
37481 * config/arm/arm-cpu-data.h: New generated file.
37482 * config/arm/arm-cpu-cdata.h: New generated file.
37484 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
37487 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
37489 (input_varpool_node): Unpack dynamically_initialized bit.
37491 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
37493 PR rtl-optimization/79032
37494 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
37495 the alignment of the adjusted memory reference against that of MODE,
37496 instead of the alignment of the original memory reference.
37498 2017-01-11 Martin Jambor <mjambor@suse.cz>
37500 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
37502 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
37503 decorated functions.
37505 2017-01-11 Richard Biener <rguenther@suse.de>
37507 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
37508 set range/nonnull info for PHI results. Do not set it on
37509 stmts marked for removal.
37511 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
37513 * expr.c (store_field): In the bitfield case, fetch the return value
37514 from the registers before applying a single big-endian adjustment.
37515 Always do a final load for a BLKmode value not larger than a word.
37517 2017-01-10 David Malcolm <dmalcolm@redhat.com>
37520 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
37521 that we correctly handle column numbers greater than
37522 LINE_MAP_MAX_COLUMN_NUMBER.
37524 2017-01-10 Martin Sebor <msebor@redhat.com>
37526 PR middle-end/78245
37527 * gimple-ssa-sprintf.c (get_destination_size): Call
37528 {init,fini}object_sizes.
37529 * tree-object-size.c (addr_object_size): Adjust.
37530 (pass_through_call): Adjust.
37531 (pass_object_sizes::execute): Adjust.
37532 * tree-object-size.h (fini_object_sizes): Declare.
37534 2017-01-10 Martin Sebor <msebor@redhat.com>
37536 PR tree-optimization/78775
37537 * builtins.c (get_size_range): Move...
37538 * calls.c: ...to here.
37539 (alloc_max_size): Accept zero argument.
37540 (operand_signed_p): Remove.
37541 (maybe_warn_alloc_args_overflow): Call get_size_range.
37542 * calls.h (get_size_range): Declare.
37544 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
37546 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
37547 from TI's devices.csv file as of September 2016.
37548 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
37550 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
37552 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
37553 * doc/invoke.texi: Likewise.
37554 * doc/md.texi: Likewise.
37555 * doc/objc.texi: Likewise.
37557 2017-01-10 Joshua Conner <joshconner@google.com>
37559 * config/arm/fuchsia-elf.h: New file.
37560 * config/fuchsia.h: New file.
37561 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
37562 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
37564 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
37566 2016-01-10 Richard Biener <rguenther@suse.de>
37568 PR tree-optimization/79034
37569 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
37570 Propagate out degenerate PHIs in the joiner.
37572 2017-01-10 Martin Liska <mliska@suse.cz>
37574 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
37575 (sort_congruence_classes_by_decl_uid): Likewise.
37576 (sort_congruence_class_groups_by_decl_uid): Likewise.
37577 (sem_item_optimizer::merge_classes): Sort class, groups in these
37578 classes and members in the groups by DECL_UID of declarations.
37579 This would make merge operations stable.
37581 2017-01-10 Martin Liska <mliska@suse.cz>
37583 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
37584 usage of m_classes_vec.
37585 (sem_item_optimizer::~sem_item_optimizer): Likewise.
37586 (sem_item_optimizer::get_group_by_hash): Likewise.
37587 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
37588 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
37589 (sem_item_optimizer::verify_classes): Likewise.
37590 (sem_item_optimizer::process_cong_reduction): Likewise.
37591 (sem_item_optimizer::dump_cong_classes): Likewise.
37592 (sem_item_optimizer::merge_classes): Likewise.
37593 * ipa-icf.h (congruence_class_hash): Rename from
37594 congruence_class_group_hash. Remove declaration of m_classes_vec.
37596 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
37598 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
37599 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
37600 * config.gcc: Add avx512vpopcntdqintrin.h.
37601 * config/i386/avx512vpopcntdqintrin.h: New.
37602 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
37603 * config/i386/i386-builtin-types.def: Add new types.
37604 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
37605 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
37606 __builtin_ia32_vpopcountq_v8di_mask): New.
37607 * config/i386/i386-c.c (ix86_target_macros_internal): Define
37608 __AVX512VPOPCNTDQ__.
37609 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
37610 (PTA_AVX512VPOPCNTDQ): Define.
