1 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
3 * invoke.texi (Wsuggest-attribute=cold): Document.
4 * common.opt (Wsuggest-attribute=cold): New
5 * ipa-pure-const.c (warn_function_cold): New function.
6 * predict.c (compute_function_frequency): Use it.
7 * predict.h (warn_function_cold): Declare.
9 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
11 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
14 2017-10-06 Martin Liska <mliska@suse.cz>
16 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
17 keyword for member functions.
18 (struct sanopt_tree_couple): New struct.
19 (struct sanopt_tree_couple_hash): New function.
20 (struct sanopt_ctx): Add new hash_map.
21 (has_dominating_ubsan_ptr_check): New function.
22 (record_ubsan_ptr_check_stmt): Likewise.
23 (maybe_optimize_ubsan_ptr_ifn): Likewise.
24 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
25 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
27 2017-10-06 Sudakshina Das <sudi.das@arm.com>
30 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
31 aarch64_simd_valid_immediate on CONST_VECTORs.
32 (aarch64_reg_or_bic_imm): Likewise.
34 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
36 PR rtl-optimization/82396
37 * haifa-sched.c (ready_sort_real): Disable qsort checking.
39 2017-10-06 Sebastian Pop <sebpop@gmail.com>
41 * graphite-dependences.c (scop_get_reads): Move code to...
42 (scop_get_must_writes): Move code to...
43 (scop_get_may_writes): Move code to...
44 (scop_get_reads_and_writes): ... here.
45 (scop_get_dependences): Call scop_get_reads_and_writes.
47 2017-10-06 Jakub Jelinek <jakub@redhat.com>
49 PR tree-optimization/82434
50 * fold-const.h (can_native_encode_type_p,
51 can_native_encode_string_p): Remove.
52 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
53 don't encode anything, just return what would be otherwise returned.
54 (native_encode_fixed, native_encode_complex, native_encode_vector):
56 (native_encode_string): Likewise. Inline by hand
57 can_native_encode_string_p.
58 (can_native_encode_type_p): Remove.
59 (can_native_encode_string_p): Remove.
60 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
61 STRING_CSTs using can_native_encode_string_p, test all
62 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
63 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
64 argument from native_encode_expr.
65 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
66 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
69 2017-10-06 Richard Biener <rguenther@suse.de>
71 PR tree-optimization/82397
72 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
73 operand_equal_p but rely on data_ref_compare_tree for detecting
75 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
76 to match up with dr_group_sort_cmp.
78 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
81 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
83 * config/s390/s390-builtin-types.def: Regenerate.
85 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
88 * config/s390/s390-builtin-types.def: Regenerate.
89 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
90 Change flag from B_VXE to B_VX.
91 (s390_vec_min_dbl): Remove B_VXE flag.
93 2017-10-06 Richard Biener <rguenther@suse.de>
95 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
96 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
97 translate_isl_ast_to_gimple::is_valid_rename,
98 translate_isl_ast_to_gimple::get_rename,
99 translate_isl_ast_to_gimple::get_def_bb_for_const,
100 translate_isl_ast_to_gimple::get_new_name,
101 translate_isl_ast_to_gimple::collect_all_ssa_names,
102 translate_isl_ast_to_gimple::copy_loop_phi_args,
103 translate_isl_ast_to_gimple::collect_all_ssa_names,
104 translate_isl_ast_to_gimple::copy_loop_phi_args,
105 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
106 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
107 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
108 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
109 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
110 translate_isl_ast_to_gimple::copy_cond_phi_args,
111 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
112 translate_isl_ast_to_gimple::edge_for_new_close_phis,
113 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
114 translate_isl_ast_to_gimple::rename_uses,
115 translate_isl_ast_to_gimple::rename_all_uses): Remove.
116 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
117 (set_rename_for_each_def): Likewise.
118 (graphite_copy_stmts_from_block): Handle debug stmt resetting
119 here. Handle rewriting SCEV analyzable uses here.
120 (copy_bb_and_scalar_dependences): Generate code for PHI
122 (graphite_regenerate_ast_isl): Adjust.
123 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
124 (add_write, add_read): New functions.
125 (build_cross_bb_scalars_def): Use it and simplify.
126 (build_cross_bb_scalars_use): Likewise.
127 (graphite_find_cross_bb_scalar_vars): Inline into...
128 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
129 simulating out-of-SSA. Compute liveout and add dependencies.
130 (build_scops): Force an empty entry block.
131 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
133 (sese_build_liveouts): Declare.
134 (sese_trivially_empty_bb_p): Likewise.
135 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
136 compute liveout and debug_liveout.
137 (sese_bad_liveouts_use): Remove.
138 (sese_reset_debug_liveouts_bb): Likewise.
139 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
140 (sese_build_liveouts): Build liveout and debug_liveout and store
142 (new_sese_info): Adjust.
143 (free_sese_info): Likewise.
144 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
145 do not build liveout here.
146 (move_sese_in_condition): Adjust region entry.
147 (scev_analyzable_p): Match up with chrec_apply requirements.
148 (sese_trivially_empty_bb_p): New.
149 * tree-into-ssa.c (get_reaching_def): Properly support generating
150 default-defs for incremental rewrite of anonymous names.
152 2017-10-06 Richard Biener <rguenther@suse.de>
154 * graphite-sese-to-poly.c (extract_affine): For casts increasing
155 precision do not perform modulo reduction.
157 2017-10-06 Richard Biener <rguenther@suse.de>
159 PR tree-optimization/82436
160 * tree-vect-slp.c (vect_supported_load_permutation_p): More
161 conservatively choose the vectorization factor when checking
162 whether we can perform the required load permutation.
163 (vect_transform_slp_perm_load): Assert when we may not fail.
165 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
167 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
168 message for incompatible -msdata=* and -mcall-* options.
170 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
172 * i386.c (ia32_multipass_dfa_lookahead): Default to issue rate
173 for post-reload scheduling.
175 2017-10-05 Tamar Christina <tamar.christina@arm.com>
177 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
179 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
181 * i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
182 to improve monte carlo in scimark.
184 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
186 * i386.c (ix86_size_cost, i386_cost, i486_cost, pentium_cost,
187 lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
188 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
189 pentium4_cost, nocona_cost): Set reassociation width to 1.
190 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
191 width to 2 for fp operations and 1 otherwise.
192 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
194 (atom_cost): Set reassociation width to 2.
195 (slm_cost, generic_cost): Set fp reassociation width to 2 and 1 otherwise.
196 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
197 (core_cost): Set fp reassociation width to 4 and vector to 2.
198 (ix86_reassociation_width): Rewrite using cost table; special case
199 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
200 and TARGET_AVX128_OPTIMAL.
201 * i386.h (processor_costs): Add
202 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
203 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
204 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
205 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
206 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
207 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
209 2017-10-05 Nathan Sidwell <nathan@acm.org>
211 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
213 2017-10-05 Tamar Christina <tamar.christina@arm.com>
215 * config/arm/arm.c (arm_test_fpu_data): New.
216 (arm_run_selftests): Call arm_test_fpu_data.
218 2017-10-04 Nathan Sidwell <nathan@acm.org>
220 * toplev.c (toplev::main): Remove excess parens on pretty_printer
222 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
224 2017-10-04 Sudakshina Das <sudi.das@arm.com>
226 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
227 check type for aarch64_simd_valid_immediate.
228 (aarch64_output_simd_mov_immediate): Update prototype.
229 (aarch64_simd_valid_immediate): Update prototype.
230 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
231 support for ORR-immediate.
232 (and<mode>3): modified pattern to add support for BIC-immediate.
233 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
234 now checks for valid immediate for BIC and ORR based on new enum
236 (aarch64_output_simd_mov_immediate): Function now used to output
237 BIC/ORR imm as well based on new enum argument.
238 * config/aarch64/constraints.md (Do): New vector immediate constraint.
240 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
241 (aarch64_reg_or_bic_imm): Likewise.
243 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
245 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
246 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
248 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
252 PR rtl-optimization/82396
253 * haifa-sched.c (autopref_multipass_init): Simplify
255 (autopref_rank_data): Simplify sort order.
256 * sched-int.h (autopref_multipass_data_): Remove
257 multi_mem_insn_p, min_offset and max_offset.
259 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
261 * doc/sourcebuild.texi: Document vect_peeling_profitable.
263 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
265 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
268 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
270 * doc/sourcebuild.texi: Document vect_long_mult.
272 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
274 PR tree-optimization/82413
275 * fold-const.c (build_range_check): Use widest_int when comparing
276 the maximum ETYPE value with HIGH.
278 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
280 PR rtl-optimization/82396
281 * haifa-sched.c (autopref_multipass_init): Simplify
283 (autopref_rank_data): Simplify sort order.
284 * sched-int.h (autopref_multipass_data_): Remove
285 multi_mem_insn_p, min_offset and max_offset.
287 2017-10-04 Jakub Jelinek <jakub@redhat.com>
289 PR tree-optimization/82381
290 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
291 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
294 PR tree-optimization/82374
295 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
296 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
297 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
298 current_function_decl to the new decl.
300 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
302 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
303 helper macro for IEEE float128 hardware built-in functions.
304 (SQRTF128_ODD): Add built-in functions with the round-to-odd
306 (TRUNCF128_ODD): Likewise.
307 (ADDF128_ODD): Likewise.
308 (SUBF128_ODD): Likewise.
309 (MULF128_ODD): Likewise.
310 (DIVF128_ODD): Likewise.
311 (FMAF128_ODD): Likewise.
312 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
313 UNSPEC_TRUNC_ROUND_TO_ODD.
314 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
315 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
316 floating point round to odd instructions.
317 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
318 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
319 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
320 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
321 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
322 (trunc<mode>sf2_hw): Change the truncate with round to odd
323 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
324 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
325 to odd hardware instructions.
326 (sub<mode>3_odd): Likewise.
327 (mul<mode>3_odd): Likewise.
328 (div<mode>3_odd): Likewise.
329 (sqrt<mode>2_odd): Likewise.
330 (fma<mode>4_odd): Likewise.
331 (fms<mode>4_odd): Likewise.
332 (nfma<mode>4_odd): Likewise.
333 (nfms<mode>4_odd): Likewise.
334 (trunc<mode>df2_odd): Change the truncate with round to odd
335 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
337 * doc/extend.texi (PowerPC built-in functions): Update documentation
338 for existing IEEE float128-bit built-in functions. Add built-in
339 functions that generate the IEEE 128-bit floating point round to
342 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
344 PR rtl-optimization/77729
345 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
346 to (X&(C1&~C2))|C2 transformations.
348 2017-10-03 Martin Jambor <mjambor@suse.cz>
350 PR tree-optimization/82363
351 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
352 mismatch, mark lacc written regardless of racc.
354 2017-10-03 Jakub Jelinek <jakub@redhat.com>
356 PR tree-optimization/82381
357 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
358 stmt_to_insert nor wheather SSA_NAMEs are default defs.
359 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
360 fallthrough into reassoc_stmt_dominates_stmt_p.
363 * combine.c (combine_instructions): Don't combine in unreachable
366 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
369 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
370 function to not use the have_cpu variable. Do not set cpu_index,
371 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
373 (rs6000_valid_attribute_p): Remove duplicate initializations of
374 old_optimize and func_optimize.
375 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
376 (rs6000_activate_target_options): Make global.
377 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
380 2017-10-02 Jakub Jelinek <jakub@redhat.com>
382 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
383 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
384 Set *poffset to 0 and *psize and *pmax_size to -1 if
385 *poffset + *psize overflows in HOST_WIDE_INT.
387 PR tree-optimization/82387
388 PR tree-optimization/82388
389 PR tree-optimization/82389
390 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
391 instead of live_bytes non-NULL.
393 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
396 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
399 2017-10-02 Richard Biener <rguenther@suse.de>
401 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
402 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
403 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
405 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
407 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
408 requested precision matches the type's.
409 * calls.c (alloc_max_size): Calculate the new candidate size as
410 a widest_int and use wi::to_widest when comparing it with the
411 current candidate size.
412 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
413 zero rather than integer_zero_node.
414 * match.pd: Check for a no-op conversion before using wi::add
415 rather than after. Use tree_to_uhwi when summing small shift
416 counts into an unsigned int.
418 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
419 Alan Hayward <alan.hayward@arm.com>
420 David Sherwood <david.sherwood@arm.com>
423 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
424 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
425 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
428 2017-10-02 Richard Biener <rguenther@suse.de>
430 PR tree-optimization/82355
431 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
432 a mapping for the enclosing loop but avoid generating one for
434 (copy_bb_and_scalar_dependences): Remove premature codegen
435 error on PHIs in blocks duplicated into multiple places.
436 * graphite-scop-detection.c
437 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
438 in the region use it as loop and nest to analyze the DR in.
439 (try_generate_gimple_bb): Likewise.
440 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
441 (add_loop_constraints): For blocks in a loop not in the region
442 create a dimension with a single iteration.
443 * sese.h (gbb_loop_at_index): Remove assert.
445 2017-10-01 Kevin Buettner <kevinb@redhat.com>
447 * omp-expand.c (adjust_context_scope): New function.
448 (expand_parallel_call): Call adjust_context_scope.
450 2017-10-01 Jeff Law <law@redhat.com>
452 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
453 dom_opt_dom_walker class with direct access to private members.
454 Add comments. Call test_for_singularity.
455 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
456 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
457 m_dummy_cond anymore.
458 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
460 (pass_dominator:execute): Build the dummy_cond here and pass it
461 to the dom_opt_dom_walker ctor.
462 (test_for_singularity): New function.
464 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
465 Maya Rashish <coypu@sdf.org>
467 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
468 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
469 (alpha*-*-netbsd*) Use nbsd_tm_file.
470 (arm*-*-netbsdelf*) Likewise.
471 (i[34567]86-*-netbsdelf*) Likewise.
472 (x86_64-*-netbsd*) Likewise.
473 (mips*-*-netbsd*) Likewise.
474 (powerpc-*-netbsd*) Likewise.
475 (sh*-*-netbsd*) Likewise.
476 (sparc-*-netbsdelf*) Likewise.
477 (sparc64-*-netbsd*) Likewise.
478 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
480 (vax-*-netbsdelf*) Likewise.
481 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
482 (UINT_FAST8_TYPE) Likewise.
483 (INT_FAST16_TYPE) Check CHAR_FAST16.
484 (UINT_FAST16_TYPE) Likewise.
486 2017-09-30 Jakub Jelinek <jakub@redhat.com>
489 * config/i386/i386.md
490 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
491 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
492 *divmodsi4_zext_2): New define_insn_and_split.
493 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
494 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
495 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
496 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
497 New define_insn_and_split.
498 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
499 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
500 operands[1] having DImode when mode is SImode.
502 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
503 always SImode for DIV and MOD in REG_EQUAL notes.
505 2017-09-29 Yury Gribov <tetra2005@gmail.com>
508 * match.pd: Fix handling of NaNs in pattern.
510 2017-09-29 Jeff Law <law@redhat.com>
512 * sbitmap.c (bitmap_bit_in_range_p): New function.
513 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
514 * tree-ssa-dse.c (live_bytes_read): New function.
515 (dse_classify_store): Ignore reads of dead bytes.
517 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
518 typos and whitespace errors.
519 * config/i386/predicates.md (address_no_seg_operand): Likewise.
520 * config/s390/s390.c (s390_emit_prologue): Likewise.
522 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
525 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
526 with a symbol for LRA.
528 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
530 PR rtl-optimization/82338
531 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
533 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
535 * genmodes.c (calc_wider_mode): Suppress qsort macro.
536 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
537 (qsort_chk): Declare.
538 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
539 (qsort_chk): New function.
541 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
543 PR tree-optimization/82337
544 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
545 phi definition if the PHI result appears in an abnormal PHI.
546 (find_basis_for_base_expr): Don't record a basis if the LHS of the
547 basis appears in an abnormal PHI.
549 2017-09-29 Richard Biener <rguenther@suse.de>
551 * graphite-isl-ast-to-gimple.c
552 (translate_isl_ast_to_gimple::set_codegen_error): New function.
553 (binary_op_to_tree): Use it.
554 (get_rename_from_scev): Likewise.
555 (copy_loop_phi_nodes): Likewise.
556 (copy_bb_and_scalar_dependences): Likewise.
557 (translate_pending_phi_nodes): Likewise.
559 2017-09-29 Jakub Jelinek <jakub@redhat.com>
562 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
563 for movabsq $(i32 << shift), r64.
565 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
567 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
568 index when encoding %esp as %rsp to avoid 0x67 prefix.
570 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
572 * config/i386/i386.md (*movsf_internal, *movdf_internal):
573 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
575 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
577 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
578 Extensions with more than 16 double VFP registers.
579 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
580 to_clear_mask and all code related to it. Replace the remaining
581 entry by a sbitmap and adapt code accordingly.
583 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
585 * brig-builtins.def: Change pure attributes to const.
587 2017-09-28 Joseph Myers <joseph@codesourcery.com>
589 * config.gcc (default_gnu_indirect_function): Default to yes for
590 sparc*-*-linux* with glibc.
592 2017-09-28 Joseph Myers <joseph@codesourcery.com>
594 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
595 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
596 when creating .init_array and .fini_array sections with priority
599 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
602 * config/aarch64/aarch64.c
603 (aarch64_builtin_support_vector_misalignment): Always return false
604 when misalignment is unknown.
606 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
608 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
609 this function to return false if the definition used by the swap
610 instruction is artificial, or if the memory address from which the
611 constant value is loaded is not represented by a base address held
612 in a register or if the base address register is a frame or stack
613 pointer. Additionally, return false if the base address of the
614 loaded constant is a SYMBOL_REF but is not considered to be a
616 (replace_swapped_load_constant): New function.
617 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
618 loaded constant vector with a load of a swapped constant vector.
620 2017-09-27 Carl Love <cel@us.ibm.com>
622 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
623 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
624 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
627 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
629 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
630 first, always call autopref_rank_data otherwise.
632 2017-09-27 Richard Biener <rguenther@suse.de>
634 * graphite-scop-detection.c (find_scop_parameters): Move
635 loop bound handling ...
636 (gather_bbs::before_dom_children): ... here, avoiding the need
637 to build scop_info->loop_nest.
638 (record_loop_in_sese): Remove.
639 * sese.h (sese_info_t::loop_nest): Remove.
640 * sese.c (new_sese_info): Do not allocate loop_nest.
641 (free_sese_info): Do not free loop_nest.
643 2017-09-27 Jakub Jelinek <jakub@redhat.com>
646 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
647 lhs from calls if the lhs has addressable type.
649 2017-09-27 Richard Biener <rguenther@suse.de>
651 * graphite.h (scop::max_alias_set): New member.
652 * graphite-scop-detection.c: Remove references to non-existing
654 (build_alias_sets): Record the maximum alias set used for drs.
655 (build_scops): Support zero as unlimited for
656 --param graphite-max-arrays-per-scop.
657 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
659 (build_poly_sr_1): ... here. Compute alias set based on the
660 maximum alias set used for drs rather than
661 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
663 2017-09-27 Richard Biener <rguenther@suse.de>
665 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
666 --param loop-block-tile-size=0 to disable tiling.
668 2017-09-27 Richard Biener <rguenther@suse.de>
670 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
671 (graphite-max-nb-scop-params): Document special value zero.
672 * domwalk.h (dom_walker::STOP): New symbolical constant.
673 (dom_walker::dom_walker): Add optional parameter for bb to
675 (dom_walker::~dom_walker): Declare.
676 (dom_walker::before_dom_children): Document STOP return value.
677 (dom_walker::m_user_bb_to_rpo): New member.
678 (dom_walker::m_bb_to_rpo): Likewise.
679 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
680 mapping here if not provided by the user.
681 (dom_walker::~dom_walker): Free bb to RPO mapping if not
682 provided by the user.
683 (dom_walker::STOP): Define.
684 (dom_walker::walk): Do not compute bb to RPO mapping here.
685 Support STOP return value from before_dom_children to stop
687 * graphite-optimize-isl.c (optimize_isl): If the schedule
688 is the same still generate code if -fgraphite-identity
689 or -floop-parallelize-all are given.
690 * graphite-scop-detection.c: Include cfganal.h.
691 (gather_bbs::gather_bbs): Get and pass through bb to RPO
693 (gather_bbs::before_dom_children): Return STOP for BBs
695 (build_scops): Compute bb to RPO mapping and pass it to
696 the domwalk. Treat --param graphite-max-nb-scop-params=0
697 as not limiting the number of params.
698 * graphite.c (graphite_initialize): Remove limit on the
699 number of basic-blocks in a function.
700 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
701 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
704 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
706 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
707 Adjust code to eliminate needing to do the shift right 32-bits
708 operation after XSCVDPSPN.
710 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
712 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
713 ((X / Y) != 0 -> X >= Y): Likewise.
715 2017-09-26 Carl Love <cel@us.ibm.com>
717 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
718 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
719 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
720 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
721 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
722 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
723 definitions and overloading.
724 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
725 statement for P9V_BUILTIN_XST_LEN_R.
726 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
727 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
728 define_expand and define_insn for the instructions and builtins.
729 * doc/extend.texi: Update the built-in documentation file for the new
731 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
732 define_insn for the instructions
734 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
737 * gcc/config/netbsd-protos.h: New file.
738 * gcc/config/netbsd.c: New file.
739 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
740 * gcc/config/t-netbsd: New file.
741 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
742 (tmake_file) Add t-netbsd.
743 (extra_objs) Add netbsd.o.
745 2017-09-26 Janus Weil <janus@gcc.gnu.org>
749 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
751 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
753 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
754 sign extension from a vector register to a GPR by doing a 32-bit
755 direct move and then an EXTSW.
756 (extendsi<mode>2 splitter): Likewise.
757 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
758 right or vector extract after doing XSCVDPSPN. Use
759 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
761 (movdi_from_sf_zero_ext): Likewise.
762 (reload_gpr_from_vsxsf): Likewise.
763 (p8_mfvsrd_4_disf): Delete, no longer used.
764 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
765 then needing to move the SFmode to a GPR to use the XSCVDPSP
766 instruction instead of FRSP and XSCVDPSPN.
767 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
768 it is adjacent to the other XSCVSPDP insns.
769 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
770 SFmode to be in traditional Altivec registers.
771 (vsx_xscvdpspn): Eliminate useless alternative constraint.
772 (vsx_xscvspdpn): Likewise.
773 (vsx_xscvspdpn_scalar): Likewise.
775 2017-09-26 Martin Jambor <mjambor@suse.cz>
777 * tree-sra.c (compare_access_positions): Put integral types first,
778 stabilize sorting of integral types, remove conditions putting
779 non-full-precision integers last.
780 (sort_and_splice_var_accesses): Disable scalarization if a
781 non-integert would be represented by a non-full-precision integer.
783 2017-09-26 Joseph Myers <joseph@codesourcery.com>
785 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
786 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
787 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
788 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
789 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
790 conditionals inside the function instead of around it. Call
791 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
792 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
794 2017-09-26 Richard Biener <rguenther@suse.de>
796 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
798 (scop_detection::build_scop_breadth): ... this. Removed.
799 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
800 (scop_detection::harmful_stmt_in_bb): Likewise.
801 (scop_detection::graphite_can_represent_stmt): Likewise.
802 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
803 (scop_detection::can_represent_loop): Remove recursion, fold in ...
804 (scop_detection::can_represent_loop_1): ... this. Removed.
805 (scop_detection::harmful_loop_in_region): Simplify after inlining
806 the above and remove more quadraticness.
807 (build_scops): Adjust.
808 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
811 2017-09-26 Jakub Jelinek <jakub@redhat.com>
814 * config/i386/i386.c (ix86_print_operand_address_as): Only test
815 REGNO (base) == SP_REG if base is a REG.
818 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
819 if it is different SSA_NAME.
820 (optimize_range_tests_cmp_bitwise): New function.
821 (optimize_range_tests): Call it.
823 2017-09-26 Richard Biener <rguenther@suse.de>
825 PR tree-optimization/82321
826 * graphite.c (canonicalize_loop_closed_ssa): Properly check
827 for the def being inside the loop.
829 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
831 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
833 * config/s390/s390-builtins.def: Fix constraint on op4.
835 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
837 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
838 independent expanders.
839 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
840 ("vec_ordered", "vec_unordered"): New expanders.
842 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
844 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
847 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
849 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
850 vec_unpacks_lo_v16qi.
851 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
853 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
855 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
856 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
857 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
859 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
861 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
863 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
865 ("*vec_slb<mode>"): New insn pattern.
866 ("vec_shr_<mode>"): New expander.
867 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
868 and force the shift count operand to V16QImode.
869 ("vec_srb<mode>"): Set shift count mode to V16QI.
871 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
873 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
874 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
875 ("vec_widen_smult_hi_<mode>"): New expander definitions.
877 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
880 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
882 2017-09-26 Richard Biener <rguenther@suse.de>
884 PR tree-optimization/82320
885 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
888 2017-09-25 Jeff Law <law@redhat.com>
890 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
891 prototype for new argument.
892 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
893 mostly extracted from rs6000_emit_allocate_stack.
894 (rs6000_emit_probe_stack_range_stack_clash): New function.
895 (rs6000_emit_allocate_stack): Call
896 rs6000_emit_probe_stack_range_stack_clash as needed.
897 (rs6000_emit_probe_stack_range): Add additional argument
898 to call to gen_probe_stack_range{si,di}.
899 (output_probe_stack_range): New.
900 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
901 (output_probe_stack_range_stack_clash): New.
902 (rs6000_emit_prologue): Emit notes into dump file as requested.
903 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
904 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
905 Add additional operand and pass it to output_probe_stack_range.
907 2017-09-25 Bin Cheng <bin.cheng@arm.com>
909 PR tree-optimization/82163
910 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
911 (checking_verify_loop_closed_ssa): New parameter.
912 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
913 (check_loop_closed_ssa_stmt): Delete.
914 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
915 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
916 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
919 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
921 * brig-builtins.def: Treat HSAIL barrier builtins as
922 setjmp/longjump style functions.
924 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
926 * target.def (constant_alignment): New hook.
927 * defaults.h (CONSTANT_ALIGNMENT): Delete.
928 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
929 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
930 * doc/tm.texi: Regenerate.
931 * targhooks.h (default_constant_alignment): Declare.
932 (constant_alignment_word_strings): Likewise.
933 * targhooks.c (default_constant_alignment): New function.
934 (constant_alignment_word_strings): Likewise.
935 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
936 instead of CONSTANT_ALIGNMENT.
937 * varasm.c (align_variable, get_variable_align, build_constant_desc)
938 (force_const_mem): Likewise.
939 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
940 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
941 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
942 (TARGET_CONSTANT_ALIGNMENT): Redefine.
943 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
945 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
946 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
947 constant_alignment_word_strings.
948 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
949 (CONSTANT_ALIGNMENT): Likewise.
950 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
951 (arm_constant_alignment): New function.
952 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
953 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
954 constant_alignment_word_strings.
955 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
956 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
957 constant_alignment_word_strings.
958 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
959 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
960 (cris_constant_alignment): New function.
961 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
962 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
963 (epiphany_constant_alignment): New function.
964 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
965 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
966 constant_alignment_word_strings.
967 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
968 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
969 constant_alignment_word_strings.
970 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
971 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
972 constant_alignment_word_strings.
973 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
974 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
975 * config/i386/i386.c (ix86_constant_alignment): Make static.
976 Use the same interface as the target hook.
977 (TARGET_CONSTANT_ALIGNMENT): Redefine.
978 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
979 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
980 constant_alignment_word_strings.
981 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
982 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
983 (TARGET_CONSTANT_ALIGNMENT): Redefine.
984 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
985 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
986 constant_alignment_word_strings.
987 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
988 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
989 constant_alignment_word_strings.
990 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
991 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
992 constant_alignment_word_strings.
993 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
994 * config/microblaze/microblaze.c (microblaze_constant_alignment):
996 (TARGET_CONSTANT_ALIGNMENT): Redefine.
997 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
998 * config/mips/mips.c (mips_constant_alignment): New function.
999 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1000 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
1001 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
1002 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1003 (mmix_constant_alignment): Make static. Use the same interface
1005 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
1006 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1007 constant_alignment_word_strings.
1008 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
1009 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1010 constant_alignment_word_strings.
1011 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
1012 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1013 constant_alignment_word_strings.
1014 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
1015 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1016 (rs6000_constant_alignment): New function.
1017 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
1018 * config/riscv/riscv.c (riscv_constant_alignment): New function.
1019 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1020 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
1021 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1022 (rs6000_constant_alignment): New function.
1023 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
1024 * config/s390/s390.c (s390_constant_alignment): New function.
1025 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1026 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
1027 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1028 constant_alignment_word_strings.
1029 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
1030 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1031 (sparc_constant_alignment): New function.
1032 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
1033 * config/spu/spu.c (spu_constant_alignment): New function.
1034 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1035 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
1036 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1037 constant_alignment_word_strings.
1038 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
1039 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1040 constant_alignment_word_strings.
1041 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
1042 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1043 constant_alignment_word_strings.
1044 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
1045 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1046 (visium_constant_alignment): New function.
1047 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
1048 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1049 (xtensa_constant_alignment): New function.
1050 * system.h (CONSTANT_ALIGNMENT): Poison.
1052 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
1054 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
1055 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
1056 (rs6000_builtin_valid_without_lhs): New helper function.
1057 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1058 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
1060 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
1062 * target.h (vec_perm_indices): Use unsigned short rather than
1064 (auto_vec_perm_indices): Likewise.
1065 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
1066 Use unsigned int rather than unsigned char.
1067 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
1069 2017-09-25 Richard Biener <rguenther@suse.de>
1071 * cfgloop.h (sort_sibling_loops): Declare.
1072 * cfgloop.c (sort_sibling_loops_cmp): New helper.
1073 (sort_sibling_loops): New function sorting the sibling loop list
1075 * graphite.c (graphite_transform_loops): Sort sibling loops.
1077 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
1079 * target.def (vec_perm_const_ok): Change sel parameter to
1081 * optabs-query.c (can_vec_perm_p): Update accordingly.
1082 * doc/tm.texi: Regenerate.
1083 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
1084 auto_vec_perm_indices and remove separate nelt field.
1085 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
1086 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
1087 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
1088 (aarch64_expand_vec_perm_const): Update accordingly.
1089 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
1090 to vec_perm_indices.
1091 * config/arm/arm.c (expand_vec_perm_d): Change perm to
1092 auto_vec_perm_indices and remove separate nelt field.
1093 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
1094 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
1095 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
1097 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
1098 to vec_perm_indices.
1099 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
1100 sel to vec_perm_indices.
1101 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
1102 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
1103 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
1105 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
1108 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
1111 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
1112 on the FUNCTION_DECL function context if it has a DIE that is a
1115 2017-09-25 Richard Biener <rguenther@suse.de>
1117 PR tree-optimization/82285
1118 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
1121 2017-09-25 Tom de Vries <tom@codesourcery.com>
1125 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
1128 2017-09-25 Richard Biener <rguenther@suse.de>
1130 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
1131 ISL errors other than isl_error_quota happen. Dump if the
1132 schedule is the same.
1133 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
1134 errors instead of aborting inside ISL.
1136 2017-09-25 Iain Sandoe <iain@codesourcery.com>
1139 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
1140 of libgcc_eh for m64.
1141 * config/i386/darwin64.h: Likewise.
1143 2017-09-25 Richard Biener <rguenther@suse.de>
1146 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
1147 attribute for incomplete types nor twice for complete ones.
1149 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
1152 * config/i386/i386.c (ix86_print_operand_address_as): Encode
1153 %esp as %rsp to avoid 0x67 prefix if there is no index or base
1156 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
1159 * config/i386/i386.opt (mprefer-avx256): Use
1160 ix86_target_flags variable.
1161 * config/i386/i386.c (ix86_target_string): Move
1162 -mprefer-avx256 to flag2_opts.
1164 2017-09-22 Jakub Jelinek <jakub@redhat.com>
1167 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
1168 and x != -1 | y != -1 into (x & y) != -1.
1170 2017-09-22 Steve Ellcey <sellcey@cavium.com>
1172 * config.gcc: Add new case statement to set
1173 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
1174 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
1175 s390x-*-linux* case statements. Added aarch64 to the list of
1176 supported architectures.
1178 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1180 PR tree-optimization/82289
1181 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
1182 STMT_VINFO_RELEVANT_P.
1184 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1185 Alan Hayward <alan.hayward@arm.com>
1186 David Sherwood <david.sherwood@arm.com>
1188 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
1189 for VR_RANGE only; don't allow VR_ANTI_RANGE.
1190 (extract_range_from_binary_expr_1): Don't call
1191 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
1193 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1194 Alan Hayward <alan.hayward@arm.com>
1195 David Sherwood <david.sherwood@arm.com>
1197 * target.def (preferred_vector_alignment): New hook.
1198 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
1200 * doc/tm.texi: Regenerate.
1201 * targhooks.h (default_preferred_vector_alignment): Declare.
1202 * targhooks.c (default_preferred_vector_alignment): New function.
1203 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
1205 (DR_TARGET_ALIGNMENT): New macro.
1206 (aligned_access_p): Update commentary.
1207 (vect_known_alignment_in_bytes): New function.
1208 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
1210 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
1211 Calculate the misalignment based on the target alignment rather than
1213 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
1214 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
1215 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
1216 the target alignment, rather than masking the element misalignment
1217 with the number of elements in a vector. Also use the target
1218 alignment when calculating the maximum number of peels.
1219 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
1220 instead of TYPE_ALIGN_UNIT.
1221 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
1222 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
1223 (vect_create_addr_base_for_vector_ref): Update call accordingly.
1224 (vect_create_data_ref_ptr): Likewise.
1225 (vect_setup_realignment): Realign by ANDing with
1226 -DR_TARGET_MISALIGNMENT.
1227 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
1228 the number of peels based on DR_TARGET_ALIGNMENT.
1229 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
1230 with the guaranteed alignment boundary when deciding whether
1232 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
1233 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
1234 (ensure_base_align): Remove stmt_info parameter. Get the
1235 target base alignment from DR_TARGET_ALIGNMENT.
1236 (vectorizable_store): Update call accordingly. Interpret
1237 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
1239 (vectorizable_load): Likewise.
1241 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1242 Alan Hayward <alan.hayward@arm.com>
1243 David Sherwood <david.sherwood@arm.com>
1245 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
1246 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
1247 (vect_enhance_data_refs_alignment): Likewise.
1249 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
1251 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
1252 error. Only quit immediately if parsing is complete.
1253 (BEGIN): Initialize fatal_err and parse_done.
1254 (begin fpu, end fpu): Check number of arguments.
1255 (begin arch, end arch): Likewise.
1256 (begin cpu, end cpu): Likewise.
1257 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
1258 (optalias): Likewise.
1260 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
1262 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
1263 * config/arm/arm-isa.h: Delete. Move definitions to ...
1264 * arm-cpus.in: ... here. Use new feature and fgroup values.
1265 * config/arm/arm.c (arm_option_override): Use lower case for feature
1267 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
1268 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
1269 * config/arm/parsecpu.awk (END): Add new command 'isa'.
1271 (print_isa_bits_for): New function.
1272 (gen_isa): New function.
1273 (gen_comm_data): Use print_isa_bits_for.
1274 (define feature): New keyword.
1275 (define fgroup): New keyword.
1276 * config/arm/t-arm (TM_H): Remove.
1277 (GTM_H): Add arm-isa.h.
1278 (arm-isa.h): Add rule to generate file.
1279 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
1280 case for feature bit names.
1282 2017-09-22 Richard Biener <rguenther@suse.de>
1284 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
1286 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
1287 print of no dependency loops ...
1288 * graphite.c (graphite_transform_loops): ... here.
1289 (canonicalize_loop_closed_ssa_form): Work from inner to outer
1291 (same_close_phi_node, remove_duplicate_close_phi,
1292 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
1293 (canonicalize_loop_closed_ssa): ... here and simplify.
1294 * graphite-optimize-isl.c: Include tree-vectorizer.h.
1295 (optimize_isl): Use dump_printf_loc to tell when we stopped
1296 optimizing because of an ISL timeout.
1298 2017-09-22 Richard Biener <rguenther@suse.de>
1300 PR tree-optimization/82291
1301 * tree-if-conv.c (predicate_mem_writes): Make sure to
1302 remove writes in blocks predicated with false.
1304 2017-09-22 Richard Biener <rguenther@suse.de>
1306 * sese.c: Include cfganal.h.
1307 (if_region_set_false_region): Remove.
1308 (create_if_region_on_edge): Likewise.
1309 (move_sese_in_condition): Re-implement without destroying
1312 2017-09-22 Richard Biener <rguenther@suse.de>
1314 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
1315 Verify both BBs contain loop PHI nodes before dispatching to
1317 (graphite_regenerate_ast_isl): Do not recompute dominators,
1318 do not verify three times. Restructure for clarity.
1319 * graphite-scop-detection.c (same_close_phi_node,
1320 remove_duplicate_close_phi, make_close_phi_nodes_unique,
1321 defined_in_loop_p, canonicalize_loop_closed_ssa,
1322 canonicalize_loop_closed_ssa_form): Simplify, remove excess
1323 checking and SSA rewrite, move to ...
1324 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
1325 (graphite_initialize): Do not pass in ctx, do not reset the
1326 SCEV cache, compute only dominators.
1327 (graphite_transform_loops): Allocate ISL ctx after
1328 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
1329 Maintain post-dominators only around build_scops.
1330 * sese.c (if_region_set_false_region): Make static. Free
1331 and recompute dominators.
1332 (move_sese_in_condition): Assert we don't get called with
1333 post-dominators computed.
1334 * sese.h (if_region_set_false_region): Remove.
1336 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
1338 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
1339 mode attribute for TARGET_AVX512VL.
1341 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
1343 * config/i386/i386.opt (mprefer-avx256): New option.
1344 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
1346 (ix86_preferred_simd_mode): Return 256-bit AVX modes
1347 for TARGET_PREFER_AVX256.
1348 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
1350 2017-09-21 Jeff Law <law@redhat.com>
1352 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
1353 Fix dump output if the only stack space is for pushed registers.
1355 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1357 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
1360 2017-09-21 Martin Sebor <msebor@redhat.com>
1363 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
1364 code (in C++) or code that triggers warnings.
1366 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
1368 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
1370 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
1372 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
1373 * sched-rgn.c: Ditto.
1374 * sel-sched-ir.c: Ditto.
1376 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
1378 * toplev.h (set_random_seed): Adjust return type.
1379 * toplev.c (init_local_tick): Move eager initialization of random_seed
1380 to get_random_seed. Adjust comment.
1381 (init_random_seed): Inline to get_random_seed, delete.
1382 (get_random_seed): Initialize random_seed lazily.
1383 (set_random_seed): Do not return previous value.
1384 (print_switch_value): Do not call get_random_seed.
1386 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
1388 * cgraph.c (delete_function_version): New, broken out from...
1389 (cgraph_node::delete_function_version): ...here. Rename to
1390 cgraph_node::delete_function_version_by_decl. Update all uses.
1391 (cgraph_node::remove): Call delete_function_version.
1393 2017-09-21 Jakub Jelinek <jakub@redhat.com>
1396 * tree-inline.c (expand_call_inline): Emit clobber stmts for
1397 VAR_DECLs to which addressable non-volatile parameters are mapped
1398 and for id->retvar after the return value assignment. Clear
1399 id->retval and id->retbnd after inlining.
1401 2017-09-21 Richard Biener <rguenther@suse.de>
1403 PR tree-optimization/82276
1404 PR tree-optimization/82244
1405 * tree-vrp.c (build_assert_expr_for): Set
1406 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
1408 (remove_range_assertions): Revert earlier change.
1410 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
1413 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
1415 2017-09-21 Richard Biener <rguenther@suse.de>
1417 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1418 Restore valid IL after code generation errors.
1419 * graphite.c (graphite_transform_loops): Diagnose code
1420 generation issues as MSG_MISSED_OPTIMIZATION and continue
1421 with processing SCOPs.
1423 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1424 Alan Hayward <alan.hayward@arm.com>
1425 David Sherwood <david.sherwood@arm.com>
1427 * calls.c (compute_argument_addresses): Use simplify_gen_binary
1428 rather than choosing between plus_constant and gen_rtx_<CODE>.
1429 * expr.c (emit_push_insn): Likewise.
1430 (expand_expr_real_2): Likewise.
1432 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1433 Alan Hayward <alan.hayward@arm.com>
1434 David Sherwood <david.sherwood@arm.com>
1436 * loop-unroll.c (split_iv): Call copy_rtx on the step.
1438 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1439 Alan Hayward <alan.hayward@arm.com>
1440 David Sherwood <david.sherwood@arm.com>
1442 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
1443 calling tree_to_uhwi.
1445 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1446 Alan Hayward <alan.hayward@arm.com>
1447 David Sherwood <david.sherwood@arm.com>
1449 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
1450 INTEGER_CST rather than a negative test for ADDR_EXPR.
1452 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1453 Alan Hayward <alan.hayward@arm.com>
1454 David Sherwood <david.sherwood@arm.com>
1456 * tree-vrp.c (extract_range_from_binary_expr_1): Check
1457 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
1459 2017-09-21 Richard Biener <rguenther@suse.de>
1461 PR tree-optimization/71351
1462 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
1463 graphite_create_new_loop_guard): Remove, fold remaining parts
1465 (translate_isl_ast_node_for): ... here and simplify.
1467 2017-09-21 Jakub Jelinek <jakub@redhat.com>
1470 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
1471 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
1472 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
1473 alternative always use QI mode, for -Os imov (=R,R) alternative
1474 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
1477 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1478 Jeff Law <law@redhat.com>
1480 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
1481 (allocate_stack_space): New function, partially extracted from
1483 (s390_emit_prologue): Track offset to most recent stack probe.
1484 Code to allocate space moved into allocate_stack_space.
1485 Dump actions when no stack is allocated.
1486 (s390_prologue_plus_offset): New function.
1487 (s390_emit_stack_probe): Likewise.
1489 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
1491 * common.opt (Wa, Wl, Wp, g, gz=): Add
1493 (gno-column-info): Remove.
1494 (gcolumn-info): Drop RejectNegative.
1496 (gno-record-gcc-switches): Remove.
1497 (grecord-gcc-switches): Drop RejectNegative.
1498 (gno-split-dwarf): Remove.
1499 (gsplit-dwarf): Drop RejectNegative.
1500 (gno-strict-dwarf): Remove.
1501 (gstrict-dwarf): Drop RejectNegative.
1502 * config/darwin.opt (gfull, gused): Add RejectNegative.
1503 * dwarf2out.c (gen_producer_string): Drop
1504 gno-record-gcc-switches handler.
1505 * optc-gen.awk: Add g to prefixes with negative forms.
1506 * opts-common.c (remapping_prefix_p): New.
1507 (find_opt): Check it.
1508 (generate_canonical_option): Test g prefix.
1509 (option_map): Add -gno- mapping.
1510 (add_misspelling_candidates): Check remapping_prefix_p.
1512 2017-09-20 Jeff Law <law@redhat.com>
1514 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
1515 thinko in stack clash protection support.
1517 * explow.c (compute_stack_clash_protection_loop_data): Use
1518 CONST_INT_P instead of explicit test. Verify object is a
1519 CONST_INT_P before looking at INTVAL.
1520 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
1521 instead of explicit test.
1523 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
1526 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
1527 address instead of to r1 and r11.
1529 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
1531 * config.gcc: Support "knm".
1532 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
1533 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1535 * config/i386/i386.c (m_KNM): Define.
1536 (processor_target_table): Add "knm".
1538 (ix86_option_override_internal): Add "knm".
1539 (ix86_issue_rate): Add PROCESSOR_KNM.
1540 (ix86_adjust_cost): Ditto.
1541 (ia32_multipass_dfa_lookahead): Ditto.
1542 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
1543 (fold_builtin_cpu): Add M_INTEL_KNM.
1544 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
1545 (processor_type): Add PROCESSOR_KNM.
1546 * config/i386/x86-tune.def: Add m_KNM.
1547 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
1549 2017-09-20 Richard Biener <rguenther@suse.de>
1551 PR tree-optimization/80213
1552 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
1553 are allowed in empty BBs as well.
1554 (canonicalize_loop_closed_ssa): Also look for other complex
1556 (scop_detection::get_sese): Include the loop-closed PHI block
1558 (scop_detection::merge_sese): Remove code adding extra blocks.
1559 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
1560 (build_scops): Assert the final returned scop is invalid.
1562 2017-09-20 Richard Biener <rguenther@suse.de>
1564 PR tree-optimization/82264
1565 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
1567 (vn_phi_lookup): Likewise.
1568 (vn_phi_insert): Likewise.
1570 2017-09-20 Jakub Jelinek <jakub@redhat.com>
1572 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
1573 that fits into uhwi or shwi, add DW_AT_const_value regardless
1574 of early_dwarf without going through RTL, using add_AT_unsigned
1577 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
1578 (DEBUG_LTO_ABBREV_SECTION): Likewise.
1579 (DEBUG_LTO_MACINFO_SECTION): Likewise.
1580 (DEBUG_MACRO_SECTION): Likewise.
1581 (DEBUG_LTO_MACRO_SECTION): Likewise.
1582 (DEBUG_STR_DWO_SECTION): Likewise.
1583 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
1584 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
1585 (DEBUG_LTO_DWO_LINE_SECTION): Define.
1586 (DEBUG_LTO_LINE_STR_SECTION): Define.
1587 (init_sections_and_labels): Initialize debug_line_str_section
1588 variable. Initialize debug_loc_section for -gdwarf-5 to
1589 DEBUG_LOCLISTS_SECTION. Formatting fixes.
1591 2017-09-20 Richard Biener <rguenther@suse.de>
1593 * graphite-sese-to-poly.c (extract_affine): Properly handle
1594 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
1596 2017-09-20 Richard Biener <rguenther@suse.de>
1598 PR tree-optimization/81373
1599 * graphite-scop-detection.c (build_cross_bb_scalars_def):
1600 Force SESE live-out defs to be handled even if they are
1603 2017-09-19 Jeff Law <law@redhat.com>
1605 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
1606 nothing for stack adjustments with REG_STACK_CHECK.
1607 * sched-deps.c (parse_add_or_inc): Reject insns with
1608 REG_STACK_CHECK from dependency breaking.
1609 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
1610 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
1611 * reg-notes.def (STACK_CHECK): New note.
1613 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
1614 (ix86_expand_prologue): Dump stack clash info as needed.
1615 Call ix86_adjust_stack_and_probe_stack_clash as needed.
1617 * function.c (dump_stack_clash_frame_info): New function.
1618 * function.h (dump_stack_clash_frame_info): Prototype.
1619 (enum stack_clash_probes): New enum.
1621 * config/alpha/alpha.c (alpha_expand_prologue): Also check
1622 flag_stack_clash_protection.
1623 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
1624 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
1625 (arm_frame_pointer_required): Likewise.
1626 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
1627 (ia64_expand_prologue): Likewise.
1628 * config/mips/mips.c (mips_expand_prologue): Likewise.
1629 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
1630 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
1631 (sparc_flat_expand_prologue): Likewise.
1632 * config/spu/spu.c (spu_expand_prologue): Likewise.
1634 * explow.c: Include "params.h".
1635 (anti_adjust_stack_and_probe_stack_clash): New function.
1636 (get_stack_check_protect): Likewise.
1637 (compute_stack_clash_protection_loop_data): Likewise.
1638 (emit_stack_clash_protection_loop_start): Likewise.
1639 (emit_stack_clash_protection_loop_end): Likewise.
1640 (allocate_dynamic_stack_space): Use get_stack_check_protect.
1641 Use anti_adjust_stack_and_probe_stack_clash.
1642 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
1643 (emit_stack_clash_protection_loop_start): Likewise.
1644 (emit_stack_clash_protection_loop_end): Likewise.
1645 * rtl.h (get_stack_check_protect): Prototype.
1646 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
1647 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
1648 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
1650 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
1652 * doc/tm.texi: Rebuilt.
1653 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
1654 get_stack_check_protect.
1655 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
1656 * config/arm/arm.c (arm_expand_prologue): Likewise.
1657 (arm_frame_pointer_required): Likewise.
1658 * config/i386/i386.c (ix86_expand_prologue): Likewise.
1659 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
1660 * config/mips/mips.c (mips_expand_prologue): Likewise.
1661 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
1662 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
1663 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
1664 (sparc_flat_expand_prologue): Likewise.
1666 * common.opt (-fstack-clash-protection): New option.
1667 * flag-types.h (enum stack_check_type): Note difference between
1668 -fstack-check= and -fstack-clash-protection.
1669 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
1670 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
1671 * toplev.c (process_options): Issue warnings/errors for cases
1672 not handled with -fstack-clash-protection.
1673 * doc/invoke.texi (-fstack-clash-protection): Document new option.
1674 (-fstack-check): Note additional problem with -fstack-check=generic.
1675 Note that -fstack-check is primarily for Ada and refer users
1676 to -fstack-clash-protection for stack-clash-protection.
1677 Document new params for stack clash protection.
1679 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
1681 * config/i386/i386.c (ix86_split_long_move): Do not handle
1682 address used for LEA in a special way.
1684 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
1686 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
1688 2017-09-19 Martin Sebor <msebor@redhat.com>
1691 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
1692 of incompatible types.
1694 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
1696 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
1697 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
1698 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1699 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
1701 2017-09-19 Richard Biener <rguenther@suse.de>
1703 PR tree-optimization/82244
1704 * tree-vrp.c (remove_range_assertions): Do not propagate
1705 a constant to abnormals but replace the assert with a copy.
1707 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
1709 PR rtl-optimization/57878
1710 PR rtl-optimization/68988
1711 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
1712 avoidance test involving non_reload_pseudos. Move frequency test
1713 below the general fragmentation avoidance test.
1715 2017-09-19 Richard Biener <rguenther@suse.de>
1717 PR tree-optimization/69728
1718 * graphite-sese-to-poly.c (schedule_error): New global.
1719 (add_loop_schedule): Handle empty domain by failing the
1721 (build_original_schedule): Handle schedule_error.
1723 2017-09-19 Richard Biener <rguenther@suse.de>
1725 * graphite-scop-detection.c (scop_detection::can_represent_loop):
1726 Do not iterate to sibling loops but only to siblings of inner
1729 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
1732 * config/m68k/m68k.md (moveq feeding equality comparison): Check
1733 that the registers are different.
1735 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
1737 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
1738 to processor_model and "amdfam17h" to arch_names_table.
1739 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
1741 2017-09-18 Jakub Jelinek <jakub@redhat.com>
1744 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
1746 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1747 Alan Hayward <alan.hayward@arm.com>
1748 David Sherwood <david.sherwood@arm.com>
1750 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
1752 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
1754 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
1755 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
1756 vect_schedule_slp_instance.
1757 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
1758 Update call to vect_slp_analyze_node_operations. Simplify return
1760 (vect_slp_analyze_bb_1): Update call accordingly.
1761 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
1762 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
1763 (vect_schedule_slp): Update call accordingly.
1765 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1766 Alan Hayward <alan.hayward@arm.com>
1767 David Sherwood <david.sherwood@arm.com>
1769 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
1770 with types that aren't in fact scalar.
1772 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1774 * tree-vect-slp.c (vect_record_max_nunits): New function,
1776 (vect_build_slp_tree_1): ...here.
1777 (vect_build_slp_tree_2): Call it for phis too.
1779 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1781 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
1782 to vect_get_vec_def_for_operand when getting the mask operand.
1784 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1785 Alan Hayward <alan.hayward@arm.com>
1786 David Sherwood <david.sherwood@arm.com>
1788 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
1791 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1792 Alan Hayward <alan.hayward@arm.com>
1793 David Sherwood <david.sherwood@arm.com>
1795 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
1796 calculation for vector booleans.
1798 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1799 Alan Hayward <alan.hayward@arm.com>
1800 David Sherwood <david.sherwood@arm.com>
1802 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
1804 (vect_transform_stmt): ...here.
1805 (vect_analyze_stmt): Use it instead of calling
1806 vectorizable_live_operation directly.
1808 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
1810 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
1811 non-SIMT targets in acc vector loops.
1813 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
1815 * configure.ac: Add arc and check if assembler supports gdwarf2.
1816 * configure: Regenerate.
1818 2017-09-18 Richard Biener <rguenther@suse.de>
1820 PR tree-optimization/82220
1821 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
1822 epilogue niters from the min_profitable_iters compute.
1824 2017-09-18 Jakub Jelinek <jakub@redhat.com>
1827 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
1828 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
1829 (ix86_init_pic_reg): Revert 2017-09-01 changes.
1831 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
1834 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
1835 switching to a new text section.
1837 2017-09-18 Richard Biener <rguenther@suse.de>
1839 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
1841 (build_alias_set): Reject aliases with no access function.
1843 2017-09-18 Richard Biener <rguenther@suse.de>
1845 PR tree-optimization/79622
1846 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
1848 (build_cross_bb_scalars_use): Likewise.
1850 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
1852 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
1854 2017-09-18 Alan Modra <amodra@gmail.com>
1857 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
1858 stack_pointer_rtx for count 0. Update comments. Break up
1859 large rtl expression.
1861 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
1863 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
1864 Increase to 20 bytes.
1865 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
1866 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
1867 or avx version of the stub.
1869 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
1872 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
1873 compute the minimum stack alignment. Also update preferred stack
1874 boundary for leaf functions.
1876 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
1878 PR tree-optimization/82228
1879 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
1882 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
1884 * common/config/nds32/nds32-common.c
1885 (nds32_option_optimization_table): Refine formatting.
1886 (nds32_option_optimization_table): Use -fsched-pressure and
1887 -fomit-frame-pointer for specific optimization level.
1889 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
1891 * config/nds32/nds32.c: Refine formatting and comments.
1892 * config/nds32/nds32.h: Likewise.
1893 * config/nds32/nds32.md: Likewise.
1894 * config/nds32/nds32-cost.c: Likewise.
1895 * config/nds32/nds32-isr.c: Likewise.
1896 * config/nds32/nds32-md-auxiliary.c: Likewise.
1897 * config/nds32/nds32-multiple.md: Likewise.
1898 * config/nds32/nds32-predicates.c: Likewise.
1900 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
1901 Jakub Jelinek <jakub@redhat.com>
1903 Add support for -std=c++2a.
1904 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
1906 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
1908 2017-09-15 Steve Ellcey <sellcey@cavium.com>
1911 * doc/extend.texi (Common Function Attributes): Add
1912 references to ARM, AArch64, and S/390 specific attributes.
1913 (Function Specific Option Pragmas): Add AArch64 and S/390
1914 to list of back ends that support the target pragma.
1916 2017-09-15 Nathan Sidwell <nathan@acm.org>
1918 * doc/standards.texi: Fix C++17 description. Update URLs for
1921 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
1923 * common.opt (Wcast-align=strict): New warning option.
1924 * doc/invoke.texi: Document -Wcast-align=strict.
1926 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
1928 * cgraph.h (cgraph_thunk_info): Add comments.
1929 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
1930 assert for VIRTUAL_* arguments stricter.
1932 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
1934 PR tree-optimization/71026
1935 * match.pd: Move RDIV patterns from fold-const.c
1936 * fold-const.c (distribute_real_division): Removed.
1937 (fold_binary_loc): Remove calls to distribute_real_divison.
1939 2017-09-15 Jakub Jelinek <jakub@redhat.com>
1941 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
1942 c++1z and gnu++1z as deprecated. Change other references to
1943 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
1944 Change -Wc++1z-compat to -Wc++17-compat.
1945 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
1946 * dwarf2out.c (highest_c_language): Handle C++17.
1947 (gen_compile_unit_die): Likewise.
1949 2017-09-15 Jakub Jelinek <jakub@redhat.com>
1951 PR rtl-optimization/82192
1952 * combine.c (make_extraction): Don't look through non-paradoxical
1953 SUBREGs or TRUNCATE if pos + len is or might be bigger than
1956 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
1957 Alan Hayward <alan.hayward@arm.com>
1958 David Sherwood <david.sherwood@arm.com>
1960 * target.def (function_arg_offset): New hook.
1961 * targhooks.h (default_function_arg_offset): Declare.
1962 * targhooks.c (default_function_arg_offset): New function.
1963 * function.c (locate_and_pad_parm): Use
1964 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
1965 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
1966 (TARGET_FUNCTION_ARG_OFFSET): ...this.
1967 * doc/tm.texi: Regenerate.
1968 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
1969 * config/spu/spu.c (spu_function_arg_offset): New function.
1970 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
1971 * system.h (FUNCTION_ARG_OFFSET): Poison.
1973 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
1974 Alan Hayard <alan.hayward@arm.com>
1975 David Sherwood <david.sherwood@arm.com>
1977 * target.def (truly_noop_truncation): New hook.
1978 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
1979 than TRULY_NOOP_TRUNCATION.
1980 * hooks.h (hook_bool_uint_uint_true): Declare.
1981 * hooks.c (hook_bool_uint_uint_true): New function.
1982 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
1983 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
1984 * doc/tm.texi: Regenerate.
1985 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
1986 rather than TRULY_NOOP_TRUNCATION in comments.
1987 (simplify_comparison): Likewise.
1988 (record_truncated_value): Likewise.
1989 * expmed.c (extract_bit_field_1): Likewise.
1990 (extract_split_bit_field): Likewise.
1991 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
1992 instead of TRULY_NOOP_TRUNCATION.
1993 * function.c (assign_parm_setup_block): Likewise.
1994 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
1995 * rtlhooks.c: Include target.h.
1996 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
1997 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
1998 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
1999 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
2000 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
2001 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
2002 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
2003 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
2004 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
2005 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
2006 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
2007 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
2008 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
2009 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
2010 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
2011 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
2012 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
2013 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
2014 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
2015 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
2016 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
2017 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
2018 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
2019 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
2020 * config/mips/mips.c (mips_truly_noop_truncation): New function.
2021 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2022 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2023 rather than TRULY_NOOP_TRUNCATION in comments.
2024 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
2025 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
2026 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
2027 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
2028 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
2029 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
2030 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
2031 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
2032 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
2033 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
2034 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
2035 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2036 rather than TRULY_NOOP_TRUNCATION in comments.
2037 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
2038 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
2039 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
2040 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
2041 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
2042 TRULY_NOOP_TRUNCATION condition.
2043 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
2044 (TRULY_NOOP_TRUNCATION): Delete.
2045 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
2046 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
2047 * config/spu/spu.c (spu_truly_noop_truncation): New function.
2048 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2049 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
2050 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
2051 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
2052 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2053 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2054 rather than TRULY_NOOP_TRUNCATION in comments.
2055 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
2056 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
2057 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
2058 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
2059 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
2060 * system.h (TRULY_NOOP_TRUNCATION): Poison.
2062 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
2065 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
2066 (*cmp_ior): Likewise.
2067 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
2068 (*ior_scc_scc_cmp): Likewise.
2069 (*and_scc_scc): Likewise.
2070 (*and_scc_scc_cmp): Likewise.
2072 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2073 Alan Hayard <alan.hayward@arm.com>
2074 David Sherwood <david.sherwood@arm.com>
2076 * target.def (can_change_mode_class): New hook.
2077 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
2078 (hard_regno_nregs): Likewise.
2079 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
2080 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
2081 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
2082 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
2083 (LOAD_EXTEND_OP): Update accordingly.
2084 * doc/tm.texi: Regenerate.