37611 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
37612 TARGET_AVX512VPOPCNTDQ_P): Define.
37613 * config/i386/i386.opt: Add mavx512vpopcntdq.
37614 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
37615 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
37617 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37619 PR middle-end/77484
37620 * predict.def (PRED_CALL): Set to 67.
37622 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
37624 * expr.c (store_field): In the bitfield case, if the value comes from
37625 a function call and is of an aggregate type returned in registers, do
37626 not modify the field mode; extract the value in all cases if the mode
37627 is BLKmode and the size is not larger than a word.
37629 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
37632 * config/i386/cpuid.h: Fix undefined behavior.
37634 2017-01-04 Jeff Law <law@redhat.com>
37636 PR tree-optimization/79007
37637 PR tree-optimization/67955
37638 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
37639 conservative for pt.null when flag_non_call_exceptions is on.
37641 2017-01-09 Jakub Jelinek <jakub@redhat.com>
37643 PR translation/79019
37644 PR translation/79020
37645 * params.def (PARAM_INLINE_MIN_SPEEDUP,
37646 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
37647 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
37649 * config/avr/avr.opt (maccumulate-args): Likewise.
37650 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
37651 * common.opt (freport-bug): Likewise.
37652 * cif-code.def (CIF_FINAL_ERROR): Likewise.
37653 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
37654 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
37655 translatable string.
37656 * config/i386/i386.c (function_value_32): Likewise.
37657 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
37658 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
37660 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
37661 * common/config/msp430/msp430-common.c (msp430_handle_option):
37663 * symtab.c (symtab_node::verify_base): Likewise.
37664 * opts.c (set_debug_level): Likewise.
37665 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
37666 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
37667 missing whitespace to translatable strings.
37668 * config/avr/avr.md (bswapsi2): Fix typo in comment.
37669 * config/sh/superh.h: Likewise.
37670 * config/i386/xopintrin.h: Likewise.
37671 * config/i386/znver1.md: Likewise.
37672 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
37673 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
37674 * double-int.h (struct double_int): Likewise.
37675 * double-int.c (div_and_round_double): Likewise.
37676 * wide-int.cc: Likewise.
37677 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
37678 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
37679 * cfgcleanup.c (crossjumps_occured): Renamed to ...
37680 (crossjumps_occurred): ... this.
37681 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
37684 PR tree-optimization/78899
37685 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
37686 returning bool return struct loop *, NULL for failure and the new
37688 (versionable_outer_loop_p): Don't version outer loop if it has
37689 dont_vectorized bit set.
37690 (tree_if_conversion): When versioning outer loop, ensure
37691 tree_if_conversion is performed also on the inner loop of the
37692 non-vectorizable outer loop copy.
37693 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
37694 LOOP_VECTORIZED in inner loop of the scalar outer loop and
37695 prevent vectorization of it.
37696 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
37697 the outer loop vectorization of the non-scalar version is attempted
37698 before vectorization of the inner loop in scalar version. If
37699 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
37700 vectorization of its inner loop.
37701 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
37702 has 2 inner loops, rename also on edges from bb whose single pred
37703 is outer_loop->header. Fix typo in function comment.
37705 2017-01-09 Martin Sebor <msebor@redhat.com>
37708 * asan.c (asan_emit_stack_protection): Increase local buffer size
37709 to avoid snprintf truncation warning.
37711 2017-01-09 Andrew Pinski <apinski@cavium.com>
37713 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
37714 to reference thunderx2t99 for the tuning structure
37715 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
37717 (thunderx2t99_extra_costs): This.
37718 * config/aarch64/aarch64-tune.md: Regenerate.
37719 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
37720 (vulcan_addrcost_table): This.
37721 (vulcan_regmove_cost): Rename to ...
37722 (thunderx2t99_regmove_cost): This.
37723 (vulcan_vector_cost): Rename to ...
37724 (thunderx2t99_vector_cost): this.
37725 (vulcan_branch_cost): Rename to ...
37726 (thunderx2t99_branch_cost): This.
37727 (vulcan_tunings): Rename to ...
37728 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
37729 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
37731 2017-01-09 Martin Jambor <mjambor@suse.cz>
37735 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
37736 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
37737 (propagate_vr_accross_jump_function): Use the above function for all
37738 value range computations for pass-through jump functions and type
37739 converasion from explicit value range values.