2085 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
2086 CANNOT_CHANGE_MODE_CLASS.
2087 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
2088 (REG_CAN_CHANGE_MODE_P): ...this new macro.
2089 * combine.c (simplify_set): Update accordingly.
2090 * emit-rtl.c (validate_subreg): Likewise.
2091 * recog.c (general_operand): Likewise.
2092 * regcprop.c (mode_change_ok): Likewise.
2093 * reload1.c (choose_reload_regs): Likewise.
2094 (inherit_piecemeal_p): Likewise.
2095 * rtlanal.c (simplify_subreg_regno): Likewise.
2096 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
2097 instead of CANNOT_CHANGE_MODE_CLASS.
2098 (reload_cse_simplify_operands): Likewise.
2099 * reload.c (push_reload): Use targetm.can_change_mode_class
2100 instead of CANNOT_CHANGE_MODE_CLASS.
2101 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
2102 REG_CANNOT_CHANGE_MODE_P.
2103 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2104 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
2105 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2106 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2107 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2108 (arm_can_change_mode_class): New function.
2109 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
2110 than CANNOT_CHANGE_MODE_CLASS in comments.
2111 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2112 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
2113 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
2114 (ix86_can_change_mode_class): ...this new function, inverting the
2115 sense of the return value.
2116 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2117 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2118 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2119 (ia64_can_change_mode_class): New function.
2120 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2121 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
2122 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
2123 (m32c_can_change_mode_class): ...this new function, inverting the
2124 sense of the return value.
2125 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2126 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2127 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
2128 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
2129 (mips_can_change_mode_class): ...this new function, inverting the
2130 sense of the return value.
2131 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2132 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2133 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2134 (msp430_can_change_mode_class): New function.
2135 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2136 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
2137 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2138 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2139 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2140 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
2141 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2142 (pa_cannot_change_mode_class): Replace with...
2143 (pa_can_change_mode_class): ...this new function, inverting the
2144 sense of the return value.
2145 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
2146 than CANNOT_CHANGE_MODE_CLASS in comments.
2147 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2148 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
2149 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2150 (pdp11_cannot_change_mode_class): Replace with...
2151 (pdp11_can_change_mode_class): ...this new function, inverting the
2152 sense of the return value.
2153 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2154 * config/powerpcspe/powerpcspe-protos.h
2155 (rs6000_cannot_change_mode_class_ptr): Delete.
2156 * config/powerpcspe/powerpcspe.c
2157 (rs6000_cannot_change_mode_class_ptr): Delete.
2158 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2159 (rs6000_option_override_internal): Assign to
2160 targetm.can_change_mode_class instead of
2161 rs6000_cannot_change_mode_class_ptr.
2162 (rs6000_cannot_change_mode_class): Replace with...
2163 (rs6000_can_change_mode_class): ...this new function, inverting the
2164 sense of the return value.
2165 (rs6000_debug_cannot_change_mode_class): Replace with...
2166 (rs6000_debug_can_change_mode_class): ...this new function.
2167 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2168 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
2169 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2170 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2171 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
2173 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
2174 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2175 (rs6000_option_override_internal): Assign to
2176 targetm.can_change_mode_class instead of
2177 rs6000_cannot_change_mode_class_ptr.
2178 (rs6000_cannot_change_mode_class): Replace with...
2179 (rs6000_can_change_mode_class): ...this new function, inverting the
2180 sense of the return value.
2181 (rs6000_debug_cannot_change_mode_class): Replace with...
2182 (rs6000_debug_can_change_mode_class): ...this new function.
2183 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2184 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
2185 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
2186 (s390_can_change_mode_class): ...this new function, inverting the
2187 sense of the return value.
2188 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2189 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2190 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
2191 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2192 (sh_cannot_change_mode_class): Replace with...
2193 (sh_can_change_mode_class): ...this new function, inverting the
2194 sense of the return value.
2195 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2196 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2197 (sparc_can_change_mode_class): New function.
2198 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2199 * config/spu/spu.c (spu_can_change_mode_class): New function.
2200 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2201 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2202 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2203 (visium_can_change_mode_class): New function.
2204 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
2206 2017-09-15 Richard Biener <rguenther@suse.de>
2208 PR tree-optimization/82217
2209 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
2210 but not undefined case.
2212 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2215 * postreload.c (reload_cse_simplify_operands): Skip
2216 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
2218 2017-09-15 Richard Biener <rguenther@suse.de>
2220 PR tree-optimization/68823
2221 * graphite-scop-detection.c (build_alias_set): If we have a
2222 possible dependence check whether we can handle them by just
2223 looking at the DRs DR_ACCESS_FNs.
2224 (build_scops): If build_alias_set fails, fail the SCOP.
2226 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
2228 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
2229 to support float128 built-in functions that require the ISA 3.0
2231 (BU_FLOAT128_3_HW): Likewise.
2232 (SQRTF128): Add support for the IEEE 128-bit square root and fma
2234 (FMAF128): Likewise.
2236 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
2237 support for built-in functions that need the ISA 3.0 IEEE 128-bit
2238 floating point instructions.
2239 (rs6000_invalid_builtin): Likewise.
2240 (rs6000_builtin_mask_names): Likewise.
2241 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
2242 (RS6000_BTM_FLOAT128_HW): Likewise.
2243 (RS6000_BTM_COMMON): Likewise.
2244 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
2246 * doc/extend.texi (RS/6000 built-in functions): Document the
2247 IEEE 128-bit floating point square root and fused multiply-add
2250 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
2252 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
2253 reg (r2) isn't in the set of registers defined in the prologue.
2255 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2256 Alan Hayward <alan.hayward@arm.com>
2257 David Sherwood <david.sherwood@arm.com>
2259 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
2260 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
2261 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
2262 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
2263 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
2265 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
2266 max_vectorization_factor.
2267 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
2269 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2270 Alan Hayward <alan.hayward@arm.com>
2271 David Sherwood <david.sherwood@arm.com>
2273 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
2274 (vect_worthwhile_without_simd_p): Declare.
2275 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
2276 (vectorizable_reduction): Use it.
2277 * tree-vect-stmts.c (vectorizable_shift): Likewise.
2278 (vectorizable_operation): Likewise.
2280 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2281 Alan Hayward <alan.hayward@arm.com>
2282 David Sherwood <david.sherwood@arm.com>
2284 * tree-vectorizer.h (vect_get_num_copies): New function.
2285 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
2286 * tree-vect-loop.c (vectorizable_reduction): Likewise.
2287 (vectorizable_induction): Likewise.
2288 (vectorizable_live_operation): Likewise.
2289 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
2290 (vectorizable_bswap): Likewise.
2291 (vectorizable_call): Likewise.
2292 (vectorizable_conversion): Likewise.
2293 (vectorizable_assignment): Likewise.
2294 (vectorizable_shift): Likewise.
2295 (vectorizable_operation): Likewise.
2296 (vectorizable_store): Likewise.
2297 (vectorizable_load): Likewise.
2298 (vectorizable_condition): Likewise.
2299 (vectorizable_comparison): Likewise.
2300 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
2302 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2303 Alan Hayward <alan.hayward@arm.com>
2304 David Sherwood <david.sherwood@arm.com>
2306 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
2307 of vect_init_vector.
2309 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2310 Alan Hayward <alan.hayward@arm.com>
2311 David Sherwood <david.sherwood@arm.com>
2313 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
2314 an inline wrapper that provides a location.
2315 (gimple_build_vector): Likewise.
2316 * gimple-fold.c (gimple_build_vector_from_val): New function.
2317 (gimple_build_vector): Likewise.
2318 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
2319 functions to build the initial value. Always return a gimple value.
2320 (get_initial_defs_for_reduction): Likewise. Only compute
2322 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
2323 vect_init_vector on the results from get_initial_def(s)_for_reduction.
2324 (vectorizable_induction): Use gimple_build_vector rather than
2327 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2328 Alan Hayward <alan.hayward@arm.com>
2329 David Sherwood <david.sherwood@arm.com>
2331 * target.h (vec_perm_indices): New typedef.
2332 (auto_vec_perm_indices): Likewise.
2333 * optabs-query.h: Include target.h
2334 (can_vec_perm_p): Take a vec_perm_indices *.
2335 * optabs-query.c (can_vec_perm_p): Likewise.
2336 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
2337 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
2338 * tree-vect-generic.c (lower_vec_perm): Likewise.
2339 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
2340 (vect_grouped_load_supported): Likewise.
2341 (vect_shift_permute_load_chain): Likewise.
2342 (vect_permute_store_chain): Use auto_vec_perm_indices.
2343 (vect_permute_load_chain): Likewise.
2344 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
2345 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
2346 Update uses of can_vec_perm_p.
2347 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
2348 mode with a number of elements. Take a vec_perm_indices *.
2349 (vect_create_epilog_for_reduction): Update accordingly.
2350 Use auto_vec_perm_indices.
2351 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
2352 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
2353 (vect_transform_slp_perm_load): Likewise.
2354 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
2355 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
2356 (vect_gen_perm_mask_checked): Likewise.
2357 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
2358 (vect_gen_perm_mask_checked): Likewise.
2359 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
2360 (vectorizable_store): Likewise.
2361 (vectorizable_load): Likewise.
2362 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
2363 (vectorizable_bswap): Likewise.
2365 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2366 Alan Hayward <alan.hayward@arm.com>
2367 David Sherwood <david.sherwood@arm.com>
2369 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
2370 * tree.c (build_vector): Likewise.
2371 (build_vector_from_ctor): Update accordingly.
2372 (build_vector_from_val): Likewise.
2373 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
2374 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
2375 * tree-vect-generic.c (add_rshift): Likewise.
2376 (expand_vector_divmod): Likewise.
2377 (optimize_vector_constructor): Likewise.
2378 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
2379 (vect_transform_slp_perm_load): Likewise.
2380 (vect_schedule_slp_instance): Likewise.
2381 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
2382 (vectorizable_call): Likewise.
2383 (vect_gen_perm_mask_any): Likewise. Add elements in order.
2384 * expmed.c (make_tree): Likewise.
2385 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
2386 a vector passed to build_vector.
2387 (fold_convert_const): Likewise.
2388 (exact_inverse): Likewise.
2389 (fold_ternary_loc): Likewise.
2390 (fold_relational_const): Likewise.
2391 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
2392 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
2393 (const_unop): Likewise. Store the reduction accumulator in a
2394 variable rather than an array.
2395 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
2396 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
2397 the new vector, rather than constructing it after the input arrays.
2398 (native_interpret_vector): Use auto_vec<tree> when building
2399 a vector passed to build_vector. Add elements in order.
2400 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
2401 auto_vec<tree> when building a vector passed to build_vector.
2402 (vect_create_epilog_for_reduction): Likewise.
2403 (vectorizable_induction): Likewise.
2404 (get_initial_def_for_reduction): Likewise. Fix indentation of
2406 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
2408 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
2409 passed to build_vector.
2411 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2412 Alan Hayward <alan.hayward@arm.com>
2413 David Sherwood <david.sherwood@arm.com>
2415 * tree-core.h (tree_base::u): Add an "nelts" field.
2416 (tree_vector): Use VECTOR_CST_NELTS as the length.
2417 * tree.c (tree_size): Likewise.
2418 (make_vector): Initialize VECTOR_CST_NELTS.
2419 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
2420 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
2421 TYPE_VECTOR_SUBPARTS.
2422 * expr.c (const_vector_mask_from_tree): Consistently use "units"
2423 as the number of units, setting it from VECTOR_CST_NELTS.
2424 (const_vector_from_tree): Likewise.
2425 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
2426 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
2427 (fold_negate_expr_1): Likewise.
2428 (fold_convert_const): Likewise.
2429 (const_binop): Likewise. Differentiate the number of output and
2431 (const_unop): Likewise.
2432 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
2433 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
2434 in cases that did the opposite.
2436 2017-09-14 Richard Biener <rguenther@suse.de>
2438 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
2441 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
2443 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
2445 2017-09-14 Jakub Jelinek <jakub@redhat.com>
2448 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
2449 if and where to split a bb, except for splitting before debug insn
2450 sequences followed by non-label real insn. Delete debug insns
2451 in between basic blocks.
2453 * combine.c (make_compound_operation_int): Formatting fixes.
2455 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
2456 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2457 * config/netbsd.h (LINK_EH_SPEC): Likewise.
2458 * config/sol2.h (LINK_EH_SPEC): Likewise.
2459 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2460 * config/s390/linux.h (LINK_SPEC): Likewise.
2461 * config/freebsd.h (LINK_EH_SPEC): Likewise.
2462 * config/openbsd.h (LINK_EH_SPEC): Likewise.
2463 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2464 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
2465 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
2466 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2467 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
2468 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
2470 2017-09-13 Jakub Jelinek <jakub@redhat.com>
2472 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
2474 (ENDFILE_LINUX_SPEC): Likewise.
2475 (LINK_EH_SPEC): Likewise.
2476 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
2477 (LINK_OS_LINUX_SPEC32): Likewise.
2478 (LINK_OS_LINUX_SPEC64): Likewise.
2479 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
2480 (LINK_OS_LINUX_SPEC): Likewise.
2482 2017-09-13 Martin Liska <mliska@suse.cz>
2485 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
2486 CASE_HIGH is NULL_TREE.
2488 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
2489 Alan Hayward <alan.hayward@arm.com>
2490 David Sherwood <david.sherwood@arm.com>
2492 * target.def (secondary_memory_needed): New hook.
2493 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
2494 instead of SECONDARY_MEMORY_NEEDED.
2495 (secondary_memory_needed_mode): Likewise.
2496 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
2497 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
2498 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
2499 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
2500 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
2501 * doc/tm.texi: Regenerate.
2502 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
2503 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
2504 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2505 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
2506 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
2507 * config/i386/i386.c (inline_secondary_memory_needed): Put the
2508 mode argument first and change the reg_class arguments to reg_class_t.
2509 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
2510 Make static. Update the call to inline_secondary_memory_needed.
2511 (ix86_register_move_cost): Update the call to
2512 inline_secondary_memory_needed.
2513 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2514 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
2516 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
2517 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2519 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
2520 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
2521 * config/mips/mips.c (mips_secondary_memory_needed): Make static
2522 and match hook interface. Add comment from mips.h.
2523 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2524 * config/mmix/mmix.md (truncdfsf2): Refer to
2525 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2527 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
2528 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
2529 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2530 (pa_secondary_memory_needed): New function.
2531 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
2532 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
2533 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2534 (pdp11_secondary_memory_needed): Make static and match hook interface.
2535 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
2536 * config/powerpcspe/powerpcspe-protos.h
2537 (rs6000_secondary_memory_needed_ptr): Delete.
2538 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
2540 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2541 (rs6000_option_override_internal): Assign to
2542 targetm.secondary_memory_needed rather than
2543 rs6000_secondary_memory_needed_ptr.
2544 (rs6000_secondary_memory_needed): Match hook interface.
2545 (rs6000_debug_secondary_memory_needed): Likewise.
2546 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
2547 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
2548 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
2549 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2550 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
2551 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
2553 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
2554 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2555 (rs6000_option_override_internal): Assign to
2556 targetm.secondary_memory_needed rather than
2557 rs6000_secondary_memory_needed_ptr.
2558 (rs6000_secondary_memory_needed): Match hook interface.
2559 (rs6000_debug_secondary_memory_needed): Likewise.
2560 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
2561 * config/s390/s390.c (s390_secondary_memory_needed): New function.
2562 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2563 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
2564 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2565 (sparc_secondary_memory_needed): New function.
2566 * lra-constraints.c (check_and_process_move): Refer to
2567 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2569 (curr_insn_transform): Likewise.
2570 (process_alt_operands): Use targetm.secondary_memory_needed
2571 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2572 (check_secondary_memory_needed_p): Likewise.
2573 (choose_split_class): Likewise.
2574 * reload.c: Unconditionally include code that was previously
2575 conditional on SECONDARY_MEMORY_NEEDED.
2576 (push_secondary_reload): Use targetm.secondary_memory_needed
2577 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2578 (push_reload): Likewise.
2579 * reload1.c: Unconditionally include code that was previously
2580 conditional on SECONDARY_MEMORY_NEEDED.
2581 (choose_reload_regs): Use targetm.secondary_memory_needed
2582 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2583 (gen_reload): Likewise.
2584 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
2586 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
2587 Alan Hayward <alan.hayward@arm.com>
2588 David Sherwood <david.sherwood@arm.com>
2590 * target.def (secondary_memory_needed_mode): New hook:
2591 * targhooks.c (default_secondary_memory_needed_mode): Declare.
2592 * targhooks.h (default_secondary_memory_needed_mode): New function.
2593 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
2594 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
2595 * doc/tm.texi: Regenerate.
2596 * lra-constraints.c (check_and_process_move): Use
2597 targetm.secondary_memory_needed_mode instead of
2598 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
2599 (curr_insn_transform): Likewise.
2600 * reload.c (get_secondary_mem): Likewise.
2601 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2602 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
2604 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2605 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2606 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
2607 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2608 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
2610 * config/powerpcspe/powerpcspe-protos.h
2611 (rs6000_secondary_memory_needed_mode): Delete.
2612 * config/powerpcspe/powerpcspe.c
2613 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2614 (rs6000_secondary_memory_needed_mode): Make static.
2615 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2616 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
2618 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
2620 (rs6000_secondary_memory_needed_mode): Make static.
2621 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2622 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
2623 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2624 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2625 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
2627 (sparc_secondary_memory_needed_mode): New function.
2628 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
2630 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
2632 * config/aarch64/constraints.md (Umq): New constraint.
2633 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
2635 (mov<mode>): Update condition.
2637 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2639 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
2640 when bitposition is the same.
2642 2017-09-13 Richard Biener <rguenther@suse.de>
2644 * dwarf2out.c (output_die_symbol): Remove.
2645 (output_die): Do not output a DIEs symbol.
2647 2017-09-13 Richard Biener <rguenther@suse.de>
2650 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
2651 default-def to avoid breaking iterator update with the weird
2652 interaction with cgraph_update_edges_for_call_stmt_node.
2654 2017-09-13 Richard Biener <rguenther@suse.de>
2656 * tree-cfg.c (verify_gimple_assign_binary): Add verification
2657 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
2658 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
2659 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
2661 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
2663 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
2664 Disable pc relative literal load irrespective of
2665 TARGET_FIX_ERR_A53_84341 for default.
2667 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
2669 * config/sparc/sparc.c (output_return): Output the source location of
2670 the insn in the delay slot, if any.
2671 (output_sibcall): Likewise.
2673 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
2676 * common.opt (-static-pie): New alias.
2677 (shared): Negate static-pie.
2678 (-no-pie): Update help text.
2680 (static-pie): New option.
2681 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
2682 -static-pie support.
2683 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
2684 (LINK_EH_SPEC): Likewise.
2685 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2686 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
2687 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
2688 * gcc.c (LINK_COMMAND_SPEC): Likewise.
2689 (init_gcc_specs): Likewise.
2690 (init_spec): Likewise.
2691 (display_help): Update help message for -pie.
2692 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
2695 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
2697 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
2698 (movdi_aarch64): Likewise.
2699 (movti_aarch64): Likewise.
2701 2017-09-12 Simon Wright <simon@pushface.org>
2704 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
2705 calculation of the minor version, always output as 0.
2707 2017-09-12 Jakub Jelinek <jakub@redhat.com>
2710 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
2711 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
2712 on it early, rather than manual conversion late. For
2713 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
2714 instead of performing manual conversion.
2716 2017-09-12 Carl Love <cel@us.ibm.com>
2718 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
2719 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
2721 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
2722 VMULOSW): Add definitions.
2723 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
2724 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
2725 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
2726 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
2727 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
2729 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
2731 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
2733 (movti_aarch64): Likewise.
2734 (movdf_aarch64): Likewise.
2735 (movtf_aarch64): Likewise.
2736 (load_pairdi): Likewise.
2737 (store_pairdi): Likewise.
2738 (load_pairdf): Likewise.
2739 (store_pairdf): Likewise.
2740 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
2741 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
2742 (ldr_got_small_<mode>): Likewise.
2743 (ldr_got_small_28k_<mode>): Likewise.
2744 (ldr_got_tiny): Likewise.
2745 * config/aarch64/iterators.md (ldst_sz): New.
2746 (ldpstp_sz): Likewise.
2747 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
2749 (thunderx_load): Split load_8 to load_16.
2750 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
2752 (thunderx2t99_storepair_basic): Split store_8 to store_16.
2753 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
2754 (xgene1_store_pair): Split store_8 to store_16.
2755 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
2756 (falkor_st_0_st_sd): Split store_8 to store_16.
2758 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
2760 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
2761 and store1/2/3/4 to store_4/8/12/16.
2762 * config/aarch64/aarch64.md: Update for rename.
2763 * config/arm/arm.md: Likewise.: Likewise.
2764 * config/arm/arm.c: Likewise.
2765 * config/arm/thumb1.md: Likewise.
2766 * config/arm/thumb2.md: Likewise.
2767 * config/arm/vfp.md: Likewise.
2768 * config/arm/arm-generic.md: Likewise.
2769 * config/arm/arm1020e.md: Likewise.
2770 * config/arm/arm1026ejs.md: Likewise.
2771 * config/arm/arm1136jfs.md: Likewise.
2772 * config/arm/arm926ejs.md: Likewise.
2773 * config/arm/cortex-a15.md: Likewise.
2774 * config/arm/cortex-a17.md: Likewise.
2775 * config/arm/cortex-a5.md: Likewise.
2776 * config/arm/cortex-a53.md: Likewise.
2777 * config/arm/cortex-a57.md: Likewise.
2778 * config/arm/cortex-a7.md: Likewise.
2779 * config/arm/cortex-a8.md: Likewise.
2780 * config/arm/cortex-a9.md: Likewise.
2781 * config/arm/cortex-m4.md: Likewise.
2782 * config/arm/cortex-m7.md: Likewise.
2783 * config/arm/cortex-r4.md: Likewise.
2784 * config/arm/exynos-m1.md: Likewise.
2785 * config/arm/fa526.md: Likewise.
2786 * config/arm/fa606te.md: Likewise.
2787 * config/arm/fa626te.md: Likewise.
2788 * config/arm/fa726te.md: Likewise.
2789 * config/arm/fmp626.md: Likewise.
2790 * config/arm/iwmmxt.md: Likewise.
2791 * config/arm/ldmstm.md: Likewise.
2792 * config/arm/marvell-pj4.md: Likewise.
2793 * config/arm/xgene1.md: Likewise.
2794 * config/aarch64/thunderx.md: Likewise.
2795 * config/aarch64/thunderx2t99.md: Likewise.
2796 * config/aarch64/falkor.md: Likewise.
2798 2017-09-12 Martin Liska <mliska@suse.cz>
2800 * attribs.c (private_lookup_attribute): New function.
2801 * attribs.h (private_lookup_attribute): Declared here.
2802 (lookup_attribute): Called from this place.
2804 2017-09-12 Richard Biener <rguenther@suse.de>
2806 PR tree-optimization/82157
2807 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
2808 stmts with side-effects.
2810 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2811 Alan Hayward <alan.hayward@arm.com>
2812 David Sherwood <david.sherwood@arm.com>
2814 * target.def (hard_regno_nregs): New hook.
2815 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
2816 * targhooks.h (default_hard_regno_nregs): Declare.
2817 * targhooks.c (default_hard_regno_nregs): New function.
2818 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
2819 (TARGET_HARD_REGNO_NREGS): ...this hook.
2820 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
2821 (CLASS_MAX_NREGS): Likewise.
2822 * doc/tm.texi: Regenerate.
2823 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
2824 instead of HARD_REGNO_NREGS.
2825 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
2826 HARD_REGNO_NREGS in the comment.
2827 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
2828 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
2829 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
2830 Return an unsigned int.
2831 (TARGET_HARD_REGNO_NREGS): Redefine.
2832 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
2833 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
2834 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
2835 (arc_hard_regno_nregs): New function.
2836 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
2837 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
2838 (arm_hard_regno_nregs): New function.
2839 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
2840 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
2841 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
2842 (TARGET_HARD_REGNO_NREGS): Redefine.
2843 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
2844 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
2845 (HARD_REGNO_NREGS): Delete.
2846 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
2847 (cr16_hard_regno_nregs): New function.
2848 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
2849 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
2850 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
2851 (cris_hard_regno_nregs): New function.
2852 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
2853 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
2854 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
2855 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
2856 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
2857 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
2858 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
2859 (frv_hard_regno_nregs): Make static. Take and return an
2861 (frv_class_max_nregs): Remove outdated copy of documentation.
2862 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
2863 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
2864 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
2865 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
2866 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
2867 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
2868 (TARGET_HARD_REGNO_NREGS): Redefine.
2869 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
2870 (CLASS_MAX_NREGS): Update comment.
2871 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
2872 (ia64_hard_regno_nregs): New function.
2873 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
2874 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
2875 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
2876 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
2877 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
2879 (m32c_hard_regno_nregs): Likewise. Make static.
2880 (TARGET_HARD_REGNO_NREGS): Redefine.
2881 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
2882 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
2883 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
2884 (m68k_hard_regno_nregs): New function.
2885 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
2886 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
2887 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
2888 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
2889 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
2890 Take and return an unsigned int.
2891 (TARGET_HARD_REGNO_NREGS): Redefine.
2892 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
2893 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
2894 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
2895 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
2896 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
2897 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
2898 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
2899 (msp430_hard_regno_nregs): Make static. Take and return an
2901 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
2902 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
2903 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
2904 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
2905 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
2906 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
2907 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
2908 (TARGET_HARD_REGNO_NREGS): Redefine.
2909 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
2910 (PA_HARD_REGNO_NREGS): ...this.
2911 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
2912 (PA_HARD_REGNO_NREGS): ...this.
2913 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
2914 (pa_hard_regno_nregs): New function.
2915 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
2916 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
2917 (pdp11_hard_regno_nregs): New function.
2918 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
2919 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
2920 (rs6000_hard_regno_nregs_hook): New function.
2921 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
2922 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
2923 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
2924 Take and return an unsigned int. Move earlier in file.
2925 (TARGET_HARD_REGNO_NREGS): Redefine.
2926 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
2927 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
2928 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
2929 (rl78_hard_regno_nregs): Make static. Take and return an
2931 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
2932 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
2933 (rs6000_hard_regno_nregs_hook): New function.
2934 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
2935 * config/rx/rx.c (rx_hard_regno_nregs): New function.
2936 (TARGET_HARD_REGNO_NREGS): Redefine.
2937 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
2938 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
2939 instead of HARD_REGNO_NREGS.
2940 (s390_hard_regno_nregs): New function.
2941 (s390_hard_regno_mode_ok): Add comment from s390.h.
2942 (TARGET_HARD_REGNO_NREGS): Redefine.
2943 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
2944 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
2945 (sh_hard_regno_nregs): New function.
2946 (sh_pass_in_reg_p): Use it.
2947 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
2948 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
2949 (sparc_hard_regno_nregs): New function.
2950 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
2951 * config/spu/spu.c (spu_hard_regno_nregs): New function.
2952 (spu_function_arg_advance): Use it, supplying a valid register number.
2953 (TARGET_HARD_REGNO_NREGS): Redefine.
2954 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
2955 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
2956 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
2957 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
2958 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
2959 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
2960 (CLASS_MAX_NREGS): Remove copy of old documentation.
2961 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
2962 (visium_hard_regno_nregs): New function.
2963 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
2964 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
2965 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
2966 xtensa_hard_regno_nregs): New function.
2967 * system.h (HARD_REGNO_NREGS): Poison.
2969 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2971 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
2972 hard_regno_nregs instead of HARD_REGNO_NREGS.
2973 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
2974 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
2975 (c6x_expand_epilogue): Likewise.
2976 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
2977 (frv_read_iacc_argument): Likewise.
2978 * config/sh/sh.c: Include regs.h.
2979 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
2980 (regs_used): Likewise.
2981 (output_stack_adjust): Likewise.
2982 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
2983 * expmed.c: Include regs.h.
2984 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
2985 * ree.c: Include regs.h.
2986 (combine_reaching_defs): Use hard_regno_nregs instead of
2988 (add_removable_extension): Likewise.
2990 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2992 * regs.h (hard_regno_nregs): Turn into a function.
2993 (end_hard_regno): Update accordingly.
2994 * caller-save.c (setup_save_areas): Likewise.
2995 (save_call_clobbered_regs): Likewise.
2996 (replace_reg_with_saved_mem): Likewise.
2997 (insert_restore): Likewise.
2998 (insert_save): Likewise.
2999 * combine.c (can_change_dest_mode): Likewise.
3000 (move_deaths): Likewise.
3001 (distribute_notes): Likewise.
3002 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
3003 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
3004 (rs6000_split_multireg_move): Likewise.
3005 (rs6000_register_move_cost): Likewise.
3006 (rs6000_memory_move_cost): Likewise.
3007 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
3008 (rs6000_split_multireg_move): Likewise.
3009 (rs6000_register_move_cost): Likewise.
3010 (rs6000_memory_move_cost): Likewise.
3011 * cselib.c (cselib_reset_table): Likewise.
3012 (cselib_lookup_1): Likewise.
3013 * emit-rtl.c (set_mode_and_regno): Likewise.
3014 * function.c (aggregate_value_p): Likewise.
3015 * ira-color.c (setup_profitable_hard_regs): Likewise.
3016 (check_hard_reg_p): Likewise.
3017 (calculate_saved_nregs): Likewise.
3018 (assign_hard_reg): Likewise.
3019 (improve_allocation): Likewise.
3020 (calculate_spill_cost): Likewise.
3021 * ira-emit.c (modify_move_list): Likewise.
3022 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
3023 (ira_hard_reg_in_set_p): Likewise.
3024 * ira.c (setup_reg_mode_hard_regset): Likewise.
3025 (clarify_prohibited_class_mode_regs): Likewise.
3026 (check_allocation): Likewise.
3027 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3028 (lra_setup_reg_renumber): Likewise.
3029 (setup_try_hard_regno_pseudos): Likewise.
3030 (spill_for): Likewise.
3031 (assign_hard_regno): Likewise.
3032 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
3033 * lra-constraints.c (in_class_p): Likewise.
3034 (lra_constraint_offset): Likewise.
3035 (simplify_operand_subreg): Likewise.
3036 (lra_constraints): Likewise.
3037 (split_reg): Likewise.
3038 (split_if_necessary): Likewise.
3039 (invariant_p): Likewise.
3040 (inherit_in_ebb): Likewise.
3041 * lra-lives.c (process_bb_lives): Likewise.
3042 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
3043 (get_hard_regs): Likewise.
3044 (do_remat): Likewise.
3045 * lra-spills.c (assign_spill_hard_regs): Likewise.
3046 * mode-switching.c (create_pre_exit): Likewise.
3047 * postreload.c (reload_combine_recognize_pattern): Likewise.
3048 * recog.c (peep2_find_free_register): Likewise.
3049 * regcprop.c (kill_value_regno): Likewise.
3050 (set_value_regno): Likewise.
3051 (copy_value): Likewise.
3052 (maybe_mode_change): Likewise.
3053 (find_oldest_value_reg): Likewise.
3054 (copyprop_hardreg_forward_1): Likewise.
3055 * regrename.c (check_new_reg_p): Likewise.
3056 (regrename_do_replace): Likewise.
3057 * reload.c (push_reload): Likewise.
3058 (combine_reloads): Likewise.
3059 (find_dummy_reload): Likewise.
3060 (operands_match_p): Likewise.
3061 (find_reloads): Likewise.
3062 (find_equiv_reg): Likewise.
3063 (reload_adjust_reg_for_mode): Likewise.
3064 * reload1.c (count_pseudo): Likewise.
3065 (count_spilled_pseudo): Likewise.
3066 (find_reg): Likewise.
3067 (clear_reload_reg_in_use): Likewise.
3068 (free_for_value_p): Likewise.
3069 (allocate_reload_reg): Likewise.
3070 (choose_reload_regs): Likewise.
3071 (reload_adjust_reg_for_temp): Likewise.
3072 (emit_reload_insns): Likewise.
3073 (delete_output_reload): Likewise.
3074 * rtlanal.c (subreg_get_info): Likewise.
3075 * sched-deps.c (sched_analyze_reg): Likewise.
3076 * sel-sched.c (init_regs_for_mode): Likewise.
3077 (mark_unavailable_hard_regs): Likewise.
3078 (choose_best_reg_1): Likewise.
3079 (verify_target_availability): Likewise.
3080 * valtrack.c (dead_debug_insert_temp): Likewise.
3081 * var-tracking.c (track_loc_p): Likewise.
3082 (emit_note_insn_var_location): Likewise.
3083 * varasm.c (make_decl_rtl): Likewise.
3084 * reginfo.c (choose_hard_reg_mode): Likewise.
3085 (init_reg_modes_target): Refer directly to
3086 this_target_regs->x_hard_regno_nregs.
3088 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3090 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
3091 instead of hard_regno_nregs.
3093 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3095 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
3096 end_hard_regno instead of hard_regno_nregs.
3097 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
3098 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
3099 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
3100 * ira-color.c (improve_allocation): Likewise.
3101 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3102 * lra-lives.c (mark_regno_live): Likewise.
3103 (mark_regno_dead): Likewise.
3104 * lra-remat.c (operand_to_remat): Likewise.
3105 * lra.c (collect_non_operand_hard_regs): Likewise.
3106 * postreload.c (reload_combine_note_store): Likewise.
3107 (move2add_valid_value_p): Likewise.
3108 * reload.c (regno_clobbered_p): Likewise.
3110 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3112 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
3114 * config/v850/v850.c (v850_reorg): Likewise.
3115 * reload.c (refers_to_regno_for_reload_p): Likewise.
3116 (find_equiv_reg): Likewise.
3117 * reload1.c (reload_reg_reaches_end_p): Likewise.
3119 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3121 * caller-save.c (add_used_regs): Use REG_NREGS instead of
3123 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
3124 * config/arm/arm.c (output_move_neon): Likewise.
3125 (arm_attr_length_move_neon): Likewise.
3126 (neon_split_vcombine): Likewise.
3127 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
3128 (c6x_mark_reg_written): Likewise.
3129 (c6x_dwarf_register_span): Likewise.
3130 * config/i386/i386.c (ix86_save_reg): Likewise.
3131 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
3132 (rws_access_reg): Likewise.
3133 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
3134 * mode-switching.c (create_pre_exit): Likewise.
3135 * ree.c (combine_reaching_defs): Likewise.
3136 (add_removable_extension): Likewise.
3137 * regcprop.c (find_oldest_value_reg): Likewise.
3138 (copyprop_hardreg_forward_1): Likewise.
3139 * reload.c (reload_inner_reg_of_subreg): Likewise.
3140 (push_reload): Likewise.
3141 (combine_reloads): Likewise.
3142 (find_dummy_reload): Likewise.
3143 (reload_adjust_reg_for_mode): Likewise.
3144 * reload1.c (find_reload_regs): Likewise.
3145 (forget_old_reloads_1): Likewise.
3146 (reload_reg_free_for_value_p): Likewise.
3147 (reload_adjust_reg_for_temp): Likewise.
3148 (emit_reload_insns): Likewise.
3149 (delete_output_reload): Likewise.
3150 * sel-sched.c (choose_best_reg_1): Likewise.
3151 (choose_best_pseudo_reg): Likewise.
3153 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3154 Alan Hayward <alan.hayward@arm.com>
3155 David Sherwood <david.sherwood@arm.com>
3157 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
3158 * target.def (slow_unaligned_access): New hook.
3159 * targhooks.h (default_slow_unaligned_access): Declare.
3160 * targhooks.c (default_slow_unaligned_access): New function.
3161 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
3162 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
3163 * doc/tm.texi: Regenerate.
3164 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
3165 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
3166 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
3168 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
3169 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
3171 (rs6000_slow_unaligned_access): New function.
3172 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
3173 (expand_block_compare): Likewise.
3174 (expand_strn_compare): Likewise.
3175 (rs6000_rtx_costs): Likewise.
3176 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
3177 (riscv_slow_unaligned_access): Likewise.
3178 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
3179 (riscv_slow_unaligned_access_p): ...this and make static.
3180 (riscv_option_override): Update accordingly.
3181 (riscv_slow_unaligned_access): New function.
3182 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
3183 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
3184 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
3185 (rs6000_slow_unaligned_access): New function.
3186 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
3187 (rs6000_rtx_costs): Likewise.
3188 * config/rs6000/rs6000-string.c (expand_block_compare)
3189 (expand_strn_compare): Use targetm.slow_unaligned_access instead
3190 of SLOW_UNALIGNED_ACCESS.
3191 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
3192 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
3193 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
3194 of SLOW_UNALIGNED_ACCESS.
3195 * expmed.c (simple_mem_bitfield_p): Likewise.
3196 * expr.c (alignment_for_piecewise_move): Likewise.
3197 (emit_group_load_1): Likewise.
3198 (emit_group_store): Likewise.
3199 (copy_blkmode_from_reg): Likewise.
3200 (emit_push_insn): Likewise.
3201 (expand_assignment): Likewise.
3202 (store_field): Likewise.
3203 (expand_expr_real_1): Likewise.
3204 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
3205 * lra-constraints.c (simplify_operand_subreg): Likewise.
3206 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
3207 * gimple-ssa-store-merging.c: Likewise in block comment at start
3209 * tree-ssa-strlen.c: Include target.h.
3210 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
3211 of SLOW_UNALIGNED_ACCESS.
3212 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
3214 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3216 PR rtl-optimization/82185
3217 * expmed.c (emit_store_flag_int): Only test tem if it has been
3220 2017-09-12 Richard Biener <rguenther@suse.de>
3223 * match.pd ((FTYPE) N CMP CST): Fix typo.
3225 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
3227 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
3229 (mips_near_type_p): Add 'short_call' attribute as a synonym
3231 * doc/extend.texi (short_call): Document new function attribute.
3233 2017-09-12 Jakub Jelinek <jakub@redhat.com>
3236 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
3237 assertion check that in the condition.
3238 (get_atomic_generic_size): Likewise. Before testing if parameter
3239 has pointer type, if it has array type, call for C++
3240 default_conversion to perform array-to-pointer conversion.
3242 2017-09-12 Richard Biener <rguenther@suse.de>
3244 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
3245 for operations we cannot scalarize.
3247 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
3249 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
3250 vectors heap vectors. Clean up comments.
3251 Make visited_bbs a reference.
3252 (profitable_jump_thread_path): Make GC
3253 vectors heap vectors. Clean up comments.
3255 (convert_and_register_jump_thread_path): Make GC vectors heap
3257 (check_subpath_and_update_thread_path): Same. Clean up comments.
3258 Make visited_bbs a reference.
3259 (handle_phi): Abstract common code to to
3260 register_jump_thread_path_if_profitable.
3261 Rename VAR_BB to DEF_BB.
3263 Make GC vectors heap vectors.
3264 Make visited_bbs a reference.
3265 (handle_assignment): Same.
3266 (register_jump_thread_path_if_profitable): New.
3267 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
3269 Make GC vectors heap vectors. Clean up comments.
3270 Make visited_bbs a reference.
3271 (find_jump_threads_backwards): Make visited_bbs live in the stack.
3272 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
3275 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
3278 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
3279 words of E_DImode object are reachable by xtensa_uimm8x4 access.
3281 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
3283 Revert r251800 and r251799.
3285 2017-09-11 Martin Jambor <mjambor@suse.cz>
3288 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
3289 arguments in advance.
3290 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
3291 use it to find predecessor edges.
3292 (naive_outof_ssa): Collect vector of predecessors.
3294 2017-09-08 Jason Merrill <jason@redhat.com>
3296 PR c++/70029 - ICE with ref-qualifier and -flto
3297 * langhooks.h (struct lang_hooks_for_types): Add
3298 copy_lang_qualifiers.
3299 * attribs.c (build_type_attribute_qual_variant): Use it.
3300 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
3302 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
3303 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
3305 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
3308 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
3309 (*mulsi3_sp64): New instruction.
3310 (mulsi3): New expander.
3312 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
3314 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
3316 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3318 * sancov.c: Include memmodel.h.
3320 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
3323 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
3326 2017-09-07 Carl Love <cel@us.ibm.com>
3328 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
3329 the sldi instruction.
3331 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
3333 * sancov.c: Include tm_p.h.
3335 2017-09-07 Jakub Jelinek <jakub@redhat.com>
3338 * output.h (switch_to_other_text_partition): New declaration.
3339 * varasm.c (switch_to_other_text_partition): New function.
3340 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
3341 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
3342 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
3343 to the other text partition before emitting LCL label and switch back
3344 after emitting the word after it.
3346 2017-09-07 Richard Biener <rguenther@suse.de>
3348 * passes.def (pass_split_crit_edges): Remove instance before PRE.
3349 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
3350 critical edges here, after loop init.
3351 (pass_data_pre): Remove PROP_no_crit_edges flags.
3352 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
3353 for valueization of call args to avoid leaking VN_TOP.
3354 (visit_use): Assert we do not visit default defs.
3355 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
3356 Use error_mark_node to more easily detect leaking VN_TOP.
3357 All default-defs are varying, not VN_TOP. Mark them visited.
3358 (run_scc_vn): Make code match comment.
3360 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
3362 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
3363 OPTION_MASK_FLOAT128_KEYWORD.
3364 (POWERPC_MASKS): Likewise.
3365 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
3366 support for the -mfloat128-type option, and make -mfloat128
3367 default on PowerPC Linux systems. Define or undefine
3368 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
3369 Define __float128 to be __ieee128 if IEEE 128-bit support is
3370 enabled, or undefine it.
3371 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
3372 Delete defining __FLOAT128_TYPE__.
3373 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
3374 -mfloat128-type option and make -mfloat128 default on PowerPC
3376 (TARGET_FLOAT128_TYPE): Likewise.
3377 (-mfloat128-type): Likewise.
3378 * config/rs6000/rs6000.c (rs6000_option_override_internal):
3379 Delete the -mfloat128-type option and make -mfloat128 default on
3380 PowerPC Linux systems. Always use __ieee128 to be the keyword for
3381 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
3382 128-bit floating point is enabled. Change tests from using
3383 -mfloat128-type to -mfloat128.
3384 (rs6000_mangle_type): Use the correct mangling for the __float128
3385 type even if normal long double is restricted to 64-bits.
3386 (floatn_mode): Enable the _Float128 type by default on VSX Linux
3388 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
3389 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
3390 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
3392 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
3393 documentation for -mfloat128.
3395 2017-09-06 Olivier Hainque <hainque@adacore.com>
3397 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
3399 2017-09-06 Wish Wu <wishwu007@gmail.com>
3400 Jakub Jelinek <jakub@redhat.com>
3402 * asan.c (initialize_sanitizer_builtins): Add
3403 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
3404 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
3405 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
3406 BT_FN_VOID_UINT64_PTR variables.
3407 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
3408 (BT_FN_VOID_UINT16_UINT16): Likewise.
3409 (BT_FN_VOID_UINT32_UINT32): Likewise.
3410 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
3411 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
3412 (BT_FN_VOID_UINT64_PTR): Likewise.
3413 * common.opt (flag_sanitize_coverage): New variable.
3414 (fsanitize-coverage=trace-pc): Remove.
3415 (fsanitize-coverage=): Add.
3416 * flag-types.h (enum sanitize_coverage_code): New enum.
3417 * fold-const.c (fold_range_test): Disable non-short-circuit
3418 optimization if flag_sanitize_coverage.
3419 (fold_truth_andor): Likewise.
3420 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
3421 * opts.c (COVERAGE_SANITIZER_OPT): Define.
3422 (coverage_sanitizer_opts): New array.
3423 (get_closest_sanitizer_option): Add OPTS argument, handle also
3424 OPT_fsanitize_coverage_.
3425 (parse_sanitizer_options): Adjusted to also handle
3426 OPT_fsanitize_coverage_.
3427 (common_handle_option): Add OPT_fsanitize_coverage_.
3428 * sancov.c (instrument_comparison, instrument_switch): New function.
3429 (sancov_pass): Add trace-cmp support.
3430 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
3431 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
3432 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
3433 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
3434 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
3435 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
3436 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
3437 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
3438 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
3439 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
3441 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
3443 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
3444 error. Only quit immediately if parsing is complete.
3445 (BEGIN): Initialize fatal_err and parse_done.
3446 (begin fpu, end fpu): Check number of arguments.
3447 (begin arch, end arch): Likewise.
3448 (begin cpu, end cpu): Likewise.
3449 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
3450 (optalias): Likewise.
3452 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
3454 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
3455 * config/arm/arm-isa.h: Delete. Move definitions to ...
3456 * arm-cpus.in: ... here. Use new feature and fgroup values.
3457 * config/arm/arm.c (arm_option_override): Use lower case for feature
3459 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
3460 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
3461 * config/arm/parsecpu.awk (END): Add new command 'isa'.
3463 (print_isa_bits_for): New function.
3464 (gen_isa): New function.
3465 (gen_comm_data): Use print_isa_bits_for.
3466 (define feature): New keyword.
3467 (define fgroup): New keyword.
3468 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
3469 (arm-isa.h): Add rule to generate file.
3470 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
3471 case for feature bit names.
3473 2017-09-06 Richard Biener <rguenther@suse.de>
3475 * tree-ssa-pre.c (NECESSARY): Remove.
3476 (create_expression_by_pieces): Do not touch pass-local flags.
3477 (insert_into_preds_of_block): Likewise.
3478 (do_pre_regular_insertion): Likewise.
3479 (eliminate_insert): Likewise.
3480 (eliminate_dom_walker::before_dom_children): Likewise.
3481 (fini_eliminate): Do not look at inserted_exprs.
3482 (mark_operand_necessary): Remove.
3483 (remove_dead_inserted_code): Replace with simple work-list
3484 algorithm based on inserted_exprs and SSA uses.
3485 (pass_pre::execute): Re-order fini_eliminate and
3486 remove_dead_inserted_code.
3488 2017-09-06 Olivier Hainque <hainque@adacore.com>
3490 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
3491 for VxWorks 7. Adjust surrounding comments.
3493 2017-09-06 Richard Biener <rguenther@suse.de>
3495 * gimple-ssa-strength-reduction.c
3496 (find_candidates_dom_walker::before_dom_children): Also allow
3499 2017-09-06 Richard Biener <rguenther@suse.de>
3501 PR tree-optimization/82108
3502 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
3503 for gap in the non-permutation SLP case.
3505 2017-09-06 Martin Jambor <mjambor@suse.cz>
3507 PR tree-optimization/82078
3508 * tree-sra.c (sort_and_splice_var_accesses): Move call to
3509 add_access_to_work_queue...
3510 (build_accesses_from_assign): ...here.
3511 (propagate_all_subaccesses): Make sure racc is the group
3512 representative, if there is one.
3514 2017-09-06 Jakub Jelinek <jakub@redhat.com>
3517 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
3520 2017-09-06 Richard Biener <rguenther@suse.de>
3522 * gimple-ssa-strength-reduction.c
3523 (find_candidates_dom_walker::before_doom_children): Use a
3524 type and not a mode check.
3526 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
3529 * config/arm/predicates.md (arm_general_adddi_operand): Create new
3531 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
3533 2017-09-05 Jeff Law <law@redhat.com>
3535 PR tree-optimization/64910
3536 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
3537 cases where we have 3 or more operands.
3539 2017-09-05 Jakub Jelinek <jakub@redhat.com>
3542 * omp-low.c (lower_omp_for): Recompute tree invariant if
3543 gimple_omp_for_initial/final is ADDR_EXPR.
3546 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
3547 into gimple val before gimplification fo the COND_EXPR.
3549 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
3551 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
3552 REGION_COPY argument.
3553 (thread_through_all_blocks): Remove unused argument to
3554 duplicate_thread_path.
3556 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3557 Alan Hayward <alan.hayward@arm.com>
3558 David Sherwood <david.sherwood@arm.com>
3560 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
3561 Take a scalar_mode rather than a machine_mode.
3562 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
3563 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
3564 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
3565 (aarch64_gen_adjusted_ldpstp): Likewise.
3566 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
3568 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3569 Alan Hayward <alan.hayward@arm.com>
3570 David Sherwood <david.sherwood@arm.com>
3572 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
3573 Take a scalar_int_mode instead of a machine_mode.
3574 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
3575 (aarch64_output_scalar_simd_mov_immediate): Likewise.
3576 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
3577 (aarch64_simd_attr_length_rglist): Delete.
3578 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
3579 a scalar_int_mode instead of a machine_mode.
3580 (aarch64_add_offset): Likewise.
3581 (aarch64_internal_mov_immediate): Likewise
3582 (aarch64_add_constant_internal): Likewise.
3583 (aarch64_add_constant): Likewise.
3584 (aarch64_movw_imm): Likewise.
3585 (aarch64_rtx_arith_op_extract_p): Likewise.
3586 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
3587 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
3588 Remove assert that the mode isn't a vector.
3589 (aarch64_output_scalar_simd_mov_immediate): Likewise.
3590 (aarch64_expand_mov_immediate): Update calls after above changes.
3591 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
3592 (aarch64_and_bitmask_imm): Check for scalar integer modes.
3593 (aarch64_move_imm): Likewise.
3594 (aarch64_can_const_movi_rtx_p): Likewise.
3595 (aarch64_strip_extend): Likewise.
3596 (aarch64_extr_rtx_p): Likewise.
3597 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
3598 a CONST_INT when the mode parameter is VOIDmode.
3599 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
3601 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3603 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
3604 * stor-layout.c (bitwise_mode_for_mode): Likewise.
3605 (bitwise_type_for_mode): Update accordingly.
3607 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3609 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
3610 * stor-layout.c (mode_for_size_tree): Likewise.
3611 (mode_for_array): Update accordingly.
3612 (layout_decl): Likewise.
3613 (compute_record_mode): Likewise. Only set the mode once.
3615 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3617 * target.def (get_mask_mode): Change return type to opt_mode.
3619 * doc/tm.texi: Regenerate.
3620 * targhooks.h (default_get_mask_mode): Return an opt_mode.
3621 * targhooks.c (default_get_mask_mode): Likewise.
3622 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
3623 * optabs-query.c (can_vec_mask_load_store_p): Update use of
3624 targetm.get_mask_mode.
3625 * tree.c (build_truth_vector_type): Likewise.
3627 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3629 * machmode.h (mode_for_vector): Return an opt_mode.
3630 * stor-layout.c (mode_for_vector): Likewise.
3631 (mode_for_int_vector): Update accordingly.
3632 (layout_type): Likewise.
3633 * config/i386/i386.c (emit_memmov): Likewise.
3634 (ix86_expand_set_or_movmem): Likewise.
3635 (ix86_expand_vector_init): Likewise.
3636 (ix86_get_mask_mode): Likewise.
3637 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
3639 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
3640 * expmed.c (extract_bit_field_1): Likewise.
3641 * expr.c (expand_expr_real_2): Likewise.
3642 * optabs-query.c (can_vec_perm_p): Likewise.
3643 (can_vec_mask_load_store_p): Likewise.
3644 * optabs.c (expand_vec_perm): Likewise.
3645 * targhooks.c (default_get_mask_mode): Likewise.
3646 * tree-vect-stmts.c (vectorizable_store): Likewise.
3647 (vectorizable_load): Likewise.
3648 (get_vectype_for_scalar_type_and_size): Likewise.
3650 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3652 * machmode.h (mode_for_int_vector): New function.
3653 * stor-layout.c (mode_for_int_vector): Likewise.
3654 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
3655 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
3656 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
3657 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
3658 (s390_expand_vcond): Likewise.
3660 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3662 * machmode.h (opt_machine_mode): New type.
3663 (opt_mode<T>): Allow construction from anything that can be
3665 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
3666 (mode_for_size): Return an opt_machine_mode.
3667 * stor-layout.c (mode_for_size): Likewise.
3668 (mode_for_size_tree): Update call accordingly.
3669 (bitwise_mode_for_mode): Likewise.
3670 (make_fract_type): Likewise.
3671 (make_accum_type): Likewise.
3672 * caller-save.c (replace_reg_with_saved_mem): Update call
3674 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3675 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3676 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3677 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3678 * expmed.c (extract_bit_field_1): Likewise.
3679 * reload.c (get_secondary_mem): Likewise.
3680 * varasm.c (assemble_integer): Likewise.
3681 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
3684 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3686 * machmode.h (decimal_float_mode_for_size): New function.
3687 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
3688 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
3689 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
3690 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
3691 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
3692 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
3694 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3696 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
3697 (get_builtin_sync_mode): Likewise.
3698 (expand_ifn_atomic_compare_exchange): Likewise.
3699 (expand_builtin_atomic_clear): Likewise.
3700 (expand_builtin_atomic_test_and_set): Likewise.
3701 (fold_builtin_atomic_always_lock_free): Likewise.
3702 * calls.c (compute_argument_addresses): Likewise.
3703 (emit_library_call_value_1): Likewise.
3704 (store_one_arg): Likewise.
3705 * combine.c (combine_instructions): Likewise.
3706 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
3707 * config/arm/arm.c (arm_function_value): Likewise.
3708 (aapcs_allocate_return_reg): Likewise.
3709 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
3710 * config/i386/i386.c (construct_container): Likewise.
3711 (ix86_gimplify_va_arg): Likewise.
3712 (ix86_expand_sse_cmp): Likewise.
3713 (emit_memmov): Likewise.
3714 (emit_memset): Likewise.
3715 (expand_small_movmem_or_setmem): Likewise.
3716 (ix86_expand_pextr): Likewise.
3717 (ix86_expand_pinsr): Likewise.
3718 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
3719 * config/microblaze/microblaze.c (microblaze_block_move_straight):
3721 * config/mips/mips.c (mips_function_value_1) Likewise.
3722 (mips_block_move_straight): Likewise.
3723 (mips_expand_ins_as_unaligned_store): Likewise.
3724 * config/powerpcspe/powerpcspe.c
3725 (rs6000_darwin64_record_arg_advance_flush): Likewise.
3726 (rs6000_darwin64_record_arg_flush): Likewise.
3727 * config/rs6000/rs6000.c
3728 (rs6000_darwin64_record_arg_advance_flush): Likewise.
3729 (rs6000_darwin64_record_arg_flush): Likewise.
3730 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
3731 (sparc_function_value_1): Likewise.
3732 * config/spu/spu.c (adjust_operand): Likewise.
3733 (spu_emit_branch_or_set): Likewise.
3734 (arith_immediate_p): Likewise.
3735 * emit-rtl.c (gen_lowpart_common): Likewise.
3736 * expr.c (expand_expr_real_1): Likewise.
3737 * function.c (assign_parm_setup_block): Likewise.
3738 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
3739 * reload1.c (alter_reg): Likewise.
3740 * stor-layout.c (mode_for_vector): Likewise.
3741 (layout_type): Likewise.
3743 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3745 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
3746 (spu_convert_move): Likewise.
3747 * lower-subreg.c (resolve_simple_move): Likewise.
3749 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3752 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
3753 define_insn to a define_expand.
3754 (altivec_vsum2sws_direct): New define_insn.
3755 (altivec_vsumsws): Convert from a define_insn to a define_expand.
3757 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
3759 * config/arm/arm.c (arm_option_params_internal): Improve setting of
3762 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
3767 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
3768 realign stack if stack alignment needed is less than incoming
3771 2017-09-05 Marek Polacek <polacek@redhat.com>
3774 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
3777 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
3779 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
3781 2017-09-05 Richard Biener <rguenther@suse.de>
3783 PR tree-optimization/82084
3784 * fold-const.c (can_native_encode_string_p): Handle wide characters.
3786 2017-09-05 Richard Biener <rguenther@suse.de>
3788 PR tree-optimization/82102
3789 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
3791 2017-09-05 Martin Liska <mliska@suse.cz>
3793 PR tree-optimization/82032
3794 * tree-cfg.c (generate_range_test): New function.
3795 * tree-cfg.h (generate_range_test): Declared here.
3796 * tree-cfgcleanup.c (convert_single_case_switch): New function.
3797 (cleanup_control_expr_graph): Use it.
3798 * tree-switch-conversion.c (try_switch_expansion): Remove
3800 (emit_case_nodes): Use generate_range_test.
3802 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
3805 * config/i386/i386.md (*<btsc><mode>_mask): Add
3806 TARGET_USE_BT to insn constraint.
3807 (*btr<mode>_mask): Ditto.
3809 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
3811 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
3812 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
3814 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3817 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
3818 TARGET_NEON and TARGET_IWMMXT.
3819 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
3820 TARGET_NEON and TARGET_IWMMXT.
3821 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
3823 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
3825 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
3826 (ix86_rewrite_tls_address): Ditto.
3827 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
3828 (ix86_rewrite_tls_address_1): Ditto.
3829 (ix86_rewrite_tls_address): Ditto.
3830 * config/i386/predicates.md (tls_address_pattern): New predicate.
3831 * config/i386/i386.md (TLS address splitter): New splitter.
3833 2017-09-04 Richard Biener <rguenther@suse.de>
3835 PR tree-optimization/82084
3836 * fold-const.h (can_native_encode_string_p): Declare.
3837 * fold-const.c (can_native_encode_string_p): Factor out from ...
3838 (native_encode_string): ... here.
3839 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
3840 vectorizing stores from constants we later cannot handle.
3842 2017-09-04 Marek Polacek <polacek@redhat.com>
3845 * doc/invoke.texi: Update -Wtautological-compare documentation.
3847 2017-09-04 Jeff Law <law@redhat.com>
3849 PR tree-optimization/64910
3850 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
3851 swap the first and last operand if the last is a constant.
3853 2017-09-04 Marek Polacek <polacek@redhat.com>
3856 * convert.c (do_narrow): When sanitizing signed integer overflows,
3857 bail out for signed types.
3858 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
3860 2017-09-04 Richard Biener <rguenther@suse.de>
3862 PR tree-optimization/82060
3863 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
3864 Move devirtualization after stmt folding and before EH/AB/noreturn
3865 cleanup to get the stmt refs canonicalized. Use a bool instead
3866 of gimple_modified_p since that doesn't work for NOPs. Schedule
3867 NOPs generated by folding for removal.
3869 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3870 Alan Hayward <alan.hayward@arm.com>
3871 David Sherwood <david.sherwood@arm.com>
3873 * coretypes.h (pad_direction): New enum.
3874 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
3875 (FUNCTION_ARG_PADDING): Likewise.
3876 * target.def (function_arg_padding): New hook.
3877 * targhooks.h (default_function_arg_padding): Declare.
3878 * targhooks.c (default_function_arg_padding): New function.
3879 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
3880 (TARGET_FUNCTION_ARG_PADDING): ...this.
3881 * doc/tm.texi: Regenerate.
3882 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
3883 instead of direction.
3884 (compute_argument_addresses): Likewise.
3885 (load_register_parameters): Likewise.
3886 (emit_library_call_value_1): Likewise.
3887 (store_one_arg): Use targetm.calls.function_arg_padding instead
3888 of FUNCTION_ARG_PADDING.
3889 (must_pass_in_stack_var_size_or_pad): Likewise.
3890 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
3891 (emit_group_store): Likewise.
3892 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
3893 instead of FUNCTION_ARG_PADDING.
3894 (emit_push_insn): Likewise, and propagate enum change throughout
3896 * function.h (direction): Delete.
3897 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
3899 * function.c (assign_parm_find_stack_rtl): Likewise.
3900 (assign_parm_setup_block_p): Likewise.
3901 (assign_parm_setup_block): Likewise.
3902 (gimplify_parameters): Likewise.
3903 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
3904 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
3906 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
3907 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3908 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
3909 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
3910 (aarch64_function_arg_padding): ...this new function.