37740 (ipcp_propagate_stage): Do not attempt to deduce types of formal
37741 parameters from TYPE_ARG_TYPES.
37742 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
37743 (ipa_write_node_info): Stream type of the actual argument.
37744 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
37746 2017-01-09 Martin Liska <mliska@suse.cz>
37749 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
37750 (lookup_compiler): Do not show error message with have_E.
37752 2017-01-09 Jakub Jelinek <jakub@redhat.com>
37754 PR tree-optimization/78938
37755 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
37756 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
37757 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
37758 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
37761 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37763 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
37766 2017-01-09 Richard Biener <rguenther@suse.de>
37768 PR tree-optimization/78997
37769 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
37770 name condition properly.
37772 2017-01-09 Richard Biener <rguenther@suse.de>
37775 * dwarf2out.c (is_cxx): New overload with context.
37776 (is_naming_typedef_decl): Use it.
37778 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37780 * invoke.texi (Option Summary): Correct spacing in option lists
37781 and add line breaks to fix over-long lines.
37783 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37785 PR middle-end/17660
37787 * extend.texi (Common Variable Attributes): Add xref to GCC
37788 Internals manual to explain mode attribute keywords.
37790 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37793 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
37794 and Preprocessor Options.
37795 (Options for Linking): Document -pthread here....
37796 (RS/6000 and PowerPC Options): ...not here.
37797 (Solaris 2 Options): ...or here.
37798 * doc/cppopts.texi: Document -pthread.
37800 2017-01-08 Martin Sebor <msebor@redhat.com>
37802 PR middle-end/77708
37803 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
37804 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
37805 New member functions.
37806 (format_directive): Used them.
37808 (pass_sprintf_length::handle_gimple_call): Same.
37809 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
37810 to avoid truncation for any argument.
37811 (extract_affine_mul): Same.
37812 * tree.c (get_file_function_name): Same.
37814 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37816 PR middle-end/77484
37817 * predict.def (PRED_INDIR_CALL): Set to 86.
37819 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
37821 PR preprocessor/54124
37822 * doc/cppopts.texi: Reformat -d subtable to list the full name
37823 of the options. Add cross-reference to the docs for the general
37824 compiler -d options.
37825 * doc/invoke.texi (Developer Options): Add cross-reference to the
37826 preprocessor-specific -d option documentation.
37828 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
37830 PR preprocessor/13498
37831 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
37832 redudant material, and reflect new command-line options.
37833 (System Headers): Likewise.
37835 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
37837 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
37838 -isystem, and -idirafter. Copy-edit.
37839 * doc/cppopts.texi: Copy-edit. Remove contradiction about
37840 default for -ftrack-macro-expansion. Delete obsolete and
37841 badly-formatted implementation details about -fdebug-cpp output.
37842 * doc/cppwarnopts.texi: Copy-edit.
37844 2017-01-07 David Malcolm <dmalcolm@redhat.com>
37847 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
37848 that the transition from a max line width >= 1<<10 to narrower
37849 lines works correctly.
37851 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
37853 * doc/options.texi (PerFunction): New.
37854 * opt-functions.awk (switch_flags): Map both Optimization and
37855 PerFunction to CL_OPTIMIZATION.
37856 * opth-gen.awk: Test for PerFunction flag along with
37858 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
37859 it only when the latter is present. Skip those that don't in
37860 the hash function generator.
37861 * common.opt (fvar-tracking): Mark as PerFunction instead of
37863 (fvar-tracking-assignments): Likewise.
37864 (fvar-tracking-assignments-toggle): Likewise.
37865 (fvar-tracking-uninit): Likewise.
37867 2017-01-07 Jakub Jelinek <jakub@redhat.com>
37869 PR translation/79018
37870 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
37873 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
37876 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
37877 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
37878 TARGET_LONG_JUMP_TABLE_OFFSETS.
37879 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
37880 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
37881 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
37882 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
37883 * config/m68k/m68k.md (tablejump expander): Likewise.
37884 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
37885 TARGET_LONG_JUMP_TABLE_OFFSETS.
37886 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
37887 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
37889 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
37890 David Holsgrove <david.holsgrove@xilinx.com>
37892 * common/config/microblaze/microblaze-common.c
37893 (TARGET_EXCEPT_UNWIND_INFO): Remove.
37894 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
37896 * config/microblaze/microblaze.c (microblaze_must_save_register)
37897 (microblaze_expand_epilogue, microblaze_return_addr): Handle
37899 (microblaze_eh_return): New function.
37900 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
37901 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
37902 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
37903 * config/microblaze/microblaze.md (eh_return): New pattern.
37905 2017-01-06 Jakub Jelinek <jakub@redhat.com>
37907 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
37908 GCC_DIAGNOSTIC_STRINGIFY): Define.
37910 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
37912 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37914 * config/arm/arm.md (<mcrr>): New.
37916 * config/arm/arm.c (arm_arch5te): New.
37917 (arm_option_override): Set arm_arch5te.
37918 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
37920 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
37921 (arm_mcrr_qualifiers): ... this. New.
37922 (MRRC_QUALIFIERS): Define to...
37923 (arm_mrrc_qualifiers): ... this. New.
37924 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
37926 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
37927 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
37928 (MRRCI, mrrc, MRRC): New.
37929 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
37930 VUNSPEC_MRRC2): New.
37932 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37934 * config/arm/arm.md (<mcr>): New.
37936 * config/arm/arm.c (arm_coproc_builtin_available): Add
37937 support for mcr, mrc, mcr2 and mrc2.
37938 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
37939 (arm_mcr_qualifiers): ... this. New.
37940 (MRC_QUALIFIERS): Define to ...
37941 (arm_mrc_qualifiers): ... this. New.
37942 (MCR_QUALIFIERS): Define to ...
37943 (arm_mcr_qualifiers): ... this. New.
37944 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
37946 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
37947 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
37948 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
37949 VUNSPEC_MRC2): New.
37951 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37953 * config/arm/arm.md (*ldc): New.
37957 * config/arm/arm.c (arm_coproc_builtin_available): Add
37958 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
37959 (arm_coproc_ldc_stc_legitimate_address): New.
37960 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
37961 'qualifier_const_pointer'.
37962 (LDC_QUALIFIERS): Define to...
37963 (arm_ldc_qualifiers): ... this. New.
37964 (STC_QUALIFIERS): Define to...
37965 (arm_stc_qualifiers): ... this. New.
37966 * config/arm/arm-protos.h
37967 (arm_coproc_ldc_stc_legitimate_address): New.
37968 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
37969 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
37970 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
37971 stc2, stcl, stc2l): New.
37972 * config/arm/constraints.md (Uz): New.
37973 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
37974 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
37975 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
37976 VUNSPEC_STC2L): New.
37978 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37980 * config/arm/arm.md (<cdp>): New.
37981 * config/arm/arm.c (neon_const_bounds): Rename this ...
37982 (arm_const_bounds): ... this.
37983 (arm_coproc_builtin_available): New.
37984 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
37985 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
37986 (CDP_QUALIFIERS): Define to...
37987 (arm_cdp_qualifiers): ... this. New.
37989 (arm_expand_builtin_args): Add case for 6 arguments.
37990 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
37991 (arm_const_bounds): ... this.
37992 (arm_coproc_builtin_available): New.
37993 * config/arm/arm_acle.h (__arm_cdp): New.
37995 * config/arm/arm_acle_builtins.def (cdp): New.
37997 * config/arm/iterators.md (CDPI,CDP,cdp): New.
37998 * config/arm/neon.md: Rename all 'neon_const_bounds' to
37999 'arm_const_bounds'.
38000 * config/arm/types.md (coproc): New.
38001 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
38002 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
38003 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
38004 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
38006 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38008 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
38009 (UBINOP_QUALIFIERS): New.
38011 (acle_builtin_data): New. Change comment.
38012 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
38013 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
38014 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
38015 arm_acle_builtins.def.
38016 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
38017 (arm_init_acle_builtins): New.
38018 (CRC32_BUILTIN): Remove.
38019 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
38020 crc32cb, crc32ch and crc32cw.
38021 (arm_init_crc32_builtins): Remove.
38022 (arm_init_builtins): Use arm_init_acle_builtins rather
38023 than arm_init_crc32_builtins.
38024 (arm_expand_acle_builtin): New.
38025 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
38026 * config/arm/arm_acle_builtins.def: New.
38028 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38030 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
38031 (arm_builtin_datum): ... this.