3911 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
3912 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3913 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
3914 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3915 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
3916 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3917 (arm_pad_arg_upward): Replace with...
3918 (arm_function_arg_padding): ...this new function.
3919 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
3921 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
3922 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
3923 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3924 (ia64_hpux_function_arg_padding): Replace with...
3925 (ia64_function_arg_padding): ...this new function. Use pad_direction
3926 instead of direction. Check for TARGET_HPUX.
3927 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
3928 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3929 (iq2000_function_arg_padding): New function.
3930 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
3931 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
3932 (mips_function_arg_padding): ...this new function.
3933 (mips_pad_reg_upward): Update accordingly.
3934 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3935 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
3936 targetm.calls.function_arg_padding.
3937 (FUNCTION_ARG_PADDING): Delete.
3938 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3939 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
3940 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3941 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
3942 (nios2_block_reg_padding): Return pad_direction instead of direction.
3943 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
3944 instead of direction.
3945 (nios2_function_arg_padding): Likewise. Make static.
3946 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3947 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
3948 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
3949 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
3950 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3951 (pa_function_arg_padding): Make static. Return pad_direction instead
3953 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
3954 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3955 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
3956 instead of direction. Use targetm.calls.function_arg_padding.
3957 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
3958 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
3959 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
3960 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
3961 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
3963 (function_arg_padding): Rename to...
3964 (rs6000_function_arg_padding): ...this. Make static. Return
3965 pad_direction instead of direction.
3966 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
3967 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
3968 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3969 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
3970 instead of direction. Use targetm.calls.function_arg_padding.
3971 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
3972 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
3973 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
3974 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
3975 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3976 (function_arg_padding): Rename to...
3977 (rs6000_function_arg_padding): ...this. Make static. Return
3978 pad_direction instead of direction.
3979 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
3980 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
3981 * config/s390/s390.c (s390_function_arg_padding): New function.
3982 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3983 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
3984 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
3985 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3986 (function_arg_padding): Rename to...
3987 (sparc_function_arg_padding): ...this. Make static. Return
3988 pad_direction instead of direction.
3989 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
3990 * config/spu/spu.c (spu_function_arg_padding): New function.
3991 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3992 * system.h (FUNCTION_ARG_PADDING): Poison.
3994 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3995 Alan Hayward <alan.hayward@arm.com>
3996 David Sherwood <david.sherwood@arm.com>
3998 * target.def (modes_tieable_p): New hook.
3999 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
4000 (TARGET_MODES_TIEABLE_P): ...this.
4001 * doc/tm.texi.in: Regenerate.
4002 * hooks.h (hook_bool_mode_mode_true): Declare.
4003 * hooks.c (hook_bool_mode_mode_true): New function.
4004 * combine.c (subst): Use targetm.modes_tieable_p instead of
4006 * dse.c (find_shift_sequence): Likewise.
4007 * expmed.c (extract_low_bits): Likewise.
4008 * lower-subreg.c: Include target.h.
4009 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
4011 * rtlanal.c (rtx_cost): Likewise.
4012 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
4013 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
4014 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
4015 (TARGET_MODES_TIEABLE_P): Redefine.
4016 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
4017 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
4018 (TARGET_MODES_TIEABLE_P): Redefine.
4019 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
4020 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
4021 (arc_modes_tieable_p): New function.
4022 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
4023 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
4024 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
4025 (arm_modes_tieable_p): Make static.
4026 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
4027 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
4028 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
4029 (TARGET_MODES_TIEABLE_P): Redefine.
4030 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
4031 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
4032 (TARGET_MODES_TIEABLE_P): Redefine.
4033 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
4034 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
4035 (cr16_modes_tieable_p): New function.
4036 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
4037 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
4038 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
4039 (TRULY_NOOP_TRUNCATION): Update comment.
4040 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
4041 (TRULY_NOOP_TRUNCATION): Update comment.
4042 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
4043 (frv_modes_tieable_p): New function.
4044 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
4045 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
4046 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
4047 (TARGET_MODES_TIEABLE_P): Redefine.
4048 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
4049 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
4050 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
4051 (TARGET_MODES_TIEABLE_P): Redefine.
4052 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
4053 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
4054 (ia64_modes_tieable_p): New function.
4055 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
4056 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
4057 (iq2000_modes_tieable_p): New function.
4058 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
4059 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
4060 (lm32_modes_tieable_p): New function.
4061 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
4062 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
4063 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
4064 (TARGET_MODES_TIEABLE_P): Redefine.
4065 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
4066 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
4067 (m32r_modes_tieable_p): New function.
4068 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
4069 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
4070 (m68k_modes_tieable_p): New function.
4071 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
4072 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
4073 (mcore_modes_tieable_p): New function.
4074 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
4075 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
4077 (TARGET_MODES_TIEABLE_P): Redefine.
4078 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
4079 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
4080 * config/mips/mips.c (mips_modes_tieable_p): Make static.
4081 (TARGET_MODES_TIEABLE_P): Redefine.
4082 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
4083 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
4084 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
4085 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
4086 (mn10300_modes_tieable_p): ...this and make static.
4087 (TARGET_MODES_TIEABLE_P): Redefine.
4088 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
4089 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
4090 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
4091 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
4092 (msp430_modes_tieable_p): Make static.
4093 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
4094 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
4095 (TARGET_MODES_TIEABLE_P): Redefine.
4096 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
4097 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
4098 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
4099 (TARGET_MODES_TIEABLE_P): Redefine.
4100 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
4101 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
4102 * config/pa/pa.c (pa_modes_tieable_p): Make static.
4103 (TARGET_MODES_TIEABLE_P): Redefine.
4104 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
4105 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
4106 (pdp11_modes_tieable_p): New function.
4107 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
4108 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
4109 (rs6000_modes_tieable_p): New function.
4110 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
4111 * config/powerpcspe/powerpcspe.md: Update comment.
4112 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
4113 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
4114 (TARGET_MODES_TIEABLE_P): Redefine.
4115 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
4116 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
4117 (rl78_modes_tieable_p): New function.
4118 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
4119 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
4120 (rs6000_modes_tieable_p): New function.
4121 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
4122 * config/rs6000/rs6000.md: Update comment.
4123 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
4124 * config/rx/rx.c (rx_modes_tieable_p): New function.
4125 (TARGET_MODES_TIEABLE_P): Redefine.
4126 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
4127 * config/s390/s390.c (s390_modes_tieable_p): New function.
4128 (TARGET_MODES_TIEABLE_P): Redefine.
4129 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
4130 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
4131 (sh_modes_tieable_p): New function.
4132 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
4133 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
4134 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
4135 (sparc_modes_tieable_p): Make static.
4136 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
4137 * config/spu/spu.c (spu_modes_tieable_p): New function.
4138 (TARGET_MODES_TIEABLE_P): Redefine.
4139 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
4140 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
4141 (TARGET_MODES_TIEABLE_P): Redefine.
4142 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
4143 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
4144 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
4145 * config/v850/v850.c (v850_modes_tieable_p): New function.
4146 (TARGET_MODES_TIEABLE_P): Redefine.
4147 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
4148 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
4149 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
4150 (visium_modes_tieable_p): New function.
4151 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
4152 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
4153 (xtensa_modes_tieable_p): New function.
4154 * system.h (MODES_TIEABLE_P): Poison.
4156 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4157 Alan Hayward <alan.hayward@arm.com>
4158 David Sherwood <david.sherwood@arm.com>
4160 * target.def (hard_regno_mode_ok): New hook.
4161 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
4162 (TARGET_HARD_REGNO_MODE_OK): ...this.
4163 * doc/tm.texi.in: Regenerate.
4164 * hooks.h (hook_bool_uint_mode_true): Declare.
4165 * hooks.c (hook_bool_uint_mode_true): New function.
4166 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
4168 * genpreds.c (write_insn_preds_c): Add an include of target.h.
4169 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
4170 instead of HARD_REGNO_MODE_OK.
4171 * caller-save.c: Include target.h.
4172 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
4174 * combine.c (can_combine_p): Likewise.
4175 (combinable_i3pat): Likewise.
4176 (can_change_dest_mode): Likewise.
4177 * expr.c (init_expr_target): Likewise.
4178 (convert_move): Likewise.
4179 (convert_modes): Likewise.
4180 * ira.c (setup_prohibited_class_mode_regs): Likewise.
4181 (setup_prohibited_mode_move_regs): Likewise.
4182 * ira.h (target_ira): Likewise.
4183 * lra-assigns.c (find_hard_regno_for_1): Likewise.
4184 * lra-constraints.c (process_alt_operands): Likewise.
4185 (split_reg): Likewise.
4186 * recog.c (peep2_find_free_register): Likewise.
4187 * ree.c (combine_reaching_defs): Likewise.
4188 * regcprop.c (maybe_mode_change): Likewise.
4189 * reginfo.c (init_reg_sets_1): Likewise.
4190 (choose_hard_reg_mode): Likewise.
4191 (simplifiable_subregs): Likewise.
4192 * regrename.c (check_new_reg_p): Likewise.
4193 * reload.c (find_valid_class): Likewise.
4194 (find_valid_class_1): Likewise.
4195 (reload_inner_reg_of_subreg): Likewise.
4196 (push_reload): Likewise.
4197 (combine_reloads): Likewise.
4198 (find_dummy_reload): Likewise.
4199 (find_reloads): Likewise.
4200 * reload1.c (find_reg): Likewise.
4201 (set_reload_reg): Likewise.
4202 (allocate_reload_reg): Likewise.
4203 (choose_reload_regs): Likewise.
4204 (reload_adjust_reg_for_temp): Likewise.
4205 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
4206 (simplify_subreg_regno): Likewise.
4207 * sel-sched.c (init_regs_for_mode): Likewise.
4208 * varasm.c (make_decl_rtl): Likewise.
4209 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
4210 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
4212 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
4213 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
4214 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4215 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
4216 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
4217 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4218 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
4219 (arc_mode_class): Delete.
4220 (HARD_REGNO_MODE_OK): Delete.
4221 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4222 (arc_hard_regno_mode_ok): Rename old array to...
4223 (arc_hard_regno_mode_ok_modes): ...this.
4224 (arc_conditional_register_usage): Update accordingly.
4225 (arc_mode_class): Make static.
4226 (arc_hard_regno_mode_ok): New function.
4227 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
4228 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
4229 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4230 (arm_hard_regno_mode_ok): Make static.
4231 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
4233 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
4234 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
4235 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
4237 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4238 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
4239 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
4240 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
4241 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
4242 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4243 * config/bfin/predicates.md (valid_reg_operand): Use
4244 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
4245 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
4246 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
4247 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4248 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
4249 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
4250 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4251 (cr16_hard_regno_mode_ok): Make static and return a bool.
4252 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
4253 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4254 (cris_hard_regno_mode_ok): New function.
4255 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
4256 (epiphany_mode_class): Delete.
4257 (HARD_REGNO_MODE_OK): Delete.
4258 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
4259 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4260 (hard_regno_mode_ok): Rename to...
4261 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
4262 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
4263 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
4265 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
4266 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
4267 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4268 (frv_hard_regno_mode_ok): Make static and return a bool.
4269 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
4271 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
4272 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
4273 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
4274 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
4276 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4277 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
4278 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
4279 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
4281 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4282 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
4283 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4284 (ia64_hard_regno_mode_ok): New function.
4285 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
4286 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4287 (iq2000_hard_regno_mode_ok): New function.
4288 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
4289 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4290 (lm32_hard_regno_mode_ok): New function.
4291 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
4292 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
4293 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
4294 instead of HARD_REGNO_MODE_OK.
4295 (m32c_hard_regno_ok): Rename to...
4296 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
4297 (m32c_cannot_change_mode_class): Update accordingly.
4298 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4299 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
4300 (m32r_mode_class): Delete.
4301 (HARD_REGNO_MODE_OK): Delete.
4302 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4303 (m32r_hard_regno_mode_ok): Rename to...
4304 (m32r_hard_regno_modes): ...this.
4305 (m32r_mode_class): Make static.
4306 (m32r_hard_regno_mode_ok): New function.
4307 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
4308 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
4309 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4310 (m68k_hard_regno_mode_ok): Make static.
4311 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
4312 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4313 (mcore_hard_regno_mode_ok): New function.
4314 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
4315 (HARD_REGNO_MODE_OK): Delete.
4316 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
4318 (microblaze_hard_regno_mode_ok_p): ...this and make static.
4319 (microblaze_hard_regno_mode_ok): New function.
4320 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4321 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
4322 (mips_hard_regno_mode_ok): Delete.
4323 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
4324 (mips_hard_regno_mode_ok_p): ...this and make static.
4325 (mips_hard_regno_mode_ok_p): Rename to...
4326 (mips_hard_regno_mode_ok_uncached): ...this.
4327 (mips_hard_regno_mode_ok): New function.
4328 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
4329 of HARD_REGNO_MODE_OK.
4330 (mips_option_override): Update after above name changes.
4331 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4332 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
4333 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
4334 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
4335 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
4336 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4337 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
4338 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
4339 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
4340 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4341 (msp430_hard_regno_mode_ok): Make static and return a bool.
4342 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
4343 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
4344 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
4346 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4347 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
4348 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
4349 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
4350 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
4351 (PA_HARD_REGNO_MODE_OK): ...this
4352 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
4353 (PA_HARD_REGNO_MODE_OK): ...this.
4354 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4355 (pa_hard_regno_mode_ok): New function.
4356 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
4357 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4358 (pdp11_hard_regno_mode_ok): New function.
4359 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
4360 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
4362 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
4364 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4365 (rs6000_hard_regno_mode_ok): Rename to...
4366 (rs6000_hard_regno_mode_ok_uncached): ...this.
4367 (rs6000_init_hard_regno_mode_ok): Update accordingly.
4368 (rs6000_hard_regno_mode_ok): New function.
4369 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
4370 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
4371 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
4372 (riscv_hard_regno_mode_ok): ...this and make static.
4373 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4374 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
4375 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
4376 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4377 (rl78_hard_regno_mode_ok): Make static and return bool.
4378 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
4379 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
4381 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
4382 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4383 (rs6000_hard_regno_mode_ok): Rename to...
4384 (rs6000_hard_regno_mode_ok_uncached): ...this.
4385 (rs6000_init_hard_regno_mode_ok): Update accordingly.
4386 (rs6000_hard_regno_mode_ok): New function.
4387 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
4388 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
4389 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4390 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
4391 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
4392 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
4393 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4394 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
4395 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
4396 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4397 (sh_hard_regno_mode_ok): Make static.
4398 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
4399 instead of HARD_REGNO_MODE_OK.
4400 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
4401 (sparc_mode_class): Delete.
4402 (HARD_REGNO_MODE_OK): Delete.
4403 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4404 (hard_regno_mode_classes): Make static.
4405 (sparc_mode_class): Likewise.
4406 (sparc_hard_regno_mode_ok): New function.
4407 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
4408 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
4409 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
4411 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4412 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
4413 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
4414 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
4415 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
4416 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4417 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
4418 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
4419 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4420 (visium_hard_regno_mode_ok): New function.
4421 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
4422 instead of HARD_REGNO_MODE_OK.
4423 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
4424 (HARD_REGNO_MODE_OK): Delete.
4425 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
4426 (xtensa_hard_regno_mode_ok_p): ...this and make static.
4427 (xtensa_option_override): Update accordingly.
4428 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4429 (xtensa_hard_regno_mode_ok): New function.
4430 * system.h (HARD_REGNO_MODE_OK): Poison.
4432 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4433 Alan Hayward <alan.hayward@arm.com>
4434 David Sherwood <david.sherwood@arm.com>
4436 * target.def (hard_regno_call_part_clobbered): New hook.
4437 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
4438 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
4439 * doc/tm.texi: Regenerate.
4440 * hooks.h (hook_bool_uint_mode_false): Declare.
4441 * hooks.c (hook_bool_uint_mode_false): New function.
4442 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4443 * cselib.c (cselib_process_insn): Use
4444 targetm.hard_regno_call_part_clobbered instead of
4445 HARD_REGNO_CALL_PART_CLOBBERED.
4446 * ira-conflicts.c (ira_build_conflicts): Likewise.
4447 * ira-costs.c (ira_tune_allocno_costs): Likewise.
4448 * lra-constraints.c (need_for_call_save_p): Likewise.
4449 * lra-lives.c: Include target.h.
4450 (check_pseudos_live_through_calls): Use
4451 targetm.hard_regno_call_part_clobbered instead of
4452 HARD_REGNO_CALL_PART_CLOBBERED.
4453 * regcprop.c: Include target.h.
4454 (copyprop_hardreg_forward_1): Use
4455 targetm.hard_regno_call_part_clobbered instead of
4456 HARD_REGNO_CALL_PART_CLOBBERED.
4457 * reginfo.c (choose_hard_reg_mode): Likewise.
4458 * regrename.c (check_new_reg_p): Likewise.
4459 * reload.c (find_equiv_reg): Likewise.
4460 * reload1.c (emit_reload_insns): Likewise.
4461 * sched-deps.c (deps_analyze_insn): Likewise.
4462 * sel-sched.c (init_regs_for_mode): Likewise.
4463 (mark_unavailable_hard_regs): Likewise.
4464 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
4465 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4466 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
4468 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4469 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4470 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
4472 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
4474 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4475 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4476 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
4478 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4479 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4480 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
4482 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4483 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
4485 * config/powerpcspe/powerpcspe.c
4486 (rs6000_hard_regno_call_part_clobbered): New function.
4487 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4488 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4489 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
4491 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4492 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4493 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
4495 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4496 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4497 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
4499 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4500 Alan Hayward <alan.hayward@arm.com>
4501 David Sherwood <david.sherwood@arm.com>
4503 * rtl.h (subreg_memory_offset): Declare.
4504 * emit-rtl.c (subreg_memory_offset): New function.
4505 * expmed.c (store_bit_field_1): Use it.
4506 * expr.c (undefined_operand_subword_p): Likewise.
4507 * simplify-rtx.c (simplify_subreg): Likewise.
4509 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
4511 PR rtl-optimization/57448
4514 * optabs.c (expand_atomic_load): Place compiler memory barriers if
4515 using atomic_load pattern.
4516 (expand_atomic_store): Likewise.
4518 2017-09-04 Jakub Jelinek <jakub@redhat.com>
4521 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
4522 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
4523 handling. Use replace_call_with_value with NULL instead of
4524 gsi_replace, unlink_stmt_vdef and release_defs.
4526 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
4529 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
4531 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4534 * rtl.h (emit_library_call_value_1): Declare.
4535 (emit_library_call): Replace declaration with a series of overloads.
4536 Remove the parameter count argument.
4537 (emit_library_call_value): Likewise.
4538 * calls.c (emit_library_call_value_1): Make global. Replace varargs
4539 with an "rtx_mode_t *".
4540 (emit_library_call_value): Delete.
4541 (emit_library_call): Likewise.
4542 * asan.c (asan_emit_stack_protection): Update calls accordingly.
4543 (asan_emit_allocas_unpoison): Likewise.
4544 * builtins.c (expand_builtin_powi): Likewise.
4545 (expand_asan_emit_allocas_unpoison): Likewise.
4546 * cfgexpand.c (expand_main_function): Likewise.
4547 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
4548 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
4549 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
4550 * config/arm/arm.c (arm_trampoline_init): Likewise.
4551 (arm_call_tls_get_addr): Likewise.
4552 (arm_expand_divmod_libfunc): Likewise.
4553 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
4554 (smulsi3_highpart): Likewise.
4555 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
4556 (c6x_expand_compare): Likewise.
4557 (c6x_expand_movmem): Likewise.
4558 * config/frv/frv.c (frv_trampoline_init): Likewise.
4559 * config/i386/i386.c (ix86_trampoline_init): Likewise.
4560 (ix86_expand_divmod_libfunc): Likewise.
4561 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
4562 (ia64_expand_compare): Likewise.
4563 (ia64_profile_hook): Likewise.
4564 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
4565 (nonlocal_goto): Likewise.
4566 (restore_stack_nonlocal): Likewise.
4567 * config/m32r/m32r.c (block_move_call): Likewise.
4568 (m32r_trampoline_init): Likewise.
4569 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
4570 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
4571 (m68k_call_m68k_read_tp): Likewise.
4572 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
4573 (microblaze_expand_divide): Likewise.
4574 * config/mips/mips.h (mips_args): Likewise.
4575 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
4576 (MIPS_ICACHE_SYNC): Likewise.
4577 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
4578 (nios2_trampoline_init): Likewise.
4579 * config/pa/pa.c (hppa_tls_call): Likewise.
4580 (pa_trampoline_init): Likewise.
4581 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
4582 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
4583 (expand_strn_compare): Likewise.
4584 (rs6000_generate_compare): Likewise.
4585 (rs6000_expand_float128_convert): Likewise.
4586 (output_profile_hook): Likewise.
4587 (rs6000_trampoline_init): Likewise.
4588 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
4589 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
4590 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
4591 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
4592 (rs6000_generate_compare): Likewise.
4593 (rs6000_expand_float128_convert): Likewise.
4594 (output_profile_hook): Likewise.
4595 (rs6000_trampoline_init): Likewise.
4596 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
4597 * config/sh/sh.c (sh_trampoline_init): Likewise.
4598 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
4599 (sparc_emit_float_lib_cmp): Likewise.
4600 (sparc32_initialize_trampoline): Likewise.
4601 (sparc64_initialize_trampoline): Likewise.
4602 (sparc_profile_hook): Likewise.
4603 * config/spu/spu.c (ea_load_store): Likewise.
4604 * config/spu/spu.md (floatunssidf2): Likewise.
4605 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
4606 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
4607 * config/visium/visium.c (expand_block_move_4): Likewise.
4608 (expand_block_move_2): Likewise.
4609 (expand_block_move_1): Likewise.
4610 (expand_block_set_4): Likewise.
4611 (expand_block_set_2): Likewise.
4612 (expand_block_set_1): Likewise.
4613 (visium_trampoline_init): Likewise.
4614 (visium_profile_hook): Likewise.
4615 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
4616 (xtensa_setup_frame_addresses): Likewise.
4617 (xtensa_trampoline_init): Likewise.
4618 * except.c (sjlj_emit_function_enter): Likewise.
4619 (sjlj_emit_function_exit): Likewise.
4620 * explow.c (allocate_dynamic_stack_space): Likewise.
4621 (probe_stack_range): Likewise.
4622 * expr.c (convert_mode_scalar): Likewise.
4623 * optabs.c (expand_binop): Likewise.
4624 (expand_twoval_binop_libfunc): Likewise.
4625 (expand_unop): Likewise.
4626 (prepare_cmp_insn): Likewise.
4627 (prepare_float_lib_cmp): Likewise.
4628 (expand_float): Likewise.
4629 (expand_fix): Likewise.
4630 (expand_fixed_convert): Likewise.
4631 (maybe_emit_sync_lock_test_and_set): Likewise.
4632 (expand_atomic_compare_and_swap): Likewise.
4633 (expand_mem_thread_fence): Likewise.
4634 (expand_atomic_fetch_op): Likewise.
4636 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
4638 * doc/generic.texi (OpenACC): Adjust URL.
4639 * doc/invoke.texi (C Dialect Options): Ditto.
4641 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
4643 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
4644 predicate for operand 1. Add (m,<S>) constraint.
4645 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
4646 Prevent memory operand 1 with register operand 2.
4648 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
4650 PR rtl-optimization/82024
4651 * combine.c (try_combine): If the combination result is a PARALLEL,
4652 and we only need to retain the SET in there that would be placed
4653 at I2, check that we can place that at I3 instead, before doing so.
4655 2017-09-01 Jakub Jelinek <jakub@redhat.com>
4658 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
4660 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
4661 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
4664 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
4665 Jeff Law <law@redhat.com>
4667 * varasm.c (bss_initializer_p): Do not put constants into .bss
4668 (categorize_decl_for_section): Handle bss_initializer_p returning
4669 false when DECL_INITIAL is NULL.
4671 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4674 * config/s390/s390.c (s390_can_inline_p): New function.
4676 2017-09-01 Jeff Law <law@redhat.com>
4678 PR tree-optimization/82052
4679 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
4680 Always initialize the returned slot after a hash table miss
4681 when INSERT is true.
4683 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
4685 * config/s390/s390.md (mem_signal_fence): Remove.
4686 * doc/md.texi (mem_signal_fence): Remove.
4687 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
4689 * target-insns.def (mem_signal_fence): Remove.
4691 2017-09-01 Jakub Jelinek <jakub@redhat.com>
4694 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
4697 * asan.c (create_odr_indicator): Strip name encoding from assembler
4698 name before appending it after __odr_asan_.
4700 2017-09-01 Martin Liska <mliska@suse.cz>
4702 PR tree-optimization/82059
4703 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
4704 frequency only when an edge is redirected.
4706 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4708 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
4709 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
4710 (arc_conditional_register_usage): Remove ARC600 lp_count
4712 (arc_file_start): Emit Tag_ARC_CPU_variation.
4713 (arc_can_use_doloop_p): New conditions to use ZOLs.
4714 (hwloop_fail): New function.
4715 (hwloop_optimize): Likewise.
4716 (hwloop_pattern_reg): Likewise.
4717 (arc_doloop_hooks): New struct, to be used with reorg_loops.
4718 (arc_reorg_loops): New function, calls reorg_loops.
4719 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
4720 (arc600_corereg_hazard): Remove ZOL checking, case handled by
4722 (arc_loop_hazard): Remove function, functionality moved into
4724 (arc_hazard): Remove arc_loop_hazard call.
4725 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
4726 into hwloop_optimize.
4727 (arc_label_align): Remove ZOL handling.
4728 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
4729 * config/arc/arc.md (doloop_begin): Remove pattern.
4730 (doloop_begin_i): Likewise.
4731 (doloop_end_i): Likewise.
4732 (doloop_fallback): Likewise.
4733 (doloop_fallback_m): Likewise.
4734 (doloop_end): Reimplement expand.
4735 (arc_lp): New pattern for LP instruction.
4736 (loop_end): New pattern.
4737 (loop_fail): Likewise.
4738 (decrement_and_branch_until_zero): Likewise.
4739 * config/arc/arc.opt (mlpc-width): New option.
4740 * doc/invoke.texi (mlpc-width): Document option.
4742 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4744 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
4745 (arc_ccfsm_advance): Fix checking for delay slots.
4746 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
4748 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4750 * config/arc/arc.md (movqi_insn): Add stores to save constant long
4752 (movhi_insn): Update store instruction constraint which are saving
4753 6-bit short immediates.
4754 (movsi_insn): Consider also short scaled load operations.
4755 (zero_extendhisi2_i): Use Usd constraint instead of T.
4756 (extendhisi2_i): Add q constraint.
4757 (arc_clzsi2): Add type and length attributes.
4758 (arc_ctzsi2): Likewise.
4759 * config/arc/constraints.md (Usc): Update constraint, the
4760 assembler can parse two relocations for a single instruction.
4762 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4764 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
4765 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
4767 2017-08-31 Olivier Hainque <hainque@adacore.com>
4769 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
4770 match as powerpc-wrs-vxworks*.
4772 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
4774 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
4775 register constraint for by-element operand.
4776 (aarch64_mls_elt_merge<mode>): Likewise.
4778 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4780 * config/arc/arc.c (arc_can_follow_jump): Check for short
4783 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4785 * config.gcc: Use g.opt for arc.
4786 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
4787 functionality moved to ...
4788 (legitimate_scaled_address_p): New function, ...here.
4789 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
4790 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
4791 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
4793 (arc_override_options): Handle G option.
4794 (arc_output_pic_addr_const): Correct function definition.
4795 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
4796 (arc_decl_anon_ns_mem_p): Delete.
4797 (arc_in_small_data_p): Overhaul this function to take into
4798 consideration the value given via G option.
4799 (arc_rewrite_small_data_1): Renamed and corrected old
4800 arc_rewrite_small_data function.
4801 (arc_rewrite_small_data): New function.
4802 (small_data_pattern): Don't use pic_offset_table_rtx.
4803 * config/arc/arc.h (CC1_SPEC): Recognize G option.
4804 * config/arc/simdext.md (movmisalignv2hi): Use
4805 prepare_move_operands function.
4807 (movmisalign*): Likewise.
4808 * doc/invoke.texi (ARC options): Document -G option.
4810 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4812 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
4814 * config/arc/arc.c (arc_print_operand): Output scalled address for
4815 sdata whenever is possible.
4816 (arc_in_small_data_p): Allow sdata for 64bit datum when double
4817 load/stores are available.
4818 (compact_sda_memory_operand): Check for the alignment required by
4819 code density instructions.
4820 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
4822 * config/arc/constraints.md (Usd): Update constraint.
4823 (Us0): New constraint.
4824 (Usc): Update constraint.
4826 2017-08-31 Richard Biener <rguenther@suse.de>
4829 * dwarf2out.c (dwarf2out_early_global_decl): Process each
4832 2017-08-31 Tamar Christina <tamar.christina@arm.com>
4834 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
4835 Resize type_signature.
4837 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4838 Alan Hayward <alan.hayward@arm.com>
4839 David Sherwood <david.sherwood@arm.com>
4841 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
4842 subregs whose inner modes can be stored in GPRs.
4843 (aarch64_classify_index): Likewise.
4845 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4846 Alan Hayward <alan.hayward@arm.com>
4847 David Sherwood <david.sherwood@arm.com>
4849 * config/aarch64/iterators.md (V_cmp_result): Rename to...
4850 (V_INT_EQUIV): ...this.
4851 (v_cmp_result): Rename to...
4852 (v_int_equiv): ...this.
4853 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
4854 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
4855 (copysign<mode>3): Likewise.
4856 (aarch64_simd_bsl<mode>_internal): Likewise.
4857 (aarch64_simd_bsl<mode>): Likewise.
4858 (vec_cmp<mode><mode>): Likewise.
4859 (vcond<mode><mode>): Likewise.
4860 (vcond<v_cmp_mixed><mode>): Likewise.
4861 (vcondu<mode><v_cmp_mixed>): Likewise.
4862 (aarch64_cm<optab><mode>): Likewise.
4863 (aarch64_cmtst<mode>): Likewise.
4864 (aarch64_fac<optab><mode>): Likewise.
4865 (vec_perm_const<mode>): Likewise.
4866 (vcond_mask_<mode><v_cmp_result>): Rename to...
4867 (vcond_mask_<mode><v_int_equiv>): ...this.
4868 (vec_cmp<mode><v_cmp_result>): Rename to...
4869 (vec_cmp<mode><v_int_equiv>): ...this.
4871 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4872 Alan Hayward <alan.hayward@arm.com>
4873 David Sherwood <david.sherwood@arm.com>
4875 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
4877 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
4878 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
4879 (UNSPEC_LD4_DREG): New unspecs.
4880 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
4881 (aarch64_ld2<mode>_dreg_be): Replace with...
4882 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
4884 (aarch64_ld3<mode>_dreg_le)
4885 (aarch64_ld3<mode>_dreg_be): Replace with...
4886 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
4888 (aarch64_ld4<mode>_dreg_le)
4889 (aarch64_ld4<mode>_dreg_be): Replace with...
4890 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
4893 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4895 PR tree-optimization/81987
4896 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
4897 insert an initializer in a location not dominated by the stride
4900 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
4902 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
4903 on the entire header of the finally block in the fallthru case.
4905 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
4907 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
4909 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
4911 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
4912 rs6000_emit_move_from_cr and call renamed function.
4913 (rs6000_emit_prologue): Call renamed functions.
4914 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
4915 movesi_from_cr, remove volatile CRs.
4917 2017-08-30 Jon Beniston <jon@beniston.com>
4918 Richard Biener <rguenther@suse.de>
4920 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
4921 of VECTOR_MODE_P check.
4922 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
4923 element vector types.
4925 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4927 * df.h (df_read_modify_subreg_p): Remove in favor of...
4928 * rtl.h (read_modify_subreg_p): ...this new function. Take a
4929 const_rtx instead of an rtx.
4930 * cprop.c (local_cprop_find_used_regs): Update accordingly.
4931 * df-problems.c (df_word_lr_mark_ref): Likewise.
4932 * ira-lives.c (mark_pseudo_reg_live): Likewise.
4933 (mark_pseudo_reg_dead): Likewise.
4934 (mark_ref_dead): Likewise.
4935 * reginfo.c (init_subregs_of_mode): Likewise.
4936 * sched-deps.c (sched_analyze_1): Likewise.
4937 * df-scan.c (df_def_record_1): Likewise.
4938 (df_uses_record): Likewise.
4939 (df_read_modify_subreg_p): Remove in favor of...
4940 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
4941 const_rtx instead of an rtx.
4943 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4944 Alan Hayward <alan.hayward@arm.com>
4945 David Sherwood <david.sherwood@arm.com>
4947 * rtl.h (partial_subreg_p): New function.
4948 * caller-save.c (save_call_clobbered_regs): Use it.
4949 * calls.c (expand_call): Likewise.
4950 * combine.c (combinable_i3pat): Likewise.
4951 (simplify_set): Likewise.
4952 (make_extraction): Likewise.
4953 (make_compound_operation_int): Likewise.
4954 (gen_lowpart_or_truncate): Likewise.
4955 (force_to_mode): Likewise.
4956 (make_field_assignment): Likewise.
4957 (reg_truncated_to_mode): Likewise.
4958 (record_truncated_value): Likewise.
4959 (move_deaths): Likewise.
4960 * cse.c (record_jump_cond): Likewise.
4961 (cse_insn): Likewise.
4962 * cselib.c (cselib_lookup_1): Likewise.
4963 * expmed.c (extract_bit_field_using_extv): Likewise.
4964 * function.c (assign_parm_setup_reg): Likewise.
4965 * ifcvt.c (noce_convert_multiple_sets): Likewise.
4966 * ira-build.c (create_insn_allocnos): Likewise.
4967 * lra-coalesce.c (merge_pseudos): Likewise.
4968 * lra-constraints.c (match_reload): Likewise.
4969 (simplify_operand_subreg): Likewise.
4970 (curr_insn_transform): Likewise.
4971 * lra-lives.c (process_bb_lives): Likewise.
4972 * lra.c (new_insn_reg): Likewise.
4973 (lra_substitute_pseudo): Likewise.
4974 * regcprop.c (mode_change_ok): Likewise.
4975 (maybe_mode_change): Likewise.
4976 (copyprop_hardreg_forward_1): Likewise.
4977 * reload.c (push_reload): Likewise.
4978 (find_reloads): Likewise.
4979 (find_reloads_subreg_address): Likewise.
4980 * reload1.c (alter_reg): Likewise.
4981 (eliminate_regs_1): Likewise.
4982 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4984 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
4986 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
4987 back to if statements, including unpack.
4989 2017-08-30 Martin Liska <mliska@suse.cz>
4992 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
4994 (func_checker::compare_asm_inputs_outputs): ... this function.
4995 (func_checker::compare_gimple_asm): Use the function to compare
4996 also ASM constrains.
4997 * ipa-icf-gimple.h: Rename the function.
4999 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5000 Alan Hayward <alan.hayward@arm.com>
5001 David Sherwood <david.sherwood@arm.com>
5003 * coretypes.h (complex_mode): New type.
5004 * gdbhooks.py (build_pretty_printer): Handle it.
5005 * machmode.h (complex_mode): New class.
5006 (complex_mode::includes_p): New function.
5007 (is_complex_int_mode): Likewise.
5008 (is_complex_float_mode): Likewise.
5009 * genmodes.c (get_mode_class): Handle complex mode classes.
5010 * function.c (expand_function_end): Use is_complex_int_mode.
5012 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5013 Alan Hayward <alan.hayward@arm.com>
5014 David Sherwood <david.sherwood@arm.com>
5016 * coretypes.h (scalar_mode_pod): New typedef.
5017 * gdbhooks.py (build_pretty_printer): Handle it.
5018 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
5019 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
5020 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
5021 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
5024 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5025 Alan Hayward <alan.hayward@arm.com>
5026 David Sherwood <david.sherwood@arm.com>
5028 * machmode.h (mode_for_vector): Take a scalar_mode instead
5030 * stor-layout.c (mode_for_vector): Likewise.
5031 * explow.c (promote_mode): Use as_a <scalar_mode>.
5032 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
5034 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5035 Alan Hayward <alan.hayward@arm.com>
5036 David Sherwood <david.sherwood@arm.com>
5038 * target.def (preferred_simd_mode): Take a scalar_mode
5039 instead of a machine_mode.
5040 * targhooks.h (default_preferred_simd_mode): Likewise.
5041 * targhooks.c (default_preferred_simd_mode): Likewise.
5042 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
5043 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
5044 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
5045 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
5046 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
5047 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
5048 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
5049 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
5051 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
5052 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
5053 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
5054 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
5055 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
5056 * doc/tm.texi: Regenerate.
5057 * optabs-query.c (can_vec_mask_load_store_p): Return false for
5060 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5061 Alan Hayward <alan.hayward@arm.com>
5062 David Sherwood <david.sherwood@arm.com>
5064 * target.def (scalar_mode_supported_p): Take a scalar_mode
5065 instead of a machine_mode.
5066 * targhooks.h (default_scalar_mode_supported_p): Likewise.
5067 * targhooks.c (default_scalar_mode_supported_p): Likewise.
5068 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
5069 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
5070 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
5071 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
5072 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
5073 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
5074 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
5075 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
5076 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
5077 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
5078 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
5079 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
5081 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
5082 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
5083 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
5084 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
5085 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
5087 * doc/tm.texi: Regenerate.
5089 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5090 Alan Hayward <alan.hayward@arm.com>
5091 David Sherwood <david.sherwood@arm.com>
5093 * coretypes.h (opt_scalar_mode): New typedef.
5094 * gdbhooks.py (build_pretty_printers): Handle it.
5095 * machmode.h (mode_iterator::get_2xwider): Add overload for
5097 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
5099 * expr.c (convert_mode_scalar): Likewise.
5100 * omp-low.c (omp_clause_aligned_alignment): Likewise.
5101 * optabs.c (expand_float): Likewise.
5102 (expand_fix): Likewise.
5103 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5105 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5106 Alan Hayward <alan.hayward@arm.com>
5107 David Sherwood <david.sherwood@arm.com>
5109 * optabs.c (expand_float): Explicitly check for scalars before
5110 using a branching expansion.
5111 (expand_fix): Likewise.
5113 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5114 Alan Hayward <alan.hayward@arm.com>
5115 David Sherwood <david.sherwood@arm.com>
5117 * expr.c (convert_mode): Split scalar handling out into...
5118 (convert_mode_scalar): ...this new function. Treat the modes
5121 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5122 Alan Hayward <alan.hayward@arm.com>
5123 David Sherwood <david.sherwood@arm.com>
5125 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
5127 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
5129 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5130 Alan Hayward <alan.hayward@arm.com>
5131 David Sherwood <david.sherwood@arm.com>
5133 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
5134 rather than a machine_mode.
5135 (fixed_from_string): Likewise.
5136 (fixed_convert): Likewise.
5137 (fixed_convert_from_int): Likewise.
5138 (fixed_convert_from_real): Likewise.
5139 (real_convert_from_fixed): Likewise.
5140 * fixed-value.c (fixed_from_double_int): Likewise.
5141 (fixed_from_string): Likewise.
5142 (fixed_convert): Likewise.
5143 (fixed_convert_from_int): Likewise.
5144 (fixed_convert_from_real): Likewise.
5145 (real_convert_from_fixed): Likewise.
5146 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
5148 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5149 Alan Hayward <alan.hayward@arm.com>
5150 David Sherwood <david.sherwood@arm.com>
5152 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
5153 of separate mode class checks. Do not allow vector modes here.
5154 (immed_wide_int_const): Use as_a <scalar_mode>.
5155 * explow.c (trunc_int_for_mode): Likewise.
5156 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
5157 (wi::shwi): Likewise.
5158 (wi::min_value): Likewise.
5159 (wi::max_value): Likewise.
5160 * dwarf2out.c (loc_descriptor): Likewise.
5161 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
5164 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5165 Alan Hayward <alan.hayward@arm.com>
5166 David Sherwood <david.sherwood@arm.com>
5168 * tree.h (SCALAR_TYPE_MODE): New macro.
5169 * expr.c (expand_expr_addr_expr_1): Use it.
5170 (expand_expr_real_2): Likewise.
5171 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
5172 (fold_convert_const_fixed_from_int): Likewise.
5173 (fold_convert_const_fixed_from_real): Likewise.
5174 (native_encode_fixed): Likewise
5175 (native_encode_complex): Likewise
5176 (native_encode_vector): Likewise.
5177 (native_interpret_fixed): Likewise.
5178 (native_interpret_real): Likewise.
5179 (native_interpret_complex): Likewise.
5180 (native_interpret_vector): Likewise.
5181 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
5182 (simd_clone_adjust_argument_types): Likewise.
5183 (simd_clone_init_simd_arrays): Likewise.
5184 (simd_clone_adjust): Likewise.
5185 * stor-layout.c (layout_type): Likewise.
5186 * tree.c (build_minus_one_cst): Likewise.
5187 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
5188 * tree-inline.c (estimate_move_cost): Likewise.
5189 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
5190 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
5191 (vectorizable_reduction): Likewise.
5192 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
5193 (vect_recog_mixed_size_cond_pattern): Likewise.
5194 (check_bool_pattern): Likewise.
5195 (adjust_bool_pattern): Likewise.
5196 (search_type_for_mask_1): Likewise.
5197 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
5198 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5199 (vectorizable_load): Likewise.
5200 (vectorizable_store): Likewise.
5201 * ubsan.c (ubsan_encode_value): Likewise.
5202 * varasm.c (output_constant): Likewise.
5204 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5205 Alan Hayward <alan.hayward@arm.com>
5206 David Sherwood <david.sherwood@arm.com>
5208 * coretypes.h (scalar_mode): New class.
5209 * machmode.h (scalar_mode): Likewise.
5210 (scalar_mode::includes_p): New function.
5211 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
5212 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
5213 * genmodes.c (get_mode_class): Handle remaining scalar modes.
5214 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
5215 * expmed.c (store_bit_field_1): Likewise.
5216 (extract_bit_field_1): Likewise.
5217 * expr.c (write_complex_part): Likewise.
5218 (read_complex_part): Likewise.
5219 (emit_move_complex_push): Likewise.
5220 (expand_expr_real_2): Likewise.
5221 * function.c (assign_parm_setup_reg): Likewise.
5222 (assign_parms_unsplit_complex): Likewise.
5223 * optabs.c (expand_binop): Likewise.
5224 * rtlanal.c (subreg_get_info): Likewise.
5225 * simplify-rtx.c (simplify_immed_subreg): Likewise.
5226 * varasm.c (output_constant_pool_2): Likewise.
5228 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5229 Alan Hayward <alan.hayward@arm.com>
5230 David Sherwood <david.sherwood@arm.com>
5232 * expmed.c (extract_high_half): Use scalar_int_mode and remove
5234 (expmed_mult_highpart_optab): Likewise.
5235 (expmed_mult_highpart): Likewise.
5237 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5238 Alan Hayward <alan.hayward@arm.com>
5239 David Sherwood <david.sherwood@arm.com>
5241 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
5242 instead of a machine_mode.
5243 (builtin_memset_read_str): Likewise.
5244 * builtins.c (c_readstr): Likewise.
5245 (builtin_memcpy_read_str): Likewise.
5246 (builtin_strncpy_read_str): Likewise.
5247 (builtin_memset_read_str): Likewise.
5248 (builtin_memset_gen_str): Likewise.
5249 (expand_builtin_signbit): Use scalar_int_mode for local variables.
5250 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
5251 instead of a machine_mode.
5252 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
5254 (make_extraction): Likewise.
5255 (try_widen_shift_mode): Take and return scalar_int_modes instead
5257 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
5258 a scalar_int_mode instead of a machine_mode.
5259 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
5260 (avr_addr_space_pointer_mode): Likewise.
5261 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
5262 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
5263 (msp430_unwind_word_mode): Likewise.
5264 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
5265 (spu_addr_space_pointer_mode): Likewise.
5266 (spu_addr_space_address_mode): Likewise.
5267 (spu_libgcc_cmp_return_mode): Likewise.
5268 (spu_libgcc_shift_count_mode): Likewise.
5269 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
5270 (rl78_addr_space_pointer_mode): Likewise.
5271 (fl78_unwind_word_mode): Likewise.
5272 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
5274 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
5275 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
5276 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
5277 (mips_valid_pointer_mode): Likewise.
5278 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
5279 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
5280 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
5282 (ft32_addr_space_address_mode): Likewise.
5283 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
5284 scalar_int_mode instead of a machine_mode.
5285 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
5287 (m32c_addr_space_address_mode): Likewise.
5288 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
5289 (rs6000_eh_return_filter_mode): Likewise.
5290 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
5291 (rs6000_eh_return_filter_mode): Likewise.
5292 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
5293 (s390_libgcc_shift_count_mode): Likewise.
5294 (s390_unwind_word_mode): Likewise.
5295 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
5297 * target.def (mode_rep_extended): Likewise.
5298 (valid_pointer_mode): Likewise.
5299 (addr_space.valid_pointer_mode): Likewise.
5300 (eh_return_filter_mode): Return a scalar_int_mode rather than
5302 (libgcc_cmp_return_mode): Likewise.
5303 (libgcc_shift_count_mode): Likewise.
5304 (unwind_word_mode): Likewise.
5305 (addr_space.pointer_mode): Likewise.
5306 (addr_space.address_mode): Likewise.
5307 * doc/tm.texi: Regenerate.
5308 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
5310 (do_jump): Use scalar_int_mode for local variables.
5311 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
5312 rather than a machine_mode.
5313 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
5314 (scompare_loc_descriptor_wide): Likewise.
5315 (scompare_loc_descriptor_narrow): Likewise.
5316 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
5318 * except.c (sjlj_emit_dispatch_table): Likewise.
5319 (expand_builtin_eh_copy_values): Likewise.
5320 * explow.c (convert_memory_address_addr_space_1): Likewise.
5321 Take a scalar_int_mode rather than a machine_mode.
5322 (convert_memory_address_addr_space): Take a scalar_int_mode rather
5323 than a machine_mode.
5324 (memory_address_addr_space): Use scalar_int_mode for local variables.
5325 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
5326 rather than a machine_mode.
5327 * expmed.c (mask_rtx): Likewise.
5328 (init_expmed_one_conv): Likewise.
5329 (expand_mult_highpart_adjust): Likewise.
5330 (extract_high_half): Likewise.
5331 (expmed_mult_highpart_optab): Likewise.
5332 (expmed_mult_highpart): Likewise.
5333 (expand_smod_pow2): Likewise.
5334 (expand_sdiv_pow2): Likewise.
5335 (emit_store_flag_int): Likewise.
5336 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
5338 (extract_low_bits): Likewise.
5339 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
5341 * expr.c (pieces_addr::adjust): Likewise.
5342 (can_store_by_pieces): Likewise.
5343 (store_by_pieces): Likewise.
5344 (clear_by_pieces_1): Likewise.
5345 (expand_expr_addr_expr_1): Likewise.
5346 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
5347 (expand_expr_real_1): Likewise.
5348 (try_casesi): Likewise.
5349 * final.c (shorten_branches): Likewise.
5350 * fold-const.c (fold_convert_const_int_from_fixed): Change the
5351 type of "mode" to machine_mode.
5352 * internal-fn.c (expand_arith_overflow_result_store): Take a
5353 scalar_int_mode rather than a machine_mode.
5354 (expand_mul_overflow): Use scalar_int_mode for local variables.
5355 * loop-doloop.c (doloop_modify): Likewise.
5356 (doloop_optimize): Likewise.
5357 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
5358 than a machine_mode.
5359 (expand_doubleword_shift_condmove): Likewise.
5360 (expand_doubleword_shift): Likewise.
5361 (expand_doubleword_clz): Likewise.
5362 (expand_doubleword_popcount): Likewise.
5363 (expand_doubleword_parity): Likewise.
5364 (expand_absneg_bit): Use scalar_int_mode for local variables.
5365 (prepare_float_lib_cmp): Likewise.
5366 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
5367 rather than a machine_mode.
5368 (convert_memory_address_addr_space): Likewise.
5369 (get_mode_bounds): Likewise.
5370 (get_address_mode): Return a scalar_int_mode rather than a
5372 * rtlanal.c (get_address_mode): Likewise.
5373 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
5374 than a machine_mode.
5375 * targhooks.c (default_mode_rep_extended): Likewise.
5376 (default_valid_pointer_mode): Likewise.
5377 (default_addr_space_valid_pointer_mode): Likewise.
5378 (default_eh_return_filter_mode): Return a scalar_int_mode rather
5379 than a machine_mode.
5380 (default_libgcc_cmp_return_mode): Likewise.
5381 (default_libgcc_shift_count_mode): Likewise.
5382 (default_unwind_word_mode): Likewise.
5383 (default_addr_space_pointer_mode): Likewise.
5384 (default_addr_space_address_mode): Likewise.
5385 * targhooks.h (default_eh_return_filter_mode): Likewise.
5386 (default_libgcc_cmp_return_mode): Likewise.
5387 (default_libgcc_shift_count_mode): Likewise.
5388 (default_unwind_word_mode): Likewise.
5389 (default_addr_space_pointer_mode): Likewise.
5390 (default_addr_space_address_mode): Likewise.
5391 (default_mode_rep_extended): Take a scalar_int_mode rather than
5393 (default_valid_pointer_mode): Likewise.
5394 (default_addr_space_valid_pointer_mode): Likewise.
5395 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
5397 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
5398 rather than a machine_mode.
5399 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
5400 for local variables.
5401 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5402 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
5403 than a machine_mode.
5405 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5406 Alan Hayward <alan.hayward@arm.com>
5407 David Sherwood <david.sherwood@arm.com>
5409 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
5410 the mode argument to scalar_int_mode.
5411 (do_jump_by_parts_zero_rtx): Likewise.
5412 (do_jump_by_parts_equality_rtx): Likewise.
5413 (do_jump_by_parts_greater): Take a mode argument.
5414 (do_jump_by_parts_equality): Likewise.
5415 (do_jump_1): Update calls accordingly.
5417 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5418 Alan Hayward <alan.hayward@arm.com>
5419 David Sherwood <david.sherwood@arm.com>
5421 * is-a.h (safe_dyn_cast): New function.
5422 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
5423 (jump_table_for_label): Likewise.
5424 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
5425 instead of an rtx_insn *.
5426 (shorten_branches): Use dyn_cast instead of LABEL_P and
5427 JUMP_TABLE_DATA_P. Use jump_table_for_label and
5428 rtx_jump_table_data::get_data_mode.
5429 (final_scan_insn): Likewise.
5431 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5432 Alan Hayward <alan.hayward@arm.com>
5433 David Sherwood <david.sherwood@arm.com>
5435 * combine.c (try_combine): Use is_a <scalar_int_mode> when
5436 trying to combine a full-register integer set with a subreg
5439 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5440 Alan Hayward <alan.hayward@arm.com>
5441 David Sherwood <david.sherwood@arm.com>
5443 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
5444 that is always either address_mode or pointer_mode.
5446 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5447 Alan Hayward <alan.hayward@arm.com>
5448 David Sherwood <david.sherwood@arm.com>
5450 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
5451 when the two are known to be equal.
5453 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5454 Alan Hayward <alan.hayward@arm.com>
5455 David Sherwood <david.sherwood@arm.com>
5457 * simplify-rtx.c (simplify_const_unary_operation): Use
5458 is_a <scalar_int_mode> instead of checking for a nonzero
5459 precision. Forcibly convert op_mode to a scalar_int_mode
5460 in that case. More clearly differentiate the operand and
5461 result modes and use the former when deciding what the value
5462 of a count-bits operation should be. Use is_int_mode instead
5463 of checking for a MODE_INT. Remove redundant check for whether
5464 this mode has a zero precision.
5466 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5467 Alan Hayward <alan.hayward@arm.com>
5468 David Sherwood <david.sherwood@arm.com>
5470 * optabs.c (widen_leading): Change the type of the mode argument
5471 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
5472 (widen_bswap): Likewise.
5473 (expand_parity): Likewise.
5474 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
5475 (expand_ffs): Likewise.
5476 (epand_unop): Check for scalar integer modes before calling the
5479 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5480 Alan Hayward <alan.hayward@arm.com>
5481 David Sherwood <david.sherwood@arm.com>
5483 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
5485 (expand_expr_real_1): Update call accordingly.
5487 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5488 Alan Hayward <alan.hayward@arm.com>
5489 David Sherwood <david.sherwood@arm.com>
5491 * expmed.c (store_bit_field_using_insv): Add op0_mode and
5492 value_mode arguments. Use scalar_int_mode internally.
5493 (store_bit_field_1): Rename the new integer mode from imode
5494 to op0_mode and use it instead of GET_MODE (op0). Update calls
5495 to store_split_bit_field, store_bit_field_using_insv and
5496 store_fixed_bit_field.
5497 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
5498 Use scalar_int_mode internally. Use a bit count rather than a mode
5499 when calculating the largest bit size for get_best_mode.
5500 Update calls to store_split_bit_field and store_fixed_bit_field_1.
5501 (store_fixed_bit_field_1): Add mode and value_mode arguments.
5502 Remove assertion that OP0 has a scalar integer mode.
5503 (store_split_bit_field): Add op0_mode and value_mode arguments.
5504 Update calls to extract_fixed_bit_field.
5505 (extract_bit_field_using_extv): Add an op0_mode argument.
5506 Use scalar_int_mode internally.
5507 (extract_bit_field_1): Rename the new integer mode from imode to
5508 op0_mode and use it instead of GET_MODE (op0). Update calls to
5509 extract_split_bit_field, extract_bit_field_using_extv and
5510 extract_fixed_bit_field.
5511 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
5512 to extract_split_bit_field and extract_fixed_bit_field_1.
5513 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
5514 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
5516 (extract_split_bit_field): Add an op0_mode argument. Update call
5517 to extract_fixed_bit_field.
5519 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5520 Alan Hayward <alan.hayward@arm.com>
5521 David Sherwood <david.sherwood@arm.com>
5523 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
5524 * explow.c (hard_function_value): Likewise.
5525 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
5526 convert_to_mode call outside the loop.
5527 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
5528 for the mode iterator. Require the mode specified by max_pieces
5530 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
5532 (copy_blkmode_to_reg): Likewise.
5533 (set_storage_via_setmem): Likewise.
5534 * optabs.c (prepare_cmp_insn): Likewise.
5535 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
5536 * stor-layout.c (finish_bitfield_representative): Likewise.
5538 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5539 Alan Hayward <alan.hayward@arm.com>
5540 David Sherwood <david.sherwood@arm.com>
5542 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
5543 * expr.c (convert_move): Use them.
5544 (convert_modes): Likewise.
5545 (store_expr_with_bounds): Likewise.
5547 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5548 Alan Hayward <alan.hayward@arm.com>
5549 David Sherwood <david.sherwood@arm.com>
5551 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
5552 parameter for the mode of "x". Remove the "known_x", "known_mode"
5553 and "known_ret" arguments. Change the type of the mode argument
5555 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
5556 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
5557 (reg_num_sign_bit_copies_for_combine): Likewise.
5558 * rtlanal.c (nonzero_bits1): Likewise.
5559 (num_sign_bit_copies1): Likewise.
5560 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
5561 (reg_num_sign_bit_copies_general): Likewise.
5562 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
5563 (reg_nonzero_bits_general): Likewise.
5565 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5566 Alan Hayward <alan.hayward@arm.com>
5567 David Sherwood <david.sherwood@arm.com>
5569 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
5570 than in subroutines. Return 1 for non-integer modes.
5571 (cached_num_sign_bit_copies): Change the type of the mode parameter
5573 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
5574 classes. Handle CONST_INT_P first and then check whether X also
5575 has a scalar integer mode. Check the same thing for inner registers
5576 of a SUBREG and for values that are being extended or truncated.
5578 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5579 Alan Hayward <alan.hayward@arm.com>
5580 David Sherwood <david.sherwood@arm.com>
5582 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
5583 in subroutines. Return the mode mask for non-integer modes.
5584 (cached_nonzero_bits): Change the type of the mode parameter
5586 (nonzero_bits1): Likewise. Remove early exit for other mode
5587 classes. Handle CONST_INT_P first and then check whether X
5588 also has a scalar integer mode.
5590 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5591 Alan Hayward <alan.hayward@arm.com>
5592 David Sherwood <david.sherwood@arm.com>
5594 * expr.c (widest_int_mode_for_size): Make the comment match the code.
5595 Return a scalar_int_mode and assert that the size is greater than
5597 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
5598 (op_by_pieces_d::op_by_pieces_d): Likewise.
5599 (op_by_pieces_d::run): Likewise.
5600 (can_store_by_pieces): Likewise.
5602 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5603 Alan Hayward <alan.hayward@arm.com>
5604 David Sherwood <david.sherwood@arm.com>
5606 * combine.c (extract_left_shift): Add a mode argument and update
5608 (make_compound_operation_int): Change the type of the mode parameter
5609 to scalar_int_mode and update the call to extract_left_shift.
5611 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5612 Alan Hayward <alan.hayward@arm.com>
5613 David Sherwood <david.sherwood@arm.com>
5615 * combine.c (simplify_and_const_int): Change the type of the mode
5616 parameter to scalar_int_mode.
5617 (simplify_and_const_int_1): Likewise. Update recursive call.
5619 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5620 Alan Hayward <alan.hayward@arm.com>
5621 David Sherwood <david.sherwood@arm.com>
5623 * combine.c (simplify_compare_const): Check that the mode is a
5624 scalar_int_mode (rather than VOIDmode) before testing its
5626 (simplify_comparison): Move COMPARISON_P handling out of the
5627 loop and restrict the latter part of the loop to scalar_int_modes.
5628 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
5629 and when considering SUBREG_REGs. Use is_int_mode instead of
5630 checking GET_MODE_CLASS against MODE_INT.
5632 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5633 Alan Hayward <alan.hayward@arm.com>
5634 David Sherwood <david.sherwood@arm.com>
5636 * combine.c (try_widen_shift_mode): Move check for equal modes to...
5637 (simplify_shift_const_1): ...here. Use scalar_int_mode for
5638 shift_unit_mode and for modes involved in scalar shifts.
5640 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5641 Alan Hayward <alan.hayward@arm.com>
5642 David Sherwood <david.sherwood@arm.com>
5644 * combine.c (force_int_to_mode): New function, split out from...
5645 (force_to_mode): ...here. Keep xmode up-to-date and use it
5646 instead of GET_MODE (x).
5648 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5649 Alan Hayward <alan.hayward@arm.com>
5650 David Sherwood <david.sherwood@arm.com>
5652 * optabs-query.h (extraction_insn::struct_mode): Change type to
5653 opt_scalar_int_mode and update comment.
5654 (extraction_insn::field_mode): Change type to scalar_int_mode.
5655 (extraction_insn::pos_mode): Likewise.
5656 * combine.c (make_extraction): Update accordingly.
5657 * optabs-query.c (get_traditional_extraction_insn): Likewise.
5658 (get_optab_extraction_insn): Likewise.
5659 * recog.c (simplify_while_replacing): Likewise.
5660 * expmed.c (narrow_bit_field_mem): Change the type of the mode
5661 parameter to opt_scalar_int_mode.
5663 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5664 Alan Hayward <alan.hayward@arm.com>
5665 David Sherwood <david.sherwood@arm.com>
5667 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
5668 to a scalar_int_mode instead of a machine_mode.