38032 (arm_init_neon_builtin): Rename to ...
38033 (arm_init_builtin): ... this. Add a new parameters PREFIX
38034 and USE_SIG_IN_NAME.
38035 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
38036 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
38037 'arm_builtin_datum'.
38038 (arm_init_vfp_builtins): Likewise.
38039 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
38040 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
38041 (arm_expand_neon_args): Rename to ...
38042 (arm_expand_builtin_args): ... this. Rename builtin_arg
38043 enum values and differentiate between ARG_BUILTIN_MEMORY
38044 and ARG_BUILTIN_NEON_MEMORY.
38045 (arm_expand_neon_builtin_1): Rename to ...
38046 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
38047 values, arm_expand_builtin_args and add bool parameter NEON.
38048 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
38049 (arm_expand_vfp_builtin): Likewise.
38050 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
38052 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38054 PR middle-end/77484
38055 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
38056 * predict.c (tree_estimate_probability_bb): Reverse direction of
38057 polymorphic call predictor.
38059 2017-01-06 David Malcolm <dmalcolm@redhat.com>
38061 * passes.c (execute_one_pass): Split out pass-skipping logic into...
38062 (determine_pass_name_match): ...this new function and...
38063 (should_skip_pass_p): ...this new function.
38065 2017-01-06 Nathan Sidwell <nathan@acm.org>
38067 * ipa-visibility.c (function_and_variable_visibility): Reformat
38068 comments and long lines. Remove extrneous if.
38069 * symtab.c (symtab_node::make_decl_local): Fix code format.
38070 (symtab_node::set_section_for_node): Fix comment typo.
38072 2017-01-06 Martin Liska <mliska@suse.cz>
38075 * lra-constraints.c: Rename invariant to lra_invariant.
38076 * predict.c (set_even_probabilities): Initialize e to NULL.
38078 2017-01-05 Martin Sebor <msebor@redhat.com>
38080 PR tree-optimization/78910
38081 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
38082 (format_integer): Correct off-by-one error in the handling
38083 of precision with negative numbers in signed conversions..
38085 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
38087 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
38089 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38091 PR tree-optimization/71016
38092 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
38093 factor_out_conditional_conversion. Formatting fix.
38094 (factor_out_conditional_conversion): Add cond_stmt argument.
38095 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
38096 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
38099 2017-01-05 David Malcolm <dmalcolm@redhat.com>
38101 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
38102 read-rtl-function.o, and selftest-rtl.o.
38103 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
38104 (selftest::aarch64_test_loading_full_dump): New function.
38105 (selftest::aarch64_run_selftests): New function.
38106 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
38107 selftest::aarch64_run_selftests.
38108 * config/i386/i386.c
38109 (selftest::ix86_test_loading_dump_fragment_1): New function.
38110 (selftest::ix86_test_loading_call_insn): New function.
38111 (selftest::ix86_test_loading_full_dump): New function.
38112 (selftest::ix86_test_loading_unspec): New function.
38113 (selftest::ix86_run_selftests): Call the new functions.
38114 * emit-rtl.c (maybe_set_max_label_num): New function.
38115 * emit-rtl.h (maybe_set_max_label_num): New decl.
38116 * function.c (instantiate_decls): Guard call to
38117 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
38118 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
38120 * gensupport.c (gen_reader::gen_reader): Pass "false"
38121 for new "compact" param of rtx_reader.
38122 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
38123 rather than an empty string for NULL strings.
38124 * read-md.c: Potentially include config.h rather than bconfig.h.
38125 Wrap include of errors.h with #ifdef GENERATOR_FILE.
38126 (have_error): New global, copied from errors.c.
38127 (md_reader::read_name): Rename to...
38128 (md_reader::read_name_1): ...this, adding "out_loc" param,
38129 and converting "missing name or number" to returning false, rather
38131 (md_reader::read_name): Reimplement in terms of read_name_1.
38132 (md_reader::read_name_or_nil): New function.
38133 (md_reader::read_string): Handle "(nil)" by returning NULL.
38134 (md_reader::md_reader): Add new param "compact".
38135 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
38136 (md_reader::read_file): New method.
38137 * read-md.h (md_reader::md_reader): Add new param "compact".
38138 (md_reader::read_file): New method.
38139 (md_reader::is_compact): New accessor.
38140 (md_reader::read_name): Convert return type from void to file_location.
38141 (md_reader::read_name_or_nil): New decl.
38142 (md_reader::read_name_1): New decl.
38143 (md_reader::m_compact): New field.
38144 (noop_reader::noop_reader): Pass "false" for new "compact" param
38146 (rtx_reader::rtx_reader): Add new "compact" param.
38147 (rtx_reader::read_rtx_operand): Make virtual and convert return
38148 type from void to rtx.
38149 (rtx_reader::read_until): New decl.
38150 (rtx_reader::handle_any_trailing_information): New virtual function.
38151 (rtx_reader::postprocess): New virtual function.
38152 (rtx_reader::finalize_string): New virtual function.
38153 (rtx_reader::m_in_call_function_usage): New field.
38154 (rtx_reader::m_reuse_rtx_by_id): New field.
38155 * read-rtl-function.c: New file.
38156 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
38157 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
38158 (selftest::verify_three_block_rtl_cfg): New decl.
38159 * read-rtl-function.h: New file.
38160 * read-rtl.c: Potentially include config.h rather than bconfig.h.
38161 For host, include function.h, memmodel.h, and emit-rtl.h.
38162 (one_time_initialization): New function.
38163 (struct compact_insn_name): New struct.
38164 (compact_insn_names): New array.
38165 (find_code): Handle insn codes in compact dumps.
38166 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
38167 (bind_subst_iter_and_attr): Likewise.
38168 (add_condition_to_string): Likewise.
38169 (add_condition_to_rtx): Likewise.
38170 (apply_attribute_uses): Likewise.
38171 (add_current_iterators): Likewise.
38172 (apply_iterators): Likewise.
38173 (initialize_iterators): Guard usage of apply_subst_iterator with
38174 #ifdef GENERATOR_FILE.
38175 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
38176 (md_reader::read_mapping): Likewise.
38177 (add_define_attr_for_define_subst): Likewise.
38178 (add_define_subst_attr): Likewise.
38179 (read_subst_mapping): Likewise.
38180 (check_code_iterator): Likewise.
38181 (rtx_reader::read_rtx): Likewise. Move one-time initialization
38183 (one_time_initialization): New function.
38184 (rtx_reader::read_until): New method.
38185 (read_flags): New function.
38186 (parse_reg_note_name): New function.
38187 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
38188 Handle reuse_rtx ids.
38189 Wrap iterator lookup within #ifdef GENERATOR_FILE.
38190 Add parsing support for RTL dumps, mirroring the special-cases in
38191 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
38192 values, and calling handle_any_trailing_information.
38193 (rtx_reader::read_rtx_operand): Convert return type from void
38194 to rtx, returning return_rtx. Handle case 'e'. Call
38195 finalize_string on XSTR and XTMPL fields.
38196 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
38197 "(nil)" values were omitted. Call the postprocess vfunc on the
38199 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
38200 class ctor. Initialize m_in_call_function_usage. Call
38201 one_time_initialization.
38202 * rtl-tests.c (selftest::test_uncond_jump): Call
38203 set_new_first_and_last_insn.
38204 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
38205 * selftest-rtl.c: New file.
38206 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
38207 (selftest::get_insn_by_uid): New decl.
38208 * selftest-run-tests.c (selftest::run_tests): Call
38209 read_rtl_function_c_tests.
38210 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
38211 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
38214 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
38216 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
38217 operands in a special way. Assert that pos+len <= mode precision.
38219 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38221 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
38222 3 argument Alias with unlimited for the negative form.
38223 (fno-vect-cost-model): Removed.
38225 2017-01-05 Martin Liska <mliska@suse.cz>
38227 * hsa-gen.c (gen_hsa_divmod): New function.
38228 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
38230 2017-01-05 Martin Liska <mliska@suse.cz>
38233 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
38236 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38238 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
38239 small constant length operands.
38241 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38243 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
38244 between loop iterations.
38246 2017-01-05 Martin Liska <mliska@suse.cz>
38249 * gimplify.c (gimplify_decl_expr): Compare to
38250 asan_poisoned_variables instread of checking flags.
38251 (gimplify_target_expr): Likewise.
38252 (gimplify_expr): Likewise.
38253 (gimplify_function_tree): Conditionally initialize
38254 asan_poisoned_variables.