5669 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
5670 (get_best_mode): Return a boolean and use a pointer argument to store
5671 the selected mode. Replace the limit mode parameter with a bit limit.
5672 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
5673 for the values returned by bit_field_mode_iterator::next_mode.
5674 (store_bit_field): Update call to get_best_mode.
5675 (store_fixed_bit_field): Likewise.
5676 (extract_fixed_bit_field): Likewise.
5677 * expr.c (optimize_bitfield_assignment_op): Likewise.
5678 * fold-const.c (optimize_bit_field_compare): Likewise.
5679 (fold_truth_andor_1): Likewise.
5680 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
5681 Update for new type of m_mode.
5682 (get_best_mode): As above.
5684 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5685 Alan Hayward <alan.hayward@arm.com>
5686 David Sherwood <david.sherwood@arm.com>
5688 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
5689 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
5690 (store_bit_field): Check is_a <scalar_int_mode> before calling
5691 strict_volatile_bitfield_p.
5692 (extract_bit_field): Likewise.
5694 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5695 Alan Hayward <alan.hayward@arm.com>
5696 David Sherwood <david.sherwood@arm.com>
5698 * target.def (cstore_mode): Return a scalar_int_mode.
5699 * doc/tm.texi: Regenerate.
5700 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
5701 * targhooks.h (default_cstore_mode): Likewise.
5702 * targhooks.c (default_cstore_mode): Likewise, using a forced
5704 * expmed.c (emit_cstore): Expect the target of the cstore to be
5707 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5708 Alan Hayward <alan.hayward@arm.com>
5709 David Sherwood <david.sherwood@arm.com>
5711 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
5713 (niter_desc): Likewise mode.
5714 (iv_analyze): Add a mode parameter.
5716 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
5717 and change its type to scalar_int_mode.
5718 * loop-iv.c: Update commentary at head of file.
5719 (iv_constant): Pass the mode paraeter before the rtx it describes
5720 and change its type to scalar_int_mode. Remove VOIDmode handling.
5721 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
5722 (iv_extend): Likewise.
5723 (shorten_into_mode): Likewise.
5724 (iv_add): Use scalar_int_mode.
5725 (iv_mult): Likewise.
5726 (iv_shift): Likewise.
5727 (canonicalize_iv_subregs): Likewise.
5728 (get_biv_step_1): Pass the outer_mode parameter before the rtx
5729 it describes and change its mode to scalar_int_mode. Also change
5730 the type of the returned inner_mode to scalar_int_mode.
5731 (get_biv_step): Likewise, turning outer_mode from a pointer
5732 into a direct parameter. Update call to get_biv_step_1.
5733 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
5734 iv_constant and get_biv_step.
5735 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
5736 and change its type to scalar_int_mode. Don't initialise iv->mode
5737 to VOIDmode and remove later checks for its still being VOIDmode.
5738 Update calls to iv_analyze_op and iv_analyze_expr. Check
5739 is_a <scalar_int_mode> when changing the mode under consideration.
5740 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
5741 Update call to iv_analyze_expr.
5742 (iv_analyze_op): Add a mode parameter. Reject subregs whose
5743 inner register is not also a scalar_int_mode. Update call to
5745 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
5746 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
5747 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
5748 separate mode class checks. Update calls to iv_analyze. Remove
5749 fix-up of VOIDmodes after iv_analyze_biv.
5750 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
5751 don't have a scalar_int_mode. Update call to biv_p.
5753 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5754 Alan Hayward <alan.hayward@arm.com>
5755 David Sherwood <david.sherwood@arm.com>
5757 * cfgexpand.c (convert_debug_memory_address): Use
5758 as_a <scalar_int_mode>.
5759 * combine.c (expand_compound_operation): Likewise.
5760 (make_extraction): Likewise.
5761 (change_zero_ext): Likewise.
5762 (simplify_comparison): Likewise.
5763 * cse.c (cse_insn): Likewise.
5764 * dwarf2out.c (minmax_loc_descriptor): Likewise.
5765 (mem_loc_descriptor): Likewise.
5766 (loc_descriptor): Likewise.
5767 * expmed.c (init_expmed_one_mode): Likewise.
5768 (synth_mult): Likewise.
5769 (emit_store_flag_1): Likewise.
5770 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
5771 of a comparison with size.
5772 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
5773 (reduce_to_bit_field_precision): Likewise.
5774 * function.c (expand_function_end): Likewise.
5775 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
5776 * loop-doloop.c (doloop_modify): Likewise.
5777 * optabs.c (expand_binop): Likewise.
5778 (expand_unop): Likewise.
5779 (expand_copysign_absneg): Likewise.
5780 (prepare_cmp_insn): Likewise.
5781 (maybe_legitimize_operand): Likewise.
5782 * recog.c (const_scalar_int_operand): Likewise.
5783 * rtlanal.c (get_address_mode): Likewise.
5784 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5785 (simplify_cond_clz_ctz): Likewise.
5786 * tree-nested.c (get_nl_goto_field): Likewise.
5787 * tree.c (build_vector_type_for_mode): Likewise.
5788 * var-tracking.c (use_narrower_mode): Likewise.
5790 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5791 Alan Hayward <alan.hayward@arm.com>
5792 David Sherwood <david.sherwood@arm.com>
5794 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
5795 * builtins.c (expand_builtin_signbit): Use it.
5796 * cfgexpand.c (expand_debug_expr): Likewise.
5797 * dojump.c (do_jump): Likewise.
5798 (do_compare_and_jump): Likewise.
5799 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
5800 * expmed.c (make_tree): Likewise.
5801 * expr.c (expand_expr_real_2): Likewise.
5802 (expand_expr_real_1): Likewise.
5803 (try_casesi): Likewise.
5804 * fold-const-call.c (fold_const_call_ss): Likewise.
5805 * fold-const.c (unextend): Likewise.
5806 (extract_muldiv_1): Likewise.
5807 (fold_single_bit_test): Likewise.
5808 (native_encode_int): Likewise.
5809 (native_encode_string): Likewise.
5810 (native_interpret_int): Likewise.
5811 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
5812 * internal-fn.c (expand_addsub_overflow): Likewise.
5813 (expand_neg_overflow): Likewise.
5814 (expand_mul_overflow): Likewise.
5815 (expand_arith_overflow): Likewise.
5816 * match.pd: Likewise.
5817 * stor-layout.c (layout_type): Likewise.
5818 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
5819 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
5820 * tree-ssanames.c (get_range_info): Likewise.
5821 * tree-switch-conversion.c (array_value_type) Likewise.
5822 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
5823 (vect_recog_divmod_pattern): Likewise.
5824 (vect_recog_mixed_size_cond_pattern): Likewise.
5825 * tree-vrp.c (extract_range_basic): Likewise.
5826 (simplify_float_conversion_using_ranges): Likewise.
5827 * tree.c (int_fits_type_p): Likewise.
5828 * ubsan.c (instrument_bool_enum_load): Likewise.
5829 * varasm.c (mergeable_string_section): Likewise.
5830 (narrowing_initializer_constant_valid_p): Likewise.
5831 (output_constant): Likewise.
5833 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5834 Alan Hayward <alan.hayward@arm.com>
5835 David Sherwood <david.sherwood@arm.com>
5837 * machmode.h (NARROWEST_INT_MODE): New macro.
5838 * expr.c (alignment_for_piecewise_move): Use it instead of
5839 GET_CLASS_NARROWEST_MODE (MODE_INT).
5840 (push_block): Likewise.
5841 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
5843 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5845 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5846 Alan Hayward <alan.hayward@arm.com>
5847 David Sherwood <david.sherwood@arm.com>
5849 * postreload.c (move2add_valid_value_p): Change the type of the
5850 mode parameter to scalar_int_mode.
5851 (move2add_use_add2_insn): Add a mode parameter and use it instead
5853 (move2add_use_add3_insn): Likewise.
5854 (reload_cse_move2add): Update accordingly.
5856 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5857 Alan Hayward <alan.hayward@arm.com>
5858 David Sherwood <david.sherwood@arm.com>
5860 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
5862 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
5863 * optabs.c (expand_unop): Likewise.
5865 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5866 Alan Hayward <alan.hayward@arm.com>
5867 David Sherwood <david.sherwood@arm.com>
5869 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
5870 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
5871 (popcount_loc_descriptor): Likewise.
5872 (bswap_loc_descriptor): Likewise.
5873 (rotate_loc_descriptor): Likewise.
5874 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
5875 calling the functions above.
5877 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5878 Alan Hayward <alan.hayward@arm.com>
5879 David Sherwood <david.sherwood@arm.com>
5881 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
5883 (try_combine): Likewise.
5884 (simplify_if_then_else): Likewise.
5885 * cse.c (cse_insn): Likewise.
5886 * dwarf2out.c (mem_loc_descriptor): Likewise.
5887 * emit-rtl.c (gen_lowpart_common): Likewise.
5888 * simplify-rtx.c (simplify_truncation): Likewise.
5889 (simplify_binary_operation_1): Likewise.
5890 (simplify_const_relational_operation): Likewise.
5891 (simplify_ternary_operation): Likewise.
5892 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
5894 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5895 Alan Hayward <alan.hayward@arm.com>
5896 David Sherwood <david.sherwood@arm.com>
5898 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
5899 * reload.c (push_reload): Likewise.
5900 (find_reloads): Likewise.
5902 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5903 Alan Hayward <alan.hayward@arm.com>
5904 David Sherwood <david.sherwood@arm.com>
5906 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
5907 (make_compound_operation_int): Likewise.
5908 (change_zero_ext): Likewise.
5909 * expr.c (convert_move): Likewise.
5910 (convert_modes): Likewise.
5911 * fwprop.c (forward_propagate_subreg): Likewise.
5912 * loop-iv.c (get_biv_step_1): Likewise.
5913 * optabs.c (widen_operand): Likewise.
5914 * postreload.c (move2add_valid_value_p): Likewise.
5915 * recog.c (simplify_while_replacing): Likewise.
5916 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5917 (simplify_binary_operation_1): Likewise. Remove redundant
5918 mode equality check.
5920 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5921 Alan Hayward <alan.hayward@arm.com>
5922 David Sherwood <david.sherwood@arm.com>
5924 * combine.c (combine_simplify_rtx): Add checks for
5925 is_a <scalar_int_mode>.
5926 (simplify_if_then_else): Likewise.
5927 (make_field_assignment): Likewise.
5928 (simplify_comparison): Likewise.
5929 * ifcvt.c (noce_try_bitop): Likewise.
5930 * loop-invariant.c (canonicalize_address_mult): Likewise.
5931 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5933 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5934 Alan Hayward <alan.hayward@arm.com>
5935 David Sherwood <david.sherwood@arm.com>
5937 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
5938 is_a <scalar_int_mode> instead of != BLKmode.
5940 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5941 Alan Hayward <alan.hayward@arm.com>
5942 David Sherwood <david.sherwood@arm.com>
5944 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
5945 instead of != VOIDmode.
5946 * combine.c (if_then_else_cond): Likewise.
5947 (change_zero_ext): Likewise.
5948 * dwarf2out.c (mem_loc_descriptor): Likewise.
5949 (loc_descriptor): Likewise.
5950 * rtlanal.c (canonicalize_condition): Likewise.
5951 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
5953 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5954 Alan Hayward <alan.hayward@arm.com>
5955 David Sherwood <david.sherwood@arm.com>
5957 * simplify-rtx.c (simplify_binary_operation_1): Use
5958 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
5960 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5961 Alan Hayward <alan.hayward@arm.com>
5962 David Sherwood <david.sherwood@arm.com>
5964 * wide-int.h (int_traits<unsigned char>) New class.
5965 (int_traits<unsigned short>) Likewise.
5966 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
5967 Use GET_MODE_UNIT_PRECISION and remove redundant test for
5969 * combine.c (set_nonzero_bits_and_sign_copies): Use
5970 is_a <scalar_int_mode>.
5971 (find_split_point): Likewise.
5972 (combine_simplify_rtx): Likewise.
5973 (simplify_logical): Likewise.
5974 (expand_compound_operation): Likewise.
5975 (expand_field_assignment): Likewise.
5976 (make_compound_operation): Likewise.
5977 (extended_count): Likewise.
5978 (change_zero_ext): Likewise.
5979 (simplify_comparison): Likewise.
5980 * dwarf2out.c (scompare_loc_descriptor): Likewise.
5981 (ucompare_loc_descriptor): Likewise.
5982 (minmax_loc_descriptor): Likewise.
5983 (mem_loc_descriptor): Likewise.
5984 (loc_descriptor): Likewise.
5985 * expmed.c (init_expmed_one_mode): Likewise.
5986 * lra-constraints.c (lra_constraint_offset): Likewise.
5987 * optabs.c (prepare_libcall_arg): Likewise.
5988 * postreload.c (move2add_note_store): Likewise.
5989 * reload.c (operands_match_p): Likewise.
5990 * rtl.h (load_extend_op): Likewise.
5991 * rtlhooks.c (gen_lowpart_general): Likewise.
5992 * simplify-rtx.c (simplify_truncation): Likewise.
5993 (simplify_unary_operation_1): Likewise.
5994 (simplify_binary_operation_1): Likewise.
5995 (simplify_const_binary_operation): Likewise.
5996 (simplify_const_relational_operation): Likewise.
5997 (simplify_subreg): Likewise.
5998 * stor-layout.c (bitwise_mode_for_mode): Likewise.
5999 * var-tracking.c (adjust_mems): Likewise.
6000 (prepare_call_arguments): Likewise.
6002 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6003 Alan Hayward <alan.hayward@arm.com>
6004 David Sherwood <david.sherwood@arm.com>
6006 * machmode.h (is_int_mode): New fuction.
6007 * combine.c (find_split_point): Use it.
6008 (combine_simplify_rtx): Likewise.
6009 (simplify_if_then_else): Likewise.
6010 (simplify_set): Likewise.
6011 (simplify_shift_const_1): Likewise.
6012 (simplify_comparison): Likewise.
6013 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
6014 * cse.c (notreg_cost): Likewise.
6015 (cse_insn): Likewise.
6016 * cselib.c (cselib_lookup_1): Likewise.
6017 * dojump.c (do_jump_1): Likewise.
6018 (do_compare_rtx_and_jump): Likewise.
6019 * dse.c (get_call_args): Likewise.
6020 * dwarf2out.c (rtl_for_decl_init): Likewise.
6021 (native_encode_initializer): Likewise.
6022 * expmed.c (emit_store_flag_1): Likewise.
6023 (emit_store_flag): Likewise.
6024 * expr.c (convert_modes): Likewise.
6025 (store_field): Likewise.
6026 (expand_expr_real_1): Likewise.
6027 * fold-const.c (fold_read_from_constant_string): Likewise.
6028 * gimple-ssa-sprintf.c (get_format_string): Likewise.
6029 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
6030 * optabs.c (expand_binop): Likewise.
6031 (expand_unop): Likewise.
6032 (expand_abs_nojump): Likewise.
6033 (expand_one_cmpl_abs_nojump): Likewise.
6034 * simplify-rtx.c (mode_signbit_p): Likewise.
6035 (val_signbit_p): Likewise.
6036 (val_signbit_known_set_p): Likewise.
6037 (val_signbit_known_clear_p): Likewise.
6038 (simplify_relational_operation_1): Likewise.
6039 * tree.c (vector_type_mode): Likewise.
6041 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6042 Alan Hayward <alan.hayward@arm.com>
6043 David Sherwood <david.sherwood@arm.com>
6045 * machmode.h (smallest_mode_for_size): Fix formatting.
6046 (smallest_int_mode_for_size): New function.
6047 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
6048 instead of smallest_mode_for_size.
6049 * combine.c (make_extraction): Likewise.
6050 * config/arc/arc.c (arc_expand_movmem): Likewise.
6051 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
6052 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
6053 * config/s390/s390.c (s390_expand_insv): Likewise.
6054 * config/sparc/sparc.c (assign_int_registers): Likewise.
6055 * config/spu/spu.c (spu_function_value): Likewise.
6056 (spu_function_arg): Likewise.
6057 * coverage.c (get_gcov_type): Likewise.
6058 (get_gcov_unsigned_t): Likewise.
6059 * dse.c (find_shift_sequence): Likewise.
6060 * expmed.c (store_bit_field_1): Likewise.
6061 * expr.c (convert_move): Likewise.
6062 (store_field): Likewise.
6063 * internal-fn.c (expand_arith_overflow): Likewise.
6064 * optabs-query.c (get_best_extraction_insn): Likewise.
6065 * optabs.c (expand_twoval_binop_libfunc): Likewise.
6066 * stor-layout.c (layout_type): Likewise.
6067 (initialize_sizetypes): Likewise.
6068 * targhooks.c (default_get_mask_mode): Likewise.
6069 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
6071 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6072 Alan Hayward <alan.hayward@arm.com>
6073 David Sherwood <david.sherwood@arm.com>
6075 * machmode.h (opt_mode::else_blk): New function.
6076 (int_mode_for_mode): Declare.
6077 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
6078 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
6080 * cfgexpand.c (expand_debug_expr): Likewise.
6081 * combine.c (gen_lowpart_or_truncate): Likewise.
6082 (gen_lowpart_for_combine): Likewise.
6083 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
6084 * config/avr/avr.c (avr_to_int_mode): Likewise.
6085 (avr_out_plus_1): Likewise.
6086 (avr_out_plus): Likewise.
6087 (avr_out_round): Likewise.
6088 * config/i386/i386.c (ix86_split_to_parts): Likewise.
6089 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
6090 (s390_expand_vcond): Likewise.
6091 * config/spu/spu.c (spu_split_immediate): Likewise.
6092 (spu_expand_mov): Likewise.
6093 * dse.c (get_stored_val): Likewise.
6094 * expmed.c (store_bit_field_1): Likewise.
6095 (convert_extracted_bit_field): Use int_mode_for_mode instead of
6097 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
6098 (extract_low_bits): Likewise.
6099 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
6100 handling rather than repeating the check.
6101 (emit_group_store): Likewise.
6102 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
6103 * optabs.c (expand_absneg_bit): Likewise.
6104 (expand_copysign_absneg): Likewise.
6105 (expand_copysign_bit): Likewise.
6106 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
6107 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
6108 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
6109 * var-tracking.c (prepare_call_arguments): Likewise.
6110 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
6111 int_mode_for_mode instead of mode_for_size.
6112 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
6114 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6115 Alan Hayward <alan.hayward@arm.com>
6116 David Sherwood <david.sherwood@arm.com>
6118 * machmode.h (int_mode_for_size): New function.
6119 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
6120 instead of mode_for_size.
6121 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
6123 * combine.c (expand_field_assignment): Use int_mode_for_size
6124 instead of mode_for_size.
6125 (make_extraction): Likewise.
6126 (simplify_shift_const_1): Likewise.
6127 (simplify_comparison): Likewise.
6128 * dojump.c (do_jump): Likewise.
6129 * dwarf2out.c (mem_loc_descriptor): Likewise.
6130 * emit-rtl.c (init_derived_machine_modes): Likewise.
6131 * expmed.c (flip_storage_order): Likewise.
6132 (convert_extracted_bit_field): Likewise.
6133 * expr.c (copy_blkmode_from_reg): Likewise.
6134 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
6135 * internal-fn.c (expand_mul_overflow): Likewise.
6136 * lower-subreg.c (simple_move): Likewise.
6137 * optabs-libfuncs.c (init_optabs): Likewise.
6138 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6139 * tree.c (vector_type_mode): Likewise.
6140 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
6141 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
6142 * tree-vect-generic.c (expand_vector_parallel): Likewise.
6143 * tree-vect-stmts.c (vectorizable_load): Likewise.
6144 (vectorizable_store): Likewise.
6146 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6147 Alan Hayward <alan.hayward@arm.com>
6148 David Sherwood <david.sherwood@arm.com>
6150 * coretypes.h (pod_mode): New type.
6151 (scalar_int_mode_pod): New typedef.
6152 * machmode.h (pod_mode): New class.
6153 (int_n_data_t::m): Change type to scalar_int_mode_pod.
6154 * genmodes.c (emit_mode_int_n): Update accordingly.
6155 * lower-subreg.h (target_lower_subreg): Change type to
6156 scalar_int_mode_pod.
6157 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
6158 scalar_int_mode_pod.
6160 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6161 Alan Hayward <alan.hayward@arm.com>
6162 David Sherwood <david.sherwood@arm.com>
6164 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
6165 machine_mode to scalar_int_mode.
6166 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
6167 (rs6000_option_override_internal): Remove cast to int.
6168 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
6169 machine_mode to scalar_int_mode.
6170 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
6171 (rs6000_option_override_internal): Remove cast to int.
6172 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
6173 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
6175 * config/s390/s390.c (s390_expand_builtin): Likewise.
6176 * coretypes.h (scalar_int_mode): New type.
6177 (opt_scalar_int_mode): New typedef.
6178 * machmode.h (scalar_int_mode): New class.
6179 (scalar_int_mode::includes_p): New function.
6180 (byte_mode): Change type to scalar_int_mode.
6181 (word_mode): Likewise.
6182 (ptr_mode): Likewise.
6183 * emit-rtl.c (byte_mode): Likewise.
6184 (word_mode): Likewise.
6185 (ptr_mode): Likewise.
6186 (init_derived_machine_modes): Update accordingly.
6187 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
6188 and MODE_PARTIAL_INT.
6189 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
6190 opt_scalar_int_mode.
6192 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6193 Alan Hayward <alan.hayward@arm.com>
6194 David Sherwood <david.sherwood@arm.com>
6196 * target.def (libgcc_floating_mode_supported_p): Take a
6198 * doc/tm.texi: Regenerate.
6199 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
6201 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
6202 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
6205 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6206 Alan Hayward <alan.hayward@arm.com>
6207 David Sherwood <david.sherwood@arm.com>
6209 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
6210 * doc/tm.texi: Regenerate.
6211 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
6212 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
6213 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
6214 * targhooks.h (default_floatn_mode): Likewise.
6215 * targhooks.c (default_floatn_mode): Likewise.
6216 * tree.c (build_common_tree_nodes): Update accordingly.
6218 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6219 Alan Hayward <alan.hayward@arm.com>
6220 David Sherwood <david.sherwood@arm.com>
6222 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
6223 (mode_iterator::iterate_p): Likewise.
6224 (mode_iterator::get_wider): Likewise.
6225 * expr.c (init_expr_target): Use opt_scalar_float_mode.
6227 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6228 Alan Hayward <alan.hayward@arm.com>
6229 David Sherwood <david.sherwood@arm.com>
6231 * coretypes.h (opt_scalar_float_mode): New typedef.
6232 * machmode.h (float_mode_for_size): New function.
6233 * emit-rtl.c (double_mode): Delete.
6234 (init_emit_once): Use float_mode_for_size.
6235 * stor-layout.c (layout_type): Likewise.
6236 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
6238 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6239 Alan Hayward <alan.hayward@arm.com>
6240 David Sherwood <david.sherwood@arm.com>
6242 * output.h (assemble_real): Take a scalar_float_mode.
6243 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
6244 * config/arm/arm.md (consttable_4): Likewise.
6245 (consttable_8): Likewise.
6246 (consttable_16): Likewise.
6247 * config/mips/mips.md (consttable_float): Likewise.
6248 * config/s390/s390.c (s390_output_pool_entry): Likewise.
6249 * varasm.c (assemble_real): Take a scalar_float_mode.
6250 (output_constant_pool_2): Update accordingly.
6251 (output_constant): Likewise.
6253 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6254 Alan Hayward <alan.hayward@arm.com>
6255 David Sherwood <david.sherwood@arm.com>
6257 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
6258 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
6259 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
6260 (native_encode_real): Likewise.
6261 (native_interpret_real): Likewise.
6262 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
6263 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6265 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6266 Alan Hayward <alan.hayward@arm.com>
6267 David Sherwood <david.sherwood@arm.com>
6269 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
6270 <scalar_float_mode>. Simplify.
6271 (gen_extend_conv_libfunc): Likewise.
6273 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6274 Alan Hayward <alan.hayward@arm.com>
6275 David Sherwood <david.sherwood@arm.com>
6277 * coretypes.h (scalar_float_mode): New type.
6278 * machmode.h (mode_traits::from_int): Use machine_mode if
6279 USE_ENUM_MODES is defined.
6280 (is_a): New function.
6282 (dyn_cast): Likewise.
6283 (scalar_float_mode): New class.
6284 (scalar_float_mode::includes_p): New function.
6285 (is_float_mode): Likewise.
6286 * gdbhooks.py (MachineModePrinter): New class.
6287 (build_pretty_printer): Use it for scalar_float_mode.
6288 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
6289 (format_helper::format_helper): Turn into a template.
6290 * genmodes.c (get_mode_class): New function.
6291 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
6292 or machine_mode if none.
6293 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
6294 as_a <scalar_float_mode>.
6295 * dwarf2out.c (mem_loc_descriptor): Likewise.
6296 (insert_float): Likewise.
6297 (add_const_value_attribute): Likewise.
6298 * simplify-rtx.c (simplify_immed_subreg): Likewise.
6299 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
6300 (expand_unop): Update accordingly.
6301 (expand_abs_nojump): Likewise.
6302 (expand_copysign_absneg): Take a scalar_float_mode.
6303 (expand_copysign_bit): Likewise.
6304 (expand_copysign): Update accordingly.
6306 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6307 Alan Hayward <alan.hayward@arm.com>
6308 David Sherwood <david.sherwood@arm.com>
6310 * coretypes.h (opt_mode): New class.
6311 * machmode.h (opt_mode): Likewise.
6312 (opt_mode::else_void): New function.
6313 (opt_mode::require): Likewise.
6314 (opt_mode::exists): Likewise.
6315 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
6316 (GET_MODE_2XWIDER_MODE): Likewise.
6317 (mode_iterator::get_wider): Update accordingly.
6318 (mode_iterator::get_2xwider): Likewise.
6319 (mode_iterator::get_known_wider): Likewise, turning into a template.
6320 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
6321 forcing a wider mode to exist.
6322 * config/cr16/cr16.h (LONG_REG_P): Likewise.
6323 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6324 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
6325 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
6326 * lower-subreg.c (init_lower_subreg): Likewise.
6327 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
6328 on the final iteration.
6329 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
6330 a wider mode exists before asking for a move pattern.
6331 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
6332 forcing a wider mode to exist.
6333 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
6334 returning false if no such mode exists.
6335 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
6336 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
6337 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
6338 Avoid checking for a MODE_INT if we already know the mode is not a
6340 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
6341 forcing a wider mode to exist.
6342 (expmed_mult_highpart_optab): Likewise.
6343 (expmed_mult_highpart): Likewise.
6344 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
6346 * lto-streamer-in.c (lto_input_mode_table): Likewise.
6347 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
6348 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
6349 * internal-fn.c (expand_mul_overflow): Update use of
6350 GET_MODE_2XWIDER_MODE.
6351 * omp-low.c (omp_clause_aligned_alignment): Likewise.
6352 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
6353 GET_MODE_WIDER_MODE.
6354 (convert_plusminus_to_widen): Likewise.
6355 * tree-switch-conversion.c (array_value_type): Likewise.
6356 * var-tracking.c (emit_note_insn_var_location): Likewise.
6357 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6358 Return false inside rather than outside the loop if no wider mode
6360 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
6361 and GET_MODE_2XWIDER_MODE
6362 (can_compare_p): Use else_void.
6363 * gdbhooks.py (OptMachineModePrinter): New class.
6364 (build_pretty_printer): Use it for opt_mode.
6366 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6367 Alan Hayward <alan.hayward@arm.com>
6368 David Sherwood <david.sherwood@arm.com>
6370 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
6371 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
6373 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6374 Alan Hayward <alan.hayward@arm.com>
6375 David Sherwood <david.sherwood@arm.com>
6377 * machmode.h (mode_traits): New structure.
6378 (get_narrowest_mode): New function.
6379 (mode_iterator::start): Likewise.
6380 (mode_iterator::iterate_p): Likewise.
6381 (mode_iterator::get_wider): Likewise.
6382 (mode_iterator::get_known_wider): Likewise.
6383 (mode_iterator::get_2xwider): Likewise.
6384 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
6385 (FOR_EACH_MODE): Likewise.
6386 (FOR_EACH_MODE_FROM): Likewise.
6387 (FOR_EACH_MODE_UNTIL): Likewise.
6388 (FOR_EACH_WIDER_MODE): Likewise.
6389 (FOR_EACH_2XWIDER_MODE): Likewise.
6390 * builtins.c (expand_builtin_strlen): Use new mode iterators.
6391 * combine.c (simplify_comparison): Likewise
6392 * config/i386/i386.c (type_natural_mode): Likewise.
6393 * cse.c (cse_insn): Likewise.
6394 * dse.c (find_shift_sequence): Likewise.
6395 * emit-rtl.c (init_derived_machine_modes): Likewise.
6396 (init_emit_once): Likewise.
6397 * explow.c (hard_function_value): Likewise.
6398 * expmed.c (extract_fixed_bit_field_1): Likewise.
6399 (extract_bit_field_1): Likewise.
6400 (expand_divmod): Likewise.
6401 (emit_store_flag_1): Likewise.
6402 * expr.c (init_expr_target): Likewise.
6403 (convert_move): Likewise.
6404 (alignment_for_piecewise_move): Likewise.
6405 (widest_int_mode_for_size): Likewise.
6406 (emit_block_move_via_movmem): Likewise.
6407 (copy_blkmode_to_reg): Likewise.
6408 (set_storage_via_setmem): Likewise.
6409 (compress_float_constant): Likewise.
6410 * omp-low.c (omp_clause_aligned_alignment): Likewise.
6411 * optabs-query.c (get_best_extraction_insn): Likewise.
6412 * optabs.c (expand_binop): Likewise.
6413 (expand_twoval_unop): Likewise.
6414 (expand_twoval_binop): Likewise.
6415 (widen_leading): Likewise.
6416 (widen_bswap): Likewise.
6417 (expand_parity): Likewise.
6418 (expand_unop): Likewise.
6419 (prepare_cmp_insn): Likewise.
6420 (prepare_float_lib_cmp): Likewise.
6421 (expand_float): Likewise.
6422 (expand_fix): Likewise.
6423 (expand_sfix_optab): Likewise.
6424 * postreload.c (move2add_use_add2_insn): Likewise.
6425 * reg-stack.c (reg_to_stack): Likewise.
6426 * reginfo.c (choose_hard_reg_mode): Likewise.
6427 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6428 * stor-layout.c (mode_for_size): Likewise.
6429 (smallest_mode_for_size): Likewise.
6430 (mode_for_vector): Likewise.
6431 (finish_bitfield_representative): Likewise.
6432 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
6433 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
6434 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6435 * var-tracking.c (prepare_call_arguments): Likewise.
6437 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6438 Alan Hayward <alan.hayward@arm.com>
6439 David Sherwood <david.sherwood@arm.com>
6441 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
6442 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
6443 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
6444 * machmode.h (mode_size): Move earlier in file.
6445 (mode_precision): Likewise.
6446 (mode_inner): Likewise.
6447 (mode_nunits): Likewise.
6448 (mode_unit_size): Likewise.
6449 (unit_unit_precision): Likewise.
6450 (mode_wider): Likewise.
6451 (mode_2xwider): Likewise.
6452 (machine_mode): New class.
6453 (mode_to_bytes): New function.
6454 (mode_to_bits): Likewise.
6455 (mode_to_precision): Likewise.
6456 (mode_to_inner): Likewise.
6457 (mode_to_unit_size): Likewise.
6458 (mode_to_unit_precision): Likewise.
6459 (mode_to_nunits): Likewise.
6460 (GET_MODE_SIZE): Use mode_to_bytes.
6461 (GET_MODE_BITSIZE): Use mode_to_bits.
6462 (GET_MODE_PRECISION): Use mode_to_precision.
6463 (GET_MODE_INNER): Use mode_to_inner.
6464 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
6465 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
6466 (GET_MODE_NUNITS): Use mode_to_nunits.
6467 * system.h (ALWAYS_INLINE): New macro.
6468 * config/powerpcspe/powerpcspe-c.c
6469 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
6470 int for arg1_mode and arg2_mode.
6472 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6473 Alan Hayward <alan.hayward@arm.com>
6474 David Sherwood <david.sherwood@arm.com>
6476 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
6477 Prefix mode names with E_ in case statements.
6478 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6479 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
6480 (aarch64_split_simd_move): Likewise.
6481 (aarch64_gen_storewb_pair): Likewise.
6482 (aarch64_gen_loadwb_pair): Likewise.
6483 (aarch64_gen_store_pair): Likewise.
6484 (aarch64_gen_load_pair): Likewise.
6485 (aarch64_get_condition_code_1): Likewise.
6486 (aarch64_constant_pool_reload_icode): Likewise.
6487 (get_rsqrte_type): Likewise.
6488 (get_rsqrts_type): Likewise.
6489 (get_recpe_type): Likewise.
6490 (get_recps_type): Likewise.
6491 (aarch64_gimplify_va_arg_expr): Likewise.
6492 (aarch64_simd_container_mode): Likewise.
6493 (aarch64_emit_load_exclusive): Likewise.
6494 (aarch64_emit_store_exclusive): Likewise.
6495 (aarch64_expand_compare_and_swap): Likewise.
6496 (aarch64_gen_atomic_cas): Likewise.
6497 (aarch64_emit_bic): Likewise.
6498 (aarch64_emit_atomic_swap): Likewise.
6499 (aarch64_emit_atomic_load_op): Likewise.
6500 (aarch64_evpc_trn): Likewise.
6501 (aarch64_evpc_uzp): Likewise.
6502 (aarch64_evpc_zip): Likewise.
6503 (aarch64_evpc_ext): Likewise.
6504 (aarch64_evpc_rev): Likewise.
6505 (aarch64_evpc_dup): Likewise.
6506 (aarch64_gen_ccmp_first): Likewise.
6507 (aarch64_gen_ccmp_next): Likewise.
6508 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
6509 (alpha_emit_xfloating_libcall): Likewise.
6510 (emit_insxl): Likewise.
6511 (alpha_arg_type): Likewise.
6512 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
6513 (arc_preferred_simd_mode): Likewise.
6514 (arc_secondary_reload): Likewise.
6515 (get_arc_condition_code): Likewise.
6516 (arc_print_operand): Likewise.
6517 (arc_legitimate_constant_p): Likewise.
6518 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6519 * config/arc/arc.md (casesi_load): Likewise.
6520 (casesi_compact_jump): Likewise.
6521 * config/arc/predicates.md (proper_comparison_operator): Likewise.
6522 (cc_use_register): Likewise.
6523 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6524 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
6525 (arm_init_iwmmxt_builtins): Likewise.
6526 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
6527 (neon_expand_vector_init): Likewise.
6528 (arm_attr_length_move_neon): Likewise.
6529 (maybe_get_arm_condition_code): Likewise.
6530 (arm_emit_vector_const): Likewise.
6531 (arm_preferred_simd_mode): Likewise.
6532 (arm_output_iwmmxt_tinsr): Likewise.
6533 (thumb1_output_casesi): Likewise.
6534 (thumb2_output_casesi): Likewise.
6535 (arm_emit_load_exclusive): Likewise.
6536 (arm_emit_store_exclusive): Likewise.
6537 (arm_expand_compare_and_swap): Likewise.
6538 (arm_evpc_neon_vuzp): Likewise.
6539 (arm_evpc_neon_vzip): Likewise.
6540 (arm_evpc_neon_vrev): Likewise.
6541 (arm_evpc_neon_vtrn): Likewise.
6542 (arm_evpc_neon_vext): Likewise.
6543 (arm_validize_comparison): Likewise.
6544 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
6545 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
6546 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
6547 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
6548 (c6x_preferred_simd_mode): Likewise.
6549 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
6550 (epiphany_rtx_costs): Likewise.
6551 * config/epiphany/predicates.md (proper_comparison_operator):
6553 * config/frv/frv.c (condexec_memory_operand): Likewise.
6554 (frv_emit_move): Likewise.
6555 (output_move_single): Likewise.
6556 (output_condmove_single): Likewise.
6557 (frv_hard_regno_mode_ok): Likewise.
6558 (frv_matching_accg_mode): Likewise.
6559 * config/h8300/h8300.c (split_adds_subs): Likewise.
6560 (h8300_rtx_costs): Likewise.
6561 (h8300_print_operand): Likewise.
6562 (compute_mov_length): Likewise.
6563 (output_logical_op): Likewise.
6564 (compute_logical_op_length): Likewise.
6565 (compute_logical_op_cc): Likewise.
6566 (h8300_shift_needs_scratch_p): Likewise.
6567 (output_a_shift): Likewise.
6568 (compute_a_shift_length): Likewise.
6569 (compute_a_shift_cc): Likewise.
6570 (expand_a_rotate): Likewise.
6571 (output_a_rotate): Likewise.
6572 * config/i386/i386.c (classify_argument): Likewise.
6573 (function_arg_advance_32): Likewise.
6574 (function_arg_32): Likewise.
6575 (function_arg_64): Likewise.
6576 (function_value_64): Likewise.
6577 (ix86_gimplify_va_arg): Likewise.
6578 (ix86_legitimate_constant_p): Likewise.
6579 (put_condition_code): Likewise.
6580 (split_double_mode): Likewise.
6581 (ix86_avx256_split_vector_move_misalign): Likewise.
6582 (ix86_expand_vector_logical_operator): Likewise.
6583 (ix86_split_idivmod): Likewise.
6584 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
6585 (ix86_build_const_vector): Likewise.
6586 (ix86_build_signbit_mask): Likewise.
6587 (ix86_match_ccmode): Likewise.
6588 (ix86_cc_modes_compatible): Likewise.
6589 (ix86_expand_branch): Likewise.
6590 (ix86_expand_sse_cmp): Likewise.
6591 (ix86_expand_sse_movcc): Likewise.
6592 (ix86_expand_int_sse_cmp): Likewise.
6593 (ix86_expand_vec_perm_vpermi2): Likewise.
6594 (ix86_expand_vec_perm): Likewise.
6595 (ix86_expand_sse_unpack): Likewise.
6596 (ix86_expand_int_addcc): Likewise.
6597 (ix86_split_to_parts): Likewise.
6598 (ix86_vectorize_builtin_gather): Likewise.
6599 (ix86_vectorize_builtin_scatter): Likewise.
6600 (avx_vpermilp_parallel): Likewise.
6601 (inline_memory_move_cost): Likewise.
6602 (ix86_tieable_integer_mode_p): Likewise.
6603 (x86_maybe_negate_const_int): Likewise.
6604 (ix86_expand_vector_init_duplicate): Likewise.
6605 (ix86_expand_vector_init_one_nonzero): Likewise.
6606 (ix86_expand_vector_init_one_var): Likewise.
6607 (ix86_expand_vector_init_concat): Likewise.
6608 (ix86_expand_vector_init_interleave): Likewise.
6609 (ix86_expand_vector_init_general): Likewise.
6610 (ix86_expand_vector_set): Likewise.
6611 (ix86_expand_vector_extract): Likewise.
6612 (emit_reduc_half): Likewise.
6613 (ix86_emit_i387_round): Likewise.
6614 (ix86_mangle_type): Likewise.
6615 (ix86_expand_round_sse4): Likewise.
6616 (expand_vec_perm_blend): Likewise.
6617 (canonicalize_vector_int_perm): Likewise.
6618 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
6619 (expand_vec_perm_1): Likewise.
6620 (expand_vec_perm_interleave3): Likewise.
6621 (expand_vec_perm_even_odd_pack): Likewise.
6622 (expand_vec_perm_even_odd_1): Likewise.
6623 (expand_vec_perm_broadcast_1): Likewise.
6624 (ix86_vectorize_vec_perm_const_ok): Likewise.
6625 (ix86_expand_vecop_qihi): Likewise.
6626 (ix86_expand_mul_widen_hilo): Likewise.
6627 (ix86_expand_sse2_abs): Likewise.
6628 (ix86_expand_pextr): Likewise.
6629 (ix86_expand_pinsr): Likewise.
6630 (ix86_preferred_simd_mode): Likewise.
6631 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
6632 * config/i386/sse.md (*andnot<mode>3): Likewise.
6633 (<mask_codefor><code><mode>3<mask_name>): Likewise.
6634 (*<code><mode>3): Likewise.
6635 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
6636 (ia64_expand_atomic_op): Likewise.
6637 (ia64_arg_type): Likewise.
6638 (ia64_mode_to_int): Likewise.
6639 (ia64_scalar_mode_supported_p): Likewise.
6640 (ia64_vector_mode_supported_p): Likewise.
6641 (expand_vec_perm_broadcast): Likewise.
6642 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
6643 (iq2000_function_arg_advance): Likewise.
6644 (iq2000_function_arg): Likewise.
6645 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
6646 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
6647 (m68k_libcall_value): Likewise.
6648 (m68k_function_value): Likewise.
6649 (sched_attr_op_type): Likewise.
6650 * config/mcore/mcore.c (mcore_output_move): Likewise.
6651 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
6653 (microblaze_function_arg): Likewise.
6654 * config/mips/mips.c (mips16_build_call_stub): Likewise.
6655 (mips_print_operand): Likewise.
6656 (mips_mode_ok_for_mov_fmt_p): Likewise.
6657 (mips_vector_mode_supported_p): Likewise.
6658 (mips_preferred_simd_mode): Likewise.
6659 (mips_expand_vpc_loongson_even_odd): Likewise.
6660 (mips_expand_vec_unpack): Likewise.
6661 (mips_expand_vi_broadcast): Likewise.
6662 (mips_expand_vector_init): Likewise.
6663 (mips_expand_vec_reduc): Likewise.
6664 (mips_expand_msa_cmp): Likewise.
6665 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
6666 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
6667 (cc_flags_for_mode): Likewise.
6668 * config/msp430/msp430.c (msp430_print_operand): Likewise.
6669 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
6670 (nds32_output_casesi_pc_relative): Likewise.
6671 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6672 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
6673 (nvptx_gen_unpack): Likewise.
6674 (nvptx_gen_pack): Likewise.
6675 (nvptx_gen_shuffle): Likewise.
6676 (nvptx_gen_wcast): Likewise.
6677 (nvptx_preferred_simd_mode): Likewise.
6678 * config/pa/pa.c (pa_secondary_reload): Likewise.
6679 * config/pa/predicates.md (base14_operand): Likewise.
6680 * config/powerpcspe/powerpcspe-c.c
6681 (altivec_resolve_overloaded_builtin): Likewise.
6682 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
6684 (rs6000_preferred_simd_mode): Likewise.
6685 (output_vec_const_move): Likewise.
6686 (rs6000_expand_vector_extract): Likewise.
6687 (rs6000_split_vec_extract_var): Likewise.
6688 (reg_offset_addressing_ok_p): Likewise.
6689 (rs6000_legitimate_offset_address_p): Likewise.
6690 (rs6000_legitimize_address): Likewise.
6691 (rs6000_emit_set_const): Likewise.
6692 (rs6000_const_vec): Likewise.
6693 (rs6000_emit_move): Likewise.
6694 (spe_build_register_parallel): Likewise.
6695 (rs6000_darwin64_record_arg_recurse): Likewise.
6696 (swap_selector_for_mode): Likewise.
6697 (spe_init_builtins): Likewise.
6698 (paired_init_builtins): Likewise.
6699 (altivec_init_builtins): Likewise.
6700 (do_load_for_compare): Likewise.
6701 (rs6000_generate_compare): Likewise.
6702 (rs6000_expand_float128_convert): Likewise.
6703 (emit_load_locked): Likewise.
6704 (emit_store_conditional): Likewise.
6705 (rs6000_output_function_epilogue): Likewise.
6706 (rs6000_handle_altivec_attribute): Likewise.
6707 (rs6000_function_value): Likewise.
6708 (emit_fusion_gpr_load): Likewise.
6709 (emit_fusion_p9_load): Likewise.
6710 (emit_fusion_p9_store): Likewise.
6711 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
6712 (fusion_gpr_mem_load): Likewise.
6713 (fusion_addis_mem_combo_load): Likewise.
6714 (fusion_addis_mem_combo_store): Likewise.
6715 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
6716 (fusion_gpr_mem_load): Likewise.
6717 (fusion_addis_mem_combo_load): Likewise.
6718 (fusion_addis_mem_combo_store): Likewise.
6719 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6721 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
6722 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
6723 (rs6000_preferred_simd_mode): Likewise.
6724 (output_vec_const_move): Likewise.
6725 (rs6000_expand_vector_extract): Likewise.
6726 (rs6000_split_vec_extract_var): Likewise.
6727 (reg_offset_addressing_ok_p): Likewise.
6728 (rs6000_legitimate_offset_address_p): Likewise.
6729 (rs6000_legitimize_address): Likewise.
6730 (rs6000_emit_set_const): Likewise.
6731 (rs6000_const_vec): Likewise.
6732 (rs6000_emit_move): Likewise.
6733 (rs6000_darwin64_record_arg_recurse): Likewise.
6734 (swap_selector_for_mode): Likewise.
6735 (paired_init_builtins): Likewise.
6736 (altivec_init_builtins): Likewise.
6737 (rs6000_expand_float128_convert): Likewise.
6738 (emit_load_locked): Likewise.
6739 (emit_store_conditional): Likewise.
6740 (rs6000_output_function_epilogue): Likewise.
6741 (rs6000_handle_altivec_attribute): Likewise.
6742 (rs6000_function_value): Likewise.
6743 (emit_fusion_gpr_load): Likewise.
6744 (emit_fusion_p9_load): Likewise.
6745 (emit_fusion_p9_store): Likewise.
6746 * config/rx/rx.c (rx_gen_move_template): Likewise.
6747 (flags_from_mode): Likewise.
6748 * config/s390/predicates.md (s390_alc_comparison): Likewise.
6749 (s390_slb_comparison): Likewise.
6750 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
6751 (s390_vector_mode_supported_p): Likewise.
6752 (s390_cc_modes_compatible): Likewise.
6753 (s390_match_ccmode_set): Likewise.
6754 (s390_canonicalize_comparison): Likewise.
6755 (s390_emit_compare_and_swap): Likewise.
6756 (s390_branch_condition_mask): Likewise.
6757 (s390_rtx_costs): Likewise.
6758 (s390_secondary_reload): Likewise.
6759 (__SECONDARY_RELOAD_CASE): Likewise.
6760 (s390_expand_cs): Likewise.
6761 (s390_preferred_simd_mode): Likewise.
6762 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
6763 * config/sh/sh.c (sh_print_operand): Likewise.
6764 (dump_table): Likewise.
6765 (sh_secondary_reload): Likewise.
6766 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6767 * config/sh/sh.md (casesi_worker_1): Likewise.
6768 (casesi_worker_2): Likewise.
6769 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
6770 (fcc_comparison_operator): Likewise.
6771 * config/sparc/sparc.c (sparc_expand_move): Likewise.
6772 (emit_soft_tfmode_cvt): Likewise.
6773 (sparc_preferred_simd_mode): Likewise.
6774 (output_cbranch): Likewise.
6775 (sparc_print_operand): Likewise.
6776 (sparc_expand_vec_perm_bmask): Likewise.
6777 (vector_init_bshuffle): Likewise.
6778 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
6779 (spu_vector_mode_supported_p): Likewise.
6780 (spu_expand_insv): Likewise.
6781 (spu_emit_branch_or_set): Likewise.
6782 (spu_handle_vector_attribute): Likewise.
6783 (spu_builtin_splats): Likewise.
6784 (spu_builtin_extract): Likewise.
6785 (spu_builtin_promote): Likewise.
6786 (spu_expand_sign_extend): Likewise.
6787 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
6788 (tilegx_simd_int): Likewise.
6789 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
6790 (tilepro_simd_int): Likewise.
6791 * config/v850/v850.c (const_double_split): Likewise.
6792 (v850_print_operand): Likewise.
6793 (ep_memory_offset): Likewise.
6794 * config/vax/vax.c (vax_rtx_costs): Likewise.
6795 (vax_output_int_move): Likewise.
6796 (vax_output_int_add): Likewise.
6797 (vax_output_int_subtract): Likewise.
6798 * config/visium/predicates.md (visium_branch_operator): Likewise.
6799 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
6800 (visium_print_operand_address): Likewise.
6801 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6802 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
6803 (xtensa_expand_conditional_branch): Likewise.
6804 (xtensa_copy_incoming_a7): Likewise.
6805 (xtensa_output_literal): Likewise.
6806 * dfp.c (decimal_real_maxval): Likewise.
6807 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
6809 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6810 Alan Hayward <alan.hayward@arm.com>
6811 David Sherwood <david.sherwood@arm.com>
6813 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
6814 (mode_nunits_inline): Likewise.
6815 (mode_inner_inline): Likewise.
6816 (mode_unit_size_inline): Likewise.
6817 (mode_unit_precision_inline): Likewise.
6818 (emit_insn_modes_h): Likewise. Also emit a #define of the
6820 (emit_mode_wider): Add an E_ prefix to mode names.
6821 (emit_mode_complex): Likewise.
6822 (emit_mode_inner): Likewise.
6823 (emit_mode_adjustments): Likewise.
6824 (emit_mode_int_n): Likewise.
6825 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
6826 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
6827 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
6828 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
6829 (CRC32_BUILTIN, ENTRY): Likewise.
6830 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
6831 (aarch64_pop_regs): Likewise.
6832 (aarch64_process_components): Likewise.
6833 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
6834 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
6835 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
6836 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
6837 * config/arm/arm.c (arm_init_libfuncs): Likewise.
6838 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
6840 * config/i386/i386-builtin.def (pcmpestr): Likewise.
6841 (pcmpistr): Likewise.
6842 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
6843 * config/mmix/mmix.c (mmix_output_condition): Likewise.
6844 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
6846 * config/rl78/rl78.c (mduc_regs): Likewise.
6847 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
6848 (htm_expand_builtin): Likewise.
6849 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
6850 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
6851 * config/xtensa/xtensa.c (print_operand): Likewise.
6852 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
6853 (NUM_MODE_VECTOR_INT): Likewise.
6854 * genoutput.c (null_operand): Likewise.
6855 (output_operand_data): Likewise.
6856 * genrecog.c (print_parameter_value): Likewise.
6857 * lra.c (debug_operand_data): Likewise.
6859 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6861 * dwarf2out.c (scompare_loc_descriptor_wide)
6862 (scompare_loc_descriptor_narrow): New functions, split out from...
6863 (scompare_loc_descriptor): ...here.
6864 * expmed.c (emit_store_flag_int): New function, split out from...
6865 (emit_store_flag): ...here.
6867 2017-08-30 Richard Biener <rguenther@suse.de>
6869 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
6870 (dwarf2out_early_finish): Move setting of AT_pubnames from
6871 early debug output to early finish.
6873 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6875 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
6876 and -mdata-region to the assembler.
6878 2017-08-30 Richard Biener <rguenther@suse.de>
6880 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
6882 (gen_subprogram_die): Add DW_AT_object_pointer only early.
6883 (dwarf2out_early_global_decl): Only generate a DIE for the
6884 abstract origin if it doesn't already exist or is a declaration DIE.
6885 (resolve_addr): Do not add the linkage name twice when
6886 generating a stub DIE for the DW_TAG_GNU_call_site target.
6888 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6890 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6891 Use machine_mode rather than int for arg1_mode.
6893 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
6896 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
6897 that the second argument of the built-in functions to unpack
6898 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
6899 switch statement instead a lot of if statements.
6900 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
6901 Allow 64-bit values to be in Altivec registers as well as
6902 traditional floating point registers.
6903 (pack<mode>, FMOVE128_VSX iterator): Likewise.
6905 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
6907 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
6908 MAX_REGS_PER_ADDRESS == 1.
6910 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
6912 * config/i386/i386.opt (flag_fentry): Do not init to -1.
6913 * config/i386/i386.c (ix86_option_override_internal): Simplify
6914 setting of opts->x_flag_entry.
6916 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6917 Jakub Jelinek <jakub@redhat.com>
6918 Richard Biener <rguenther@suse.de>
6920 PR tree-optimization/81503
6921 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
6922 folded constant fits in the target type; reorder tests for clarity.
6924 2017-08-29 Martin Liska <mliska@suse.cz>
6926 * passes.def: Include pass_lower_switch.
6927 * stmt.c (dump_case_nodes): Remove and move to
6928 tree-switch-conversion.
6929 (case_values_threshold): Likewise.
6930 (expand_switch_as_decision_tree_p): Likewise.
6931 (emit_case_decision_tree): Likewise.
6932 (expand_case): Likewise.
6933 (balance_case_nodes): Likewise.
6934 (node_has_low_bound): Likewise.
6935 (node_has_high_bound): Likewise.
6936 (node_is_bounded): Likewise.
6937 (emit_case_nodes): Likewise.
6938 (struct simple_case_node): New struct.
6939 (add_case_node): Remove.
6940 (emit_case_dispatch_table): Use vector instead of case_list.
6941 (reset_out_edges_aux): Remove.
6942 (compute_cases_per_edge): Likewise.
6943 (expand_case): Build list of simple_case_node.
6944 (expand_sjlj_dispatch_table): Use it.
6945 * tree-switch-conversion.c (struct case_node): Moved from
6946 stmt.c and adjusted.
6947 (emit_case_nodes): Likewise.
6948 (node_has_low_bound): Likewise.
6949 (node_has_high_bound): Likewise.
6950 (node_is_bounded): Likewise.
6951 (case_values_threshold): Likewise.
6952 (reset_out_edges_aux): Likewise.
6953 (compute_cases_per_edge): Likewise.
6954 (add_case_node): Likewise.
6955 (dump_case_nodes): Likewise.
6956 (balance_case_nodes): Likewise.
6957 (expand_switch_as_decision_tree_p): Likewise.
6958 (emit_jump): Likewise.
6959 (emit_case_decision_tree): Likewise.
6960 (try_switch_expansion): Likewise.
6961 (do_jump_if_equal): Likewise.
6962 (emit_cmp_and_jump_insns): Likewise.
6963 (fix_phi_operands_for_edge): New function.
6964 (record_phi_operand_mapping): Likewise.
6965 (class pass_lower_switch): New pass.
6966 (pass_lower_switch::execute): New function.
6967 (make_pass_lower_switch): Likewise.
6968 (conditional_probability):
6969 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
6970 * tree-pass.h: Add make_pass_lower_switch.
6972 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6975 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
6978 2017-08-29 Richard Biener <rguenther@suse.de>
6980 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
6981 we do not add a DW_AT_inline attribute twice.
6982 (gen_subprogram_die): Remove code setting DW_AT_inline on
6983 DECL_ABSTRACT_P nodes.
6985 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
6987 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
6988 calls to internal functions.
6989 (gimplify_modify_expr): Likewise.
6990 * tree-call-cdce.c (use_internal_fn): Likewise.
6991 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
6992 (convert_to_divmod): Set the nothrow flag.
6993 * tree-if-conv.c (predicate_mem_writes): Likewise.
6994 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
6995 (vectorizable_call): Likewise.
6996 (vectorizable_store): Likewise.
6997 (vectorizable_load): Likewise.
6998 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
6999 (vect_recog_mask_conversion_pattern): Likewise.
7001 2017-08-29 Martin Liska <mliska@suse.cz>
7004 * gcc.c (driver_handle_option): Add new argument.
7005 * opts-common.c (handle_option): Pass
7006 target_option_override_hook.
7007 * opts-global.c (lang_handle_option): Add new option.
7008 (set_default_handlers): Add new argument.
7009 (decode_options): Likewise.
7010 * opts.c (target_handle_option): Likewise.
7011 (common_handle_option): Call target_option_override_hook.
7012 * opts.h (struct cl_option_handler_func): Add hook for
7013 target option override.
7014 (struct cl_option_handlers): Likewise.
7015 (set_default_handlers): Add new argument.
7016 (decode_options): Likewise.
7017 (common_handle_option): Likewise.
7018 (target_handle_option): Likewise.
7019 * toplev.c (toplev::main): Pass targetm.target_option.override
7022 2017-08-29 Richard Biener <rguenther@suse.de>
7023 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7025 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
7026 life to the active subtree.
7028 2017-08-28 Jeff Law <law@redhat.com>
7030 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
7031 derive_equivalences.
7032 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
7034 (edge_info::derive_equivalences): New private member function
7036 * tree-ssa-dom.c (class edge_info): Changed from a struct
7037 to a class. Add ctor/dtor, methods and data members.
7038 (edge_info::edge_info): Renamed from allocate_edge_info.
7039 Initialize additional members.
7040 (edge_info::~edge_info): New.
7041 (free_dom_edge_info): Delete the edge info.
7042 (record_edge_info): Use new class & associated member functions.
7043 Tighten forms for testing for edge equivalences.
7044 (record_temporary_equivalences): Iterate over the simple
7045 equivalences rather than assuming there's only one per edge.
7046 (cprop_into_successor_phis): Iterate over the simple
7047 equivalences rather than assuming there's only one per edge.
7048 (optimize_stmt): Use operand_equal_p rather than pointer
7049 equality for mini-DSE code.
7051 2017-08-28 Nathan Sidwell <nathan@acm.org>
7053 * gcc.c (execute): Fold SIGPIPE handling into switch
7054 statement. Adjust internal error message.
7056 2017-08-28 Richard Biener <rguenther@suse.de>
7059 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
7060 Do nothing for removed DIEs.
7062 2017-08-28 Richard Biener <rguenther@suse.de>
7064 PR tree-optimization/81977
7065 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
7068 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
7071 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
7072 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
7073 using targetm.gen_mem_thread_fence.
7075 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
7078 * config/i386/i386.md (*<btsc><mode>): Change operand 2
7079 predicate to register_operand. Reorder operands.
7080 (*btr<mode>): Ditto.
7081 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
7082 (*btr<mode>_mask): Ditto.
7084 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
7086 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
7087 * config/rs6000/xmmintrin.h: New file.
7088 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
7090 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7093 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
7094 parameter and_insn and return it.
7095 (recombine_lvx_pattern): Insert a copy to ensure availability of
7096 the base register of the copied masking operation at the point of
7097 the instruction replacement.
7098 (recombine_stvx_pattern): Likewise.
7100 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
7102 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
7103 undocumented switches.
7104 (-mpower9-dform-vector): Likewise.
7105 (-mpower9-dform): Likewise.
7106 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
7107 comments to delete references to -mpower9-dform* switches.
7108 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
7109 Delete reference to -mpower9-dform* switches, test for
7110 -mpower9-vector instead.
7111 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
7112 (OTHER_P9_VECTOR_MASKS): Likewise.
7113 (POWERPC_MASKS): Likewise.
7114 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
7115 tests against -mpower9-dform* to -mpower9-vector. Delete code
7116 that checked for -mpower9-dform* consistancy with other options.
7117 Add test for -mpower9-misc to enable other power9 switches.
7118 (rs6000_init_hard_regno_mode_ok): Likewise.
7119 (rs6000_option_override_internal): Likewise.
7120 (rs6000_emit_prologue): Likewise.
7121 (rs6000_emit_epilogue): Likewise.
7122 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
7123 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
7124 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
7126 (emit_fusion_p9_store): Likewise.
7127 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
7128 resetting these macros if the assembler does not support ISA 3.0
7130 (TARGET_P9_DFORM_VECTOR): Likewise.
7131 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
7132 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
7134 2017-08-25 Alan Modra <amodra@gmail.com>
7136 PR rtl-optimization/81747
7137 * cse.c (cse_extended_basic_block): Don't attempt to record
7138 equivalences for degenerate conditional branches that branch
7139 to their fall-through.