38256 2017-01-04 Jeff Law <law@redhat.com>
38258 PR tree-optimizatin/78812
38259 * rtl.h (contains_mem_rtx_p): Prototype.
38260 * ifcvt.c (containts_mem_rtx_p): Move from here to...
38261 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
38262 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
38263 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
38264 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
38266 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38268 * input.c (assert_char_at_range): Default-initialize actual_range.
38270 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38272 * df-scan.c (df_ref_create_structure): Make regno unsigned,
38273 to match the caller.
38275 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38277 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
38278 insns after final jump in test to emit dummy move.
38280 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38282 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
38283 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
38285 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38287 * multiple_target.c (create_dispatcher_calls): Init e_next.
38288 * tree-ssa-loop-split.c (split_loop): Init border.
38289 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
38292 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
38297 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
38298 (altivec_register_operand): Do not return true if the operand
38299 contains a SUBREG mixing SImode and SFmode.
38300 (vsx_register_operand): Likewise.
38301 (vsx_reg_sfsubreg_ok): New predicate.
38302 (vfloat_operand): Do not return true if the operand contains a
38303 SUBREG mixing SImode and SFmode.
38304 (vint_operand): Likewise.
38305 (vlogical_operand): Likewise.
38306 (gpc_reg_operand): Likewise.
38307 (int_reg_operand): Likewise.
38308 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
38309 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
38310 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
38312 (rs6000_emit_move_si_sf_subreg): New helper function.
38313 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
38314 fixup SUBREGs involving SImode and SFmode.
38315 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
38316 numbers for the new peephole2 optimization.
38317 (peephole2 for SFmode unions): New peephole2 to optimize cases in
38318 the GLIBC math library that do AND/IOR/XOR operations on single
38319 precision floating point.
38320 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
38321 target macros to say whether we need to avoid SUBREGs mixing
38323 (TARGET_ALLOW_SF_SUBREG): Likewise.
38324 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
38325 (UNSPEC_SI_FROM_SF): Likewise.
38326 (iorxor): Change spacing.
38327 (and_ior_xor): New iterator for AND, IOR, and XOR.
38328 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
38329 (movdi_from_sf_zero_ext): Likewise.
38330 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
38331 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
38332 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
38333 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
38334 (fms<mode>4): Likewise.
38335 (fnma<mode>4): Likewise.
38336 (fnms<mode>4): Likewise.
38337 (nfma<mode>4): Likewise.
38338 (nfms<mode>4): Likewise.
38340 2017-01-04 Marek Polacek <polacek@redhat.com>
38343 * doc/invoke.texi: Document -Wpointer-compare.
38345 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38347 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
38350 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
38351 descriptions for -gdwarf-5 and emit them as uleb128 instead of
38354 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
38357 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
38358 documentation of the powerpc_popcntb_ok attribute.
38359 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
38360 code to issue warning messages if a requested CPU configuration is
38361 not supported by the binary (assembler and loader) toolchain.
38362 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
38363 made to define a built-in function that has been disabled.
38364 (paired_init_builtins): Add assertion to prevent ICE if attempt is
38365 made to define a built-in function that has been disabled.
38366 (altivec_init_builtins): Add comment explaining why definition
38367 of the DST built-in functions is not preceded by an assertion
38368 check. Add assertions to prevent ICE if attempts are made to
38369 define an altivec predicate or an abs* built-in function that has
38371 (htm_init_builtins): Add comment explaining why definition of the
38372 htm built-in functions is not preceded by an assertion check.
38374 2017-01-04 Jeff Law <law@redhat.com>
38376 PR tree-optimizatin/67955
38377 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
38378 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
38379 the points-to solution does not include pt_null. Use DECL_PT_UID
38382 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
38384 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
38385 Use gen_int_mode instead of gen_lopwart for const_int operands.
38387 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38389 PR tree-optimization/71563
38390 * match.pd: Simplify X << Y into X if Y is known to be 0 or
38391 out of range value - has low bits known to be zero.
38393 2017-01-04 Alan Modra <amodra@gmail.com>
38395 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
38396 * configure: Regenerate.
38397 * config.in: Regenerate.