7141 2017-08-24 Martin Sebor <msebor@redhat.com>
7144 * gimple-fold.c (size_must_be_zero_p): New function.
7145 (gimple_fold_builtin_memory_op): Call it.
7147 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
7149 * config/rs6000/mm_malloc.h: New file.
7151 2017-08-24 Bin Cheng <bin.cheng@arm.com>
7153 PR tree-optimization/81913
7154 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
7155 analysis when either IVs in condition can wrap.
7157 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
7159 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
7160 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
7162 2017-08-24 Richard Biener <rguenther@suse.de>
7165 * targhooks.c (default_target_can_inline_p): Properly
7166 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
7167 is present and always compare.
7168 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
7169 infer -mfpmath=sse from TARGET_SSE_P.
7170 (ix86_can_inline_p): Properly use target_option_default_node when
7171 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
7173 2017-08-24 Richard Biener <rguenther@suse.de>
7176 * dwarf2out.c (output_die): Handle flag_generate_offload like
7178 (output_comp_unit): Likewise.
7179 (gen_array_type_die): Likewise.
7180 (dwarf2out_early_finish): Likewise.
7181 (note_variable_value_in_expr): Likewise.
7182 (dwarf2out_finish): Likewise. Adjust assert.
7183 * cgraphunit.c (symbol_table::compile): Move setting of
7184 flag_generate_offload earlier ...
7185 (symbol_table::finalize_compilation_unit): ... here, before
7186 early debug finalization.
7188 2017-08-24 Richard Biener <rguenther@suse.de>
7190 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
7191 and ipa-fnsummary.h.
7192 (ix86_can_inline_p): When ix86_fpmath flags do not match
7193 check whether the callee uses FP math at all.
7195 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
7198 * tree-ssanames.c (get_nonzero_bits): Use element_precision
7199 instead of TYPE_PRECISION.
7201 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
7202 Alan Hayward <alan.hayward@arm.com>
7203 David Sherwood <david.sherwood@arm.com>
7205 * combine.c (make_extraction): Use subreg_offset_from_lsb.
7207 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
7209 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
7211 (ix86_frame::stack_realign_allocate): New field.
7212 (struct machine_frame_state): Modify comments.
7213 (machine_frame_state::sp_realigned_fp_end): New field.
7214 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
7216 (sp_valid_at): Add assertion to assure no attempt to access invalid
7217 offset of a realigned stack.
7218 (fp_valid_at): Likewise.
7219 (choose_baseaddr): Modify comments.
7220 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
7221 ix86_expand_prologue.
7222 (ix86_expand_prologue): Modify stack realignment and allocation.
7223 (ix86_expand_epilogue): Modify comments.
7224 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
7225 avx2_runtime, avx512f, and avx512f_runtime.
7227 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
7229 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
7230 (mstackrealign): Do not init to -1.
7231 * config/i386/i386.c (ix86_option_override_internal):
7232 Check opts_set, not opts when setting default value of
7233 opts->x_ix86_force_align_arg_pointer.
7235 2017-08-23 Richard Biener <rguenther@suse.de>
7237 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
7238 lang_hooks.decl_printable_name.
7239 * print-rtl-function.c (print_rtx_function): Likewise.
7240 * tree-pretty-print.c (dump_function_header): Likewise.
7242 2017-08-23 Richard Biener <rguenther@suse.de>
7245 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
7246 -g0 at compile-time.
7248 2017-08-23 Tamar Christina <tamar.christina@arm.com>
7251 * doc/sourcebuild.texi (Other hardware attributes):
7254 2017-08-23 Tamar Christina <tamar.christina@arm.com>
7257 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
7258 Fix single-use check.
7260 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
7262 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
7264 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
7266 * doc/install.texi: Modify to add more details on running selected
7269 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
7271 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
7272 is combined with -mabi=ms.
7273 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
7276 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7278 PR tree-optimization/81488
7279 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
7280 and cached_basis fields.
7281 (MAX_SPREAD): New constant.
7282 (alloc_cand_and_find_basis): Initialize new fields.
7283 (clear_visited): New function.
7284 (create_phi_basis_1): Rename from create_phi_basis, set visited
7285 and cached_basis fields.
7286 (create_phi_basis): New wrapper function.
7287 (phi_add_costs_1): Rename from phi_add_costs, add spread
7288 parameter, set visited field, short-circuit when limits reached.
7289 (phi_add_costs): New wrapper function.
7290 (record_phi_increments_1): Rename from record_phi_increments, set
7292 (record_phi_increments): New wrapper function.
7293 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
7294 (phi_incr_cost): New wrapper function.
7295 (all_phi_incrs_profitable_1): Rename from
7296 all_phi_incrs_profitable, set visited field.
7297 (all_phi_incrs_profitable): New wrapper function.
7299 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
7300 Alan Hayward <alan.hayward@arm.com>
7301 David Sherwood <david.sherwood@arm.com>
7303 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
7304 that takes the outer and inner modes.
7305 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
7306 comparison as the canonical test for a paradoxical subreg.
7307 * combine.c (simplify_set): Use paradoxical_subreg_p.
7308 (make_extraction): Likewise.
7309 (force_to_mode): Likewise.
7310 (rtx_equal_for_field_assignment_p): Likewise.
7311 (gen_lowpart_for_combine): Likewise.
7312 (simplify_comparison): Likewise.
7313 * cse.c (equiv_constant): Likewise.
7314 * expmed.c (store_bit_field_1): Likewise.
7315 * final.c (alter_subreg): Likewise.
7316 * fwprop.c (propagate_rtx): Likewise.
7317 (forward_propagate_subreg): Likewise.
7318 * ira-conflicts.c (ira_build_conflicts): Likewise.
7319 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
7320 * lra-constraints.c (curr_insn_transform): Likewise.
7321 (split_reg): Likewise.
7322 * lra-eliminations.c (move_plus_up): Likewise.
7323 (lra_eliminate_regs_1): Likewise.
7324 * recog.c (general_operand): Likewise.
7325 * ree.c (combine_reaching_defs): Likewise.
7326 * reload.c (push_reload): Likewise.
7327 (find_reloads): Likewise.
7328 * reload1.c (elimination_effects): Likewise.
7329 (compute_reload_subreg_offset): Likewise.
7330 (choose_reload_regs): Likewise.
7331 * rtlanal.c (subreg_lsb_1): Likewise.
7332 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7333 (simplify_subreg): Likewise.
7334 * var-tracking.c (track_loc_p): Likewise.
7335 * emit-rtl.c (byte_lowpart_offset): Likewise.
7336 (paradoxical_subreg_p): Delete out-of-line definition.
7338 2017-08-22 Jeff Law <law@redhat.com>
7340 PR tree-optimization/81741
7341 PR tree-optimization/71947
7342 * tree-ssa-dom.c: Include tree-inline.h.
7343 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
7344 equivalences if one is more expensive to compute than the other.
7345 * tree-ssa-scopedtables.h (class const_or_copies): Make
7346 record_const_or_copy_raw method private.
7347 (class avail_exprs_stack): New method simplify_binary_operation.
7348 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
7349 avail_exprs_stack::simplify_binary_operation as needed.
7350 (avail_exprs_stack::simplify_binary_operation): New function.
7352 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
7354 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
7355 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
7356 (DOT_SYMBOLS): Likewise.
7357 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
7358 (RELOCATABLE_NEEDS_FIXUP): Likewise.
7359 (RS6000_ABI_NAME): Likewise.
7360 (TARGET_CMODEL): Likewise.
7361 (TOC_SECTION_ASM_OP): Likewise.
7362 (SET_CMODEL): New macro.
7363 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
7365 2017-08-22 Richard Biener <rguenther@suse.de>
7367 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
7368 to allow for free-lang-data replacements similar to verify_type_variant.
7370 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
7371 Alan Hayward <alan.hayward@arm.com>
7372 David Sherwood <david.sherwood@arm.com>
7374 * config/aarch64/aarch64.md (casesi): Use DImode rather than
7375 VOIDmode for the LABEL_REF.
7377 2017-08-22 Richard Biener <rguenther@suse.de>
7379 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
7381 2017-08-22 Richard Biener <rguenther@suse.de>
7383 * common.opt (feliminate-dwarf2-dups): Ignore.
7384 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
7385 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
7386 same_die_p_wrap, compute_section_prefix,
7387 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
7388 (comdat_symbol_id, comdat_symbol_number): Likewise.
7389 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
7391 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
7392 (output_die): Mark unreachable path unreachable.
7393 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
7394 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
7395 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
7396 (dwarf2out_early_finish): Likewise.
7398 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
7400 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
7402 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
7405 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
7406 not VAR_P. Filter attribute warnings with OPT_Wattributes.
7407 (avr_attribute_table) <io, io_low, address>: Initialize
7408 .decl_required with true.
7410 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
7412 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
7413 undocumented debugging options.
7414 (-mvsx-scalar-double): Likewise.
7415 (-mallow-df-permute): Likewise.
7416 (-mvectorize-builtins): Likewise.
7417 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
7418 (rs6000_builtin_vectorized_function): Likewise.
7419 (rs6000_builtin_md_vectorized_function): Likewise.
7420 (rs6000_opt_vars): Likewise.
7422 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
7425 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
7426 (*btrq_imm): Rename from *btrq.
7427 (*btcq_imm): Rename from *btcq.
7428 (btsc): New code attribute.
7429 (*<btsc><mode>): New insn pattern.
7430 (*btr<mode>): Ditto.
7431 (*<btsc><mode>_mask): New insn_and_split pattern.
7432 (*btr<mode>_mask): Ditto.
7434 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7435 Alan Hayward <alan.hayward@arm.com>
7436 David Sherwood <david.sherwood@arm.com>
7438 * function.c (pad_below): Simplify padding calculation.
7440 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7441 Alan Hayward <alan.hayward@arm.com>
7442 David Sherwood <david.sherwood@arm.com>
7444 * target.def (function_prologue): Remove frame size argument.
7445 (function_epilogue): Likewise.
7446 * doc/tm.texi: Regenerate.
7447 * final.c (final_start_function): Update call to function_prologue.
7448 (final_end_function): Update call to function_epilogue.
7449 (default_function_pro_epilogue): Remove frame size argument.
7450 * output.h (default_function_pro_epilogue): Likewise.
7451 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
7452 (arm_output_function_prologue): Likewise.
7453 * config/frv/frv.c (frv_function_prologue): Likewise.
7454 (frv_function_epilogue): Likewise.
7455 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
7456 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
7457 (ia64_output_function_epilogue): Likewise.
7458 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
7459 (m32r_output_function_epilogue): Likewise.
7460 * config/microblaze/microblaze.c (microblaze_function_prologue)
7461 (microblaze_function_epilogue): Likewise.
7462 * config/mips/mips.c (mips_output_function_prologue): Likewise.
7463 (mips_output_function_epilogue): Likewise.
7464 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
7465 (mmix_target_asm_function_epilogue): Likewise.
7466 * config/msp430/msp430.c (msp430_start_function): Likewise.
7467 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
7468 (nds32_asm_function_epilogue): Likewise.
7469 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
7470 * config/pa/pa.c (pa_output_function_prologue): Likewise.
7471 (pa_output_function_epilogue): Likewise.
7472 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
7473 (rs6000_output_function_epilogue): Likewise.
7474 * config/rl78/rl78.c (rl78_start_function): Likewise.
7475 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
7476 (rs6000_output_function_epilogue): Likewise.
7477 * config/rx/rx.c (rx_output_function_prologue): Likewise.
7478 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
7479 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
7480 (sparc_asm_function_epilogue): Likewise.
7482 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7484 * tree.h (type_has_mode_precision_p): New function.
7485 * convert.c (convert_to_integer_1): Use it.
7486 * expr.c (expand_expr_real_2): Likewise.
7487 (expand_expr_real_1): Likewise.
7488 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
7489 * match.pd: Likewise.
7490 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
7491 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
7492 * tree-tailcall.c (process_assignment): Likewise.
7493 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7494 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
7495 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
7496 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7497 (vectorizable_assignment): Likewise.
7498 (vectorizable_shift): Likewise.
7499 (vectorizable_operation): Likewise.
7500 * tree-vrp.c (register_edge_assert_for_2): Likewise.
7502 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
7504 * match.pd: Add pow (C, x) simplification.
7506 2017-08-21 Richard Biener <rguenther@suse.de>
7508 PR tree-optimization/81900
7509 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
7510 for blocks with abnormal predecessors.
7511 (compute_antic): Do not set visited flag prematurely.
7513 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
7516 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
7518 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7520 * stor-layout.h (vector_type_mode): Move to...
7521 * tree.h (vector_type_mode): ...here.
7522 * stor-layout.c (vector_type_mode): Move to...
7523 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
7525 2017-08-21 Richard Biener <rguenther@suse.de>
7527 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
7528 register_external_die hooks.
7529 (debug_false_tree_charstarstar_uhwistar): Declare.
7530 (debug_nothing_tree_charstar_uhwi): Likewise.
7531 * debug.c (do_nothing_debug_hooks): Adjust.
7532 (debug_false_tree_charstarstar_uhwistar): New do nothing.
7533 (debug_nothing_tree_charstar_uhwi): Likewise.
7534 * dbxout.c (dbx_debug_hooks): Adjust.
7535 (xcoff_debug_hooks): Likewise.
7536 * sdbout.c (sdb_debug_hooks): Likewise.
7537 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
7538 * dwarf2out.c (macinfo_label_base): New global.
7539 (dwarf2out_register_external_die): New function for the
7540 register_external_die hook.
7541 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
7542 (dwarf2_debug_hooks): Use them.
7543 (dwarf2_lineno_debug_hooks): Adjust.
7544 (struct die_struct): Add with_offset flag.
7545 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
7546 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
7547 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
7548 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
7549 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
7550 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
7551 defining section names for the early LTO debug variants.
7552 (reset_indirect_string): New helper.
7553 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
7554 (print_dw_val): Add support for offsetted symbol references.
7555 (get_ultimate_context): Split out from is_cxx.
7556 (is_cxx): Use get_ultimate_context.
7557 (is_fortran): Add decl overload.
7558 (compute_comp_unit_symbol): Split out worker from
7559 compute_section_prefix.
7560 (compute_section_prefix): Call compute_comp_unit_symbol and
7561 set comdat_type_p here.
7562 (output_die): Skip DIE symbol output for the LTO added one.
7563 Handle DIE symbol references with offset.
7564 (output_comp_unit): Guard section name mangling properly.
7565 For LTO debug sections emit a symbol at the section beginning
7566 which we use to refer to its DIEs.
7567 (add_abstract_origin_attribute): For DIEs registered via
7568 dwarf2out_register_external_die directly refer to the early
7569 DIE rather than indirectly through the shadow one we created.
7570 Remove obsolete call to dwarf2out_abstract_function for
7571 non-function/block origins.
7572 (gen_array_type_die): When generating early LTO debug do
7573 not emit DW_AT_string_length.
7574 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
7575 late when in LTO. As suggested place a gcc_unreachable for
7576 the DECL_ABSTRACT_P case.
7577 (gen_subprogram_die): Avoid another specification DIE
7578 for early built declarations/definitions for the late LTO case.
7579 (gen_variable_die): Add type references for late duplicated VLA dies
7581 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
7582 we have the abstract instance already.
7583 (process_scope_var): Adjust decl DIE contexts in LTO which
7584 first puts them in limbo.
7585 (gen_decl_die): Do not generate type DIEs late apart from
7586 types for VLAs or for decls we do not yet have a DIE. Do not
7587 call dwarf2out_abstract_function late.
7588 (dwarf2out_early_global_decl): Make sure to create DIEs
7589 for abstract instances of a decl first.
7590 (dwarf2out_late_global_decl): Adjust comment.
7591 (output_macinfo_op): With multiple macro sections use
7592 macinfo_label_base to distinguish labels.
7593 (output_macinfo): Likewise. Update macinfo_label_base.
7594 Pass in the line info label.
7595 (note_variable_value_in_expr): When generating LTO resolve
7596 all variable values here by generating DIEs as needed.
7597 (init_sections_and_labels): Add early LTO debug flag parameter
7598 and generate different sections and names if set. Add generation
7599 counter for the labels so we can have multiple of them.
7600 (reset_dies): Helper to allow DIEs to be output multiple times.
7601 (dwarf2out_finish): When outputting DIEs to the fat part of an
7602 LTO object first reset DIEs.
7603 (dwarf2out_early_finish): Output early DIEs when generating LTO.
7604 (modified_type_die): Check for decl_ultimate_origin being self
7606 (gen_type_die_with_usage): Likewise.
7607 (gen_typedef_die): Allow decl_ultimate_origin being self.
7608 (set_decl_abstract_flags): Remove.
7609 (set_block_abstract_flags): Likewise.
7610 (dwarf2out_abstract_function): Treat the early generated DIEs
7611 as the abstract copy and only add DW_AT_inline and
7612 DW_AT_artificial here and call set_decl_origin_self.
7613 If the DIE has an abstract origin don't do anything.
7614 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
7615 if we have none yet (Go fails to build one, PR78628).
7616 (variably_modified_type_p): Prevent endless recursion for Ada
7617 cyclic pointer types.
7618 * lto-streamer-in.c: Include debug.h.
7619 (dref_queue): New global.
7620 (lto_read_tree_1): Stream in DIE references.
7621 (lto_input_tree): Register DIE references.
7622 (input_function): Stream DECL_DEBUG_ARGS.
7623 * lto-streamer-out.c: Include debug.h.
7624 (lto_write_tree_1): Output DIE references.
7625 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
7626 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
7627 (output_function): Stream DECL_DEBUG_ARGS.
7628 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
7629 Stream DECL_ABSTRACT_ORIGIN.
7630 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
7631 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
7632 DECL_CONTEXT for file-scope decls.
7633 * lto-streamer.h (struct dref_entry): Declare.
7634 (dref_queue): Likewise.
7635 * cfgexpand.c (pass_expand::execute): Do not call the
7636 outlining_inline_function hook here.
7637 * lto-wrapper.c (debug_obj): New global.
7638 (tool_cleanup): Unlink it if required.
7639 (debug_objcopy): New function.
7640 (run_gcc): Handle early debug sections in the IL files by
7641 extracting them to separate files, partially linkin them and
7642 feeding the result back as result to the linker.
7643 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
7644 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
7645 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
7646 sections into a separate segment.
7647 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
7649 (darwin_asm_dwarf_section): Likewise.
7650 (darwin_asm_output_dwarf_offset): Likewise.
7651 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
7653 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7654 Alan Hayward <alan.hayward@arm.com>
7655 David Sherwood <david.sherwood@arm.com>
7657 * read-md.h (md_reader::record_potential_iterator_use): Replace
7658 pointer argument with an rtx and an index.
7659 * read-rtl.c (iterator_group::apply_iterator): Likewise.
7660 (apply_mode_iterator): Likewise.
7661 (apply_code_iterator): Likewise.
7662 (apply_int_iterator): Likewise.
7663 (apply_subst_iterator): Likewise.
7664 (record_iterator_use): Likewise.
7665 (record_attribute_use): Likewise.
7666 (md_reader::record_potential_iterator_use): Likewise. Update calls
7667 to record_iterator_use and apply_iterator.
7668 (iterator_use): Replace ptr with x and index.
7669 (attribute_use): Likewise.
7670 (apply_attribute_uses): Update calls to apply_iterator.
7671 (apply_iterators): Likewise. Update initialization of iterator_use.
7672 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
7673 and record_potential_iterator_use.
7674 (rtx_reader::read_rtx_operand): Likewise.
7676 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7677 Alan Hayward <alan.hayward@arm.com>
7678 David Sherwood <david.sherwood@arm.com>
7680 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
7683 2017-08-21 Richard Biener <rguenther@suse.de>
7686 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
7687 at struct end conservatively when comparing common bases.
7689 2017-08-21 Richard Biener <rguenther@suse.de>
7691 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
7692 (mem_ref_in_stmt): Remove.
7693 (determine_max_movement): Use ref index to get at the reference.
7694 (invariantness_dom_walker::before_dom_children): Deal with
7695 lim data already initialized.
7696 (gather_mem_refs_stmt): Initialize lim data and record ref index.
7698 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
7700 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
7701 (TARGET_ISA_ROUND): Ditto.
7702 (TARGET_ROUND): Ditto.
7703 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
7704 * config/i386/i386.md: Ditto.
7705 * config/i386/sse.md: Ditto.
7706 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
7707 with OPTION_MASK_ISA_SSE4_1.
7709 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
7712 * doc/extend.texi (x86 Built-in Functions): Correct the name of
7713 __builtin_ia32_lzcnt_u16.
7715 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
7718 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
7719 (rs6000_set_current_function): Rewrite function to use it.
7721 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
7724 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
7725 and TYPE_WARN_IF_NOT_ALIGN.
7726 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
7727 (handle_warn_if_not_align): New.
7728 (place_union_field): Call handle_warn_if_not_align.
7729 (place_field): Call handle_warn_if_not_align.
7730 Copy TYPE_WARN_IF_NOT_ALIGN.
7731 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
7732 (layout_type): Likewise.
7733 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
7735 (tree_decl_common): Add warn_if_not_align.
7736 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
7737 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
7738 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
7739 (DECL_WARN_IF_NOT_ALIGN): Likewise.
7740 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
7741 * doc/extend.texi: Document warn_if_not_aligned attribute.
7742 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
7744 2017-08-17 Martin Liska <mliska@suse.cz>
7747 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
7748 (get_data_dependence): Use it as pointer type.
7749 (distribute_loop): Likewise.
7751 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7753 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
7754 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
7755 (p8_vmrgow_<mode>_direct): New define_insn.
7756 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
7757 handle endianness for vmrgew and vmrgow permute patterns.
7759 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
7761 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
7762 * config/rs6000/rs6000-cpus.def: Remove comment.
7763 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
7764 (POWERPC_MASKS): Likewise.
7765 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
7766 use of TARGET_VSX_TIMODE.
7767 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
7768 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
7769 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
7770 (rs6000_option_override_internal): Remove dead code.
7771 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
7772 (rs6000_legitimize_reload_address): Likewise.
7773 (rs6000_legitimate_address_p): Likewise.
7774 (rs6000_opt_masks): Delete "vsx-timode".
7775 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
7776 from function comment.
7777 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
7778 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
7779 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
7781 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
7782 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
7783 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
7786 (define_peephole2): Likewise.
7788 2017-08-17 Martin Sebor <msebor@redhat.com>
7791 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
7792 past the end of an array.
7793 (test_pp_format): Add test cases.
7795 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
7797 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
7798 missing ECF_NOTHROW flags.
7800 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
7803 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
7804 operands residing in integer registers.
7805 (*vsx_le_perm_load_<mode>): Likewise.
7806 (*vsx_le_perm_store_<mode>): Likewise.
7807 (define_peephole2): Add peepholes to optimize the above.
7809 2017-08-17 Marek Polacek <polacek@redhat.com>
7812 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
7813 to mimic what shorten_compare did. Change the return type to bool.
7814 (fold_cond_expr_with_comparison): Update call to
7815 operand_equal_for_comparison_p.
7816 (fold_ternary_loc): Likewise.
7818 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
7820 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
7822 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
7824 2017-08-17 Richard Biener <rguenther@suse.de>
7826 * tree-ssa-structalias.c (solve_graph): When propagating
7827 to successors update the graphs succ edges and avoid duplicate work.
7829 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
7832 * config/i386/i386.c (ix86_option_override_internal): Save target
7833 specific options after ix86_stack_protector_guard_reg was changed.
7835 2017-08-17 Richard Biener <rguenther@suse.de>
7837 PR tree-optimization/81827
7838 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
7839 (new_var_info): Initialize it conservatively.
7840 (get_call_vi): Mark register vars.
7841 (new_scalar_tmp_constraint_exp): Likewise.
7842 (handle_rhs_call): Likewise.
7843 (handle_const_call): Likewise.
7844 (create_function_info_for): Likewise.
7845 (solve_constraints): Sort varinfos to separate register from
7846 non-register vars to pack points-to solution bitmaps during
7849 2017-08-17 Marek Polacek <polacek@redhat.com>
7851 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
7853 2017-08-17 Richard Biener <rguenther@suse.de>
7855 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
7856 to true when overflow is undefined and we saturated the result.
7858 2017-08-17 Alan Modra <amodra@gmail.com>
7861 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
7862 Don't use store multiple if only one reg needs saving.
7863 (interesting_frame_related_regno): New function.
7864 (rs6000_frame_related): Don't emit frame info for regs that
7866 (rs6000_emit_epilogue): Likewise.
7868 2017-08-16 Nathan Sidwell <nathan@acm.org>
7870 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
7871 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
7872 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
7873 * tree.c (free_lang_data_in_type): Use else-if chain. Always
7874 clear TYPE_LANG_1. Remove obsolete member-function stripping.
7875 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
7876 (verify_type): Adjust for TYPE_BINFO move.
7877 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
7878 process TYPE_BINFO directly.
7879 (hash_tree): Likewise.
7880 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
7882 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
7885 2017-08-16 David Malcolm <dmalcolm@redhat.com>
7887 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
7889 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
7892 * config/i386/i386.md (*anddi_1_btr): Change predicates of
7893 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
7894 Add ix86_binary_operator_ok to insn constraint.
7895 (*iordi_1_bts): Ditto.
7896 (*xordi_1_btc): Ditto.
7897 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
7898 Update corresponding peephole2 pattern.
7902 2017-08-16 Bin Cheng <bin.cheng@arm.com>
7904 PR tree-optimization/81832
7905 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
7906 copy loop header which has IFN_LOOP_DIST_ALIAS call.
7908 2017-08-16 Marek Polacek <polacek@redhat.com>
7911 * fold-const.c (fold_indirect_ref_1): Restore original behavior
7912 regarding size_zero_node.
7914 2017-08-16 Martin Liska <mliska@suse.cz>
7917 * config.gcc: Respect previously set extra_objs in case
7920 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
7922 PR tree-optimization/81835
7923 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
7924 the phi SSA_NAME. Check that the condition in a COND_EXPR does
7925 not depend on the phi.
7927 2017-08-16 Alan Modra <amodra@gmail.com>
7929 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
7932 2017-08-16 Alan Modra <amodra@gmail.com>
7934 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
7935 (save_reg_p): ..into this. Update all callers.
7936 (first_reg_to_save): Simplify.
7938 2017-08-16 Alan Modra <amodra@gmail.com>
7940 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
7943 2017-08-15 Joseph Myers <joseph@codesourcery.com>
7947 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
7948 constant count if that count is less than 32.
7950 2017-08-15 Nathan Sidwell <nathan@acm.org>
7952 * gcc.c (execute): Emit friendlier message if inferior is killed
7953 by an external cause.
7955 2017-08-15 Richard Biener <rguenther@suse.de>
7957 PR tree-optimization/81790
7958 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
7959 CONSTRUCTORs from simplifying and VN.
7961 2017-08-14 Martin Sebor <msebor@redhat.com>
7963 * builtin-attrs.def: Add comments.
7965 2017-08-14 Martin Sebor <msebor@redhat.com>
7968 * doc/extend.texi (attribute nonstring): Document new attribute.
7970 2017-08-14 Martin Sebor <msebor@redhat.com>
7973 * tree-diagnostic.c (default_tree_printer): Handle %G.
7974 * gimple-pretty-print.h (percent_G_format): Declare new function.
7975 * gimple-pretty-print.c (percent_G_format): Define.
7976 * tree-pretty-print.c (percent_K_format): Add argument.
7978 2017-08-14 Martin Sebor <msebor@redhat.com>
7980 PR translation/79998
7981 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
7982 Remove a stray space.
7984 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
7987 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
7988 (*iordi_1_bts): Ditto.
7989 (*xordi_1_btc): Ditto.
7991 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7994 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
7995 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7997 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
7998 quoted strings, and make more translator-friendly.
7999 (darwin_rs6000_override_options): Likewise.
8000 (rs6000_option_override_internal): Likewise.
8001 (rs6000_return_in_memory): Fix overlong line.
8002 (init_cmulative_args): Use quoted strings, and make more
8003 translator-friendly.
8004 (rs6000_pass_by_reference): Fix overlong line.
8005 (def_builtin): Use quoted strings.
8006 (altivec_expand_predicate_builtin): Use quoted strings, and make
8007 more translator-friendly.
8008 (htm_expand_builtin): Use quoted strings.
8009 (cpu_expand_builtin): Use quoted strings, and make more
8010 translator-friendly.
8011 (altivec_expand_builtin): Likewise.
8012 (paired_expand_predicate_builtin): Likewise.
8013 (rs6000_invalid_builtin): Likewise.
8014 (builtin_function_type): Use quoted strings.
8015 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
8016 more translator-friendly.
8017 (rs6000_trampoline_init): Likewise.
8018 (rs6000_handle_altivec_attribute): Likewise.
8019 (rs6000_inner_target_options): Use quoted strings.
8020 (rs6000_disable_incompatible_switches): Likewise.
8021 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
8022 strings, and make more translator-friendly.
8023 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
8025 2017-08-14 Bin Cheng <bin.cheng@arm.com>
8027 PR tree-optimization/81799
8028 * tree-loop-distribution.c (version_loop_by_alias_check): Force
8029 cond_expr to simple gimple operand.
8031 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
8034 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
8036 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
8040 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
8041 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
8042 TARGET_GASISR_PROLOGUES.
8043 * config/avr/avr.c (avr_option_override): Same.
8044 (avr_pass_pre_proep::execute): Same.
8046 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
8049 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
8050 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
8052 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
8054 * config/i386/i386.md (*load_tp_<mode>): Redefine as
8055 define_insn_and_split. Split to a memory load from 0 in
8056 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
8057 using PTR mode iterator.
8058 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
8059 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
8060 (*add_tp_<mode>): Redefine as define_insn_and_split.
8061 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
8062 address space. Merge with *add_tp_x32 using PTR mode iterator.
8063 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
8064 Split to an add with a memory load from 0 in
8065 DEFAULT_TLS_SEG_REG address space.
8067 2017-08-12 Andrew Pinski <apinski@cavium.com>
8069 * config/aarch64/aarch64-option-extensions.def (rdma):
8070 Fix feature string to what Linux prints out in /proc/cpuinfo.
8072 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
8075 * dwarf2out.c (modified_type_die): For C typedef types that have
8076 an ultimate origin, process the ultimate origin instead of the
8078 (gen_typedef_die): Assert that input DECLs have no ultimate
8080 (gen_type_die_with_usage): For typedef variants that have an
8081 ultimate origin, just call gen_decl_die on the original DECL.
8082 (process_scope_var): Avoid creating DIEs for local typedefs and
8083 concrete static variables.
8085 2017-08-12 Alan Modra <amodra@gmail.com>
8089 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
8090 match gnu-user.h startfile.
8091 (ENDFILE_LINUX_SPEC): Similarly.
8093 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
8096 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
8098 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
8100 2017-08-11 Tamar Christina <tamar.christina@arm.com>
8101 * config/aarch64/aarch64.md (mov<mode>): Change.
8102 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
8103 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
8104 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
8106 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
8108 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
8109 for storage order barriers.
8111 2017-08-11 Martin Liska <mliska@suse.cz>
8114 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
8115 variables of void type.
8117 2017-08-11 Martin Liska <mliska@suse.cz>
8119 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
8120 TARGET_SUPPORTS_ALIASES.
8121 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
8122 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
8123 (optimize_weakref): Likewise.
8124 * symtab.c (symtab_node::noninterposable_alias): Likewise.
8125 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
8126 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
8128 2017-08-11 Martin Liska <mliska@suse.cz>
8131 * config/i386/i386.c (make_resolver_func): Do complete
8132 refactoring of the function.
8134 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
8137 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
8138 * config/i386/i386.c (ix86_stack_protect_guard): Use
8139 ix86_stack_protect_guard_symbol_str to generate varible declaration.
8140 * doc/invoke.texi (x86 Options): Document
8141 -mstack-protector-guard-symbol= option.
8143 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
8145 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
8146 * config/i386/i386.c (ix86_split_stack_guard): New function.
8147 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
8148 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
8149 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
8150 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
8151 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
8152 (split_stack_space_check): Call ix86_split_stack_guard.
8154 2017-08-10 Martin Sebor <msebor@redhat.com>
8156 * print-tree.c (print_node): Print location using the established
8158 Replace spaces with colons.
8159 (debug_raw, debug): Ditto.
8161 2017-08-10 Martin Sebor <msebor@redhat.com>
8164 * pretty-print.c (pp_format): Correct the handling of %s precision.
8166 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
8169 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
8171 (ix86_finalize_stack_frame_flags): This. Also clear
8172 frame_pointer_needed if -fno-omit-frame-pointer is used without
8174 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
8175 with ix86_finalize_stack_frame_flags.
8176 (ix86_expand_epilogue): Likewise.
8177 (ix86_expand_split_stack_prologue): Likewise.
8178 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
8180 2017-08-10 Martin Liska <mliska@suse.cz>
8183 * c-attribs.c (handle_target_attribute):
8184 Report warning for an empty string argument of target attribute.
8186 2017-08-09 Jakub Jelinek <jakub@redhat.com>
8189 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
8191 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
8192 or DECL_NONLOCAL labels.
8193 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
8194 or DECL_NONLOCAL labels here.
8196 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
8198 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
8199 to indicate when early gimple folding has been disabled.
8200 (rs6000_gimple_fold_builtin): Add debug content.
8201 (rs6000_invalid_builtin): Fix whitespace.
8202 (rs6000_expand_builtin): Fix whitespace.
8203 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
8205 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
8208 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
8209 SAVE_MULTIPLE if not all the registers that saves, should be saved.
8211 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
8213 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
8214 (qdf24xx): Likewise.
8215 * config/aarch64/aarch64.md: Include falkor.md.
8216 * config/aarch64/falkor.md: New.
8218 2017-08-09 Marek Polacek <polacek@redhat.com>
8221 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
8222 * diagnostic.c (emit_diagnostic): Add a comment.
8223 (emit_diagnostic_valist): New function.
8225 2017-08-09 Marek Polacek <polacek@redhat.com>
8228 * input.c (make_location): New overload.
8229 * input.h (make_location): Declare.
8231 2017-08-08 Alan Modra <amodra@gmail.com>
8232 H.J. Lu <hongjiu.lu@intel.com>
8235 * gcc.c (NO_PIE_SPEC): Delete.
8236 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
8238 (LINK_PIE_SPEC): ..to here.
8239 (LINK_COMMAND_SPEC): Support -no-pie.
8240 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
8241 chain of crtbegin*.o selection, update for PIE_SPEC changes and
8243 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
8244 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
8245 (ENDFILE_CRTEND_SPEC): Similarly.
8247 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
8250 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
8251 (mstack-protector-guard-offset=): Ditto.
8252 * config/i386/i386.c (ix86_option_override): Handle
8253 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
8255 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
8256 ix86_stack_protect_guard_offset variables.
8257 (TARGET_STACK_PROTECT_GUARD): Always define.
8258 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
8259 and -mstack-protector-guard-offset= options.
8261 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8263 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
8264 boundary case for the last candidate.
8266 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8268 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
8269 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
8271 2017-08-08 Tamar Christina <tamar.christina@arm.com>
8274 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
8275 * config/aarch64/aarch64-builtins.c
8276 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
8277 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
8278 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
8280 2017-08-08 Tamar Christina <tamar.christina@arm.com>
8281 Andrew Pinski <pinskia@gmail.com>
8284 * internal-fn.def (XORSIGN): New.
8285 * optabs.def (xorsign_optab): New.
8286 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
8287 (convert_expand_mult_copysign): New.
8288 (pass_optimize_widening_mul::execute): Call
8289 convert_expand_mult_copysign.
8291 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8293 PR tree-optimization/81354
8294 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
8295 Insert on edges rather than explicitly creating landing pads.
8296 (analyze_candidates_and_replace): Commit edge inserts.
8298 2017-08-08 Richard Biener <rguenther@suse.de>
8301 * tree-ssa-loop-niter.c: Include tree-dfa.h.
8302 (expand_simple_operations): Also look through ADDR_EXPRs with
8303 MEM_REF bases treating them as POINTER_PLUS_EXPR.
8305 2017-08-08 Richard Biener <rguenther@suse.de>
8307 PR tree-optimization/81723
8308 * tree-vect-slp.c (struct bst_traits): New hash traits.
8309 (bst_fail): New global.
8310 (vect_build_slp_tree_2): New worker, split out from ...
8311 (vect_build_slp_tree): ... this now wrapping it with using
8312 bst_fail set to cache SLP tree build fails. Properly handle
8314 (vect_analyze_slp_instance): Allocate and free bst_fail.
8316 2017-08-08 Martin Liska <mliska@suse.cz>
8319 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
8320 LABEL_DECLs that can be from a different function.
8322 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8324 PR tree-optimization/81744
8325 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
8326 loop's number of iterations.
8328 2017-08-08 Martin Liska <mliska@suse.cz>
8330 * asan.c: Include header files.
8331 * attribs.c (build_decl_attribute_variant): New function moved
8333 (build_type_attribute_qual_variant): Likewise.
8334 (cmp_attrib_identifiers): Likewise.
8335 (simple_cst_list_equal): Likewise.
8336 (omp_declare_simd_clauses_equal): Likewise.
8337 (attribute_value_equal): Likewise.
8338 (comp_type_attributes): Likewise.
8339 (build_type_attribute_variant): Likewise.
8340 (lookup_ident_attribute): Likewise.
8341 (remove_attribute): Likewise.
8342 (merge_attributes): Likewise.
8343 (merge_type_attributes): Likewise.
8344 (merge_decl_attributes): Likewise.
8345 (merge_dllimport_decl_attributes): Likewise.
8346 (handle_dll_attribute): Likewise.
8347 (attribute_list_equal): Likewise.
8348 (attribute_list_contained): Likewise.
8349 * attribs.h (lookup_attribute): New function moved from tree.[ch].
8350 (lookup_attribute_by_prefix): Likewise.
8351 * bb-reorder.c: Include header files.
8352 * builtins.c: Likewise.
8353 * calls.c: Likewise.
8354 * cfgexpand.c: Likewise.
8355 * cgraph.c: Likewise.
8356 * cgraphunit.c: Likewise.
8357 * convert.c: Likewise.
8358 * dwarf2out.c: Likewise.
8359 * final.c: Likewise.
8360 * fold-const.c: Likewise.
8361 * function.c: Likewise.
8362 * gimple-expr.c: Likewise.
8363 * gimple-fold.c: Likewise.
8364 * gimple-pretty-print.c: Likewise.
8365 * gimple.c: Likewise.
8366 * gimplify.c: Likewise.
8367 * hsa-common.c: Likewise.
8368 * hsa-gen.c: Likewise.
8369 * internal-fn.c: Likewise.
8370 * ipa-chkp.c: Likewise.
8371 * ipa-cp.c: Likewise.
8372 * ipa-devirt.c: Likewise.
8373 * ipa-fnsummary.c: Likewise.
8374 * ipa-inline.c: Likewise.
8375 * ipa-visibility.c: Likewise.
8377 * lto-cgraph.c: Likewise.
8378 * omp-expand.c: Likewise.
8379 * omp-general.c: Likewise.
8380 * omp-low.c: Likewise.
8381 * omp-offload.c: Likewise.
8382 * omp-simd-clone.c: Likewise.
8383 * opts-global.c: Likewise.
8384 * passes.c: Likewise.
8385 * predict.c: Likewise.
8386 * sancov.c: Likewise.
8387 * sanopt.c: Likewise.
8388 * symtab.c: Likewise.
8389 * toplev.c: Likewise.
8390 * trans-mem.c: Likewise.
8391 * tree-chkp.c: Likewise.
8392 * tree-eh.c: Likewise.
8393 * tree-into-ssa.c: Likewise.
8394 * tree-object-size.c: Likewise.
8395 * tree-parloops.c: Likewise.
8396 * tree-profile.c: Likewise.
8397 * tree-ssa-ccp.c: Likewise.
8398 * tree-ssa-live.c: Likewise.
8399 * tree-ssa-loop.c: Likewise.
8400 * tree-ssa-sccvn.c: Likewise.
8401 * tree-ssa-structalias.c: Likewise.
8402 * tree-ssa.c: Likewise.
8403 * tree-streamer-in.c: Likewise.
8404 * tree-vectorizer.c: Likewise.
8405 * tree-vrp.c: Likewise.
8407 * ubsan.c: Likewise.
8408 * varasm.c: Likewise.
8409 * varpool.c: Likewise.
8410 * tree.c: Remove functions moved to attribs.[ch].
8412 * config/aarch64/aarch64.c: Add attrs.h header file.
8413 * config/alpha/alpha.c: Likewise.
8414 * config/arc/arc.c: Likewise.
8415 * config/arm/arm.c: Likewise.
8416 * config/avr/avr.c: Likewise.
8417 * config/bfin/bfin.c: Likewise.
8418 * config/c6x/c6x.c: Likewise.
8419 * config/cr16/cr16.c: Likewise.
8420 * config/cris/cris.c: Likewise.
8421 * config/darwin.c: Likewise.
8422 * config/epiphany/epiphany.c: Likewise.
8423 * config/fr30/fr30.c: Likewise.
8424 * config/frv/frv.c: Likewise.
8425 * config/ft32/ft32.c: Likewise.
8426 * config/h8300/h8300.c: Likewise.
8427 * config/i386/winnt.c: Likewise.
8428 * config/ia64/ia64.c: Likewise.
8429 * config/iq2000/iq2000.c: Likewise.
8430 * config/lm32/lm32.c: Likewise.
8431 * config/m32c/m32c.c: Likewise.
8432 * config/m32r/m32r.c: Likewise.
8433 * config/m68k/m68k.c: Likewise.
8434 * config/mcore/mcore.c: Likewise.
8435 * config/microblaze/microblaze.c: Likewise.
8436 * config/mips/mips.c: Likewise.
8437 * config/mmix/mmix.c: Likewise.
8438 * config/mn10300/mn10300.c: Likewise.
8439 * config/moxie/moxie.c: Likewise.
8440 * config/msp430/msp430.c: Likewise.
8441 * config/nds32/nds32-isr.c: Likewise.
8442 * config/nds32/nds32.c: Likewise.
8443 * config/nios2/nios2.c: Likewise.
8444 * config/nvptx/nvptx.c: Likewise.
8445 * config/pa/pa.c: Likewise.
8446 * config/pdp11/pdp11.c: Likewise.
8447 * config/powerpcspe/powerpcspe.c: Likewise.
8448 * config/riscv/riscv.c: Likewise.
8449 * config/rl78/rl78.c: Likewise.
8450 * config/rx/rx.c: Likewise.
8451 * config/s390/s390.c: Likewise.
8452 * config/sh/sh.c: Likewise.
8453 * config/sol2.c: Likewise.
8454 * config/sparc/sparc.c: Likewise.
8455 * config/spu/spu.c: Likewise.
8456 * config/stormy16/stormy16.c: Likewise.
8457 * config/tilegx/tilegx.c: Likewise.
8458 * config/tilepro/tilepro.c: Likewise.
8459 * config/v850/v850.c: Likewise.
8460 * config/vax/vax.c: Likewise.
8461 * config/visium/visium.c: Likewise.
8462 * config/xtensa/xtensa.c: Likewise.
8464 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8467 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
8468 constraints since the -mupper-regs-* switches have been
8470 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
8471 into a vector from a double word element that was extracted from
8472 another vector, and eliminate extra XXPERMDI instructions.
8473 (vsx_concat_<mode>_2): Likewise.
8474 (vsx_concat_<mode>_3): Likewise.
8475 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
8476 concat to allow optimizing inserts from previous extracts.
8478 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
8480 * config/i386/i386.c (ix86_stack_protect_guard): Generate
8481 memory reference to a SSP offset in TLS address space.
8482 (ix86_print_operand) <case '@'>: Remove.
8483 (ix86_print_operand_punct_valid_p): Remove '@' code.
8484 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
8486 (stack_tls_protect_set_<mode>): Remove.
8487 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
8488 (stack_tls_protect_test_<mode>): Remove.
8489 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
8491 2017-08-07 Olivier Hainque <hainque@adacore.com>
8494 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
8496 2017-08-07 Douglas Rupp <rupp@adacore.com>
8498 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
8499 variable was referenced as multidir in command.
8501 2017-08-07 Jakub Jelinek <jakub@redhat.com>
8504 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
8507 2017-08-07 Martin Liska <mliska@suse.cz>
8509 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
8510 * config/rl78/rl78.c: Add include of attribs.h.
8511 * config/sh/sh.c: Likewise.
8512 * config/v850/v850.c: Likewise.
8514 2017-08-07 Tom de Vries <tom@codesourcery.com>
8517 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
8519 2017-08-07 Martin Liska <mliska@suse.cz>
8521 * config/mips/mips.c: Include attribs.h.
8523 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
8526 * doc/invoke.texi: Document change in behvaior for -Ofast for
8529 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
8531 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
8533 (aarch64_pop_regs): Likewise.
8534 (aarch64_gen_load_pair): Likewise.
8535 (aarch64_save_callee_saves): Likewise.
8536 (aarch64_restore_callee_saves): Likewise.
8538 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8540 * config/i386/i386.c: Revert the last change.
8542 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8545 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
8547 (ix86_finalize_stack_frame_flags): This. Also clear
8548 frame_pointer_needed if -fno-omit-frame-pointer is used without
8550 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
8551 with ix86_finalize_stack_frame_flags.
8552 (ix86_expand_epilogue): Likewise.
8553 (ix86_expand_split_stack_prologue): Likewise.
8555 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8558 * config/i386/i386.c (get_builtin_code_for_version): Set priority
8559 to P_AES for Westmere.
8561 2017-08-07 Jonathan Yong <10walls@gmail.com>
8563 * config/i386/mingw.opt (fset-stack-executable): Removed.
8564 * config/i386/cygming.opt (fset-stack-executable): Moved
8566 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
8568 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
8570 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
8572 2017-08-07 Marek Polacek <polacek@redhat.com>
8575 * fold-const.c (fold_indirect_ref_1): Check type_domain.
8577 2017-08-07 Martin Liska <mliska@suse.cz>
8579 * attribs.h (canonicalize_attr_name): New function.
8580 (cmp_attribs): Move from c-format.c and adjusted.
8581 (is_attribute_p): Moved from tree.h.
8582 * tree-inline.c: Add new includes.
8583 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
8584 (private_is_attribute_p): Remove.
8585 (private_lookup_attribute): Likewise.
8586 (private_lookup_attribute_by_prefix): Simplify.
8587 (remove_attribute): Use is_attribute_p.
8588 * tree.h: Remove removed declarations.
8590 2017-08-07 Jakub Jelinek <jakub@redhat.com>
8593 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
8594 instead of computing it in the function. Formatting fix.
8595 (expand_case): Don't rely on default_edge being the first edge,
8596 clear it if removing it, pass default_edge to
8597 emit_case_dispatch_table.
8598 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
8601 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
8603 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
8604 insn in the function, emit NOP after the insn.
8606 2017-08-06 Tom de Vries <tom@codesourcery.com>
8608 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
8611 2017-08-06 Tom de Vries <tom@codesourcery.com>
8613 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
8616 2017-08-04 Yury Gribov <tetra2005@gmail.com>
8618 PR tree-optimization/57371
8619 * match.pd: New pattern.
8621 2017-08-04 Marek Polacek <polacek@redhat.com>
8624 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
8625 perform the computation in offset_int.
8627 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8629 PR tree-optimization/81136
8630 * tree-vectorizer.h: Include tree-hash-traits.h.
8631 (vec_base_alignments): New typedef.
8632 (vec_info): Add a base_alignments field.
8633 (vect_record_base_alignments): Declare.
8634 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
8636 (DR_IS_CONDITIONAL_IN_STMT): New macro.
8637 (create_data_ref): Add an is_conditional_in_stmt argument.
8638 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
8639 the is_conditional_in_stmt field.
8640 (data_ref_loc): Add an is_conditional_in_stmt field.
8641 (get_references_in_stmt): Set the is_conditional_in_stmt field.
8642 (find_data_references_in_stmt): Update call to create_data_ref.
8643 (graphite_find_data_references_in_stmt): Likewise.
8644 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
8645 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
8646 (vect_record_base_alignment): New function.
8647 (vect_record_base_alignments): Likewise.
8648 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
8649 for nested statements even if we fail to compute a misalignment.
8650 Use pooled base alignments for unconditional references.
8651 (vect_find_same_alignment_drs): Compare base addresses instead
8653 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
8654 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
8656 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8658 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
8659 Add an explicit name for the enum. Use auto_vec for slp_instances
8661 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
8663 (_bb_vec_info): Add a constructor and destructor.
8664 (vinfo_for_stmt): Return NULL for uids of -1 as well.
8665 (destroy_loop_vec_info): Delete.
8666 (vect_destroy_datarefs): Likewise.
8667 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
8668 (vec_info::vec_info): New function.
8669 (vec_info::~vec_info): Likewise.
8670 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
8671 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
8672 destroy_loop_vec_info.
8673 * tree-vect-loop.c (new_loop_vec_info): Replace with...
8674 (_loop_vec_info::_loop_vec_info): ...this.
8675 (destroy_loop_vec_info): Replace with...
8676 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
8677 the stmt_vec_infos. Leave handling of vec_info information to its
8678 destructor. Remove explicit vector releases.
8679 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
8680 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
8681 * tree-vect-slp.c (new_bb_vec_info): Replace with...
8682 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
8683 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
8684 (destroy_bb_vec_info): Replace with...
8685 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
8686 information to its destructor.
8687 (vect_slp_analyze_bb_1): Use new and delete instead of
8688 new_bb_vec_info and destroy_bb_vec_info.
8689 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
8692 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8694 * tree-data-ref.h (subscript): Add access_fn field.
8695 (data_dependence_relation): Add could_be_independent_p.
8696 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
8697 (same_access_functions): Move to tree-data-ref.c.
8698 * tree-data-ref.c (ref_contains_union_access_p): New function.
8699 (access_fn_component_p): Likewise.
8700 (access_fn_components_comparable_p): Likewise.
8701 (dr_analyze_indices): Add a reference to access_fn_component_p.
8702 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
8704 (constant_access_functions): Likewise.
8705 (add_other_self_distances): Likewise.
8706 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
8707 (initialize_data_dependence_relation): Use XCNEW and remove
8708 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
8709 of access functions that have the same type. Allow the
8710 subsequence to end with different bases in some circumstances.
8711 Record the chosen access functions in SUB_ACCESS_FN.
8712 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
8713 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
8714 (subscript_dependence_tester_1): Likewise dra and drb.
8715 (build_classic_dist_vector): Update calls accordingly.
8716 (subscript_dependence_tester): Likewise.
8717 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
8718 DDR_COULD_BE_INDEPENDENT_P.
8719 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
8720 comp_alias_ddrs instead of may_alias_ddrs.
8721 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
8723 (vect_analyze_data_ref_dependence): Use it if
8724 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
8725 distance vectors if that fails.
8726 (dependence_distance_ge_vf): New function.
8727 (vect_prune_runtime_alias_test_list): Use it. Don't clear
8728 LOOP_VINFO_MAY_ALIAS_DDRS.
8730 2017-08-04 Richard Biener <rguenther@suse.de>
8733 * fold-const.c (fold_binary_loc): Properly restrict
8734 minus_var0 && minus_var1 case when associating undefined overflow
8737 2017-08-04 Tom de Vries <tom@codesourcery.com>
8739 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
8741 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8743 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8744 Don't start diagnostic messages with a capital letter.
8745 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8747 (rs6000_invalid_builtin): Likewise.
8748 (rs6000_trampoline_init): Likewise.
8750 2017-08-03 Jakub Jelinek <jakub@redhat.com>
8753 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
8754 after setting changeable df flags.
8756 2017-08-03 Richard Biener <rguenther@suse.de>
8758 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
8759 up if the use is in USE - X.
8761 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
8763 * toplev.c (dumpfile.h): New include.
8764 (internal_error_reentered): New static function. Use it...
8765 (internal_error_function): ...here to handle reentered internal_error.
8767 2017-08-03 Richard Biener <rguenther@suse.de>
8770 * fold-const.c (split_tree): Add minus_var and minus_con
8771 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
8772 here but always use minus_*.
8773 (associate_trees): Assert we never associate with MINUS_EXPR
8774 and NULL first operand. Do not recurse for PLUS_EXPR operands
8775 when associating as MINUS_EXPR either.
8776 (fold_binary_loc): Track minus_var and minus_con.
8778 2017-08-03 Tom de Vries <tom@codesourcery.com>
8781 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
8782 ACCEL_COMPILER, apply finish_options on
8783 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
8785 2017-08-03 Tom de Vries <tom@codesourcery.com>
8788 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
8789 function_entry_patch_area_size > 0.
8791 2017-08-03 Jakub Jelinek <jakub@redhat.com>
8794 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
8795 instead of 10??LU, perform unit multiplication in wide_int,
8796 don't change alloc_object_size_limit if the limit is larger
8799 PR tree-optimization/81655
8800 PR tree-optimization/81588
8801 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
8802 the case when ranges[i].low and high are 1 for unsigned type with
8806 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
8807 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
8809 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8811 * tree-vrp.h: Add include guard.
8813 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
8816 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
8817 (ud2): New insn pattern.
8818 * config/i386/i386.c (ix86_expand_epilogue):
8819 For naked functions, generate ud2 instead of trap insn.
8821 2017-08-02 Marek Polacek <polacek@redhat.com>
8824 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
8826 2017-08-02 Tom de Vries <tom@codesourcery.com>
8827 Cesar Philippidis <cesar@codesourcery.com>
8829 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
8830 Add missing edge probabilities.
8832 2017-08-02 Tamar Christina <tamar.christina@arm.com>
8834 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
8837 2017-08-02 Jakub Jelinek <jakub@redhat.com>
8840 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
8841 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
8842 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
8843 prologue_seq sequences - if any.
8845 2017-08-02 Richard Biener <rguenther@suse.de>
8847 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
8848 via vectors if supported, integer extracts via punning if supported
8849 or otherwise vector extracts.
8851 2017-08-02 Richard Biener <rguenther@suse.de>
8853 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
8855 (bitmap_insert_into_set): ... this.
8857 2017-08-02 Richard Biener <rguenther@suse.de>
8859 PR tree-optimization/81633
8861 2015-08-17 Alan Hayward <alan.hayward@arm.com>
8863 PR tree-optimization/71752
8864 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
8866 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
8868 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
8869 (machine_function::call_ms2sysv_pad_out): Remove field.
8870 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
8871 (ix86_compute_frame_layout): Likewise.
8873 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
8876 * config/i386/i386.c (ix86_set_func_type): Disallow naked
8877 attribute with interrupt attribute.
8879 2017-08-01 Andrew Pinski <apinski@cavium.com>
8881 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
8882 BIT_INSERT_EXPR's operand 1
8883 to see if the types precision matches.
8885 2017-08-01 Martin Liska <mliska@suse.cz>
8888 * builtins.c (expand_builtin_memcpy_args): Remove.
8889 (expand_builtin_memcpy): Call newly added function
8890 expand_builtin_memory_copy_args.
8891 (expand_builtin_memcpy_with_bounds): Likewise.
8892 (expand_builtin_mempcpy): Remove last argument.
8893 (expand_builtin_mempcpy_with_bounds): Likewise.
8894 (expand_builtin_memory_copy_args): New function created from
8895 expand_builtin_mempcpy_args with small modifications.
8896 (expand_builtin_mempcpy_args): Remove.
8897 (expand_builtin_stpcpy): Remove unused argument.
8898 (expand_builtin): Likewise.
8899 (expand_builtin_with_bounds): Likewise.
8901 2017-08-01 Martin Liska <mliska@suse.cz>
8904 Make mempcpy more optimal (PR middle-end/70140).
8906 2017-08-01 Jakub Jelinek <jakub@redhat.com>
8909 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
8910 __builtin_vec_cmpne verify both arguments are compatible vectors
8911 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
8912 verify arg1_type is a pointer or array type. For __builtin_vec_st,
8913 move computation of aligned to after checking the argument types.
8917 * config/rs6000/vsx.md (vextract_fp_from_shorth,
8918 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
8921 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
8922 Doug Rupp <rupp@adacore.com>
8923 Olivier Hainque <hainque@adacore.com>
8925 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
8926 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
8928 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
8929 for TARGET_OS_CPP_BUILTIN.
8930 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
8931 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
8933 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
8934 passing required abi options to the assembler for EABI configurations.
8935 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
8936 of .text.hot and .text.unlikely sections for kernel modules when
8937 using ARM style exceptions.
8938 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
8939 options. Add EXTRA_CC1_SPEC.
8940 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
8942 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
8944 (ARM_TARGET2_DWARF_FORMAT): Define.
8945 * config/arm/t-vxworks: Adjust multilib control to removal of the
8946 Diab command line options.
8948 2017-08-01 Martin Liska <mliska@suse.cz>
8950 PR gcov-profile/81561
8951 * gcov.c (unblock): Make unblocking safe as we need to preserve
8952 index correspondence of blocks and block_lists.
8954 2017-08-01 Richard Biener <rguenther@suse.de>
8956 PR tree-optimization/81181
8957 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8958 (compute_antic): ... end of iteration here.
8960 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
8962 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
8963 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
8964 (ftree-slp-vectorize): Likewise.
8965 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
8966 can no longer be set independent of flag_tree_loop_vectorize.
8967 * omp-general.c (emp_max_vf): Likewise.
8968 * opts.c (enable_fdo_optimizations): Remove references to
8969 flag_tree_vectorize, these are now implicit.
8970 (common_handle_option): Remove handling for OPT_ftree_vectorize,
8971 and leave it for the options machinery.
8973 2017-08-01 Martin Liska <mliska@suse.cz>
8976 * builtins.c (expand_builtin_memcpy_args): Remove.
8977 (expand_builtin_memcpy): Call newly added function
8978 expand_builtin_memory_copy_args.
8979 (expand_builtin_memcpy_with_bounds): Likewise.
8980 (expand_builtin_mempcpy): Remove last argument.
8981 (expand_builtin_mempcpy_with_bounds): Likewise.
8982 (expand_builtin_memory_copy_args): New function created from
8983 expand_builtin_mempcpy_args with small modifications.
8984 (expand_builtin_mempcpy_args): Remove.
8985 (expand_builtin_stpcpy): Remove unused argument.
8986 (expand_builtin): Likewise.
8987 (expand_builtin_with_bounds): Likewise.
8989 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
8992 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
8993 print "ds:" only for immediates in generic address space.
8995 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
8998 * config/i386/i386.c (ix86_funciton_naked): New prototype.
8999 (ix86_function_ok_for_sibcall): Return false for naked functions.
9001 2017-08-01 Richard Biener <rguenther@suse.de>
9003 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
9004 (compute_antic): Seed worklist with exit block predecessors.
9005 * cfganal.c (dfs_find_deadend): For a cycle return the source
9006 of the edge closing it.
9008 2017-08-01 Tamar Christina <tamar.christina@arm.com>
9010 * config/aarch64/aarch64.c
9011 (aarch64_can_const_movi_rtx_p): Move 0 check.
9013 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9015 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
9016 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
9018 * match.pd: Ditto in address comparison pattern.
9020 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9022 PR tree-optimization/81627
9023 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
9024 closed ssa form for store-store chain.
9026 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9028 PR tree-optimization/81620
9029 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
9030 for store-store chain.
9032 2017-08-01 Jakub Jelinek <jakub@redhat.com>
9034 PR tree-optimization/81588
9035 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
9036 ranges[i].in_p, invert comparison code ccode. For >/>=,
9037 swap rhs1 and rhs2 and comparison code unconditionally,
9038 for </<= don't do that. Don't swap rhs1/rhs2 again if
9039 ranges[i].in_p, instead invert comparison code ccode if
9040 opcode or oe->rank is BIT_IOR_EXPR.
9043 * optabs.def (vec_extract_optab, vec_init_optab): Change from
9044 a direct optab to conversion optab.
9045 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
9046 with GET_MODE_INNER as last argument instead of optab_handler.
9047 * expmed.c (extract_bit_field_1): Likewise. Use vector from
9048 vector extraction if possible and optab is available.
9049 * expr.c (store_constructor): Use convert_optab_handler instead
9050 of optab_handler. Use vector initialization from smaller
9051 vectors if possible and optab is available.
9052 * tree-vect-stmts.c (vectorizable_load): Likewise.
9053 * doc/md.texi (vec_extract, vec_init): Document that the optabs
9055 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
9056 of vec_init from half-sized vectors with the same element mode.
9057 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
9058 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
9059 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
9060 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
9061 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
9062 after mode in gen_vec_extract* calls.
9063 (vec_extract<mode>): Renamed to ...
9064 (vec_extract<mode><ssescalarmodelower>): ... this.
9065 (vec_extract<mode><ssehalfvecmodelower>): New expander.
9066 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
9067 element mode after mode in gen_vec_init* calls.
9068 (VEC_INIT_HALF_MODE): New mode iterator.
9069 (vec_init<mode>): Renamed to ...
9070 (vec_init<mode><ssescalarmodelower>): ... this.
9071 (vec_init<mode><ssehalfvecmodelower>): New expander.
9072 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
9073 (vec_extractv2sfsf): ... this.
9074 (vec_initv2sf): Renamed to ...
9075 (vec_initv2sfsf): ... this.
9076 (vec_extractv2si): Renamed to ...
9077 (vec_extractv2sisi): ... this.
9078 (vec_initv2si): Renamed to ...
9079 (vec_initv2sisi): ... this.
9080 (vec_extractv4hi): Renamed to ...
9081 (vec_extractv4hihi): ... this.
9082 (vec_initv4hi): Renamed to ...
9083 (vec_initv4hihi): ... this.
9084 (vec_extractv8qi): Renamed to ...
9085 (vec_extractv8qiqi): ... this.
9086 (vec_initv8qi): Renamed to ...
9087 (vec_initv8qiqi): ... this.
9088 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
9089 (vec_init<mode>): Renamed to ...
9090 (vec_init<mode><VEC_base_l>): ... this.
9091 (vec_extract<mode>): Renamed to ...
9092 (vec_extract<mode><VEC_base_l>): ... this.
9093 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
9094 (vec_initv2sfsf): ... this.
9095 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
9096 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
9097 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
9098 element mode after mode in gen_vec_init* calls.
9099 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
9100 (vec_init<mode><Vel>): ... this.
9101 (vec_extract<mode>): Renamed to ...
9102 (vec_extract<mode><Vel>): ... this.
9103 * config/aarch64/iterators.md (Vel): New mode attribute.
9104 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
9105 Add element mode after mode in gen_vec_extract* calls.
9106 * config/s390/vector.md (non_vec_l): New mode attribute.
9107 (vec_extract<mode>): Renamed to ...
9108 (vec_extract<mode><non_vec_l>): ... this.
9109 (vec_init<mode>): Renamed to ...
9110 (vec_init<mode><non_vec_l>): ... this.
9111 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
9112 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
9114 * config/arm/iterators.md (V_elem_l): New mode attribute.
9115 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
9116 (vec_extract<mode><V_elem_l>): ... this.
9117 (vec_extractv2di): Renamed to ...
9118 (vec_extractv2didi): ... this.
9119 (vec_init<mode>): Renamed to ...
9120 (vec_init<mode><V_elem_l>): ... this.
9121 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
9122 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
9123 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
9124 Add element mode after gen_vec_extract* calls.
9125 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
9126 (vec_init<mode><unitmode>): ... this.
9127 (vec_extract<mode>): Renamed to ...
9128 (vec_extract<mode><unitmode>): ... this.
9129 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
9130 (vec_init<mode><unitmode>): ... this.
9131 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
9132 (vec_initv2sfsf): ... this.
9133 (vec_extractv2sf): Renamed to ...
9134 (vec_extractv2sfsf): ... this.
9135 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
9136 Add element mode after gen_vec_extract* calls.
9137 * config/mips/mips.md (unitmode): New mode iterator.
9138 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
9139 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
9140 * config/spu/spu.md (inner_l): New mode attribute.
9141 (vec_init<mode>): Renamed to ...
9142 (vec_init<mode><inner_l>): ... this.
9143 (vec_extract<mode>): Renamed to ...
9144 (vec_extract<mode><inner_l>): ... this.
9145 * config/sparc/sparc.md (veltmode): New mode iterator.
9146 (vec_init<VMALL:mode>): Renamed to ...
9147 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
9148 * config/ia64/vect.md (vec_initv2si): Renamed to ...
9149 (vec_initv2sisi): ... this.
9150 (vec_initv2sf): Renamed to ...
9151 (vec_initv2sfsf): ... this.
9152 (vec_extractv2sf): Renamed to ...
9153 (vec_extractv2sfsf): ... this.
9154 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
9155 (vec_init<mode>): Renamed to ...
9156 (vec_init<mode><VEC_base_l>): ... this.
9157 (vec_extract<mode>): Renamed to ...
9158 (vec_extract<mode><VEC_base_l>): ... this.
9159 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
9160 (vec_initv2sfsf): ... this.
9161 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
9162 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
9163 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
9164 gen_vec_init* calls.
9166 2017-08-01 Richard Biener <rguenther@suse.de>
9168 PR tree-optimization/81297
9169 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
9170 TREE_OVERFLOW from INTEGER_CSTs.
9172 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
9174 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
9176 2017-07-31 Carl Love <cel@us.ibm.com>
9178 * config/rs6000/rs6000-c: Add support for built-in functions
9179 vector signed char vec_xl_be (signed long long, signed char *);
9180 vector unsigned char vec_xl_be (signed long long, unsigned char *);
9181 vector signed int vec_xl_be (signed long long, signed int *);
9182 vector unsigned int vec_xl_be (signed long long, unsigned int *);
9183 vector signed long long vec_xl_be (signed long long, signed long long *);
9184 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
9185 vector signed short vec_xl_be (signed long long, signed short *);
9186 vector unsigned short vec_xl_be (signed long long, unsigned short *);
9187 vector double vec_xl_be (signed long long, double *);
9188 vector float vec_xl_be (signed long long, float *);
9189 * config/rs6000/altivec.h (vec_xl_be): Add #define.
9190 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
9191 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
9193 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
9194 (altivec_expand_builtin): Add switch statement to call
9195 altivec_expand_xl_be for each builtin.
9196 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
9197 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
9198 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
9199 __builtin_vsx_le_be_v16qi.
9200 * doc/extend.texi: Update the built-in documentation file for the
9201 new built-in functions.
9203 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
9206 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
9208 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
9210 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9212 * config.gcc: Add z14.
9213 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
9214 CPU model numbers for z13s and z14.
9215 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
9217 * config/s390/s390-opts.h (enum processor_type): Rename
9218 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
9219 * config/s390/s390.c (processor_table): Add field for CPU name to
9220 be passed to Binutils.
9221 (s390_asm_output_machine_for_arch): Use the new field in
9222 processor_table for Binutils.
9223 (s390_expand_builtin): Replace arch12 with z14.
9224 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
9225 (s390_get_sched_attrmask): Likewise.
9226 (s390_get_unit_mask): Likewise.
9227 * config/s390/s390.opt: Add z14 to processor_type enum.
9229 2017-07-31 Martin Jambor <mjambor@suse.cz>
9232 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
9233 regardless of optimization level.
9235 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
9236 Martin Liska <mliska@suse.cz>
9238 * predict.def: Remove old comment and adjust probability.
9239 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
9242 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
9245 * config/i386/i386.c (ix86_function_naked): New function.
9246 (ix86_can_use_return_insn_p): Return false for naked functions.
9247 (ix86_expand_prologue): Skip prologue for naked functions.
9248 (ix86_expand_epilogue): Skip epilogue for naked functions
9249 and emit trap instruction.
9250 (ix86_warn_func_return): New function.
9251 (ix86_attribute_table): Add "naked" attribute specification.
9252 (TARGET_WARN_FUNC_RETURN): Define.
9253 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
9255 2017-07-31 Martin Liska <mliska@suse.cz>
9257 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
9258 (dump_gimple_bb_header): Always dump BB info.
9259 (pp_cfg_jump): Do not append info about BB when dumping a jump.
9261 2017-07-31 Martin Liska <mliska@suse.cz>
9264 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
9265 also with current_function_decl non-null equality.
9267 2017-07-31 Jakub Jelinek <jakub@redhat.com>
9270 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
9271 change type to the element type, instead add eltype variable and
9272 use it where we are interested in the element type.
9274 PR tree-optimization/81603
9275 * ipa-polymorphic-call.c
9276 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
9277 offset arithmetic in offset_int, bail out if the resulting bit offset
9278 doesn't fit into shwi.
9280 2017-07-31 Martin Liska <mliska@suse.cz>
9282 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
9283 (gimplify_save_expr): Fix comment.
9285 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
9288 * config/i386/i386.c (ix86_function_arg): Update arguments for
9290 (ix86_compute_frame_layout): Set the initial stack offset to
9291 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
9292 INCOMING_FRAME_SP_OFFSET.
9293 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
9294 stack before exception handler returns.
9295 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
9296 the 'ERROR_CODE' for exception handler.
9298 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
9300 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
9301 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
9302 (ASM_OUTPUT_REG_POP): Ditto.
9303 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
9304 instead of asm_fprintf to output pure string.
9306 2017-07-29 Jakub Jelinek <jakub@redhat.com>
9308 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
9309 to imported_module_or_decl hook.
9310 (debug_nothing_tree_tree_tree_bool): Remove.
9311 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
9312 * debug.c (do_nothing_debug_hooks): Use
9313 debug_nothing_tree_tree_tree_bool_bool instead of
9314 debug_nothing_tree_tree_tree_bool.
9315 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
9316 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
9317 * sdbout.c (sdb_debug_hooks): Likewise.
9318 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
9319 (gen_namespace_die): Add DW_AT_export_symbols attribute if
9321 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
9322 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
9323 attribute, don't add anything.
9325 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9327 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
9328 (fold_build2_stat_loc): Likewise.
9329 (fold_build3_stat_loc): Likewise.
9330 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
9331 (fold_build1_loc): Remove macro.
9332 (fold_build2_loc): Likewise.
9333 (fold_build3_loc): Likewise.
9335 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9337 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
9338 (gimple_build_debug_bind_source_stat): Likewise.
9339 * gimple.h (gimple_build_debug_bind): Remove macro.
9340 (gimple_build_debug_bind_source): Likewise.
9342 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9344 * bitmap.c (bitmap_alloc): Adjust.
9345 (bitmap_gc_alloc): Likewise.
9346 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
9348 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9350 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
9351 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
9352 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
9353 (bitmap_gc_alloc_stat): Likewise.
9354 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
9356 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9358 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
9359 * rtl.h (shallow_copy_rtx): Remove macro.
9361 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9363 * emit-rtl.c (gen_raw_REG): Adjust.
9364 * gengenrtl.c (gendef): Likewise.
9365 * rtl.c (rtx_alloc_stat): Remove _stat from name.
9366 * rtl.h (rtx_alloc): Remove macro.
9368 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9370 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
9371 (build_tree_list_stat): Likewise.
9372 * tree.h (build_tree_list): Remove macro.
9373 (build_tree_list_vec): Likewise.
9375 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9377 * tree.c (make_vector_stat): Remove _stat from name.
9378 (build_vector_stat): Likewise.
9379 * tree.h (make_vector_stat): Remove macro.
9380 (build_vector_stat): Likewise.
9382 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9384 * tree.h (build_var_debug_value): Remove prototype.
9386 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9388 * tree.c (tree_cons_stat): Remove _stat from name.
9389 * tree.h (tree_cons): Remove macro.
9391 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9393 * tree.c (build_vl_exp_stat): Remove _stat from name.
9394 * tree.h (build_vl_exp): Remove macro.
9396 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9398 * tree.c (build_decl_stat): Remove _stat from name.
9399 * tree.h (build_decl): Remove macro.
9401 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9403 * gimple.c (gimple_build_with_ops_stat): Adjust.
9404 (gimple_alloc_stat): Remove _stat from name.
9405 * gimple.h (gimple_alloc): Remove macro.
9407 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9409 * tree.c (make_tree_vec_stat): Remove _stat from name.
9410 (grow_tree_vec_stat): Likewise.
9411 * tree.h (make_tree_vec_stat): Adjust prototype.
9412 (grow_tree_vec_stat): Likewise.
9413 (make_tree_vec): Remove macro.
9414 (grow_tree_vec): Likewise.
9416 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9418 * fold-const.c (fold_build1_stat_loc): Adjust.
9419 (fold_build2_stat_loc): Likewise.
9420 (fold_build3_stat_loc): Likewise.
9421 * tree.c (build0_stat): Remove _stat from name.
9422 (build1_stat): Likewise.
9423 (build2_stat): Likewise.
9424 (build3_stat): Likewise.
9425 (build4_stat): Likewise.
9426 (build5_stat): Likewise.
9427 * tree.h (build1_loc): Remove macro, and rename _stat function
9429 (build2_loc): Likewise.
9430 (build3_loc): Likewise.
9431 (build4_loc): Likewise.
9432 (build5_loc): Likewise.
9434 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9436 * tree.c (make_int_cst_stat): Remove _stat from name.
9437 * tree.h (make_int_cst_stat): Adjust prototype.
9438 (make_int_cst): Remove macro.
9440 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9442 * tree.c (make_tre_binfo_stat): Remove _stat from name.
9443 * tree.h (make_tree_binfo_stat): Adjust prototype.
9444 (make_tree_binfo): Remove.
9446 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9448 * tree.c (copy_node_stat): Rename to copy_node.
9449 (build_distinct_type_copy): Adjust.
9450 * tree.h (copy_node_stat): Adjust prototype.
9451 (copy_node): Remove macro.
9453 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9455 * tree.c (make_node_stat): rename to make_node.
9456 (build_tree_list_stat): Adjust.
9457 (build0_stat): Likewise.
9458 (build2_stat): Likewise.
9459 (build3_stat): Likewise.
9460 (build4_stat): Likewise.
9461 (build5_stat): Likewise.
9462 (build_decl_stat): Likewise.
9463 * tree.h (make_node_stat): Adjust prototype.
9464 (make_node): remove macro.
9466 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
9468 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
9469 (PPC_FEATURE2_SCV): Likewise.
9470 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
9472 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9474 * config/aarch64/aarch64.c
9475 (aarch64_internal_mov_immediate): Add new special pattern.
9476 * config/aarch64/aarch64.md (*movdi_aarch64):
9477 Add reg/32bit const mov case.
9479 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9480 Richard Sandiford <richard.sandiford@linaro.org>
9482 * config/aarch64/aarch64.md (mov<mode>): Generalize.
9483 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
9484 Add integer and movi cases.
9485 (movi-split-hf-df-sf split, fp16): New.
9486 (enabled): Added TARGET_FP_F16INST.
9487 * config/aarch64/iterators.md (GPF_HF): New.
9488 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
9490 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9492 * config/aarch64/aarch64.c
9493 (aarch64_simd_container_mode): Add prototype.
9494 (aarch64_expand_mov_immediate): Add HI support.
9495 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
9496 (aarch64_can_const_movi_rtx_p): New.
9497 (aarch64_preferred_reload_class):
9498 Remove restrictions of using FP registers for certain SIMD operations.
9499 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
9500 (aarch64_valid_floating_const): Add integer move validation.
9501 (aarch64_simd_imm_scalar_p): Remove.
9502 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
9503 (aarch64_legitimate_constant_p): Expand list of supported cases.
9504 * config/aarch64/aarch64-protos.h
9505 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
9506 (aarch64_reinterpret_float_as_int): New.
9507 (aarch64_simd_imm_scalar_p): Remove.
9508 * config/aarch64/constraints.md (Uvi): New.
9509 (Dd): Split into Ds and new Dd.
9510 * config/aarch64/aarch64.md (*movsi_aarch64):
9512 (*movdi_aarch64): Add SIMD mov case.
9514 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9516 * tree-predcom.c: (struct chain): Handle store-store chain in which
9517 stores for elimination only store loop invariant values.
9518 (execute_pred_commoning_chain): Ditto.
9519 (prepare_initializers_chain_store_elim): Ditto.
9520 (prepare_finalizers): Ditto.
9521 (is_inv_store_elimination_chain): New function.
9522 (initialize_root_vars_store_elim_1): New function.
9524 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9526 * tree-predcom.c: Revise general description of the pass.
9527 (enum chain_type): New enum type for store elimination.
9528 (struct chain): New field supporting store elimination.
9529 (struct component): Ditto.
9530 (dump_chain): Dump store-stores chain.
9531 (release_chain): Release resources.
9532 (split_data_refs_to_components): Compute and create component
9533 contains only stores for elimination.
9534 (get_chain_last_ref_at): New function.
9535 (make_invariant_chain): Initialization.
9536 (make_rooted_chain): Specify chain type in parameter and record it.
9537 (add_looparound_copies): Skip for store-stores chain.
9538 (determine_roots_comp): Compute type of chain and pass it to
9540 (initialize_root_vars_store_elim_2): New function.
9541 (finalize_eliminated_stores): New function.
9542 (remove_stmt): Handle store for elimination.
9543 (execute_pred_commoning_chain): Execute predictive commoning on
9545 (determine_unroll_factor): Skip unroll for store-stores chain.
9546 (prepare_initializers_chain_store_elim): New function.
9547 (prepare_initializers_chain): Hanlde store-store chain.
9548 (prepare_finalizers_chain, prepare_finalizers): New function.
9549 (tree_predictive_commoning_loop): Return integer value indicating
9550 if loop is unrolled or lcssa form is corrupted.
9551 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
9553 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9555 * tree-predcom.c (initialize_root): Delete.
9556 (execute_pred_commoning_chain): Initialize root vars and replace
9557 reference of non-combined chain directly, rather than call above
9560 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9562 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
9563 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
9565 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9567 * tree-predcom.c (struct chain): New field init_seq.
9568 (release_chain): Release init_seq.
9569 (prepare_initializers_chain): Record intialization stmts in above
9571 (insert_init_seqs): New function.
9572 (tree_predictive_commoning_loop): Call insert_init_seqs.
9574 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9576 * tree-predcom.c (determine_roots_comp): Skip trivial components.
9578 2017-07-28 Richard Biener <rguenther@suse.de>
9580 * match.pd: Remove superfluous :c.
9581 * genmatch.c (simplify::id): Add member.
9582 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
9584 (current_id): New global.
9585 (dt_node::parent): Move from ...
9586 (dt_operand::parent): ... here. Add for_id member.
9587 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
9588 (decision_tree::find_node): Relax order requirement when
9589 merging DT_TRUE nodes to ones inbetween the current simplify
9590 and the one we try to merge with. Add diagnostic whenever
9591 we need to enforce pattern order by not merging.
9592 (decision_tree::insert): Set current_id.
9593 (decision_tree::print_node): Dump parent node and for_id.
9594 (parser::last_id): Add member.
9595 (parser::push_simplify): Assign unique id.
9596 (parser::parser): Initialize last_id.
9598 2017-07-28 Martin Liska <mliska@suse.cz>
9601 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
9602 gimple_build_debug_bind.
9604 2017-07-28 Richard Biener <rguenther@suse.de>
9606 PR tree-optimization/81502
9607 * match.pd: Add pattern combining BIT_INSERT_EXPR with
9609 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
9611 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
9612 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
9613 for BIT_FIELD_REF args.
9614 * fold-const.c (make_bit_field_ref): Likewise.
9615 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
9617 2017-07-28 Jakub Jelinek <jakub@redhat.com>
9620 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
9621 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
9622 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
9623 Or it into SANITIZER_UNDEFINED.
9624 * ubsan.c: Include gimple-fold.h and varasm.h.
9625 (ubsan_expand_ptr_ifn): New function.
9626 (instrument_pointer_overflow): New function.
9627 (maybe_instrument_pointer_overflow): New function.
9628 (instrument_object_size): Formatting fix.
9629 (pass_ubsan::execute): Call instrument_pointer_overflow
9630 and maybe_instrument_pointer_overflow.
9631 * internal-fn.c (expand_UBSAN_PTR): New function.
9632 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
9633 * sanitizer.def (__ubsan_handle_pointer_overflow,
9634 __ubsan_handle_pointer_overflow_abort): New builtins.
9635 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
9636 * internal-fn.def (UBSAN_PTR): New internal function.
9637 * opts.c (sanitizer_opts): Add pointer-overflow.
9638 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
9639 * fold-const.c (build_range_check): Compute pointer range check in
9640 integral type if pointer arithmetics would be needed. Formatting
9643 2017-07-28 Martin Liska <mliska@suse.cz>
9646 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
9647 parameters that are of a variable-length.
9649 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
9651 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
9653 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
9654 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
9655 (CRT_CALL_STATIC_FUNCTION): Likewise.
9656 (ASM_DEFAULT_SPEC): New define.
9657 (ASM_SPEC32): Likewise.
9658 (ASM_SPEC64): Likewise.
9659 (ASM_SPEC_COMMON): Likewise.
9660 (ASM_SPEC): Likewise.
9661 (INVALID_64BIT): Likewise.
9662 (LINK_OS_DEFAULT_SPEC): Likewise.
9663 (LINK_OS_SPEC32): Likewise.
9664 (LINK_OS_SPEC64): Likewise.
9665 (POWERPC_LINUX): Likewise.
9666 (PTRDIFF_TYPE): Likewise.
9667 (RESTORE_FP_PREFIX): Likewise.
9668 (RESTORE_FP_SUFFIX): Likewise.
9669 (SAVE_FP_PREFIX): Likewise.
9670 (SAVE_FP_SUFFIX): Likewise.
9671 (SIZE_TYPE): Likewise.
9672 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
9673 (TARGET_64BIT): Likewise.
9674 (TARGET_64BIT): Likewise.
9675 (TARGET_AIX): Likewise.
9676 (WCHAR_TYPE_SIZE): Likewise.
9677 (WCHAR_TYPE): Undefine.
9678 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
9679 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
9680 (CPP_OS_RTEMS_SPEC): Delete.
9681 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
9682 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
9684 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
9686 2017-07-28 Jakub Jelinek <jakub@redhat.com>
9688 PR tree-optimization/81578
9689 * tree-parloops.c (build_new_reduction): Bail out if
9690 reduction_code isn't one of the standard OpenMP reductions.
9691 Move the details printing after that decision.
9693 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9695 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
9696 related to reload_in_progress.
9697 (splat_input_operand): Likewise.
9698 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
9700 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
9702 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
9703 (TARGET_INSTANTIATE_DECLS): Likewise.
9704 (legitimate_indexed_address_p): Delete reload_in_progress code.
9705 (rs6000_debug_legitimate_address_p): Likewise.
9706 (rs6000_eliminate_indexed_memrefs): Likewise.
9707 (rs6000_emit_le_vsx_store): Likewise.
9708 (rs6000_emit_move_si_sf_subreg): Likewise.
9709 (rs6000_emit_move): Likewise.
9710 (register_to_reg_type): Likewise.
9711 (rs6000_pre_atomic_barrier): Likewise.
9712 (rs6000_machopic_legitimize_pic_address): Likewise.
9713 (rs6000_allocate_stack_temp): Likewise.
9714 (rs6000_address_for_fpconvert): Likewise.
9715 (rs6000_address_for_altivec): Likewise.
9716 (rs6000_secondary_memory_needed_rtx): Delete function.
9717 (rs6000_check_sdmode): Likewise.
9718 (rs6000_alloc_sdmode_stack_slot): Likewise.
9719 (rs6000_instantiate_decls): Likewise.
9720 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
9721 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
9722 Delete reload_in_progress.
9723 (*vec_reload_and_plus_<mptrsize>): Likewise.
9724 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
9725 (vsx_div_v2di): Likewise.
9726 (vsx_udiv_v2di): Likewise.
9728 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9730 * config/rs6000/rs6000.opt (mlra): Replace with stub.
9731 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
9732 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
9733 (rs6000_debug_reg_global): Delete print of LRA status.
9734 (rs6000_option_override_internal): Delete dead LRA related code.
9735 (rs6000_lra_p): Delete function.
9736 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
9738 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
9740 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
9741 * config/riscv/rtems.h: New file.
9743 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9744 Sudakshina Das <sudi.das@arm.com>
9746 * config/aarch64/aarch64.md
9747 (define_split for and<mode>3nr_compare): Move
9748 non aarch64_logical_operand to a register.
9749 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
9750 register immediate operand to a register.
9751 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
9753 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9756 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
9758 2017-07-27 Richard Biener <rguenther@suse.de>
9760 PR tree-optimization/81573
9761 PR tree-optimization/81494
9762 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
9763 multi defuse cycle case.
9765 2017-07-27 Richard Biener <rguenther@suse.de>
9767 PR tree-optimization/81571
9768 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
9771 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
9773 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
9774 earlier and only if MASK_FPU is set. Adjust formatting.
9776 2017-07-27 Martin Liska <mliska@suse.cz>
9778 * opt-functions.awk: Add validation of value of Init.
9779 * optc-gen.awk: Pass new argument.
9781 2017-07-27 Martin Liska <mliska@suse.cz>
9783 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
9784 Fix wrong condition.
9786 2017-07-27 Martin Liska <mliska@suse.cz>
9788 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
9789 BBs and edges seen by autoFDO.
9791 2017-07-27 Richard Biener <rguenther@suse.de>
9793 PR tree-optimization/81502
9794 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
9795 with incompatible but same sized type.
9796 (execute_update_addresses_taken): Likewise.
9798 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
9800 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
9801 flag_tree_loop_vectorize rather than flag_tree_vectorize.
9803 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9806 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
9807 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
9808 Change s_operand to memory_operand.
9810 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
9812 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
9813 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
9814 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
9815 Emit instructions rather than returning an expression. Handle TFmode
9816 and KFmode by casting to TImode.
9817 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
9818 (rs6000_emit_le_vsx_store): Likewise.
9819 * config/rs6000/vsx.md (VSX_TI): New iterator.
9820 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
9821 (*vsx_le_undo_permute_<mode>): Likewise.
9822 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
9823 emit the split sequence.
9824 (*vsx_le_perm_store_<mode>): Likewise.
9826 2017-07-27 Jakub Jelinek <jakub@redhat.com>
9828 PR tree-optimization/81555
9829 PR tree-optimization/81556
9830 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
9831 if true, force CHANGED for the recursive invocation.
9832 (reassociate_bb): Remember original length of ops array, pass
9833 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
9835 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
9836 attributes for noipa attribute. For naked attribute use
9837 lookup_attribute first before lookup_attribute_spec.
9838 * final.c (rest_of_handle_final): Disable IPA RA for functions with
9840 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
9841 for functions with noipa attribute.
9842 (cgraph_externally_visible_p): Return true for functions with noipa
9844 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
9845 for functions with noipa attribute.
9846 * doc/extend.texi: Document noipa function attribute.
9847 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
9848 also for functions with noipa attribute.
9849 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
9851 2017-07-26 Andrew Pinski <apinski@cavium.com>
9853 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
9854 vec_unalign_load_cost and vec_unalign_store_cost.
9856 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9858 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
9859 -mvsx-small-integer option.
9860 (ISA_3_0_MASKS_IEEE): Likewise.
9861 (OTHER_VSX_VECTOR_MASKS): Likewise.
9862 (POWERPC_MASKS): Likewise.
9863 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
9864 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
9865 code, only testing for DImode being allowed in non-VSX floating
9867 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
9868 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
9870 (rs6000_option_override_internal): Delete -mvsx-small-integer.
9871 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
9872 TARGET_P8_VECTOR test.
9873 (rs6000_secondary_reload_simple_move): Likewise.
9874 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
9875 since TARGET_P9_VECTOR was already tested.
9876 (rs6000_opt_masks): Remove -mvsx-small-integer.
9877 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
9878 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9880 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
9881 test for TARGET_VEXTRACTUB was used, and that uses
9883 (p9 extract splitter): Likewise.
9884 (vsx_extract_<mode>_di_p9): Likewise.
9885 (vsx_extract_<mode>_store_p9): Likewise.
9886 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
9887 for TARGET_P9_VECTOR was used. Delete code that is now dead with
9888 the elimination of TARGET_VSX_SMALL_INTEGER.
9889 (vsx_extract_<mode>_p8): Likewise.
9890 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
9891 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
9892 (vsx_set_<mode>_p9): Likewise.
9893 (vsx_set_v4sf_p9): Likewise.
9894 (vsx_set_v4sf_p9_zero): Likewise.
9895 (vsx_insert_extract_v4sf_p9): Likewise.
9896 (vsx_insert_extract_v4sf_p9_2): Likewise.
9897 * config/rs6000/rs6000.md (sign extend splitter): Change
9898 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
9899 (floatsi<mode>2_lfiwax_mem): Likewise.
9900 (floatunssi<mode>2_lfiwzx_mem): Likewise.
9901 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
9902 since a test for TARGET_P9_VECTOR was used.
9903 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9904 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
9905 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9906 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
9907 TARGET_P8_VECTOR test.
9908 (fix_trunc<mode>si2_stfiwx): Likewise.
9909 (fix_trunc<mode>si2_internal): Likewise.
9910 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
9911 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9913 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
9914 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
9915 TARGET_P8_VECTOR test.
9916 (fixuns_trunc<mode>si2_stfiwx): Likewise.
9917 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
9918 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9920 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
9921 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
9922 since a test for TARGET_P9_VECTOR was used.
9923 (splitter for loading small constants): Likewise.
9925 2017-07-26 Andrew Pinski <apinski@cavium.com>
9927 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
9930 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
9933 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
9934 (fp_valid_at): Likewise.
9936 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
9938 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
9939 (qdf24xx_addrcost_table): Likewise.
9940 (cortexa57_tunings): Update to use generic_branch_cost.
9941 (cortexa72_tunings): Likewise.
9942 (cortexa73_tunings): Likewise.
9943 (qdf24xx_tunings): Likewise.
9945 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
9947 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
9948 (thunderx2t99_branch_cost): Likewise.
9949 (cortexa35_tunings): Update to use generic_branch_cost.
9950 (cortexa53_tunings): Likewise.
9951 (cortexa57_tunings): Likewise.
9952 (cortexa72_tunings): Likewise.
9953 (cortexa73_tunings): Likewise.
9954 (thunderx2t99_tunings): Likewise.
9956 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
9958 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
9959 (sparc_option_override): Honour MASK_FSMULD.
9960 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
9961 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
9962 * config/sparc/sparc.opt (mfsmuld): New option.
9963 * doc/invoke.texi (mfsmuld): Document option.
9965 2017-07-26 Marek Polacek <polacek@redhat.com>
9968 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
9970 2017-07-26 Richard Biener <rguenther@suse.de>
9972 * gimple-match-head.c (do_valueize): Return OP if valueize
9974 (get_def): New helper to get at the def stmt of a SSA name
9976 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
9977 do_valueize to get at the def stmt.
9978 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
9980 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
9983 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
9985 2017-07-26 Martin Liska <mliska@suse.cz>
9988 * function.c (expand_function_start): Make expansion of
9989 nonlocal_goto_save_area after parm_birth_insn.
9991 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
9993 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
9994 from all CPU target flags enable members.
9996 2017-07-26 Richard Biener <rguenther@suse.de>
9998 * genmatch.c (dt_simplify::gen): Make iterator vars const.
9999 (decision_tree::gen): Make 'type' const.
10000 (write_predicate): Likewise.
10002 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
10004 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
10005 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
10006 (rs6000_option_override_internal): Likewise.
10007 (rs6000_expand_vector_set): Likewise.
10008 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
10009 (TARGET_UPPER_REGS_SF): Likewise.
10010 (TARGET_UPPER_REGS_DI): Likewise.
10011 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
10012 (TARGET_DIRECT_MOVE_64BIT): Likewise.
10013 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
10014 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10015 (Splitters for DI constants in Altivec registers): Likewise.
10016 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
10017 (vsx_set_v4sf_p9): Likewise.
10018 (vsx_set_v4sf_p9_zero): Likewise.
10019 (vsx_insert_extract_v4sf_p9): Likewise.
10020 (vsx_insert_extract_v4sf_p9_2): Likewise.
10022 2017-07-25 Carl Love <cel@us.ibm.com>
10024 * doc/extend.texi: Update the built-in documentation file for the
10025 existing built-in functions
10026 vector signed char vec_cnttz (vector signed char);
10027 vector unsigned char vec_cnttz (vector unsigned char);
10028 vector signed short vec_cnttz (vector signed short);
10029 vector unsigned short vec_cnttz (vector unsigned short);
10030 vector signed int vec_cnttz (vector signed int);
10031 vector unsigned int vec_cnttz (vector unsigned int);
10032 vector signed long long vec_cnttz (vector signed long long);
10033 vector unsigned long long vec_cnttz (vector unsigned long long);
10035 2017-07-25 Andrew Pinski <apinski@cavium.com>
10037 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
10038 accesses where the use is for the first operand of a BIT_INSERT.
10040 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
10043 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
10044 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
10046 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
10048 * config/i386/gstabs.h: Delete.
10049 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
10051 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
10053 * config/i386/i386.c (ix86_decompose_address): Do not check for
10054 register RTX when looking at index_reg or base_reg.
10055 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
10057 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
10059 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
10060 to update EH info here.
10062 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10064 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
10066 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10068 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
10070 2017-07-25 Torsten Duwe <duwe@suse.de>
10072 * common.opt: Introduce -fpatchable-function-entry
10073 command line option, and its variables function_entry_patch_area_size
10074 and function_entry_patch_area_start.
10075 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
10076 including a two-value parser.
10077 * target.def (print_patchable_function_entry): New target hook.
10078 * targhooks.h (default_print_patchable_function_entry): New function.
10079 * targhooks.c (default_print_patchable_function_entry): Likewise.
10080 * toplev.c (process_options): Switch off IPA-RA if
10081 patchable function entries are being generated.
10082 * varasm.c (assemble_start_function): Look at the
10083 patchable-function-entry command line switch and current
10084 function attributes and maybe generate NOP instructions by
10085 calling the print_patchable_function_entry hook.
10086 * doc/extend.texi: Document patchable_function_entry attribute.
10087 * doc/invoke.texi: Document -fpatchable_function_entry
10088 command line option.
10089 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
10091 * doc/tm.texi: Re-generate.
10093 2017-07-25 Jakub Jelinek <jakub@redhat.com>
10096 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
10097 TARGET_AVX512DQ rather than TARGET_AVX512BW.
10099 2017-07-25 Tamar Christina <tamar.christina@arm.com>
10101 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
10103 2017-07-25 Richard Biener <rguenther@suse.de>
10105 PR tree-optimization/81455
10106 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
10107 not walk in cycles when looking for guards.
10109 2017-07-25 Richard Biener <rguenther@suse.de>
10111 PR tree-optimization/81529
10112 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
10113 when optimizing backedge uses.
10115 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
10117 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
10119 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
10120 to dl_section_ref. On AIX, append an expression to subtract
10121 the size of the section length to dl_section_ref.
10123 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
10125 * configure.ac: If any of the config.* scripts fail, exit 1.
10126 * configure: Regenerate.
10128 2017-07-25 Richard Biener <rguenther@suse.de>
10130 PR middle-end/81546
10131 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
10132 of immediate uses, be more verbose on errors.
10134 2017-07-25 Richard Biener <rguenther@suse.de>
10136 PR tree-optimization/81510
10137 * tree-vect-loop.c (vect_is_simple_reduction): When the
10138 reduction stmt is not inside the loop bail out.
10140 2017-07-25 Richard Biener <rguenther@suse.de>
10142 PR tree-optimization/81303
10143 * tree-vect-loop-manip.c (vect_loop_versioning): Build
10144 profitability check against LOOP_VINFO_NITERSM1.
10146 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10148 * domwalk.c (cmp_bb_postorder): Simplify.
10149 (sort_bbs_postorder): New function. Use it...
10150 (dom_walker::walk): ...here to optimize common cases.
10152 2017-07-25 Martin Liska <mliska@suse.cz>
10155 * ipa-visibility.c (function_and_variable_visibility): Make the
10156 redirection just on target that supports aliasing.
10157 Fix GNU coding style.
10159 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
10162 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
10164 * config/arm/rtems.h: Likewise.
10165 * config/bfin/rtems.h: Likewise.
10166 * config/i386/rtemself.h: Likewise.
10167 * config/lm32/rtems.h: Likewise.
10168 * config/m32c/rtems.h: Likewise.
10169 * config/m68k/rtemself.h: Likewise.
10170 * config/microblaze/rtems.h: Likewise.
10171 * config/mips/rtems.h: Likewise.
10172 * config/moxie/rtems.h: Likewise.
10173 * config/nios2/rtems.h: Likewise.
10174 * config/powerpcspe/rtems.h: Likewise.
10175 * config/rs6000/rtems.h: Likewise.
10176 * config/rtems.h: Likewise.
10177 * config/sh/rtems.h: Likewise.
10178 * config/sh/rtemself.h: Likewise.
10179 * config/sparc/rtemself.h: Likewise.
10181 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
10184 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
10185 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
10186 * tree-ssa-structalias.c (alias_get_name): Same.
10188 2017-07-25 Bin Cheng <bin.cheng@arm.com>
10191 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
10192 instructions if no du chain is found.
10194 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
10196 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
10198 2017-07-25 Richard Biener <rguenther@suse.de>
10200 PR middle-end/81505
10201 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
10204 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
10206 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
10207 upper-regs options.
10208 (ISA_2_7_MASKS_SERVER): Likewise.
10209 (ISA_3_0_MASKS_IEEE): Likewise.
10210 (OTHER_P8_VECTOR_MASKS): Likewise.
10211 (OTHER_VSX_VECTOR_MASKS): Likewise.
10212 (POWERPC_MASKS): Likewise.
10213 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
10214 duplicate list of options.
10215 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
10216 explicit -mupper-regs options.
10217 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
10218 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
10219 alias for -mupper-regs-df.
10220 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
10221 (rs6000_init_hard_regno_mode_ok): Likewise.
10222 (rs6000_option_override_internal): Likewise.
10223 (rs6000_opt_masks): Likewise.
10224 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
10225 options in terms of whether -mvsx or -mpower8-vector was used.
10226 (TARGET_UPPER_REGS_DI): Likewise.
10227 (TARGET_UPPER_REGS_SF): Likewise.
10228 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
10229 -mupper-regs-* options.
10231 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
10233 * passes.c (emergency_dump_function): Print some empty lines and a
10234 header before the RTL dump.
10236 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
10238 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
10240 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
10243 * config/aarch64/aarch64.c (aarch64_classify_symbol):
10244 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
10246 2017-07-24 Carl Love <cel@us.ibm.com>
10248 * config/rs6000/rs6000-c.c: Add support for built-in functions
10249 vector float vec_extract_fp32_from_shorth (vector unsigned short);
10250 vector float vec_extract_fp32_from_shortl (vector unsigned short);
10251 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
10252 vec_extract_fp_from_shortl): Add defines for the two builtins.
10253 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
10254 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
10256 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
10257 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
10258 * doc/extend.texi: Update the built-in documentation file for the
10259 new built-in function.
10261 2017-07-24 Jakub Jelinek <jakub@redhat.com>
10264 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
10266 * doc/generic.texi: Likewise.
10267 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
10268 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
10270 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
10272 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
10273 (aarch64_mls_elt_merge<mode>): Likewise.
10275 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
10277 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
10278 having __cxa_atexit.
10280 2017-07-23 Michael Collison <michael.collison@arm.com>
10282 * config/arm/arm.c (arm_option_override): Deprecate
10283 use of -mstructure-size-boundary.
10284 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
10285 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
10287 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10290 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
10291 Reduce cost estimate for direct moves.
10293 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
10296 * config/i386/i386.c (ix86_option_override_internal): Disable
10297 BMI, BMI2 and TBM instructions for -m16.
10299 2017-07-21 Carl Love <cel@us.ibm.com>
10301 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10302 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10303 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10304 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10305 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10306 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10307 VMULOSW): New enum "unspec" values.
10308 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10309 altivec_vmulosw): New patterns.
10310 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10311 VMULOSW): Add definitions.
10313 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
10315 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
10316 (qdf24xx): Likewise.
10317 * config/aarch64/aarch64-options-extensions.def (rdma); New.
10318 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
10319 (AARCH64_FL_V8_1): Renumber.
10320 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
10321 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
10322 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
10323 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
10324 rdma to feature modifiers list.
10326 2017-07-21 Yury Gribov <tetra2005@gmail.com>
10328 PR middle-end/56727
10329 * ipa-visibility (function_and_variable_visibility): Convert
10330 recursive PLT call to direct call if appropriate.
10332 2017-07-21 Andrew Pinski <apinski@cavium.com>
10334 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
10335 operand 1 to see if the types precision matches.
10336 * fold-const.c (operand_equal_p): Likewise.
10338 2017-07-21 Richard Biener <rguenther@suse.de>
10340 PR tree-optimization/81303
10341 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
10342 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
10343 (vect_peeling_hash_get_lowest_cost): Adjust.
10344 (vect_enhance_data_refs_alignment): Likewise. Use
10345 vect_get_peeling_costs_all_drs to compute the penalty for no
10346 peeling to match up costs.
10348 2017-07-21 Richard Biener <rguenther@suse.de>
10350 PR tree-optimization/81500
10351 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
10352 we didn't identify a reduction path.
10354 2017-07-21 Tom de Vries <tom@codesourcery.com>
10355 Cesar Philippidis <cesar@codesourcery.com>
10357 PR gcov-profile/81442
10358 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
10361 2017-07-21 Tom de Vries <tom@codesourcery.com>
10364 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
10366 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
10367 nvptx_override_options_after_change.
10369 2017-07-21 Ulrich Drepper <drepper@redhat.com>
10371 * dwarf2out.c (output_file_names): Avoid double testing for
10372 dwarf_version >= 5.
10374 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
10376 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
10378 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
10380 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
10382 (try_crossjump_to_edge): Do not punt on partitioned functions.
10384 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
10386 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
10387 Put all BBs reachable only via paths crossing cold region to cold
10389 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
10391 2016-07-21 Richard Biener <rguenther@suse.de>
10393 PR tree-optimization/81303
10394 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
10395 into account prologue and epilogue iterations when raising
10396 min_profitable_iters to sth at least covering one vector iteration.
10398 2017-07-21 Tamar Christina <tamar.christina@arm.com>
10400 * config/arm/arm.c (arm_test_cpu_arch_dat):
10403 2017-07-20 Nathan Sidwell <nathan@acm.org>
10405 Remove TYPE_METHODS.
10406 * tree.h (TYPE_METHODS): Delete.
10407 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
10408 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
10409 (dbxout_type_methods): Scan TYPE_FIELDS.
10410 (dbxout_type): Don't check TYPE_METHODS here.
10411 * function.c (use_register_for_decl): Always ignore register for
10412 class types when not optimizing.
10413 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
10414 * tree.c (free_lang_data_in_type): Stitch out member functions and
10415 templates from TYPE_FIELDS.
10416 (build_distinct_type_copy, verify_type_variant,
10417 verify_type): Member fns are on TYPE_FIELDS.
10418 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
10419 * tree-pretty-print.c (dump_generic_node): Likewise.
10421 2017-07-20 Jakub Jelinek <jakub@redhat.com>
10424 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
10425 V2TImode and V4TImode.
10426 (ix86_expand_vector_extract): Likewise.
10427 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
10428 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
10429 (ssescalarmode): Handle V4TImode and V2TImode.
10430 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
10431 (*vec_extractv2ti, *vec_extractv4ti): New insns.
10432 (VEXTRACTI128_MODE): New mode iterator.
10433 (splitter for *vec_extractv?ti first element): New.
10434 (VEC_INIT_MODE): New mode iterator.
10435 (vec_init<mode>): Consolidate 3 expanders into one using
10436 VEC_INIT_MODE mode iterator.
10438 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
10440 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
10441 non_spilled_static_chain_regno_p.
10443 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
10445 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
10447 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
10449 * bb-reorder.c (connect_traces): Allow copying of blocks within
10452 2017-07-20 Richard Biener <rguenther@suse.de>
10454 * gimple.h (gimple_phi_result): Add gphi * overload.
10455 (gimple_phi_result_ptr): Likewise.
10456 (gimple_phi_arg): Likewise. Adjust index assert to only
10457 allow actual argument accesses rather than all slots available
10459 (gimple_phi_arg_def): Add gphi * overload.
10460 * tree-phinodes.c (make_phi_node): Initialize only actual
10462 (resize_phi_node): Clear memory not covered by old node,
10463 do not initialize excess argument slots.
10464 (reserve_phi_args_for_new_edge): Initialize new argument slot
10467 2017-07-20 Bin Cheng <bin.cheng@arm.com>
10469 PR tree-optimization/81388
10471 2016-07-21 Bin Cheng <bin.cheng@arm.com>
10473 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
10474 by removing computation of may_be_zero.
10476 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10477 Tom de Vries <tom@codesourcery.com>
10479 PR middle-end/81030
10480 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
10481 when gimple level profile disagrees with what RTL expander did.
10483 2017-07-20 Richard Biener <rguenther@suse.de>
10485 PR tree-optimization/61171
10486 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
10487 (vect_analyze_stmt): Add slp instance parameter.
10488 (vectorizable_reduction): Likewise.
10489 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
10490 (vect_is_simple_reduction): Deal with chains not detected
10491 as SLP reduction chain, specifically not properly associated
10492 chains containing a mix of plus/minus.
10493 (get_reduction_op): Remove.
10494 (get_initial_defs_for_reduction): Simplify, pass in whether
10495 this is a reduction chain, pass in the SLP node for the PHIs.
10496 (vect_create_epilog_for_reduction): Get the SLP instance as
10498 (vectorizable_reduction): Get the SLP instance as arg.
10499 During analysis remember the SLP node with the PHIs in the
10500 instance. Simplify getting at the vectorized reduction PHIs.
10501 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
10502 through SLP instance.
10503 (vect_slp_analyze_operations): Likewise.
10504 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
10505 (vect_transform_stmt): Likewise.
10507 2017-07-20 Tom de Vries <tom@codesourcery.com>
10509 PR tree-optimization/81489
10510 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
10511 read of phi arg location to before loop that modifies phi.
10513 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10515 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
10518 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10520 PR middle-end/81331
10521 * except.c (execute): Fix ordering issue.
10523 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
10525 PR rtl-optimization/81423
10526 * combine.c (make_compound_operation_int): Don't try to optimize
10527 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
10529 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
10531 PR rtl-optimization/81423
10532 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
10533 with a constant that is -1 in the truncated to mode.
10535 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10537 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
10538 (determine_unlikely_bbs): ... here.
10539 * predict.h (propagate_unlikely_bbs_forward): Declare.
10540 * cfgexpand.c (pass_expand::execute): Use it.
10541 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
10543 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
10544 propagate_unlikely_bbs_forward.
10546 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10548 PR middle-end/81331
10549 * except.c (maybe_add_nop_after_section_switch): New function.
10552 2017-07-19 Tom de Vries <tom@codesourcery.com>
10554 * gimple.h (gimple_phi_set_arg): Make assert more strict.
10556 2017-07-19 Tom de Vries <tom@codesourcery.com>
10558 * gimple.h (gimple_phi_arg): Make assert more strict.
10560 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
10562 * config.gcc (powerpc*-*-*): Add mmintrin.h.
10563 * config/rs6000/mmintrin.h: New file.
10564 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
10566 2017-07-19 Jakub Jelinek <jakub@redhat.com>
10568 PR tree-optimization/81346
10569 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
10571 2017-07-19 Tom de Vries <tom@codesourcery.com>
10573 * config/nvptx/nvptx.md (VECIM): Add V2DI.
10575 2017-07-19 Tom de Vries <tom@codesourcery.com>
10577 * config/nvptx/nvptx-modes.def: Add V2DImode.
10578 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
10579 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
10580 (nvptx_output_mov_insn): Handle lack of mov.b128.
10581 (nvptx_print_operand): Handle 'H' and 'L' codes.
10582 (nvptx_vector_mode_supported): Allow V2DImode.
10583 (nvptx_preferred_simd_mode): New function.
10584 (nvptx_data_alignment): New function.
10585 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
10586 nvptx_preferred_simd_mode.
10587 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
10589 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
10591 2017-07-19 Tom de Vries <tom@codesourcery.com>
10593 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
10594 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
10595 (nvptx_vector_mode_supported): New function. Allow V2SImode.
10596 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
10597 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
10598 (mov<VECIM>_insn): New define_insn.
10599 (define_expand "mov<VECIM>): New define_expand.
10601 2017-07-19 Tom de Vries <tom@codesourcery.com>
10603 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
10605 2017-07-19 Jakub Jelinek <jakub@redhat.com>
10607 PR tree-optimization/81346
10608 * fold-const.h (fold_div_compare, range_check_type): Declare.
10609 * fold-const.c (range_check_type): New function.
10610 (build_range_check): Use range_check_type.
10611 (fold_div_compare): No longer static, rewritten into
10612 a match.pd helper function.
10613 (fold_comparison): Don't call fold_div_compare here.
10614 * match.pd (X / C1 op C2): New optimization using fold_div_compare
10615 as helper function.
10617 2017-07-19 Nathan Sidwell <nathan@acm.org>
10619 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
10620 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
10621 * tree.c (find_decls_types_r, verify_type): Use
10622 TYPE_{MIN,MAX}_VALUE_RAW.
10623 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
10624 (hash_tree): Likewise.
10625 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
10627 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
10630 2017-07-18 Tom de Vries <tom@codesourcery.com>
10632 PR middle-end/81464
10633 * omp-expand.c (expand_omp_for_static_chunk): Handle
10634 equal-argument loop exit phi.
10636 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
10639 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
10640 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
10641 operand 2 predicate.
10642 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
10643 operand 2 predicate.
10644 (ror,rol -> rorx splitters): Use const_int_operand as
10645 operand 2 predicate.
10647 2017-06-18 Richard Biener <rguenther@suse.de>
10649 PR tree-optimization/81410
10650 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
10651 the gap in the ! slp_perm SLP case after each group.
10653 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10655 PR middle-end/81463
10656 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
10659 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10661 PR middle-end/81462
10662 * predict.c (set_even_probabilities): Cleanup; do not affect
10663 probabilities that are already known.
10664 (combine_predictions_for_bb): Call even when count is set.
10666 2017-07-18 Nathan Sidwell <nathan@acm.org>
10668 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
10671 2017-07-18 Bin Cheng <bin.cheng@arm.com>
10674 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
10675 optimization for loop niter analysis.
10677 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
10680 * config/avr/avr.c (avr_optimize_casesi): Don't use
10681 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
10683 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10685 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
10686 body_cost_vec from _vect_peel_extended_info.
10687 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
10688 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
10691 2017-07-18 Bin Cheng <bin.cheng@arm.com>
10693 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
10695 2017-07-18 Richard Biener <rguenther@suse.de>
10697 PR tree-optimization/80620
10698 PR tree-optimization/81403
10699 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
10700 info when re-using a VN table entry.
10702 2017-07-18 Richard Biener <rguenther@suse.de>
10704 PR tree-optimization/81418
10705 * tree-vect-loop.c (vectorizable_reduction): Properly compute
10706 vectype_in. Verify that with lane-reducing reduction operations
10707 we have a single def-use cycle.
10709 2017-07-17 Carl Love <cel@us.ibm.com>
10711 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
10713 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10714 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10715 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10716 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10717 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10718 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10719 VMULOSW): New enum "unspec" values.
10720 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
10721 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
10722 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10723 altivec_vmulosw): New patterns.
10724 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10725 VMULOSW): Add definitions.
10726 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
10728 * config/alpha/alpha.c: Include predict.h.
10730 2017-07-17 Yury Gribov <tetra2005@gmail.com>
10732 * tree-vrp.c (compare_assert_loc): Fix comparison function
10733 to return predictable results.
10735 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10737 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
10739 (subdi3): Likewise.
10740 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
10741 * doc/invoke.texi (mexpand-adddi): Update text.
10743 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10745 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
10746 that also clobbers the CC register. The old expand code is moved
10748 (*arc_clzsi2): ... here.
10749 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
10750 the CC register. The old expand code is moved to ...
10751 (arc_ctzsi2): ... here.
10753 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10755 * config/arc/arc.opt (mindexed-loads): Use initial value
10756 TARGET_INDEXED_LOADS_DEFAULT.
10757 (mauto-modify-reg): Use initial value
10758 TARGET_AUTO_MODIFY_REG_DEFAULT.
10759 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
10760 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
10761 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
10762 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
10764 2017-07-17 Martin Liska <mliska@suse.cz>
10767 * opts.c (finish_options): Do not allow -fgnu-tm
10768 w/ -fsanitize={kernel-,}address. Say sorry.
10770 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10773 * tree-loop-distribution.c (classify_partition): Only assert on
10774 numer of iterations.
10775 (merge_dep_scc_partitions): Delete prameter. Update function call.
10776 (distribute_loop): Remove code handling loop with unknown niters.
10777 (pass_loop_distribution::execute): Skip loop with unknown niters.
10779 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10782 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
10783 function sort_partitions_by_post_order.
10785 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10787 PR tree-optimization/81374
10788 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
10789 the max index of basic blocks, rather than number of basic blocks.
10791 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10793 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
10795 (arc_legitimate_pic_operand_p): Likewise.
10796 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
10798 (arc_needs_pcl_p): Likewise.
10799 (arc_legitimate_pc_offset_p): Likewise.
10800 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
10801 function is also used in constrains.md.
10802 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
10803 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
10804 PLUS. Only return true/false in known cases, otherwise assert.
10805 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
10806 is already called in arc_legitimate_constant_p.
10807 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
10809 (LEGITIMATE_PIC_OPERAND_P): Use
10810 arc_raw_symbolic_reference_mentioned_p function.
10811 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
10816 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10817 Andrew Burgess <andrew.burgess@embecosm.com>
10819 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
10820 (arc_return_address_register): New function.
10821 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
10822 (arc_handle_fndecl_attribute): Add naked attribute.
10823 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
10824 (TARGET_WARN_FUNC_RETURN): Likewise.
10825 (arc_allocate_stack_slots_for_args): New function.
10826 (arc_warn_func_return): Likewise.
10827 (machine_function): Change type fn_type.
10828 (arc_compute_function_type): Consider new naked function type,
10829 change function return type.
10830 (arc_must_save_register): Adapt to handle new
10831 arc_compute_function_type's return type.
10832 (arc_expand_prologue): Likewise.
10833 (arc_expand_epilogue): Likewise.
10834 (arc_return_address_regs): Delete.
10835 (arc_return_address_register): New function.
10836 (arc_epilogue_uses): Use above function.
10837 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
10838 (arc_function_type): Change encoding, add naked type.
10839 (ARC_INTERRUPT_P): Change to handle the new encoding.
10840 (ARC_FAST_INTERRUPT_P): Likewise.
10841 (ARC_NORMAL_P): Define.
10842 (ARC_NAKED_P): Likewise.
10843 (arc_compute_function_type): Delete prototype.
10844 * config/arc/arc.md (in_ret_delay_slot): Use
10845 arc_return_address_register function.
10846 (simple_return): Likewise.
10847 (p_return_i): Likewise.
10849 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10851 PR tree-optimization/81428
10852 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
10853 can't be built for those types.
10855 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
10857 Remove stuff dead since r239246.
10859 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
10860 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
10861 (avr_inform_devices): Remove dead stuff.
10863 2017-07-17 Tamar Christina <tamar.christina@arm.com>
10865 * config/arm/arm_neon.h: Fix softp typo.
10867 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10869 PR tree-optimization/81365
10870 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
10871 aggregate moves onto bb predecessor edges, make sure there are no
10872 loads that could alias the lhs in between the start of bb and the
10875 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
10878 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
10879 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
10880 [LSHIFTRT, outer_code = TRUNCATE]: Same.
10882 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10884 PR tree-optimization/81396
10885 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
10886 (init_symbolic_number): Initialize it to 1.
10887 (perform_symbolic_merge): Add n_ops from both operands into the new
10889 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
10890 without base_addr as useless if they need more than one operation.
10891 (bswap_replace): Handle !bswap case for NULL base_addr.
10893 2017-07-17 Tom de Vries <tom@codesourcery.com>
10896 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
10899 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
10901 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
10902 conditional builtin define __FIX_LEON3FT_B2BST.
10904 2017-07-17 Daniel Cederman <cederman@gaisler.com>
10906 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
10907 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
10910 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
10912 PR rtl-optimization/81424
10913 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
10914 to remove potential trapping from operands if -fnon-call-exceptions.
10916 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10918 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
10919 profile_proability for scalling.
10920 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
10922 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10924 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
10926 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10928 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
10929 fixpoint arithmetics.
10931 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10933 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
10934 fixpoint arithmetics.
10936 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10938 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
10939 fixpoint arithmetics.
10941 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10943 * profile-count.h (profile_probability::from_reg_br_prob_note,
10944 profile_probability::to_reg_br_prob_note): New functions.
10945 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
10946 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
10947 * predict.c (probability_reliable_p): Update.
10948 (edge_probability_reliable_p): Update.
10949 (br_prob_note_reliable_p): Update.
10950 (invert_br_probabilities): Update.
10951 (add_reg_br_prob_note): New function.
10952 (combine_predictions_for_insn): Update.
10953 * asan.c (asan_clear_shadow): Update.
10954 * cfgbuild.c (compute_outgoing_frequencies): Update.
10955 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
10956 (update_br_prob_note): Update.
10957 (rtl_verify_edges): Update.
10958 (purge_dead_edges): Update.
10959 (fixup_reorder_chain): Update.
10960 * emit-rtl.c (try_split): Update.
10961 * ifcvt.c (cond_exec_process_insns): Update.
10962 (cond_exec_process_if_block): Update.
10963 (dead_or_predicable): Update.
10964 * internal-fn.c (expand_addsub_overflow): Update.
10965 (expand_neg_overflow): Update.
10966 (expand_mul_overflow): Update.
10967 * loop-doloop.c (doloop_modify): Update.
10968 * loop-unroll.c (compare_and_jump_seq): Update.
10969 * optabs.c (emit_cmp_and_jump_insn_1): Update.
10970 * predict.h: Update.
10971 * reorg.c (mostly_true_jump): Update.
10973 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
10974 * config/alpha/alpha.c (emit_unlikely_jump): Update.
10975 * config/arc/arc.c: (emit_unlikely_jump): Update.
10976 * config/arm/arm.c: (emit_unlikely_jump): Update.
10977 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
10978 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
10979 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
10980 (ix86_print_operand): Update.
10981 (ix86_split_fp_branch): Update.
10982 (predict_jump): Update.
10983 * config/ia64/ia64.c (ia64_print_operand): Update.
10984 * config/mmix/mmix.c (mmix_print_operand): Update.
10985 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
10986 (rs6000_expand_split_stack_prologue): Update.
10987 * config/rs6000/rs6000.c: Update.