38399 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38402 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
38403 a substring of the message, but strcmp with the whole message. Ifdef
38404 ENABLE_NLS, translate the message first using dgettext.
38406 2017-01-03 Jeff Law <law@redhat.com>
38408 PR tree-optimizatin/78856
38409 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
38410 (mark_threaded_blocks): Remove code to truncate thread paths that
38411 cross multiple loop headers. Instead invalidate the cached loop
38412 iteration information and handle case of a thread path walking
38413 into an irreducible region.
38415 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
38418 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
38419 assertions. Add support for doing the signbit if the IEEE 128-bit
38420 floating point value is in a GPR.
38421 * config/rs6000/rs6000.md (Fsignbit): Delete.
38422 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
38423 Update the length attribute if the value is in a GPR.
38424 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
38425 the sign or zero extension instruction, since the value is always 0/1.
38426 (signbit<mode>2_dm2): Delete using <Fsignbit>.
38429 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
38430 extracting SImode to a GPR register so that we can generate a
38431 store, limit the vector to be in a traditional Altivec register
38432 for the vextuwrx instruction.
38434 2017-01-03 Ian Lance Taylor <iant@google.com>
38436 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
38438 2017-01-03 Martin Sebor <msebor@redhat.com>
38440 PR tree-optimization/78696
38441 * gimple-ssa-sprintf.c (format_floating): Correct handling of
38442 precision. Use MPFR for %f for greater fidelity. Correct handling
38444 (pass_sprintf_length::compute_format_length): Set width and precision
38445 specified by asrerisk to void_node for vararg functions.
38446 (try_substitute_return_value): Adjust dump output.
38448 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
38450 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
38452 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
38454 * doc/invoke.texi (SPARC options): Document -mlra as the default.
38455 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
38456 -mlra/-mno-lra was passed to the compiler.
38458 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
38460 PR rtl-optimization/65618
38461 * emit-rtl.c (try_split): Move initialization of "before" and
38462 "after" to just before the call to emit_insn_after_setloc.
38464 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
38466 * doc/md.texi (Standard Names): Remove reference to Java frontend.
38468 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
38470 * dwarf2out.c (gen_enumeration_type_die): When
38471 -gno-strict-dwarf, add a DW_AT_encoding attribute.
38473 2017-01-03 Jakub Jelinek <jakub@redhat.com>
38475 PR tree-optimization/78965
38476 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
38477 Change first argument from const call_info & to call_info &. For %n
38478 set info.nowrite to false.
38480 PR middle-end/78901
38481 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
38482 possibly throwing calls.
38484 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
38485 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
38486 and fns handling, rather than in a separate case SSA_NAME.
38488 2017-01-02 Jeff Law <law@redhat.com>
38490 * config/darwin-driver.c (darwin_driver_init): Const-correctness
38491 fixes for first_period and second_period variables.
38493 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
38496 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
38497 (*insvqi_1): New insn pattern.
38498 (*insvqi_1_mem_rex64): Ditto.
38499 (*insvqi_2): Ditto.
38500 (*insvqi_3): Rename from *insvqi.
38502 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
38504 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
38506 * doc/cfg.texi (Edges): Remove reference to Java.
38507 (Maintaining the CFG): Ditto.
38509 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38511 PR middle-end/77674
38512 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
38513 transparent aliases.
38515 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38517 PR middle-end/77484
38518 * predict.def (PRED_CALL): Update hitrate.
38519 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
38520 * predict.c (tree_estimate_probability_bb): Split CALL predictor
38521 into direct/indirect/polymorphic variants.
38523 2017-01-01 Jakub Jelinek <jakub@redhat.com>
38525 Update copyright years.
38527 * gcc.c (process_command): Update copyright notice dates.
38528 * gcov-dump.c (print_version): Ditto.
38529 * gcov.c (print_version): Ditto.
38530 * gcov-tool.c (print_version): Ditto.
38531 * gengtype.c (create_file): Ditto.
38532 * doc/cpp.texi: Bump @copying's copyright year.
38533 * doc/cppinternals.texi: Ditto.
38534 * doc/gcc.texi: Ditto.
38535 * doc/gccint.texi: Ditto.
38536 * doc/gcov.texi: Ditto.
38537 * doc/install.texi: Ditto.
38538 * doc/invoke.texi: Ditto.
38540 Copyright (C) 2017 Free Software Foundation, Inc.
38542 Copying and distribution of this file, with or without modification,
38543 are permitted in any medium without royalty provided the copyright
38544 notice and this notice are preserved.