10988 * config/s390/s390.c (s390_expand_vec_strlen): Update.
10989 (s390_expand_vec_movstr): Update.
10990 (s390_expand_cs_tdsi): Update.
10991 (s390_expand_split_stack_prologue): Update.
10992 * config/sh/sh.c (sh_print_operand): Update.
10993 (expand_cbranchsi4): Update.
10994 (expand_cbranchdi4): Update.
10995 * config/sparc/sparc.c (output_v9branch): Update.
10996 * config/spu/spu.c (get_branch_target): Update.
10997 (ea_load_store_inline): Update.
10998 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
10999 * config/tilepro/tilepro.c: Update.
11001 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
11003 * gimplify.c (mostly_copy_tree_r): Revert latest change.
11004 (gimplify_save_expr): Likewise.
11006 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11008 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
11010 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11012 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
11014 * timevar.def (TV_IPA_FNSUMMARY): Define.
11016 2017-07-16 Daniel Cederman <cederman@gaisler.com>
11018 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
11019 to back store errata sensitive sequence from being generated.
11020 (sqrtdf2_fix): Likewise.
11022 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11024 * tree-ssa-threadupdate.c (compute_path_counts,
11025 update_joiner_offpath_counts): Use profile_probability.
11027 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
11030 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11032 * config/arm/arm-c.c (arm_cpu_builtins): Define
11033 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
11035 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
11037 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11038 array entries to represent __ieee128 versions of the
11039 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
11040 scalar_extract_sig, and scalar_insert_exp built-in functions.
11041 (altivec_resolve_overloaded_builtin): Add special case handling
11042 for the __builtin_scalar_insert_exp function, as represented by
11043 the P9V_BUILTIN_VEC_VSIEDP constant.
11044 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
11045 exponent support for __ieee128 argument.
11046 (VSESQP): Add scalar extract signature support for __ieee128
11048 (VSTDCNQP): Add scalar test negative support for __ieee128
11050 (VSIEQP): Add scalar insert exponent support for __int128 argument
11051 with __ieee128 result.
11052 (VSIEQPF): Add scalar insert exponent support for __ieee128
11053 argument with __ieee128 result.
11054 (VSTDCQP): Add scalar test data class support for __ieee128
11056 (VSTDCNQP): Add overload support for scalar test negative with
11057 __ieee128 argument.
11058 (VSTDCQP): Add overload support for scalar test data class
11059 __ieee128 argument.
11060 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
11061 UNSPEC_VSX_SXSIGDP.
11062 (UNSPEC_VSX_SIEXPQP): New constant.
11063 (xsxexpqp): New insn for VSX scalar extract exponent quad
11065 (xsxsigqp): New insn for VSX scalar extract significand quad
11067 (xsiexpqpf): New insn for VSX scalar insert exponent quad
11068 precision with floating point argument.
11069 (xststdcqp): New expand for VSX scalar test data class quad
11071 (xststdcnegqp): New expand for VSX scalar test negative quad
11073 (xststdcqp): New insn to match expansions for VSX scalar test data
11074 class quad precision and VSX scalar test negative quad precision.
11075 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
11076 special case operand checking to enforce that second operand of
11077 VSX scalar test data class with quad precision argument is a 7-bit
11079 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
11080 prototypes and descriptions of __ieee128 versions of
11081 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
11082 scalar_test_data_class, and scalar_test_neg built-in functions.
11084 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11086 PR tree-optimization/81162
11087 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
11088 replace a negate with an add.
11090 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
11092 * doc/invoke.texi (arm/-mcpu): Document +crypto.
11094 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11096 * config/arm/arm-c.c (arm_cpu_builtins): Define
11097 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
11099 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11101 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
11102 (armv8-r): Set ARM Cortex-R52 as default CPU.
11103 * config/arm/arm-tables.opt: Regenerate.
11104 * config/arm/arm-tune.md: Regenerate.
11105 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
11107 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
11108 extension for -mcpu=cortex-r52.
11110 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11112 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
11113 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
11114 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
11115 (fp-armv8): Define it as FP_ARMv8 only.
11116 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
11117 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
11119 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
11120 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
11121 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
11122 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
11123 than TARGET_FPU_ARMV8.
11124 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
11125 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
11126 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
11128 * config/arm/neon.md (neon_vrint): Likewise.
11129 (neon_vcvt): Likewise.
11130 (neon_<fmaxmin_op><mode>): Likewise.
11131 (<fmaxmin><mode>3): Likewise.
11132 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
11133 * config/arm/predicates.md (arm_cond_move_operator): Check against
11134 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
11136 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
11138 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
11139 to top of function.
11141 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11143 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
11144 loop in comment with memset.
11146 2017-07-14 Martin Liska <mliska@suse.cz>
11148 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
11149 * dwarf2out.c (is_java): Remove the function.
11150 (output_pubname): Remove usage of the function.
11151 (lower_bound_default): Remove usage of DW_LANG_Java.
11152 (gen_compile_unit_die): Likewise.
11153 * gcc.c: Remove compiler defaults for .java and .zip files.
11154 * gimple-expr.c (remove_suffix): Change as there's no longer
11155 extension than 4-letter one.
11156 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
11157 (gimplify_save_expr): Likewise.
11158 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
11159 as it's possible even for other languages than Java.
11160 * langhooks.h (struct lang_hooks): Remove Java from a comment.
11161 * lto-opts.c (lto_write_options): Remove reference to Java.
11162 * opts.c (strip_off_ending): Update file extension handling.
11163 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
11164 * tree-eh.c (lower_resx): Likewise.
11165 * tree.c (free_lang_data_in_type): Remove dead code.
11166 (find_decls_types_r): Likewise.
11167 (build_common_builtin_nodes): Remove Java from a comment.
11168 (verify_type): Remove dead code.
11169 * varasm.c (assemble_external): Remove Java from a comment.
11171 2017-07-14 Martin Liska <mliska@suse.cz>
11173 * opts.c (finish_options): Add quotes.
11174 (common_handle_option): Likewise.
11176 2017-07-14 Martin Liska <mliska@suse.cz>
11178 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
11179 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
11180 Remove N_SO_PASCAL.
11181 * dwarf2out.c (lower_bound_default): Do not handle
11183 (gen_compile_unit_die): Likewise.
11184 * gcc.c: Remove default extension binding for GNU Pascal.
11185 * stmt.c: Remove Pascal language from a comment.
11186 * xcoffout.c: Likewise.
11188 2017-07-13 David Malcolm <dmalcolm@redhat.com>
11191 * diagnostic-show-locus.c (fixit_cmp): New function.
11192 (layout::layout): Sort m_fixit_hints.
11193 (column_range::column_range): Assert that the values are valid.
11194 (struct char_span): New struct.
11195 (correction::overwrite): New method.
11196 (struct source_line): New struct.
11197 (line_corrections::add_hint): Add assertions. Reimplement memcpy
11198 calls in terms of classes source_line and char_span, and
11199 correction::overwrite.
11200 (selftest::test_overlapped_fixit_printing_2): New function.
11201 (selftest::diagnostic_show_locus_c_tests): Call it.
11203 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
11205 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
11206 early if there is no lhs.
11208 2017-07-13 Martin Liska <mliska@suse.cz>
11210 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
11211 (gen_reference_type_die): Likewise.
11212 * stor-layout.c: Remove Pascal-related comment.
11214 2017-07-13 Martin Liska <mliska@suse.cz>
11216 * opts.c (finish_options): Add quotes to error messages.
11217 (parse_sanitizer_options): Likewise.
11219 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
11221 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
11223 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
11225 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
11227 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
11229 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
11231 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
11233 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11236 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
11237 provides the hardware capability bits, define the macro
11238 __BUILTIN_CPU_SUPPORTS__.
11239 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
11240 if GLIBC does not provide the hardware capability bits. Add a
11241 gcc_unreachable call if the built-in cpu function is neither
11242 __builtin_cpu_is nor __builtin_cpu_supports.
11243 (rs6000_get_function_versions_dispatcher): Change the warning
11244 that an old GLIBC is used which does not export the capability
11245 bits to be an error.
11246 * doc/extend.texi (target_clones attribute): Document the
11247 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
11248 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
11249 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
11250 the macros defined by GCC if the newer GLIBC is available.
11252 2017-07-12 Jeff Law <law@redhat.com>
11254 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
11255 remaining includes slightly.
11256 * config/riscv/riscv-builtins.c: Include profile-count.h.
11258 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
11261 * config/avr/avr.c (avr_set_current_function): In diagnostic
11262 messages: Quote keywords and (parts of) identifiers.
11263 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
11266 2017-07-12 Carl Love <cel@us.ibm.com>
11268 * config/rs6000/rs6000-c.c: Add support for built-in functions
11269 vector bool char vec_revb (vector bool char);
11270 vector bool short vec_revb (vector short char);
11271 vector bool int vec_revb (vector bool int);
11272 vector bool long long vec_revb (vector bool long long);
11273 * doc/extend.texi: Update the built-in documentation file for the
11274 new built-in functions.
11276 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11278 * config/s390/s390.md: Remove movcc splitter.
11280 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11282 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
11283 load/store on condition.
11285 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
11288 * config/avr/avr.c (avr_encode_section_info)
11289 [progmem && !TREE_READONLY]: Error if progmem object needs
11292 2017-07-11 Michael Collison <michael.collison@arm.com>
11294 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
11297 2017-07-11 Carl Love <cel@us.ibm.com>
11299 * config/rs6000/rs6000-c.c: Add support for builtins
11300 vector unsigned int vec_parity_lsbb (vector signed int);
11301 vector unsigned int vec_parity_lsbb (vector unsigned int);
11302 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
11303 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
11304 vector unsigned long long vec_parity_lsbb (vector signed long long);
11305 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
11306 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
11307 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
11308 * doc/extend.texi: Update the built-in documentation file for the
11309 new built-in functions.
11311 2017-07-11 David Malcolm <dmalcolm@redhat.com>
11313 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
11314 (layout::m_primary_loc): New field.
11315 (layout::layout): Initialize new field. Move location filtering
11316 logic from here to...
11317 (layout::maybe_add_location_range): ...this new method. Add
11318 support for filtering to just the lines already specified by other
11320 (layout::will_show_line_p): New method.
11321 (gcc_rich_location::add_location_if_nearby): New method.
11322 (selftest::test_add_location_if_nearby): New test function.
11323 (selftest::diagnostic_show_locus_c_tests): Call it.
11324 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
11327 2017-07-11 Tom de Vries <tom@codesourcery.com>
11329 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
11330 (bb_first_real_insn): New function.
11331 (nvptx_single): Add extra initialization of broadcasted condition
11334 2017-07-11 Nathan Sidwell <nathan@acm.org>
11336 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
11338 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
11340 * doc/extend.texi (AVR Function Attributes): Remove weblink to
11341 Binutils doc as TEXI will mess them up.
11342 * doc/invoke.texi (AVR Options): Same here.
11344 2017-07-11 Daniel Cederman <cederman@gaisler.com>
11346 * config/sparc/sparc.opt (mfix-ut700): New option.
11347 (mfix-gr712rc): Likewise.
11348 (sparc_fix_b2bst): New variable.
11349 * doc/invoke.texi (SPARC options): Document them.
11350 (ARM options): Fix warnings.
11351 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
11352 instructions to prevent sequences that can trigger the store-store
11353 errata for certain LEON3FT processors.
11354 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
11355 (sparc_option_override): Set sparc_fix_b2bst appropriately.
11356 * config/sparc/sparc.md (fix_b2bst): New attribute.
11357 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
11359 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
11362 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
11364 (*rsqrtsf2_sse): Ditto.
11366 (div<mode>3): Macroize insn from divdf3 and divsf3
11367 using MODEF mode iterator.
11369 2017-07-10 Martin Sebor <msebor@redhat.com>
11371 PR tree-optimization/80397
11372 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
11373 instead of testing for equality to INTEGER_TYPE.
11375 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
11377 * config.gcc: Remove uclibc from arc target spec.
11379 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
11381 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
11383 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11386 * lto-wrapper.c (remove_option): New function.
11387 (merge_and_complain): Merge PIC/PIE options more realistically.
11389 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
11391 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
11395 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
11396 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
11397 * config.in: Regenerate.
11398 * configure: Regenerate.
11399 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
11400 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
11401 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
11402 (TARGET_GASISR_PROLOGUES): ...target mask.
11403 * common/config/avr/avr-common.c
11404 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
11405 Set -mgas-isr-prologues.
11406 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
11407 INSERT_PASS_BEFORE for it.
11408 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
11409 * config/avr/avr.c (avr_option_override)
11410 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
11411 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
11412 (avr_attribute_table) <no_gccisr>: Add new function attribute.
11413 (avr_set_current_function) <is_no_gccisr>: Init machine field.
11414 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
11416 (make_avr_pass_pre_proep): New function.
11417 (emit_push_sfr) <treg>: Add argument to function and use it
11418 instead of TMP_REG.
11419 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
11420 and set machine->gasisr.yes.
11421 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
11422 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
11423 __gcc_isr.n_pushed to .L__stack_usage.
11424 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
11425 (avr_asm_final_postscan_insn): ...this new static function.
11426 * config/avr/avr.h (machine_function)
11427 <is_no_gccisr, use_L__stack_usage>: New fields.
11428 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
11429 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
11430 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
11431 (gasisr, *gasisr): New expander and insn.
11432 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
11433 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
11434 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
11436 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
11438 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
11441 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
11443 Move jump-tables out of .text again.
11446 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
11447 (ASM_OUTPUT_ADDR_VEC): New function.
11448 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
11449 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
11450 INSN_ADDRESSes as asm comment.
11451 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
11452 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
11453 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
11454 * config/avr/avr.md (*tablejump): Adjust comment.
11455 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
11456 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
11458 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
11459 (avr_output_addr_vec): New proto.
11460 (avr_log_t) <insn_addresses>: New field.
11462 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
11465 * config/i386/i386.c (ix86_function_arg_advance): Set
11466 outgoing_args_on_stack to true if there are outgoing arguments
11468 (ix86_function_arg): Likewise.
11469 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
11470 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
11471 * config/i386/i386.h (machine_function): Add
11472 outgoing_args_on_stack.
11474 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
11476 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
11477 supporting pthreds.
11478 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
11480 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
11482 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
11483 (REAL_H): Remove $(MACHMODE_H).
11484 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
11486 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
11487 $(MACHMODE_H) and double-int.h.
11488 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
11490 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
11493 2017-07-07 Andrew Pinski <apinski@cavium.com>
11495 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
11496 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
11498 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
11500 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
11501 Add warning if GCC was not configured to link against a GLIBC that
11502 exports the hardware capability bits.
11503 (make_resolver_func): Make resolver function private and not a
11504 COMDAT function. Create the name with clone_function_name instead
11505 of make_unique_name.
11508 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
11509 correct operand in doing the split.
11511 2017-07-07 Carl Love <cel@us.ibm.com>
11513 * config/rs6000/rs6000-c: Add support for built-in function
11514 vector unsigned short vec_pack_to_short_fp32 (vector float,
11516 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
11517 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
11518 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
11519 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
11520 (convert_4f32_8i16): Add define_expand.
11521 * doc/extend.texi: Update the built-in documentation file for the
11522 new built-in function.
11524 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11526 * config/sparc/m8.md: New file.
11527 * config/sparc/sparc.md: Include m8.md.
11529 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11531 * config/sparc/sparc.opt: New option -mvis4b.
11532 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
11533 (sparc_option_override): Handle VIS4B.
11534 (enum sparc_builtins): Define
11535 SPARC_BUILTIN_DICTUNPACK{8,16,32},
11536 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
11537 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
11538 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
11539 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
11540 (check_constant_argument): New function.
11541 (sparc_vis_init_builtins): Define builtins
11542 __builtin_vis_dictunpack{8,16,32},
11543 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
11544 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
11545 __builtin_vis_fpcmpde{8,16,32}shl and
11546 __builtin_vis_fpcmpur{8,16,32}shl.
11547 (sparc_expand_builtin): Check that the constant operands to
11548 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
11549 constant and in range.
11550 * config/sparc/sparc-c.c (sparc_target_macros): Handle
11552 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
11553 (SPARC_IMM5_P): Likewise.
11554 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
11555 (enabled): Handle vis4b.
11556 (UNSPEC_DICTUNPACK): New unspec.
11557 (UNSPEC_FPCMPSHL): Likewise.
11558 (UNSPEC_FPUCMPSHL): Likewise.
11559 (UNSPEC_FPCMPDESHL): Likewise.
11560 (UNSPEC_FPCMPURSHL): Likewise.
11561 (cpu_feature): New CPU feature `vis4b'.
11562 (dictunpack{8,16,32}): New insns.
11563 (FPCSMODE): New mode iterator.
11564 (fpcscond): New code iterator.
11565 (fpcsucond): Likewise.
11566 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
11567 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
11568 (fpcmpde{8,16,32}{si,di}shl): Likewise.
11569 (fpcmpur{8,16,32}{si,di}shl): Likewise.
11570 * config/sparc/constraints.md: Define constraints `q' for unsigned
11571 2-bit integer constants and `t' for unsigned 5-bit integer
11573 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
11575 (imm5_operand_dictunpack16): Likewise.
11576 (imm5_operand_dictunpack32): Likewise.
11577 (imm2_operand): Likewise.
11578 * doc/invoke.texi (SPARC Options): Document -mvis4b.
11579 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
11580 ditunpack* and fpcmp*shl builtins.
11582 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11584 * config.gcc: Handle m8 in --with-{cpu,tune} options.
11585 * config.in: Add HAVE_AS_SPARC6 define.
11586 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
11588 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
11590 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
11591 (CPP_CPU_SPEC): Handle m8.
11592 (ASM_CPU_SPEC): Likewise.
11593 * config/sparc/sparc-opts.h (enum processor_type): Add
11595 * config/sparc/sparc.c (m8_costs): New struct.
11596 (sparc_option_override): Handle TARGET_CPU_m8.
11597 (sparc32_initialize_trampoline): Likewise.
11598 (sparc64_initialize_trampoline): Likewise.
11599 (sparc_issue_rate): Likewise.
11600 (sparc_register_move_cost): Likewise.
11601 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
11602 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
11603 (ASM_CPU64_DEFAULT_SPEC): Likewise.
11604 (CPP_CPU_SPEC): Handle M8.
11605 (ASM_CPU_SPEC): Likewise.
11606 (AS_M8_FLAG): Define.
11607 * config/sparc/sparc.md: Add m8 to the cpu attribute.
11608 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
11609 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
11611 * configure: Regenerate.
11612 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
11615 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11617 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
11619 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
11620 ("*movdi_insn_sp32"): Do not set v3pipe.
11621 ("*movsi_insn"): Likewise.
11622 ("*movdi_insn_sp64"): Likewise.
11623 ("*movsf_insn"): Likewise.
11624 ("*movdf_insn_sp32"): Likewise.
11625 ("*movdf_insn_sp64"): Likewise.
11626 ("*zero_extendsidi2_insn_sp64"): Likewise.
11627 ("*sign_extendsidi2_insn"): Likewise.
11628 ("*mov<VM32:mode>_insn"): Likewise.
11629 ("*mov<VM64:mode>_insn_sp64"): Likewise.
11630 ("*mov<VM64:mode>_insn_sp32"): Likewise.
11631 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
11632 ("<vlop:code><VL:mode>3"): Likewise.
11633 ("*not_<vlop:code><VL:mode>3"): Likewise.
11634 ("*nand<VL:mode>_vis"): Likewise.
11635 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
11636 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
11637 ("one_cmpl<VL:mode>2"): Likewise.
11638 ("faligndata<VM64:mode>_vis"): Likewise.
11639 ("alignaddrsi_vis"): Likewise.
11640 ("alignaddrdi_vis"): Likweise.
11641 ("alignaddrlsi_vis"): Likewise.
11642 ("alignaddrldi_vis"): Likewise.
11643 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
11644 ("bmaskdi_vis"): Likewise.
11645 ("bmasksi_vis"): Likewise.
11646 ("bshuffle<VM64:mode>_vis"): Likewise.
11647 ("cmask8<P:mode>_vis"): Likewise.
11648 ("cmask16<P:mode>_vis"): Likewise.
11649 ("cmask32<P:mode>_vis"): Likewise.
11650 ("pdistn<P:mode>_vis"): Likewise.
11651 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
11653 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11655 * config/sparc/sparc.md ("subtype"): New insn attribute.
11656 ("*wrgsr_sp64"): Set insn subtype.
11657 ("*rdgsr_sp64"): Likewise.
11658 ("alignaddrsi_vis"): Likewise.
11659 ("alignaddrdi_vis"): Likewise.
11660 ("alignaddrlsi_vis"): Likewise.
11661 ("alignaddrldi_vis"): Likewise.
11662 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
11663 ("fexpand_vis"): Likewise.
11664 ("fpmerge_vis"): Likewise.
11665 ("faligndata<VM64:mode>_vis"): Likewise.
11666 ("bshuffle<VM64:mode>_vis"): Likewise.
11667 ("cmask8<P:mode>_vis"): Likewise.
11668 ("cmask16<P:mode>_vis"): Likewise.
11669 ("cmask32<P:mode>_vis"): Likewise.
11670 ("fchksm16_vis"): Likewise.
11671 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
11672 ("fmean16_vis"): Likewise.
11673 ("fp<plusminus_insn>64_vis"): Likewise.
11674 ("<plusminus_insn>v8qi3"): Likewise.
11675 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
11676 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
11677 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
11678 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
11679 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
11680 ("*movqi_insn"): Likewise.
11681 ("*movhi_insn"): Likewise.
11682 ("*movsi_insn"): Likewise.
11683 ("movsi_pic_gotdata_op"): Likewise.
11684 ("*movdi_insn_sp32"): Likewise.
11685 ("*movdi_insn_sp64"): Likewise.
11686 ("movdi_pic_gotdata_op"): Likewise.
11687 ("*movsf_insn"): Likewise.
11688 ("*movdf_insn_sp32"): Likewise.
11689 ("*movdf_insn_sp64"): Likewise.
11690 ("*zero_extendhisi2_insn"): Likewise.
11691 ("*zero_extendqihi2_insn"): Likewise.
11692 ("*zero_extendqisi2_insn"): Likewise.
11693 ("*zero_extendqidi2_insn"): Likewise.
11694 ("*zero_extendhidi2_insn"): Likewise.
11695 ("*zero_extendsidi2_insn_sp64"): Likewise.
11696 ("ldfsr"): Likewise.
11697 ("prefetch_64"): Likewise.
11698 ("prefetch_32"): Likewise.
11699 ("tie_ld32"): Likewise.
11700 ("tie_ld64"): Likewise.
11701 ("*tldo_ldub_sp32"): Likewise.
11702 ("*tldo_ldub1_sp32"): Likewise.
11703 ("*tldo_ldub2_sp32"): Likewise.
11704 ("*tldo_ldub_sp64"): Likewise.
11705 ("*tldo_ldub1_sp64"): Likewise.
11706 ("*tldo_ldub2_sp64"): Likewise.
11707 ("*tldo_ldub3_sp64"): Likewise.
11708 ("*tldo_lduh_sp32"): Likewise.
11709 ("*tldo_lduh1_sp32"): Likewise.
11710 ("*tldo_lduh_sp64"): Likewise.
11711 ("*tldo_lduh1_sp64"): Likewise.
11712 ("*tldo_lduh2_sp64"): Likewise.
11713 ("*tldo_lduw_sp32"): Likewise.
11714 ("*tldo_lduw_sp64"): Likewise.
11715 ("*tldo_lduw1_sp64"): Likewise.
11716 ("*tldo_ldx_sp64"): Likewise.
11717 ("*mov<VM32:mode>_insn"): Likewise.
11718 ("*mov<VM64:mode>_insn_sp64"): Likewise.
11719 ("*mov<VM64:mode>_insn_sp32"): Likewise.
11721 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11723 * config/sparc/sparc.md ("type"): New insn type viscmp.
11724 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
11726 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
11727 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
11728 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
11729 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
11731 ("n7_vis_logical_11cycle"): Likewise.
11732 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
11733 * config/sparc/niagara2.md ("niag3_vis": Likewise.
11734 * config/sparc/niagara.md ("niag_vis"): Likewise.
11735 * config/sparc/ultra3.md ("us3_fga"): Likewise.
11736 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
11738 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11740 * config/sparc/sparc.md: New instruction type `bmask'.
11741 (bmaskdi_vis): Use the `bmask' type.
11742 (bmasksi_vis): Likewise.
11743 * config/sparc/ultra3.md (us3_array): Likewise.
11744 * config/sparc/niagara7.md (n7_array): Likewise.
11745 * config/sparc/niagara4.md (n4_array): Likewise.
11746 * config/sparc/niagara2.md (niag2_vis): Likewise.
11747 (niag3_vis): Likewise.
11748 * config/sparc/niagara.md (niag_vis): Likewise.
11750 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11752 * ipa-comdats.c: Remove optimize check from gate.
11753 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
11754 for functions not optimized.
11755 (ipa_fn_summary_read): Skip optimize check.
11756 (ipa_fn_summary_write): Likewise.
11757 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
11759 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
11761 (can_inline_edge_p): Check flag_pcc_struct_return for match.
11762 (check_callers): Give up on caller which is not optimized.
11763 (inline_small_functions): Likewise.
11764 (ipa_inline): Do not give up when not optimizing.
11765 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
11766 away unoptimizes cdtors.
11767 (whole_program_function_and_variable_visibility): Do
11768 ipa_discover_readonly_nonaddressable_vars in LTO mode.
11769 * ipa.c (process_references): Do not check optimize.
11770 (symbol_table::remove_unreachable_nodes): Update optimize check.
11771 (set_writeonly_bit): Update optimize check.
11772 (pass_ipa_cdtor_merge::gate): Do not check optimize.
11773 (pass_ipa_single_use::gate): Remove.
11775 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11777 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
11778 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
11779 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
11780 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
11781 permute_load, permute_store, adjust_extract, adjust_splat,
11782 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
11783 replace_swap_with_copy, dump_swap_insn_table,
11784 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
11785 recombine_lvx_pattern, recombine_stvx_pattern,
11786 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
11787 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
11788 to file rs6000-p8swap.c.
11789 * config/rs6000/rs6000-p8swap.c: New file.
11790 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
11791 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
11792 and rs6000*-*-* targets.
11794 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11796 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
11798 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11800 * lto-wrapper.c (merge_and_complain): Do not merge
11801 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
11802 fsigned_zeros, ftrapping_math, fwrapv.
11803 (append_compiler_options): Do not track these options.
11804 (append_linker_options): Likewie
11806 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11808 * cgraphunit.c (cgraph_node::finalize_function): When
11809 !flag_toplevel_reorde set no_reorder flag.
11810 (varpool_node::finalize_decl): Likewise.
11811 (symbol_table::compile): Drop no toplevel reorder path.
11813 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11815 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
11816 edges; zero probability is not better than uninitialized.
11818 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
11820 * asan.h (asan_sanitize_allocas_p): Declare.
11821 * asan.c (asan_sanitize_allocas_p): New function.
11822 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
11823 (handle_builtin_alloca): Likewise.
11824 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
11825 if !asan_sanitize_allocas_p.
11826 * params.def (asan-instrument-allocas): Add new option.
11827 * params.h (ASAN_PROTECT_ALLOCAS): Define.
11828 * opts.c (common_handle_option): Disable allocas sanitization for
11831 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
11833 * asan.c: Include gimple-fold.h.
11834 (get_last_alloca_addr): New function.
11835 (handle_builtin_stackrestore): Likewise.
11836 (handle_builtin_alloca): Likewise.
11837 (asan_emit_allocas_unpoison): Likewise.
11838 (get_mem_refs_of_builtin_call): Add new parameter, remove const
11839 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
11840 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
11841 (instrument_builtin_call): Pass gimple iterator to
11842 get_mem_refs_of_builtin_call.
11843 (last_alloca_addr): New global.
11844 * asan.h (asan_emit_allocas_unpoison): Declare.
11845 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
11846 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
11847 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
11848 if function calls alloca.
11849 * gimple-fold.c (replace_call_with_value): Remove static keyword.
11850 * gimple-fold.h (replace_call_with_value): Declare.
11851 * internal-fn.c: Include asan.h.
11852 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
11853 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
11855 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11857 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
11858 (C_SELFTEST_FLAGS): New.
11859 (CPP_SELFTEST_FLAGS): New.
11860 (SELFTEST_DEPS): New, from deps of s-selftest.
11861 (C_SELFTEST_DEPS): New, from deps of s-selftest.
11862 (CPP_SELFTEST_DEPS): New.
11863 (selftest): Add dependency on s-selftest-c++.
11864 (s-selftest): Rename to...
11865 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
11866 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
11867 than SELFTEST_FLAGS.
11868 (selftest-gdb): Rename to...
11869 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
11871 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
11872 (selftest-valgrind): Rename to...
11873 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
11875 (selftest-valgrind): Reintroduce as an alias for
11876 selftest-c-valgrind.
11877 (s-selftest-c++): New.
11878 (selftest-c++-gdb): New.
11879 (selftest-c++-valgrind): New.
11881 2017-07-06 Olivier Hainque <hainque@adacore.com>
11883 * gcc.c (process_command): When deciding if undefined variables
11884 should be ignored when processing specs, accept "gcc -v" as well.
11886 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11888 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
11889 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
11891 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
11893 * config/arm/arm-cpus.in (armv8-r): Add new entry.
11894 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
11895 * config/arm/arm-tables.opt: Regenerate.
11896 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
11898 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
11900 2017-07-06 Carl Love <cel@us.ibm.com>
11902 * ChangeLog: Clean up from mid air collision
11904 2017-07-06 Carl Love <cel@us.ibm.com>
11906 * config/rs6000/rs6000-c.c: Add support for built-in functions
11907 vector signed int vec_subc (vector signed int, vector signed int);
11908 vector signed __int128 vec_subc (vector signed __int128,
11909 vector signed __int128);
11910 vector unsigned __int128 vec_subc (vector unsigned __int128,
11911 vector unsigned __int128);
11912 vector signed int vec_sube (vector signed int, vector signed int,
11913 vector signed int);
11914 vector unsigned int vec_sube (vector unsigned int,
11915 vector unsigned int,
11916 vector unsigned int);
11917 vector signed __int128 vec_sube (vector signed __int128,
11918 vector signed __int128,
11919 vector signed__int128);
11920 vector unsigned __int128 vec_sube (vector unsigned __int128,
11921 vector unsigned __int128,
11922 vector unsigned __int128);
11923 vector signed int vec_subec (vector signed int, vector signed int,
11924 vector signed int);
11925 vector unsigned int vec_subec (vector unsigned int,
11926 vector unsigned int,
11927 vector unsigned int);
11928 vector signed __int128 vec_subec (vector signed __int128,
11929 vector signed __int128,
11930 vector signed__int128);
11931 vector unsigned __int128 vec_subec (vector unsigned __int128,
11932 vector unsigned __int128,
11933 vector unsigned __int128);
11934 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
11935 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
11936 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
11937 BU_ALTIVEC_OVERLOAD_X definitions.
11938 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
11939 * doc/extend.texi: Update the built-in documentation file for the new
11940 built-in functions.
11942 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11945 * diagnostic-show-locus.c (layout::layout): Use start and finish
11946 spelling location for the start and finish of each range.
11947 * genmatch.c (linemap_client_expand_location_to_spelling_point):
11948 Add unused aspect param.
11949 * input.c (expand_location_1): Add "aspect" param, and use it
11950 to access the correct part of the location.
11951 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
11953 (expand_location_to_spelling_point): Likewise.
11954 (linemap_client_expand_location_to_spelling_point): Add "aspect"
11955 param, and pass it to expand_location_1.
11957 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
11959 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
11960 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
11961 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
11962 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
11963 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
11964 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
11965 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
11966 _mm_maskz_getmant_ss): New intrinsics.
11967 (__builtin_ia32_getexpss128_mask): Changed to ...
11968 __builtin_ia32_getexpss128_round ... this.
11969 (__builtin_ia32_getexpsd128_mask): Changed to ...
11970 __builtin_ia32_getexpsd128_round ... this.
11971 * config/i386/i386-builtin-types.def
11972 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
11973 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
11974 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
11975 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
11976 __builtin_ia32_getmantss_mask_round): New builtins.
11977 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
11978 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
11979 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
11980 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
11981 * config/i386/sse.md
11982 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
11983 avx512f_sgetexp<mode><mask_scalar_name>
11984 <round_saeonly_scalar_name> ... this.
11985 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
11986 %0, %1, %2<round_saeonly_op3>}): Changed to ...
11987 vgetexp<ssescalarmodesuffix>
11988 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11989 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
11990 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
11991 avx512f_vgetmant<mode><mask_scalar_name>
11992 <round_saeonly_scalar_name> ... this.
11993 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
11994 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
11995 vgetmant<ssescalarmodesuffix>
11996 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
11997 %0<mask_scalar_operand4>, %1, %2
11998 <round_saeonly_scalar_mask_op4>, %3} ... this.
11999 * config/i386/subst.md (mask_scalar_operand4,
12000 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
12001 round_saeonly_scalar_nimm_predicate): New subst attributes.
12003 2017-07-06 Julia Koval <julia.koval@intel.com>
12005 * config/i386/i386.c (ix86_erase_embedded_rounding):
12006 Remove code for old rounding pattern.
12008 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
12010 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
12012 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
12014 * doc/sourcebuild.texi (Test Directives, Variants of
12015 dg-require-support): Add documentation for dg-require-stack-check.
12017 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
12019 * config/i386/subst.md (mask_scalar, round_scalar,
12020 round_saeonly_scalar): New meta-templates.
12021 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
12022 round_scalar_mask_operand3, round_scalar_mask_op3,
12023 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
12024 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
12025 round_saeonly_scalar_constraint,
12026 round_saeonly_scalar_prefix): New subst attribute.
12027 * config/i386/sse.md
12028 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
12029 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
12030 <round_scalar_name> ... this.
12031 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
12032 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
12033 <round_scalar_name> ... this.
12034 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
12035 <sse>_vm<code><mode>3<mask_scalar_name>
12036 <round_saeonly_scalar_name> ... this.
12037 (v<plusminus_mnemonic><ssescalarmodesuffix>
12038 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
12039 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
12040 v<plusminus_mnemonic><ssescalarmodesuffix>
12041 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12042 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
12043 (v<multdiv_mnemonic><ssescalarmodesuffix>
12044 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
12045 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
12046 v<multdiv_mnemonic><ssescalarmodesuffix>
12047 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12048 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
12049 (v<maxmin_float><ssescalarmodesuffix>
12050 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
12051 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
12052 v<maxmin_float><ssescalarmodesuffix>
12053 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12054 %0<mask_scalar_operand3>, %1, %<iptr>2
12055 <round_saeonly_scalar_mask_op3>} ... this.
12057 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
12059 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
12060 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
12062 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
12063 Alan Hayward <alan.hayward@arm.com>
12064 David Sherwood <david.sherwood@arm.com>
12066 * combine.c (simplify_if_then_else): Remove "enum" before
12068 * compare-elim.c (can_eliminate_compare): Likewise.
12069 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
12071 (aarch64_lookup_simd_builtin_type): Likewise.
12072 (aarch64_simd_builtin_type): Likewise.
12073 (aarch64_init_simd_builtin_types): Likewise.
12074 (aarch64_simd_expand_args): Likewise.
12075 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
12077 (aarch64_reverse_mask): Likewise.
12078 (aarch64_simd_emit_reg_reg_move): Likewise.
12079 (aarch64_gen_adjusted_ldpstp): Likewise.
12080 (aarch64_ccmp_mode_to_code): Likewise.
12081 (aarch64_operands_ok_for_ldpstp): Likewise.
12082 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12083 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
12085 (aarch64_min_divisions_for_recip_mul): Likewise.
12086 (aarch64_reassociation_width): Likewise.
12087 (aarch64_get_condition_code_1): Likewise.
12088 (aarch64_simd_emit_reg_reg_move): Likewise.
12089 (aarch64_simd_attr_length_rglist): Likewise.
12090 (aarch64_reverse_mask): Likewise.
12091 (aarch64_operands_ok_for_ldpstp): Likewise.
12092 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12093 (aarch64_gen_adjusted_ldpstp): Likewise.
12094 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
12096 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
12097 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
12098 (arm_lookup_simd_builtin_type): Likewise.
12099 (arm_simd_builtin_type): Likewise.
12100 (arm_init_simd_builtin_types): Likewise.
12101 (arm_expand_builtin_args): Likewise.
12102 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
12103 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
12104 (ft32_setup_incoming_varargs): Likewise.
12105 (ft32_function_arg): Likewise.
12106 (ft32_function_arg_advance): Likewise.
12107 (ft32_pass_by_reference): Likewise.
12108 (ft32_arg_partial_bytes): Likewise.
12109 (ft32_valid_pointer_mode): Likewise.
12110 (ft32_addr_space_pointer_mode): Likewise.
12111 (ft32_addr_space_legitimate_address_p): Likewise.
12112 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
12114 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
12115 (ix86_emit_outlined_ms2sysv_restore): Likewise.
12116 (iamcu_alignment): Likewise.
12117 (canonicalize_vector_int_perm): Likewise.
12118 (ix86_noce_conversion_profitable_p): Likewise.
12119 (ix86_mpx_bound_mode): Likewise.
12120 (ix86_operands_ok_for_move_multiple): Likewise.
12121 * config/microblaze/microblaze-protos.h
12122 (microblaze_expand_conditional_branch_reg): Likewise.
12123 * config/microblaze/microblaze.c
12124 (microblaze_expand_conditional_branch_reg): Likewise.
12125 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12127 (rs6000_reassociation_width): Likewise.
12128 (rs6000_invalid_binary_op): Likewise.
12129 (fusion_p9_p): Likewise.
12130 (emit_fusion_p9_load): Likewise.
12131 (emit_fusion_p9_store): Likewise.
12132 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
12134 (riscv_hard_regno_mode_ok_p): Likewise.
12135 (riscv_address_insns): Likewise.
12136 (riscv_split_symbol): Likewise.
12137 (riscv_legitimize_move): Likewise.
12138 (riscv_function_value): Likewise.
12139 (riscv_hard_regno_nregs): Likewise.
12140 (riscv_expand_builtin): Likewise.
12141 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
12142 (riscv_build_integer): Likewise.
12143 (riscv_split_integer): Likewise.
12144 (riscv_legitimate_constant_p): Likewise.
12145 (riscv_cannot_force_const_mem): Likewise.
12146 (riscv_regno_mode_ok_for_base_p): Likewise.
12147 (riscv_valid_base_register_p): Likewise.
12148 (riscv_valid_offset_p): Likewise.
12149 (riscv_valid_lo_sum_p): Likewise.
12150 (riscv_classify_address): Likewise.
12151 (riscv_legitimate_address_p): Likewise.
12152 (riscv_address_insns): Likewise.
12153 (riscv_load_store_insns): Likewise.
12154 (riscv_force_binary): Likewise.
12155 (riscv_split_symbol): Likewise.
12156 (riscv_force_address): Likewise.
12157 (riscv_legitimize_address): Likewise.
12158 (riscv_move_integer): Likewise.
12159 (riscv_legitimize_const_move): Likewise.
12160 (riscv_legitimize_move): Likewise.
12161 (riscv_address_cost): Likewise.
12162 (riscv_subword): Likewise.
12163 (riscv_output_move): Likewise.
12164 (riscv_canonicalize_int_order_test): Likewise.
12165 (riscv_emit_int_order_test): Likewise.
12166 (riscv_function_arg_boundary): Likewise.
12167 (riscv_pass_mode_in_fpr_p): Likewise.
12168 (riscv_pass_fpr_single): Likewise.
12169 (riscv_pass_fpr_pair): Likewise.
12170 (riscv_get_arg_info): Likewise.
12171 (riscv_function_arg): Likewise.
12172 (riscv_function_arg_advance): Likewise.
12173 (riscv_arg_partial_bytes): Likewise.
12174 (riscv_function_value): Likewise.
12175 (riscv_pass_by_reference): Likewise.
12176 (riscv_setup_incoming_varargs): Likewise.
12177 (riscv_print_operand): Likewise.
12178 (riscv_elf_select_rtx_section): Likewise.
12179 (riscv_save_restore_reg): Likewise.
12180 (riscv_for_each_saved_reg): Likewise.
12181 (riscv_register_move_cost): Likewise.
12182 (riscv_hard_regno_mode_ok_p): Likewise.
12183 (riscv_hard_regno_nregs): Likewise.
12184 (riscv_class_max_nregs): Likewise.
12185 (riscv_memory_move_cost): Likewise.
12186 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
12187 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
12188 (rl78_addr_space_address_mode): Likewise.
12189 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12191 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12192 (rs6000_reassociation_width): Likewise.
12193 (rs6000_invalid_binary_op): Likewise.
12194 (fusion_p9_p): Likewise.
12195 (emit_fusion_p9_load): Likewise.
12196 (emit_fusion_p9_store): Likewise.
12197 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
12198 (ok_for_simple_move_operands): Likewise.
12199 (ok_for_simple_move_strict_operands): Likewise.
12200 (ok_for_simple_arith_logic_operands): Likewise.
12201 (visium_legitimize_reload_address): Likewise.
12202 (visium_select_cc_mode): Likewise.
12203 (output_cbranch): Likewise.
12204 (visium_split_double_move): Likewise.
12205 (visium_expand_copysign): Likewise.
12206 (visium_expand_int_cstore): Likewise.
12207 (visium_expand_fp_cstore): Likewise.
12208 * config/visium/visium.c (visium_pass_by_reference): Likewise.
12209 (visium_function_arg): Likewise.
12210 (visium_function_arg_advance): Likewise.
12211 (visium_libcall_value): Likewise.
12212 (visium_setup_incoming_varargs): Likewise.
12213 (visium_legitimate_constant_p): Likewise.
12214 (visium_legitimate_address_p): Likewise.
12215 (visium_legitimize_address): Likewise.
12216 (visium_secondary_reload): Likewise.
12217 (visium_register_move_cost): Likewise.
12218 (visium_memory_move_cost): Likewise.
12219 (prepare_move_operands): Likewise.
12220 (ok_for_simple_move_operands): Likewise.
12221 (ok_for_simple_move_strict_operands): Likewise.
12222 (ok_for_simple_arith_logic_operands): Likewise.
12223 (visium_function_value_1): Likewise.
12224 (rtx_ok_for_offset_p): Likewise.
12225 (visium_legitimize_reload_address): Likewise.
12226 (visium_split_double_move): Likewise.
12227 (visium_expand_copysign): Likewise.
12228 (visium_expand_int_cstore): Likewise.
12229 (visium_expand_fp_cstore): Likewise.
12230 (visium_split_cstore): Likewise.
12231 (visium_select_cc_mode): Likewise.
12232 (visium_split_cbranch): Likewise.
12233 (output_cbranch): Likewise.
12234 (visium_print_operand_address): Likewise.
12235 * expmed.c (flip_storage_order): Likewise.
12236 * expmed.h (emit_cstore): Likewise.
12237 (flip_storage_order): Likewise.
12238 * genrecog.c (validate_pattern): Likewise.
12239 * hsa-gen.c (gen_hsa_addr): Likewise.
12240 * internal-fn.c (expand_arith_overflow): Likewise.
12241 * ira-color.c (allocno_copy_cost_saving): Likewise.
12242 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12243 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
12244 (process_invariant_for_inheritance): Likewise.
12245 * lra-eliminations.c (move_plus_up): Likewise.
12246 * omp-low.c (lower_oacc_reductions): Likewise.
12247 * simplify-rtx.c (simplify_subreg): Likewise.
12248 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
12249 (TARGET_CHKP_BOUND_MODE): Likewise..
12250 * targhooks.c (default_chkp_bound_mode): Likewise.
12251 (default_setup_incoming_vararg_bounds): Likewise.
12252 * targhooks.h (default_chkp_bound_mode): Likewise.
12253 (default_setup_incoming_vararg_bounds): Likewise.
12254 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
12255 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
12256 (have_whole_vector_shift): Likewise.
12257 * tree-vect-stmts.c (vectorizable_load): Likewise.
12258 * doc/tm.texi: Regenerate.
12260 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
12262 Graceful degrade if Binutils PR21472 is not available.
12265 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
12266 .rodata in flash test fails.
12267 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
12268 * confgure: Regenerate.
12269 * config.in: Regenerate.
12270 * config/avr/avr.c (avr_asm_named_section)
12271 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
12272 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
12273 (avr_asm_init_sections): Same.
12275 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12277 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
12278 (fma<VH:mode>4_intrinsic): Likewise.
12279 (*fmsub<VCVTF:mode>4): Likewise.
12280 (*fmsub<VH:mode>4_intrinsic): Likewise.
12282 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
12285 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
12286 Don't depend on "optimize > 0".
12287 (out_movhi_r_mr, out_movqi_mr_r): Same.
12288 (out_movhi_mr_r, out_movqi_r_mr): Same.
12289 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
12290 io_address_operand on "optimize > 0".
12292 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12294 * tree-loop-distribution.c: Add general explanantion on the pass.
12295 (generate_loops_for_partition): Mark distributed loop.
12296 (pg_add_dependence_edges): New parameter. Handle alias data
12297 dependence specially and record it in the parameter if asked.
12298 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
12299 (init_partition_graph_vertices, add_partition_graph_edge): New.
12300 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
12301 (free_partition_graph_vdata, build_partition_graph): New.
12302 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
12303 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
12304 (data_ref_segment_size, latch_dominated_by_data_ref): New.
12305 (compute_alias_check_pairs, version_loop_by_alias_check): New.
12306 (version_for_distribution_p, finalize_partitions): New.
12307 (distribute_loop): Handle alias data dependence specially. Factor
12308 out loop fusion code as functions and call these functions.
12310 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12312 * tree-loop-distribution.c (classify_partition): New parameter and
12313 better handle reduction statement.
12314 (rdg_build_partitions): Revise comment.
12315 (distribute_loop): Compute statements in all partitions and pass it
12316 to classify_partition.
12318 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12320 * tree-loop-distribution.c (enum partition_type): New.
12321 (struct partition): New field type.
12322 (partition_merge_into): Add parameter. Update partition type.
12323 (data_dep_in_cycle_p, update_type_for_merge): New functions.
12324 (build_rdg_partition_for_vertex): Compute partition type.
12325 (rdg_build_partitions): Dump partition type.
12326 (distribute_loop): Update calls to partition_merge_into.
12328 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12330 * tree-loop-distribution.c (struct ddr_hasher): New.
12331 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
12333 (classify_partition): Call get_data_dependence.
12334 (pg_add_dependence_edges): Ditto.
12335 (distribute_loop): Release data dependence hash table.
12337 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12339 * tree-loop-distribution.c (ref_base_address): Delete.
12340 (similar_memory_accesses): Rename ...
12341 (share_memory_accesses): ... to this. Check if partitions access
12342 the same memory reference.
12343 (distribute_loop): Call share_memory_accesses.
12345 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12347 * tree-loop-distribution.c (struct partition): New field recording
12348 its data reference.
12349 (partition_alloc, partition_free): Init and release data refs.
12350 (partition_merge_into): Merge data refs.
12351 (build_rdg_partition_for_vertex): Collect data refs for partition.
12352 (pg_add_dependence_edges): Change parameters from vector to bitmap.
12354 (distribute_loop): Remve data refs from vertice data of partition
12357 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12359 * tree-loop-distribution.c (params.h): Include header file.
12360 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
12361 (datarefs_vec): New global var.
12362 (create_rdg_vertices): Use datarefs_vec directly.
12363 (free_rdg): Don't free data references.
12364 (build_rdg): Update use. Don't free data references.
12365 (distribute_loop): Compute global variable for data references.
12366 Bail out if there are too many data references.
12368 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12370 * tree-loop-distribution.c (loop_nest): New global var.
12371 (build_rdg): Use loop directly, rather than loop nest.
12372 (pg_add_dependence_edges): Remove loop nest parameter. Use global
12374 (distribute_loop): Compute global variable loop nest. Update use.
12376 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12378 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
12379 (partition_merge_into): New parameter. Dump reason for fusion.
12380 (distribute_loop): Update use of partition_merge_into.
12382 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12384 * tree-loop-distribution.c (bb_top_order_index): New.
12385 (bb_top_order_index_size, bb_top_order_cmp): New.
12386 (stmts_from_loop): Use topological order.
12387 (pass_loop_distribution::execute): Compute and release topological
12388 order for basic blocks.
12390 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12392 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
12395 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12397 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
12398 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
12399 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
12400 * internal-fn.def (LOOP_DIST_ALIAS): New.
12401 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
12402 (fold_loop_internal_call): ... this.
12403 (vect_loop_dist_alias_call): New function.
12404 (set_uid_loop_bbs): Call fold_loop_internal_call.
12405 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
12408 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
12411 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
12412 Require dead FLAGS_REG at the beginning of a peephole.
12414 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
12417 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
12418 arguments in the call to __builtin_ia32_sbb_u32.
12419 (_subborrow_u64): Swap _X and _Y arguments in the call to
12420 __builtin_ia32_sbb_u64.
12422 2017-07-04 Jakub Jelinek <jakub@redhat.com>
12425 * tree-vrp.c (compare_assert_loc): Turn into a function template
12426 with stable template parameter. Only test if a->e is NULL,
12427 !a->e == !b->e has been verified already. Use e == NULL or
12428 e != NULL instead of e or ! e tests. If stable is true, don't use
12429 iterative_hash_expr, on the other side allow a or b or both NULL
12430 and sort the NULLs last.
12431 (process_assert_insertions): Sort using compare_assert_loc<false>
12432 instead of compare_assert_loc, later sort using
12433 compare_assert_loc<true> before calling process_assert_insertions_for
12434 in a loop. Use break instead of continue once seen NULL pointer.
12436 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
12438 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
12439 Cortex-R7 and Cortex-R8 processors.
12441 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12443 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
12444 uninitialized while src is not.
12446 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
12448 * common/config/arm/arm-common.c: Adjust include path for
12450 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
12451 (arm-cpu.h): Create in build directory. Adjust dependency rules.
12452 (arm-cpu-data.h): Likewise.
12453 (arm-cpu-cdata.h): Likewise.
12454 * config/arm/arm-cpu.h: Delete.
12455 * config/arm/arm-cpu-cdata.h: Delete.
12456 * config/arm/arm-cpu-data.h: Delete.
12458 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
12460 * config/arm/arm-cpus.in (cortex-a55): New.
12461 (cortex-a75): Likewise.
12462 (cortex-a75.cortex-a55): Likewise.
12463 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
12465 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
12466 * config/arm/arm-cpu-cdata.h: Regenerate.
12467 * config/arm/arm-cpu-data.h: Regenerate.
12468 * config/arm/arm-cpu.h: Regenerate.
12469 * config/arm/arm-tables.opt: Regenerate.
12470 * config/arm/arm-tune.md: Regenerate.
12472 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12474 * haifa-sched.c (sched_create_recovery_edges): Update profile.
12476 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12478 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
12481 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
12483 PR tree-optimization/81292
12484 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
12485 full_string_p, also call adjust_related_strinfos if the adjustment
12486 is simple, otherwise invalidate related strinfos.
12488 2017-07-04 Martin Liska <mliska@suse.cz>
12491 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
12492 newly created variable as DECL_IGNORED_P.
12494 2017-07-04 Martin Liska <mliska@suse.cz>
12497 * ipa-inline.c (inline_small_functions):
12498 Use xstrdup_for_dump.
12500 2017-07-04 Tom de Vries <tom@codesourcery.com>
12502 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
12504 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
12507 * config/darwin.c (darwin_function_switched_text_sections):
12510 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
12512 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
12514 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
12516 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
12518 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12520 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
12521 min_profitable_iters, and th as inclusive lower bounds.
12522 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
12523 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
12524 for min_profitable_iters and min_profitable_estimate.
12525 (vect_transform_loop): Treat th as an inclusive lower bound.
12526 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
12528 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
12531 * config/darwin.c (darwin_function_switched_text_sections):
12532 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
12533 in two pieces, and suppress the use of buf.
12535 2017-07-03 Nathan Sidwell <nathan@acm.org>
12537 * hash-table.h (hash_table_mod1): Fix indentation.
12539 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12541 PR middle-end/81290
12542 * predict.c (force_edge_cold): Be more careful about propagation
12544 * profile-count.h (profile_probability::guessed,
12545 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
12547 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
12549 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
12551 * doc/invoke.texi (rcpc architecture extension): Document it.
12553 2017-07-03 Richard Biener <rguenther@suse.de>
12555 PR tree-optimization/60510
12556 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
12557 the scalar reduction PHI and use it.
12558 (vectorizable_reduction): Properly guard the single_defuse_cycle
12559 path for non-SLP reduction chains where we cannot use it.
12560 Rework reduc_def/index and vector type deduction. Rework
12561 vector operand gathering during reduction op code-gen.
12562 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
12563 chains dissolve the chain and leave it to non-SLP reduction
12566 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12568 * tree-data-ref.h (dr_alignment): Declare.
12569 * tree-data-ref.c (dr_alignment): New function.
12570 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
12571 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
12573 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
12575 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12577 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
12578 and base_misalignment fields.
12579 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
12580 * tree-data-ref.c: Include builtins.h.
12581 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
12582 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
12583 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
12584 * tree-vect-data-refs.c: Include tree-cfg.h.
12585 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
12586 fields instead of calculating an alignment here.
12587 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
12588 innermost_loop_behavior fields.
12590 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12592 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
12594 (DR_STEP_ALIGNMENT): New macro.
12595 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
12596 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
12597 (create_data_ref): Print it.
12598 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
12599 to tell whether the step preserves vector (mis)alignment.
12600 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
12601 Move the check for an integer step and generalise to all INTEGER_CST.
12602 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
12603 Print the outer step alignment.
12605 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12607 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
12608 with offset_alignment.
12609 (DR_ALIGNED_TO): Delete.
12610 (DR_OFFSET_ALIGNMENT): New macro.
12611 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
12612 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
12613 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
12614 (create_data_ref): Likewise.
12615 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
12616 (vect_analyze_data_refs): Likewise.
12617 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
12618 creating dummy innermost behavior.
12620 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12622 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
12623 with a "innermost_loop_behavior *" and refeence tree.
12624 * tree-data-ref.c (dr_analyze_innermost): Likewise.
12625 (create_data_ref): Update call accordingly.
12626 * tree-predcom.c (find_looparound_phi): Likewise.
12628 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12630 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
12631 fields with dr_wrt_vec_loop.
12632 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
12633 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
12634 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
12635 (vect_dr_behavior): New function.
12636 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
12637 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
12638 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
12639 track whether the step preserves the misalignment.
12640 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
12641 Use vect_dr_behavior.
12642 (vect_setup_realignment): Update call accordingly.
12643 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
12644 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
12645 call to vect_create_addr_base_for_vector_ref.
12646 (vect_create_cond_for_align_checks): Likewise.
12647 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
12648 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
12649 (vect_recog_mask_conversion_pattern): Likewise.
12650 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
12651 (new_stmt_vec_info): Remove redundant zeroing.
12653 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
12655 * common/config/arm/arm-common.c (arm_be8_option): New function.
12656 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
12657 (ISA_ARMv6): Add isa_bit_be8.
12658 * config/arm/arm.h (arm_be8_option): Add prototype.
12659 (BE8_SPEC_FUNCTION): New define.
12660 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
12661 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
12662 (mlittle-endian): Similarly.
12663 (mbe8, mbe32): New options.
12664 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
12665 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
12667 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12669 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
12671 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12673 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
12674 (cleanup_tree_cfg_bb): Use it.
12675 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
12677 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
12679 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12682 * loop-doloop.c (add_test): Update profile.
12684 2017-07-03 Martin Liska <mliska@suse.cz>
12687 * sanopt.c (rewrite_usage_of_param): New function.
12688 (sanitize_rewrite_addressable_params): Likewise.
12689 (pass_sanopt::execute): Call rewrite_usage_of_param.
12691 2017-07-03 Richard Biener <rguenther@suse.de>
12693 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
12694 back to using VIEW_CONVERT_EXPR.
12696 2017-07-03 Martin Liska <mliska@suse.cz>
12699 * doc/extend.texi: Document when a resolver function is
12700 generated for target_clones.
12702 2017-07-03 Martin Liska <mliska@suse.cz>
12704 * asan.c (asan_emit_stack_protection): Unpoison just red zones
12705 and shadow memory of auto variables which are subject of
12706 use-after-scope sanitization.
12707 (asan_expand_mark_ifn): Add do set only when is_poison.
12709 2016-07-03 Richard Biener <rguenther@suse.de>
12711 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
12713 (vect_force_simple_reduction): Record reduction def -> phi mapping.
12714 (vectorizable_reduction): Perform reduction PHI creation when
12715 visiting a reduction PHI and adjust and simplify code generation
12716 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
12717 (vect_transform_loop): Visit reduction PHIs.
12718 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
12719 defs into the SLP tree.
12720 (vect_build_slp_tree): Reduction defs terminate the recursion.
12721 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
12723 (vect_get_vec_defs_for_stmt_copy): Export.
12724 (vect_get_vec_defs): Likewise.
12725 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
12727 (vect_get_vec_defs_for_stmt_copy): Declare.
12728 (vect_get_vec_defs): Likewise.
12730 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12732 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
12733 parameter with a "loop" parameter and use it instead of the
12734 loop containing DR_STMT. Don't check simple_iv when doing
12735 BB analysis. Describe the two analysis modes in the comment.
12737 2017-07-03 Tom de Vries <tom@codesourcery.com>
12739 PR tree-optimization/69468
12740 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
12741 (find_same_succ_bb): Handle ignore_edge_flags.
12743 2017-07-03 Tom de Vries <tom@codesourcery.com>
12745 PR tree-optimization/81192
12746 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
12748 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
12750 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
12752 2017-07-03 Tom de Vries <tom@codesourcery.com>
12754 PR tree-optimization/81192
12755 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
12756 BB_SAME_SUCC (bb) == NULL.
12758 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12760 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
12763 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12765 * dumpfile.c: Include profile-count.h
12766 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
12768 (insert_cond_bb): Update profile.
12769 * tree-cfg.h (insert_cond_bb): Update prototype.
12770 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
12771 * tree-dump.c: Do not include tree-cfg.
12773 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12775 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
12777 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12779 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
12782 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12784 * tree-complex.c (expand_complex_div_wide): update profile.
12786 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12787 Alan Hayward <alan.hayward@arm.com>
12788 David Sherwood <david.sherwood@arm.com>
12790 * Makefile.in (MACHMODE_H): Remove insn-modes.h
12791 (CORETYPES_H): New define.
12792 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
12793 (insn-modes-inline.h, s-modes-inline-h): New rules.
12794 (generated_files): Add insn-modes-inline.h.
12795 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
12796 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
12797 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
12798 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
12799 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
12800 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
12801 (build/gencodes.o, build/genconditions.o): Likewise.
12802 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
12803 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
12804 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
12805 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
12806 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
12807 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
12808 * coretypes.h: Include everything up to real.h for generators.
12809 Include insn-modes.h first. Include wide-int-print.h after
12810 wide-int.h. Include insn-modes-inline.h and then machmode.h.
12811 * machmode.h: Don't include insn-modes.h here.
12812 * function-tests.c: Remove includes of signop.h, machmode.h,
12813 double-int.h and wide-int.h.
12815 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
12817 * optc-save-gen.awk: Likewise.
12818 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
12819 * godump.c: Remove include of wide-int-print.h.
12820 * pretty-print.h: Likewise.
12821 * wide-int-print.cc: Likewise.
12822 * wide-int.cc: Likewise.
12823 * hash-map-tests.c: Remove include of signop.h.
12824 * hash-set-tests.c: Likewise.
12825 * rtl-tests.c: Likewise.
12826 * mkconfig.sh: Remove include of machmode.h.
12827 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
12829 (emit_insn_modes_inline_h): ...this new function. Emit the code
12830 into an insn-modes-inline.h header file, adding appropriate
12831 include guards and end comments.
12832 (emit_insn_modes_c_header): Remove include of machmode.h.
12833 (emit_min_insn_modes_c_header): Include coretypes.h rather than
12835 (main): Handle -i flag and call emit_insn_modes_inline_h when
12838 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12840 * tree-ssa-strlen.c (strinfo): Rename the length field to
12841 nonzero_chars. Add a full_string_p field.
12842 (compare_nonzero_chars, zero_length_string_p): New functions.
12843 (get_addr_stridx): Add an offset_out parameter.
12844 Use compare_nonzero_chars.
12845 (get_stridx): Update accordingly. Use compare_nonzero_chars.
12846 (new_strinfo): Update after above changes to strinfo.
12847 (set_endptr_and_length): Set full_string_p.
12848 (get_string_length): Update after above changes to strinfo.
12849 (unshare_strinfo): Update call to new_strinfo.
12850 (maybe_invalidate): Likewise.
12851 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
12852 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
12853 as a uhwi instead of an shwi. Update after above changes to
12854 strinfo and new_strinfo.
12855 (zero_length_string): Assert that chainsi contains full strings.
12856 Use zero_length_string_p. Update call to new_strinfo.
12857 (adjust_related_strinfos): Update after above changes to strinfo.
12858 Copy full_string_p from origsi.
12859 (adjust_last_stmt): Use zero_length_string_p.
12860 (handle_builtin_strlen): Update after above changes to strinfo and
12861 new_strinfo. Install the lhs as the string length if the previous
12862 entry didn't describe a full string.
12863 (handle_builtin_strchr): Update after above changes to strinfo
12865 (handle_builtin_strcpy): Likewise.
12866 (handle_builtin_strcat): Likewise.
12867 (handle_builtin_malloc): Likewise.
12868 (handle_pointer_plus): Likewise.
12869 (handle_builtin_memcpy): Likewise. Track nonzero characters
12870 that aren't necessarily followed by a nul terminator.
12871 (handle_char_store): Likewise.
12873 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12875 PR tree-optimization/80769
12876 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
12877 for malloc and calloc. Document the new invariant that all related
12878 strinfos have delayed lengths or none do.
12879 (verify_related_strinfos): Move earlier in file.
12880 (set_endptr_and_length): New function, split out from...
12881 (get_string_length): ...here. Also set the lengths of related
12883 (zero_length_string): Assert that chainsi has known (rather than
12885 (adjust_related_strinfos): Likewise.
12887 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12889 PR tree-optimization/81136
12890 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
12891 assert that two references with the same misalignment have the same
12892 compile-time misalignment if those compile-time misalignments
12895 2017-07-01 Andi Kleen <ak@linux.intel.com>
12897 * print-tree.c (print_node): Print all attributes.
12899 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12901 * cfg.c (scale_bbs_frequencies): New function.
12902 * cfg.h (scale_bbs_frequencies): Declare it.
12903 * cfgloopanal.c (single_likely_exit): Cleanup.
12904 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
12906 (scale_loop_profile): Likewise.
12907 (loop_version): Likewise.
12908 (create_empty_loop_on_edge): Update.
12909 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
12910 scale_loop_frequencies, scale_loop_profile, loopify,
12911 loop_version): Update prototypes.
12912 * modulo-sched.c (sms_schedule): Update.
12913 * predict.c (unlikely_executed_edge_p): Also check probability.
12914 (probably_never_executed_edge_p): Fix typo.
12915 * tree-if-conv.c (version_loop_for_if_conversion): Update.
12916 * tree-parloops.c (gen_parallel_loop): Update.
12917 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
12918 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
12919 * tree-ssa-loop-split.c (split_loop): Update.
12920 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
12921 * tree-vect-loop-manip.c (vect_do_peeling): Update.
12922 (vect_loop_versioning): Update.
12923 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
12925 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12927 * trans-mem.c (split_bb_make_tm_edge): Update profile.
12929 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12931 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
12932 to keep profile consistent.
12934 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12936 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
12937 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
12938 * profile-count.h (max_safe_multiplier): Make unsigned.
12939 (profile_count::guessed_zero): New.
12941 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12943 * bb-reorder.c (fix_up_crossing_landing_pad,
12944 fix_crossing_conditional_branches): Use make_single_succ_edge
12945 to keep profile consistent.
12947 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12949 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
12952 2017-07-01 Jakub Jelinek <jakub@redhat.com>
12955 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
12956 the right scopes, make sure cond_jump isn't preserved between multiple
12957 iterations. Search for fallthru edge whenever there are 3+ edges and
12958 use find_fallthru_edge for it.
12960 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12962 Patch by Alexander Monakov <amonakov@ispras.ru>
12963 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
12964 probabilities consistently.
12966 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12968 * pa.c (pa_expand_compare_and_swap_loop): Update call of
12969 emit_cmp_and_jump_insns.
12971 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12974 * tree-inline.c (expand_call_inline): Combine profile statuses.
12976 2017-06-30 Andrew Pinski <apinski@cavium.com>
12978 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
12979 fold_stmt returned true.
12981 2017-06-30 Nathan Sidwell <nathan@acm.org>
12983 * ggc.h (empty_string): Delete.
12984 * cfgexpand.c (expand_asm_stmt): Use plain "".
12985 * optabs.c (expand_asm_memory_barrier): Likewise.
12986 * stringpool.c (empty_string): Delete.
12987 (digit_vector, digit_string): Delete.
12988 (ggc_alloc_string): Use plain "", don't optimize single digit
12989 strings. Use ggc_alloc_atomic.
12991 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
12993 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
12994 comparison set and one other set, use the cost of the non-comparison
12997 2017-06-30 Nathan Sidwell <nathan@acm.org>
12999 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
13002 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
13004 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
13005 loops. Remove now unneeded calls to gimple_switch_set_label() that
13006 just set removed labels to NULL_TREE.
13008 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
13010 * tree-ssanames.c (set_range_info_raw): Abstract from ...
13011 (set_range_info): ...here. Only call set_range_info_raw if domain
13013 (set_nonzero_bits): Call set_range_info_raw.
13014 * tree-ssanames.h (set_range_info_raw): New.
13016 2017-06-30 Jakub Jelinek <jakub@redhat.com>
13019 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
13020 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
13021 of nonimmediate_operand and <store_mask_constraint> instead of m
13022 for the input operand. For V8FI iterator, always split if input
13023 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
13024 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
13025 <store_mask_predicate> instead of register_operand and
13026 <store_mask_constraint> instead of v for the input operand. Make
13027 sure both operands aren't MEMs for if not <mask_applied>.
13029 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
13031 * lto-wrapper.c (copy_file) Close both file descriptors before
13034 2017-06-30 Martin Liska <mliska@suse.cz>
13037 * multiple_target.c (create_dispatcher_calls): Make ifunc
13038 also for function that don't have calls or are not referenced.
13040 2017-06-30 Richard Biener <rguenther@suse.de>
13042 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
13043 analyze the first scalar stmt. Move vector type computation
13044 for the BB case here from ...
13045 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
13046 live operation processing in the SLP case properly.
13048 2017-06-30 Richard Biener <rguenther@suse.de>
13050 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
13052 2017-06-30 Martin Liska <mliska@suse.cz>
13055 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
13056 before BUILT_IN_UNWIND_RESUME when ASAN is used.
13058 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
13060 * doc/invoke.texi (AArch64): Add missing options and remove redundant
13063 2017-06-30 Richard Biener <rguenther@suse.de>
13065 PR tree-optimization/81249
13066 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
13067 condition reduction result to original scalar type.
13069 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13071 * profile-count.h (enum profile_quality): Fix typos and whitespace
13074 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13076 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
13077 type for branch probabilities.
13079 2017-06-29 Julian Brown <julian@codesourcery.com>
13080 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13082 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
13083 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
13084 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
13085 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
13087 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13089 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
13090 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
13091 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
13092 CC usage from generic code to here.
13093 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
13094 CC usage into the target macros.
13096 2017-06-29 Maya Rashish <coypu@sdf.org>
13098 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
13101 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13103 * arm/arm-builtins.c: Include profile-count.h
13104 * except.c (sjlj_emit_function_enter): Use
13105 profile_probability::unlikely.
13107 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13109 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
13110 and tocrel_offset be pointer args rather than implicitly using
13112 (legitimate_constant_pool_address_p, rs6000_emit_move,
13113 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
13114 tocrel_offset and use in toc_relative_expr_p call.
13115 (print_operand, print_operand_address): Use static tocrel_base_oac
13116 and tocrel_offset_oac.
13117 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
13120 2017-06-29 Maya Rashish <coypu@sdf.org>
13122 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
13124 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
13126 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
13127 objects, take into account only the alignment of 'op0' and 'mode1' if
13130 2017-06-29 Steve Ellcey <sellcey@cavium.com>
13132 * ccmp.c (ccmp_tree_comparison_p): New function.
13133 (ccmp_candidate_p): Update to use above function.
13134 (get_compare_parts): New function.
13135 (expand_ccmp_next): Update to use new functions.
13136 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
13138 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
13139 take mode as argument.
13140 * ccmp.h (expand_ccmp_expr): Add mode as argument.
13141 * expr.c (expand_expr_real_1): Pass mode as argument.
13143 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
13145 * combine.c (combine_instructions): Print insns to dump_file, together
13148 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13150 * asan.c (asan_emit_stack_protection): Update.
13151 (create_cond_insert_point): Update.
13152 * auto-profile.c (afdo_propagate_circuit): Update.
13153 * basic-block.h (struct edge_def): Turn probability to
13154 profile_probability.
13155 (EDGE_FREQUENCY): Update.
13156 * bb-reorder.c (find_traces_1_round): Update.
13157 (better_edge_p): Update.
13158 (sanitize_hot_paths): Update.
13159 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
13160 (make_single_succ_edge): Update.
13161 (check_bb_profile): Update.
13162 (dump_edge_info): Update.
13163 (update_bb_profile_for_threading): Update.
13164 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
13165 probabilitycount to 0.
13166 * cfgbuild.c (compute_outgoing_frequencies): Update.
13167 * cfgcleanup.c (try_forward_edges): Update.
13168 (outgoing_edges_match): Update.
13169 (try_crossjump_to_edge): Update.
13170 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
13171 (expand_gimple_tailcall): Update.
13172 (construct_init_block): Use make_single_succ_edge.
13173 (construct_exit_block): Use make_single_succ_edge.
13174 * cfghooks.c (verify_flow_info): Update.
13175 (redirect_edge_succ_nodup): Update.
13176 (split_edge): Update.
13177 (account_profile_record): Update.
13178 * cfgloopanal.c (single_likely_exit): Update.
13179 * cfgloopmanip.c (scale_loop_profile): Update.
13180 (set_zero_probability): Remove.
13181 (duplicate_loop_to_header_edge): Update.
13182 * cfgloopmanip.h (loop_version): Update prototype.
13183 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
13184 (force_nonfallthru_and_redirect): Update.
13185 (update_br_prob_note): Update.
13186 (rtl_verify_edges): Update.
13187 (purge_dead_edges): Update.
13188 (rtl_lv_add_condition_to_bb): Update.
13189 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
13190 * cgraphunit.c (init_lowered_empty_function): Update.
13191 (cgraph_node::expand_thunk): Update.
13192 * cilk-common.c: Include profile-count.h
13193 * dojump.c (inv): Remove.
13194 (jumpifnot): Update.
13195 (jumpifnot_1): Update.
13196 (do_jump_1): Update.
13198 (do_jump_by_parts_greater_rtx): Update.
13199 (do_compare_rtx_and_jump): Update.
13200 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
13201 do_jump_1. do_compare_rtx_and_jump): Update prototype.
13202 * dwarf2cfi.c: Include profile-count.h
13203 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
13204 (sjlj_emit_dispatch_table): Likewise.
13205 * explow.c: Include profile-count.h
13206 * expmed.c (emit_store_flag_force): Update.
13207 (do_cmp_and_jump): Update.
13208 * expr.c (compare_by_pieces_d::generate): Update.
13209 (compare_by_pieces_d::finish_mode): Update.
13210 (emit_block_move_via_loop): Update.
13211 (store_expr_with_bounds): Update.
13212 (store_constructor): Update.
13213 (expand_expr_real_2): Update.
13214 (expand_expr_real_1): Update.
13215 * expr.h (try_casesi, try_tablejump): Update prototypes.
13216 * gimple-pretty-print.c (dump_probability): Update.
13217 (dump_profile): New.
13218 (dump_gimple_label): Update.
13219 (dump_gimple_bb_header): Update.
13220 * graph.c (draw_cfg_node_succ_edges): Update.
13221 * hsa-gen.c (convert_switch_statements): Update.
13222 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
13223 (find_if_case_1): Update.
13224 (find_if_case_2): Update.
13225 * internal-fn.c (expand_arith_overflow_result_store): Update.
13226 (expand_addsub_overflow): Update.
13227 (expand_neg_overflow): Update.
13228 (expand_mul_overflow): Update.
13229 (expand_vector_ubsan_overflow): Update.
13230 * ipa-cp.c (good_cloning_opportunity_p): Update.
13231 * ipa-split.c (split_function): Use make_single_succ_edge.
13232 * ipa-utils.c (ipa_merge_profiles): Update.
13233 * loop-doloop.c (add_test): Update.
13234 (doloop_modify): Update.
13235 * loop-unroll.c (compare_and_jump_seq): Update.
13236 (unroll_loop_runtime_iterations): Update.
13237 * lra-constraints.c (lra_inheritance): Update.
13238 * lto-streamer-in.c (input_cfg): Update.
13239 * lto-streamer-out.c (output_cfg): Update.
13240 * mcf.c (adjust_cfg_counts): Update.
13241 * modulo-sched.c (sms_schedule): Update.
13242 * omp-expand.c (expand_omp_for_init_counts): Update.
13243 (extract_omp_for_update_vars): Update.
13244 (expand_omp_ordered_sink): Update.
13245 (expand_omp_for_ordered_loops): Update.
13246 (expand_omp_for_generic): Update.
13247 (expand_omp_for_static_nochunk): Update.
13248 (expand_omp_for_static_chunk): Update.
13249 (expand_cilk_for): Update.
13250 (expand_omp_simd): Update.
13251 (expand_omp_taskloop_for_outer): Update.
13252 (expand_omp_taskloop_for_inner): Update.
13253 * omp-simd-clone.c (simd_clone_adjust): Update.
13254 * optabs.c (expand_doubleword_shift): Update.
13255 (expand_abs): Update.
13256 (emit_cmp_and_jump_insn_1): Update.
13257 (expand_compare_and_swap_loop): Update.
13258 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
13259 * predict.c (predictable_edge_p): Update.
13260 (edge_probability_reliable_p): Update.
13261 (set_even_probabilities): Update.
13262 (combine_predictions_for_insn): Update.
13263 (combine_predictions_for_bb): Update.
13264 (propagate_freq): Update.
13265 (estimate_bb_frequencies): Update.
13266 (force_edge_cold): Update.
13267 * profile-count.c (profile_count::dump): Add missing space into dump.
13268 (profile_count::debug): Add newline.
13269 (profile_count::differs_from_p): Explicitly convert to unsigned.
13270 (profile_count::stream_in): Update.
13271 (profile_probability::dump): New member function.
13272 (profile_probability::debug): New member function.
13273 (profile_probability::differs_from_p): New member function.
13274 (profile_probability::differs_lot_from_p): New member function.
13275 (profile_probability::stream_in): New member function.
13276 (profile_probability::stream_out): New member function.
13277 * profile-count.h (profile_count_quality): Rename to ...
13278 (profile_quality): ... this one.
13279 (profile_probability): New.
13280 (profile_count): Update.
13281 * profile.c (compute_branch_probabilities): Update.
13282 * recog.c (peep2_attempt): Update.
13283 * sched-ebb.c (schedule_ebbs): Update.
13284 * sched-rgn.c (find_single_block_region): Update.
13285 (compute_dom_prob_ps): Update.
13286 (schedule_region): Update.
13287 * sel-sched-ir.c (compute_succs_info): Update.
13288 * stmt.c (struct case_node): Update.
13289 (do_jump_if_equal): Update.
13290 (get_outgoing_edge_probs): Update.
13291 (conditional_probability): Update.
13292 (emit_case_dispatch_table): Update.
13293 (expand_case): Update.
13294 (expand_sjlj_dispatch_table): Update.
13295 (emit_case_nodes): Update.
13296 * targhooks.c: Update.
13297 * tracer.c (better_p): Update.
13298 (find_best_successor): Update.
13299 * trans-mem.c (expand_transaction): Update.
13300 * tree-call-cdce.c: Update.
13301 * tree-cfg.c (gimple_split_edge): Upate.
13302 (move_sese_region_to_fn): Upate.
13303 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
13304 * tree-eh.c (lower_resx): Upate.
13305 (cleanup_empty_eh_move_lp): Upate.
13306 * tree-if-conv.c (version_loop_for_if_conversion): Update.
13307 * tree-inline.c (copy_edges_for_bb): Update.
13308 (copy_cfg_body): Update.
13309 * tree-parloops.c (gen_parallel_loop): Update.
13310 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
13311 (gimple_gen_time_profiler): Update.
13312 * tree-ssa-dce.c (remove_dead_stmt): Update.
13313 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
13314 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
13315 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
13316 (unloop_loops): Update.
13317 (try_peel_loop): Update.
13318 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13319 * tree-ssa-loop-split.c (connect_loops): Update.
13320 (split_loop): Update.
13321 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
13322 (hoist_guard): Update.
13323 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
13324 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
13325 (value_replacement): Update.
13326 * tree-ssa-reassoc.c (branch_fixup): Update.
13327 * tree-ssa-tail-merge.c (replace_block_by): Update.
13328 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
13329 (create_edge_and_update_destination_phis): Update.
13330 (compute_path_counts): Update.
13331 (recompute_probabilities): Update.
13332 (update_joiner_offpath_counts): Update.
13333 (freqs_to_counts_path): Update.
13334 (duplicate_thread_path): Update.
13335 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
13336 (struct switch_conv_info): Update.
13337 (gen_inbound_check): Update.
13338 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
13339 (vect_do_peeling): Update.
13340 (vect_loop_versioning): Update.
13341 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13342 (optimize_mask_stores): Update.
13343 * ubsan.c (ubsan_expand_null_ifn): Update.
13344 * value-prof.c (gimple_divmod_fixed_value): Update.
13345 (gimple_divmod_fixed_value_transform): Update.
13346 (gimple_mod_pow2): Update.
13347 (gimple_mod_pow2_value_transform): Update.
13348 (gimple_mod_subtract): Update.
13349 (gimple_mod_subtract_transform): Update.
13350 (gimple_ic): Update.
13351 (gimple_stringop_fixed_value): Update.
13352 (gimple_stringops_transform): Update.
13353 * value-prof.h: Update.
13355 2017-06-29 Carl Love <cel@us.ibm.com>
13357 * config/rs6000/rs6000-c.c: Add support for built-in functions
13358 vector signed int vec_signed (vector float);
13359 vector signed long long vec_signed (vector double);
13360 vector signed int vec_signed2 (vector double, vector double);
13361 vector signed int vec_signede (vector double);
13362 vector signed int vec_signedo (vector double);
13363 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
13364 instruction generator.
13365 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
13366 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
13367 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
13369 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
13370 vunsignede_v2df): Add define_expands.
13371 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
13372 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
13373 VEC_UNSIGNEDO): Add definitions.
13374 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
13375 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
13376 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
13377 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
13378 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
13379 * config/rs6000/altivec.h (vec_signed, vec_signed2,
13380 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
13381 vec_unsignede, vec_unsignedo): Add builtin defines.
13382 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
13384 * doc/extend.texi: Update the built-in documentation file for the
13385 new built-in functions.
13387 2017-06-29 Richard Biener <rguenther@suse.de>
13389 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
13390 reduction chains to LOOP_VINFO_REDUCTIONS.
13391 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
13392 SLP reductions after processing reduction chains.
13394 2017-06-29 Nathan Sidwell <nathan@acm.org>
13396 * builtins.c (fold_builtin_FUNCTION): Use
13397 lang_hooks.decl_printable_name.
13399 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
13401 PR middle-end/81194
13402 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
13403 with only one label.
13404 * stmt.c (expand_case): Assert NCASES is greater than one.
13406 2017-06-29 Richard Biener <rguenther@suse.de>
13408 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
13410 (group_case_labels): Likewise.
13411 (find_taken_edge): Push sanity checking on val to workers...
13412 (find_taken_edge_cond_expr): ... here
13413 (find_taken_edge_switch_expr): ... and here, handle cases
13414 with just a default label.
13415 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
13416 (group_case_labels): Likewise.
13417 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
13418 group_case_labels does anything cleanup the CFG again.
13420 2017-06-29 Bin Cheng <bin.cheng@arm.com>
13422 PR tree-optimization/81196
13423 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
13424 exit condition comparing two IVs.
13426 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
13428 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
13429 profile to the dummy entry at the end of the list of architectures.
13430 * config/arm/arm-cpu-cdata.h: Regenerated.
13432 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13433 Michael Collison <michael.collison@arm.com>
13436 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
13438 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
13439 (*aarch64_reg_<mode>3_minus_mask): New pattern.
13440 (*aarch64_<optab>_reg_di3_mask2): New pattern.
13441 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
13442 of shift when the shift amount is masked with constant equal to
13443 the size of the mode.
13444 * config/aarch64/predicates.md (subreg_lowpart_operator): New
13447 2017-06-29 Martin Liska <mliska@suse.cz>
13449 * config/i386/i386.opt: Change range from [1,5] to [0,5].
13451 2017-06-29 Yury Gribov <tetra2005@gmail.com>
13454 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
13457 (edge_growth_cache_entry::edge_growth_cache_entry): New
13459 (reset_edge_growth_cache): Update to use constructor.
13461 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13463 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
13464 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
13465 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
13467 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
13469 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
13470 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
13472 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
13474 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
13475 (*-linux-uclibc*): Add t-uclibc tmake_file.
13476 * config/t-musl: New.
13477 * config/t-uclibc: New.
13479 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
13481 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
13483 (gen_comm_data): Emit architectural setting of arch_prof.
13484 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
13486 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
13487 (armv8-m.base, armv8-m.main): Likewise.
13488 * arm-protos.h (arm_build_target): Add profile field.
13489 (arch_option): Likewise.
13490 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
13492 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
13493 arm_active_target.profile.
13495 2017-06-28 Richard Biener <rguenther@suse.de>
13497 PR middle-end/81227
13498 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
13499 TYPE_OVERFLOW_WRAPS.
13500 * match.pd (negate_expr_p): Likewise.
13501 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
13502 fold_build2, not fold_binary.
13504 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
13506 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
13507 Convert memory address to Pmode.
13508 (aarch64_print_operand): Assert MEM operands are always Pmode.
13510 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
13513 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
13514 Remove redundant if.
13515 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
13516 * config/arm/aarch-common-protos.h
13517 (aarch_forward_to_shift_is_not_shifted_re): Remove.
13518 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
13520 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
13523 * multiple_target.c (create_dispatcher_calls): Set the default
13524 clone to be static, not public.
13526 2017-06-28 Richard Biener <rguenther@suse.de>
13528 * tree-vect-loop.c (vectorizable_reduction): Move special
13529 cond reduction IV var creation ...
13530 (vect_create_epilog_for_reduction): ... here. Remove induction_index
13531 parameter. Use STMT_VINFO_VECTYPE.
13532 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
13535 2017-06-28 Martin Liska <mliska@suse.cz>
13538 * ipa-visibility.c (non_local_p): Handle visibility.
13540 2017-06-28 Martin Liska <mliska@suse.cz>
13543 * common.opt: Add IntegerRange to various options.
13544 * opt-functions.awk (integer_range_info): New function.
13545 * optc-gen.awk: Add integer_range_info to cl_options struct.
13546 * opts-common.c (decode_cmdline_option): Handle
13547 CL_ERR_INT_RANGE_ARG.
13548 (cmdline_handle_error): Likewise.
13549 * opts.c (print_filtered_help): Show valid interval in
13550 when --help is provided.
13551 * opts.h (struct cl_option): Add range_min and range_max fields.
13552 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
13554 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
13556 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
13557 (x * C EQ/NE y * C): New transformation.
13559 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
13561 * genmultilib (combination_space): Accept '+' in option names.
13563 2017-06-28 Martin Liska <mliska@suse.cz>
13566 * asan.c (instrument_derefs): Bail out inner references
13567 that are hard register variables.
13569 2017-06-28 Jakub Jelinek <jakub@redhat.com>
13572 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
13573 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
13575 2017-06-28 Richard Biener <rguenther@suse.de>
13577 * tree-vectorizer.h (vect_get_vec_defs): Remove.
13578 (vect_get_slp_defs): Adjust.
13579 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
13581 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
13583 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
13584 get_initial_defs_for_reduction instead of vect_get_vec_defs.
13585 (vectorizable_reduction): Adjust.
13586 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
13588 (vect_get_slp_defs): Likewise.
13589 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
13590 (vectorizable_bswap): Adjust.
13591 (vectorizable_call): Likewise.
13592 (vectorizable_conversion): Likewise.
13593 (vectorizable_assignment): Likewise.
13594 (vectorizable_shift): Likewise.
13595 (vectorizable_operation): Likewise.
13596 (vectorizable_store): Likewise.
13597 (vectorizable_condition): Likewise.
13598 (vectorizable_comparison): Likewise.
13600 2017-06-28 Michael Collison <michael.collison@arm.com>
13603 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
13605 (arm_gen_movmemqi): Removed unused variable 'i'.
13606 Convert 'for' loop into 'while' loop.
13607 (arm_expand_prologue): Remove last unnecessary set of insn.
13608 (thumb_pop): Remove unused variable 'pushed_words'.
13609 (thumb_exit): Remove last unnecessary set of regs_to_pop.
13611 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13613 * config/s390/predicates.md: Use s390_rel_address_ok_p.
13614 * config/s390/s390-protos.h: Add prototype of
13615 s390_rel_address_ok_p.
13616 * config/s390/s390.c (s390_got_symbol): New function.
13617 (s390_rel_address_ok_p): New function.
13618 (legitimize_pic_address): Use s390_rel_address_ok_p.
13619 (s390_load_got): Use s390_got_symbol.
13620 (s390_option_override): Issue error if
13621 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
13622 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
13624 * config/s390/s390.opt: New option mpic-data-is-text-relative.
13626 2017-06-27 Andrew Pinski <apinski@cavium.com>
13628 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
13629 (X * copysign (1.0, X)): New pattern.
13630 (X * copysign (1.0, -X)): New pattern.
13631 (copysign (-1.0, CST)): New pattern.
13633 2017-06-27 Joseph Myers <joseph@codesourcery.com>
13635 * genmultilib (combination_space): Remove variable.
13636 Validate reuse rules against regular expression for any sequence
13637 of multilib options in any order.
13639 2017-06-27 Michael Collison <michael.collison@arm.com>
13641 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
13642 call aarch64_split_simd_combine.
13643 * (aarch64_combine_internal<mode>): Delete pattern.
13644 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
13645 Allow register and subreg operands.
13647 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13649 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
13650 specific need, just fallback on defaults.
13651 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
13653 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13654 Olivier Hainque <hainque@adacore.com>
13656 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
13658 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
13659 targets. Pick a default if no particular attempt applied.
13660 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
13663 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13665 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
13666 (x86_64-wrs-vxworks7): Likewise.
13668 2017-06-27 Marek Polacek <polacek@redhat.com>
13671 * ubsan.c (instrument_null): Check get_base_address's result for null.
13673 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
13675 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
13677 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
13679 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
13680 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
13681 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
13682 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
13683 New function types.
13684 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
13685 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
13686 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
13687 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
13688 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
13689 BUILT_IN_FEUPDATEENV): New builtins.
13690 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
13691 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
13692 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
13693 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
13695 (builtin_structptr_types): Adjust size.
13696 * tree.c (builtin_structptr_types): Add four entries.
13698 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13699 Olivier Hainque <hainque@adacore.com>
13701 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
13702 (TLS_SYM): New local macro, forcing reference to __tls__ on
13703 link command lines for VxWorks 7 RTPs, triggering initialization
13705 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
13706 OS features TLS support, true for RTPs on VxWorks 7.
13707 * config/vxworks.c (vxworks_override_options): Setup emutls
13710 2017-06-27 Jakub Jelinek <jakub@redhat.com>
13712 * predict.c (test_prediction_value_range): Use -1U instead of -1
13713 to avoid narrowing conversion warning.
13714 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
13715 to avoid narrowing conversion warning.
13716 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
13718 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
13720 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13722 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
13723 64bit configurations.
13724 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
13725 (SIZE_TYPE): Likewise.
13726 * config/vxworks.c (vxworks_emutls_var_fields): Use
13727 long_unsigned_type_node instead of unsigned_type_node as the offset
13728 field type, which is "pointer" mode in emutls.c.
13730 2017-06-27 Jakub Jelinek <jakub@redhat.com>
13733 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
13735 PR middle-end/81207
13736 * gimple-fold.c (replace_call_with_call_and_fold): Handle
13737 gimple_vuse copying separately from gimple_vdef copying.
13739 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13741 * value-prof.c (free_hist): Remove call to memset and the enclosing if
13744 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
13745 Olivier Hainque <hainque@adacore.com>
13747 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
13748 for all vxworks7 targets.
13749 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
13750 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
13751 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
13752 variations for VX6/VX7 and 32/64bits later on in ...
13753 (VXWORKS_LIB_SPEC): Leverage new macros.
13754 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
13755 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
13757 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
13759 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
13760 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
13762 2017-06-26 Carl Love <cel@us.ibm.com>
13764 * config/rs6000/rs6000-c.c: Add support for built-in functions
13765 vector bool char vec_reve (vector bool char);
13766 vector signed char vec_reve (vector signed char);
13767 vector unsigned char vec_reve (vector unsigned char);
13768 vector bool int vec_reve (vector bool int);
13769 vector signed int vec_reve (vector signed int);
13770 vector unsigned int vec_reve (vector unsigned int);
13771 vector bool long long vec_reve (vector bool long long);
13772 vector signed long long vec_reve (vector signed long long);
13773 vector unsigned long long vec_reve (vector unsigned long long);
13774 vector bool short vec_reve (vector bool short);
13775 vector signed short vec_reve (vector signed short);
13776 vector double vec_reve (vector double);
13777 vector float vec_reve (vector float);
13778 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
13779 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
13780 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
13781 (altivec_vreve): New pattern.
13782 * config/rs6000/altivec.h (vec_reve): New define.
13783 * doc/extend.texi (vec_rev): Update the built-in documentation file
13784 for the new built-in functions.
13786 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13788 PR tree-optimization/71815
13789 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
13791 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
13793 (slsr_process_phi): Likewise.
13794 (replace_uncond_cands_and_profitable_phis): Don't replace a
13795 multiply candidate with a stride of 1 (copy or cast).
13796 (phi_incr_cost): Call uses_consumed_by_stmt rather than
13798 (lowest_cost_path): Likewise.
13799 (total_savings): Likewise.
13801 2017-06-26 Richard Biener <rguenther@suse.de>
13804 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
13805 Use def_builtin_pure for all gather builtins.
13807 2017-06-26 Richard Biener <rguenther@suse.de>
13809 PR tree-optimization/81203
13810 * tree-tailcall.c (find_tail_calls): Do not move stmts into
13811 non-dominating BBs.
13813 2017-06-26 Marek Polacek <polacek@redhat.com>
13816 * doc/invoke.texi: Document -Wmultistatement-macros.
13818 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
13820 * doc/sourcebuild.texi (ARM-specific attributes): Document new
13821 arm_neon_ok_no_float_abi effective target.
13823 2017-06-26 Richard Biener <rguenther@suse.de>
13825 PR tree-optimization/80928
13826 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
13827 (copy_bbs): Set BB_DUPLICATED flag early.
13828 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
13830 (execute_on_shrinking_pred): Likewise.
13831 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
13832 BB_DUPLICATED blocks.
13833 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
13834 iterate over all PHIs considering removal of *gsi.
13836 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
13838 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
13841 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13843 * config/rs6000/rs6000-string.c: (expand_block_clear,
13844 do_load_for_compare, select_block_compare_mode,
13845 compute_current_alignment, expand_block_compare,
13846 expand_strncmp_align_check, expand_strn_compare,
13847 expand_block_move, rs6000_output_load_multiple)
13848 Move functions related to string/block move/compare
13849 to a separate file.
13850 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
13851 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
13852 for this function which is now used in two files.
13853 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
13854 * config.gcc: Add rs6000-string.o to extra_objs for
13855 targets powerpc*-*-* and rs6000*-*-*.
13857 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
13860 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
13861 32-bit, since indexed is not valid for DImode.
13862 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
13863 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
13864 (define_peephole2 for Altivec d-form load): Add 32-bit support.
13865 (define_peephole2 for Altivec d-form store): Likewise.
13868 * multiple_target.c (create_dispatcher_calls): Only create the
13869 dispatcher call if the function is the default clone of a
13870 versioned function.
13872 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
13874 PR middle-end/80902
13875 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
13876 a call, force the call to not be a tail call.
13878 2017-06-23 Jeff Law <law@redhat.com>
13880 * doc/contrib.texi: Add entry for Steven Pemberton's work on
13883 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
13885 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
13886 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
13887 handling for early expansion of vector shifts (sl,sr,sra,rl).
13888 (builtin_function_type): Add vector shift right instructions
13889 to the unsigned argument list.
13891 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
13893 rtl-optimizatoin/79286
13894 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
13895 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
13896 trap. PIC register plus a const unspec without offset can never trap.
13898 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
13900 * tree.h (builtin_structptr_type): New type.
13901 (builtin_structptr_types): Declare new array.
13902 * tree.c (builtin_structptr_types): New array.
13903 (free_lang_data, build_common_tree_nodes): Use it.
13905 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
13908 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
13911 2017-06-22 Matt Turner <mattst88@gmail.com>
13913 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
13914 Lake models to skylake case. Assume skylake for unknown
13915 models with clflushopt.
13917 2017-06-22 Jeff Law <law@redhat.com>
13919 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
13920 frame sizes that do not satisfy aarch64_uimm12_shift.
13922 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
13924 * profile-count.h (apply_probability,
13925 apply_scale, probability_in): Fix checks for zero.
13927 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13929 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
13930 * doc/cppdiropts.texi (-I @var{dir}): Document it.
13932 2016-06-22 Richard Biener <rguenther@suse.de>
13934 * tree-vect-loop.c (vect_model_reduction_cost): Handle
13935 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
13936 REDUC_MAX_EXPR support.
13937 (vectorizable_reduction): Likewise.
13938 (vect_create_epilog_for_reduction): Likewise.
13940 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
13942 * match.pd (A / (1 << B) -> A >> B): New.
13943 * generic-match-head.c: Include optabs-tree.h.
13944 * gimple-match-head.c: Likewise.
13945 * optabs-tree.h (target_supports_op_p): New.
13946 * optabs-tree.c (target_supports_op_p): New.
13948 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13950 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
13951 $gcc_cv_ld --help output.
13952 (gcc_cv_ld_demangle): Likewise.
13953 (gcc_cv_ld_eh_frame_hdr): Likewise.
13954 (gcc_cv_ld_pie): Likewise.
13955 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
13956 (gcc_cv_ld_buildid): Likewise.
13957 (gcc_cv_ld_sysroot): Likewise.
13958 (ld_bndplt_support): Likewise.
13959 (ld_pushpopstate_support): Likewise.
13960 * configure: Regenerate.
13961 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
13963 2017-06-21 Jakub Jelinek <jakub@redhat.com>
13966 * config/i386/sse.md (round<mode>2): Renumber match_dup and
13967 operands indexes to avoid gap between operands and match_dups.
13969 2017-06-21 Andrew Pinski <apinski@cavium.com>
13971 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
13972 Increment Arith_shift and Arith_shift_reg by 1.
13973 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
13975 * config/aarch64/aarch64.c (thunderx_tunings): Enable
13976 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
13977 (aarch64_strip_extend): Add new argument and test for it.
13978 (aarch64_cheap_mult_shift_p): New function.
13979 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
13980 add a cost if it is true.
13981 Update calls to aarch64_strip_extend.
13982 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
13984 2017-06-21 Andrew Pinski <apinski@cavium.com>
13986 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
13988 (thunderxt88): Likewise.
13989 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
13990 (thunderx_prefetch_tune): New variable.
13991 (thunderx2t99_prefetch_tune): Update for the correct values.
13992 (thunderxt88_tunings): New variable.
13993 (thunderx_tunings): Use thunderx_prefetch_tune instead of
13994 generic_prefetch_tune.
13995 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
13997 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13999 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
14000 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
14001 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
14002 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
14003 (aarch64_atomic_cas<mode>, GPI): Likewise.
14005 2017-06-21 Martin Liska <mliska@suse.cz>
14007 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
14008 statements on cold and hot labels.
14009 * predict.c (tree_estimate_probability_bb): Remove the
14010 prediction from this place.
14012 2017-06-21 Martin Liska <mliska@suse.cz>
14014 PR tree-optimization/79489
14015 * gimplify.c (maybe_add_early_return_predict_stmt): New
14017 (gimplify_return_expr): Call the function.
14018 * predict.c (tree_estimate_probability_bb): Remove handling
14020 * predict.def: Update comment about early return predictor.
14021 * gimple-predict.h (is_gimple_predict): New function.
14022 * predict.def: Change default value of early return to 66.
14023 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
14025 * passes.def: Put pass_strip_predict_hints to the beginning of
14028 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
14030 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
14031 FUNCTION_DECL declarations.
14032 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
14034 (dwaf2out_decl): Likewise.
14035 * godump.c (go_early_global_decl): Skip call to the real debug hook
14036 for FUNCTION_DECL declarations.
14037 * passes.c (rest_of_decl_compilation): Skip call to the
14038 early_global_decl debug hook for FUNCTION_DECL declarations, unless
14039 -fdump-go-spec is passed.
14041 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
14043 * config/i386/i386.c (struct builtin_isa): New field pure_p.
14044 Reorder for compactness.
14045 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
14046 (def_builtin_pure, def_builtin_pure2): New functions.
14047 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
14049 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
14051 * match.pd (nop_convert): New predicate.
14052 ((A +- CST1) +- CST2): Allow some NOP conversions.
14054 2017-06-21 Jakub Jelinek <jakub@redhat.com>
14057 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
14058 with ctors/dtors if GOVD_SHARED is set.
14060 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14062 * config/aarch64/aarch64.md (movti_aarch64):
14063 Emit mov rather than orr.
14064 (movtf_aarch64): Likewise.
14065 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
14066 Emit mov rather than orr.
14068 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14070 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
14071 Swap alternatives, make integer dup more expensive.
14073 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14075 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
14076 Return true for non-tls symbols.
14078 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
14080 * config/aarch64/aarch64-cores.def (cortex-a55): New.
14081 (cortex-a75): Likewise.
14082 (cortex-a75.cortex-a55): Likewise.
14083 * config/aarch64/aarch64-tune.md: Regenerate.
14084 * doc/invoke.texi (-mtune): Document new values for -mtune.
14086 2017-06-21 Tom de Vries <tom@codesourcery.com>
14088 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
14089 stack_size feature.
14090 (Effective-Target Keywords, Other attributes): Suggest using
14091 dg-add-options stack_size feature to get stack limit in stack_size
14092 effective target documentation.
14094 2017-06-21 Julian Brown <julian@codesourcery.com>
14095 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14097 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
14098 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
14099 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
14101 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
14102 attribute type list for neon_multiply.
14103 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
14104 attribute type list for neon_multiply.
14105 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
14106 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
14107 attribute type list for neon_multiply.
14108 * config/arm/types.md (crypto_pmull): Add.
14109 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
14110 attribute type list.
14112 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
14114 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
14117 2017-06-20 Jakub Jelinek <jakub@redhat.com>
14119 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
14120 to make sure not to dereference a NULL cost_classes_ptr pointer.
14122 2017-06-20 Carl Love <cel@us.ibm.com>
14124 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14125 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14126 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14127 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14128 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14129 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
14130 VMULOSW): New enum "unspec" values.
14131 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
14132 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
14133 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
14134 altivec_vmulosw): New patterns.
14135 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14136 VMULOSW): Add definitions.
14138 2017-06-20 Julia Koval <julia.koval@intel.com>
14140 * config/i386/i386.c: Fix rounding expand for new pattern.
14141 * config/i386/subst.md: Fix pattern (parallel -> unspec).
14143 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14145 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
14146 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
14148 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14150 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
14153 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14155 * config/aarch64/aarch64-cores.def: Rearrange to sort by
14156 architecture, then by implementer ID.
14157 * config/aarch64/aarch64-tune.md: Regenerate.
14159 2017-06-20 Richard Biener <rguenther@suse.de>
14161 PR middle-end/81097
14162 * fold-const.c (split_tree): Fold to type before negating.
14164 2017-06-20 David Malcolm <dmalcolm@redhat.com>
14166 * diagnostic-show-locus.c
14167 (selftest::test_fixit_deletion_affecting_newline): New function.
14168 (selftest::diagnostic_show_locus_c_tests): Call it.
14170 2017-06-20 Andreas Schwab <schwab@suse.de>
14173 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
14176 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14178 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
14179 __ARM_FEATURE_COPROC according to support.
14181 2017-06-20 Jakub Jelinek <jakub@redhat.com>
14183 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
14184 Rewritten to avoid overflow for > 32-bit pointers.
14187 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
14188 by removing enum keyword.
14189 (ubsan_type_descriptor): Likewise. Formatting fix.
14192 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
14193 splitter): Require TARGET_SSE2 in the condition.
14195 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
14198 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
14199 for doing vector set of SFmode on ISA 3.0.
14200 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
14201 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
14203 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
14204 SFmode value into a V4SF variable that was extracted from another
14205 V4SF variable without converting the element to double precision
14206 and back to single precision vector format.
14207 (vsx_insert_extract_v4sf_p9_2): Likewise.
14209 2017-06-19 Jakub Jelinek <jakub@redhat.com>
14211 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
14212 in UWHI to avoid undefined overflow.
14215 * ubsan.h (enum ubsan_encode_value_phase): New.
14216 (ubsan_encode_value): Change second argument to
14217 enum ubsan_encode_value_phase with default value of
14218 UBSAN_ENCODE_VALUE_GENERIC.
14219 * ubsan.c (ubsan_encode_value): Change second argument to
14220 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
14221 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
14222 create_tmp_var_raw instead of create_tmp_var and use a
14224 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
14225 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
14226 ubsan_encode_value callers.
14229 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
14230 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
14231 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
14233 2017-06-19 Richard Biener <rguenther@suse.de>
14235 PR middle-end/81118
14236 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
14237 estimates if we changed anything.
14239 2017-06-19 Richard Biener <rguenther@suse.de>
14241 PR tree-optimization/80887
14242 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
14243 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
14244 simplified lookups, then reset mprts_hook.
14245 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
14247 (try_to_simplify): Likewise.
14249 2017-06-19 Martin Liska <mliska@suse.cz>
14252 * gimplify.c (gimplify_switch_expr):
14253 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
14255 2017-06-19 Martin Liska <mliska@suse.cz>
14257 * doc/install.texi: Document that PGO runs in 4 stages.
14259 2017-06-19 Martin Liska <mliska@suse.cz>
14262 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
14263 to dispatcher function name.
14264 * multiple_target.c (replace_function_decl): New function.
14265 (create_dispatcher_calls): Redirect both edges and references.
14267 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
14269 * profile-count.c (profile_count::dump): Dump quality.
14270 (profile_count::differs_from_p): Update for unsigned val.
14271 * profile-count.h (profile_count_quality): New enum.
14272 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
14274 2017-06-19 Richard Biener <rguenther@suse.de>
14276 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
14277 struct function as arg.
14278 (estimate_numbers_of_iterations): Export overload with loop arg.
14279 (free_numbers_of_iterations_estimates_loop): Use an overload of
14280 free_numbers_of_iterations_estimates instead.
14281 * tree-cfg.c (remove_bb): Adjust.
14282 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
14283 * tree-parloops.c (gen_parallel_loop): Likewise.
14284 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
14286 (tree_unroll_loops_completely): Likewise.
14287 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
14288 Use an overload instead and export.
14289 (estimated_loop_iterations): Adjust.
14290 (max_loop_iterations): Likewise.
14291 (likely_max_loop_iterations): Likewise.
14292 (estimate_numbers_of_iterations): Take struct function as arg
14294 (loop_exits_before_overflow): Adjust.
14295 (free_numbers_of_iterations_estimates_loop): Use an overload.
14296 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
14297 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
14299 2017-06-19 Richard Biener <rguenther@suse.de>
14302 * ipa-prop.c (find_constructor_constant_at_offset): Handle
14303 RANGE_EXPR conservatively.
14305 2017-06-16 Carl Love <cel@us.ibm.com>
14307 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14308 definitions for vec_float, vec_float2, vec_floato,
14309 vec_floate built-ins.
14310 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
14311 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
14313 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
14314 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
14315 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
14316 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
14317 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
14318 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
14319 vec_floato): Add builtin defines.
14320 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
14321 Update the built-in documentation file for the new built-in
14324 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14326 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
14327 (mthumb): Mark as the negative of -marm.
14329 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14331 * doc/invoke.texi (ARM Options, -mcpu): Document supported
14333 (ARM Options, -mtune): Document that this accepts the same
14334 extension options as -mcpu.
14335 (ARM Options, -mfpu): Document addition of -mfpu=auto.
14337 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14339 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
14340 permitted extensions.
14342 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14344 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
14345 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
14346 (armv8-m.main): Add option +nodsp.
14347 * config/arm/arm-cpu-cdata.h: Regenerated.
14349 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14351 * config/arm/t-fuchsia: New file.
14352 * config.gcc (arm*-*-fuchsia*): Use it.
14354 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14356 * config/arm/t-symbian: Rewrite for new option infrastructure.
14358 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14360 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
14361 (MULTILIB_REQUIRED): Likewise.
14363 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14365 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
14366 (MULTILIB_RESUE): Likewise.
14367 (MULTILIB_MATCHES): Likewise.
14368 (MULTLIB_REQUIRED): Likewise.
14370 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14372 * config/arm/t-rtems: Rewrite for new option framework.
14374 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14376 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
14377 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
14378 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
14379 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
14380 * config/arm/t-multilib: ... here.
14381 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
14382 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
14383 armv7-a and armv8*-a when A-profile libraries have not been built.
14384 * config/arm/t-rmprofile: Rewrite.
14386 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14388 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
14389 with a backslash. Remove the backslash after substituting unescaped
14391 * doc/fragments.texi (MULTILIB_REUSE): Document it.
14393 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14395 * config.gcc: (arm*-*-*): When building a-profile libraries, force
14396 the driver to pass through the default setting of -mfloat-abi.
14397 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
14399 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
14400 (all_feat_combs): New rule.
14401 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
14403 * config/arm/t-aprofile: Rewrite.
14405 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14407 * config/arm/arm.h (FPUTYPE_AUTO): Define.
14408 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
14409 fpu is not specified by the user/command-line.
14410 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
14411 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
14412 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
14413 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
14414 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
14415 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
14417 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14419 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
14420 * config/arm/t-arm-elf: Rewritten.
14422 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14424 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
14425 have some floating-point instructions.
14426 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
14427 (TARGET_MAYBE_HARD_FLOAT): New macro.
14428 * config/arm/arm-builtins.c (arm_init_builtins): Use
14429 TARGET_MAYBE_HARD_FLOAT.
14430 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
14432 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14434 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
14435 (configargs.h): Include it.
14436 (arm_print_hint_for_fpu_option): New function.
14437 (arm_parse_fpu_option): New function.
14438 (candidate_extension): New class.
14439 (arm_canon_for_multilib): New function.
14440 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
14441 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
14442 (ARCH_CANONICAL_SPECS): New macro.
14443 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
14445 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14447 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
14448 are set after handling multilib fragments. Set target_cpu_default2
14451 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14453 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
14455 (arm*-*-*): Set target_cpu_default2 to a quoted string.
14456 * config/arm/parsecpu.awk (check_cpu): Validate any extension
14458 (check_arch): Likewise.
14459 * config/arm/arm.c (arm_configure_build_target): Handle
14460 TARGET_CPU_DEFAULT being a string constant. Scan any feature
14461 options in the default.
14463 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14465 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
14466 when an option is an alias of another.
14467 * config/arm/parsecpu.awk (optalias): New parser token.
14468 (gen_comm_data): Mark non-alias options as such. Emit entries
14469 for extension aliases.
14470 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
14471 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
14472 (armv6kz, armv6zk, armv6t2): Likewise.
14473 (armv7): Make vfpv3-d16 an alias.
14474 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
14476 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
14477 Sort in canonical order.
14478 (armv8-a): Sort in canonical order.
14479 (armv8.1-a, armv8.2-a): Likewise.
14480 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
14482 (cortex-a9): Sort in canonical order.
14483 * config/arm/arm.c (selftests.h): Include it.
14484 (arm_test_cpu_arch_data): New function.
14485 (arm_run_self_tests): New function.
14486 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
14487 (targetm): Move declaration to the end of the file.
14488 * arm-cpu-cdata.h: Regenerated.
14490 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14492 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
14493 call to target_mode_check describing the type of option passed.
14494 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
14495 (arm_target_thumb_only): Use arm_parse_arch_option_name or
14496 arm_parse_cpu_option_name to match parameters against list of
14498 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
14499 arm_arch_core_flags data structure.
14500 * config/arm/arm-cpu_cdata.h: Regenerated.
14502 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14504 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
14506 (arm_print_hint_for_cpu_option): Likewise.
14507 (arm_print_hint_for_arch_option): Likewise.
14508 (arm_parse_cpu_option_name): Likewise.
14509 (arm_parse_arch_option_name): Likewise.
14510 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
14511 of entries in the all_fpus list.
14512 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
14513 (arm_parse_cpu_option_name): Declare.
14514 (arm_parse_arch_option_name): Declare.
14515 (arm_parse_option_features): Declare.
14516 (arm_intialize_isa): Declare.
14517 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
14519 (gen_comm_data): ... here. Make definitions non-static.
14520 * config/arm/arm-cpu-data.h: Regenerated.
14521 * config/arm/arm-cpu-cdata.h: Regenerated.
14523 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14525 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
14526 (cpu_arch_extension): New structure.
14527 (cpu_arch_option, arch_option, cpu_option): New structures.
14528 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
14529 architecture types.
14530 (gen_data): Generate new format data tables.
14531 * config/arm/arm.c (cpu_tune): New structure.
14532 (cpu_option, processors): Delete.
14533 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
14534 (arm_print_hint_for_cpu_option): ... this and ...
14535 (arm_print_hint_for_arch_option): ... this.
14536 (arm_parse_arch_cpu_name): Delete. Replace with ...
14537 (arm_parse_cpu_option_name): ... this and ...
14538 (arm_parse_arch_option_name): ... this.
14539 (arm_unrecognized_feature): Change type of target parameter to
14541 (arm_parse_arch_cpu_features): Delete. Replace with ...
14542 (arm_parse_option_features): ... this.
14543 (arm_configure_build_target): Rework to use new configuration data
14545 (arm_print_tune_info): Rework for new configuration data tables.
14546 * config/arm/arm-cpu-data.h: Regenerated.
14547 * config/arm/arm-cpu.h: Regenerated.
14549 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14551 * Makefile.in (OBJS): Move sbitmap.o from here ...
14552 (OBJS-libcommon): ... to here.
14554 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14556 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
14557 (ISA_ALL_CRYPTO): New macro.
14558 (ISA_ALL_SIMD): New macro
14559 (ISA_ALL_FP): New macro.
14560 * config/arm/arm.c (fpu_bitlist): Update initializer.
14561 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
14563 (arm9e): Add fpu. Add option for nofp
14564 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
14565 (arm926ej-s, arm1026ej-s): Likewise.
14566 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
14567 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
14568 neon-fp16, neon-vfpv4, nofp and nosimd.
14569 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
14570 (cortex-a8): Add fpu. Add option for nofp.
14571 (cortex-a9): Add fpu. Add options for nosimd and nofp.
14572 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
14573 (cortex-r4f): Add fpu.
14574 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
14575 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
14577 (cortex-r8): Likewise.
14578 (cortex-m4): Add fpu. Add option for nofp.
14579 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
14580 (cortex-a17.cortex-a7): Likewise.
14581 (cortex-a32): Add fpu. Add options for crypto and nofp.
14582 (cortex-a35, cortex-a53): Likewise.
14583 (cortex-a57): Add fpu. Add option for crypto.
14584 (cortex-a72, cortex-a73): Likewise.
14585 (exynos-m1): Likewise.
14586 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
14587 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
14588 (cortex-m33): Add fpu. Add option for nofp.
14589 * config/arm/arm-cpu-cdata.h: Regenerated
14590 * config/arm/arm-cpu-data.h: Regenerated.
14592 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14594 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
14595 (armv5te, armv5tej): Likewise.
14596 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
14597 (armv7): Add options fp and vfpv3-d16.
14598 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
14599 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
14601 (armv7ve): Likewise.
14602 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
14603 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
14604 (armv8-a): Add nocrypto option.
14605 (armv8.1-a, armv8.2-a): Likewise.
14606 (armv8-m.main): add options fp, fp.dp and nofp.
14608 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14610 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
14612 (armv8-a+crc): Delete.
14613 (armv8.1-a): Add options simd, crypto and nofp.
14614 (armv8.2-a): Add options fp16, simd, crypto and nofp.
14615 (armv8.2-a+fp16): Delete.
14616 (armv8-m.main): Add option dsp.
14617 (armv8-m.main+dsp): Delete.
14618 (cortex-a8): Add fpu. Add nofp option.
14619 (cortex-a9): Add fpu. Add nofp and nosimd options.
14620 * config/arm/parsecpu.awk (gen_data): Generate option tables and
14621 link to main cpu and architecture data structures.
14622 (gen_comm_data): Only put isa attributes from the main architecture
14624 (option): New statement for architecture and CPU entries.
14625 * arm.c (struct cpu_option): New structure.
14626 (struct processors): Add entry for options.
14627 (arm_unrecognized_feature): New function.
14628 (arm_parse_arch_cpu_name): Ignore any characters after the first
14630 (arm_parse_arch_cpu_feature): New function.
14631 (arm_configure_build_target): Separate out any CPU and architecture
14632 features and parse separately. Don't error out if -mfpu=auto is
14633 used with only an architecture string.
14634 (arm_print_asm_arch_directives): New function.
14635 (arm_file_start): Call it.
14636 * config/arm/arm-cpu-cdata.h: Regenerated.
14637 * config/arm/arm-cpu-data.h: Likewise.
14638 * config/arm/arm-tables.opt: Likewise.
14640 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14642 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
14643 assembler when it is not -mfpu=auto.
14645 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14647 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
14648 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
14649 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
14650 (ASM_CPU_SPEC): Rewrite.
14651 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
14652 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
14653 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
14654 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
14655 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
14656 copied string is NUL-terminated. Also strip any characters prefixed
14658 (arm_rewrite_selected_arch): New function.
14659 (arm_rewrite_march): New function.
14661 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14663 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
14664 (x_arm_cpu_string, x_arm_tune_string): Likewise.
14665 (march, mcpu, mtune): Convert to string-based options.
14666 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
14667 (arm_parse_arch_cpu_name): New function.
14668 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
14669 identify selected architecture or CPU.
14670 (arm_option_save): New function.
14671 (TARGET_OPTION_SAVE): Redefine.
14672 (arm_option_restore): Restore string options.
14673 (arm_option_print): Print string options.
14675 2017-06-16 Martin Sebor <msebor@redhat.com>
14677 PR tree-optimization/80933
14678 PR tree-optimization/80934
14679 * builtins.c (fold_builtin_3): Do not handle bcmp here.
14680 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
14681 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
14682 (gimple_fold_builtin): Call them.
14684 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14686 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
14687 as unlikely; update profile.
14689 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14691 * predict.c (force_edge_cold): Handle declaring edges impossible
14694 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14696 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
14698 (try_unroll_loop_completely): Fix reporting.
14700 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14702 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
14704 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
14707 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
14708 if given a non-constant argument for an intrinsic which requires a
14711 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14713 * profile.c (compare_freqs): New function.
14714 (branch_prob): Sort edge list.
14715 (find_spanning_tree): Assume that the list is priority sorted.
14717 2017-06-16 Richard Biener <rguenther@suse.de>
14719 PR tree-optimization/81090
14720 * passes.def (pass_record_bounds): Remove.
14721 * tree-pass.h (make_pass_record_bounds): Likewise.
14722 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
14723 make_pass_record_bounds): Likewise.
14724 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
14725 not free niter estimates at the beginning but at the end.
14726 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
14728 2017-06-16 Richard Biener <rguenther@suse.de>
14730 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
14731 initializer to workaround ICE in host GCC 4.8.
14733 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14735 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
14737 (clone_inlined_nodes): Update.
14739 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14741 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
14742 prefetch settings, and enable prefetching by default at -O3.
14744 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14746 * config/aarch64/aarch64.c (aarch64_override_options_internal):
14747 Set flag_prefetch_loop_arrays according to tuning data.
14749 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14751 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
14752 New tune structure.
14753 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
14754 [Unrelated to main purpose of the patch] Place the pointer field last
14755 to enable type checking errors when tune structure are wrongly merged.
14756 * config/aarch64/aarch64.c (generic_prefetch_tune,)
14757 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
14758 (thunderx2t99_prefetch_tune): New tune constants.
14759 (tune_params *_tunings): Update all tunings (no functional change).
14760 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
14761 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
14762 from tunings structures.
14764 2017-06-16 Jakub Jelinek <jakub@redhat.com>
14767 * ubsan.c (instrument_null): Add T argument, use it instead
14768 of computing it based on IS_LHS.
14769 (instrument_object_size): Likewise.
14770 (pass_ubsan::execute): Adjust instrument_null and
14771 instrument_object_size callers to pass gimple_get_lhs or
14772 gimple_assign_rhs1 result to it. Use instrument_null instead of
14773 calling get_base_address and instrument_mem_ref. Handle
14774 aggregate call arguments for object-size sanitization.
14776 2017-06-16 Yury Gribov <tetra2005@gmail.com>
14778 PR tree-optimization/81089
14779 * tree-vrp.c (is_masked_range_test): Validate operands of
14782 2017-06-15 Martin Sebor <msebor@redhat.com>
14785 * dumpfile.c (dump_register): Avoid calling memset to initialize
14786 a class with a default ctor.
14787 * gcc.c (struct compiler): Remove const qualification.
14788 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
14789 * hash-table.h: Ditto.
14790 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
14792 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
14793 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
14795 * params.h (struct param_info): Make struct members non-const.
14796 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
14797 with default initialization.
14798 * vec.h (vec_copy_construct, vec_default_construct): New helper
14800 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
14801 with vec_copy_construct.
14802 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
14803 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
14804 * doc/invoke.texi (-Wclass-memaccess): Document.
14806 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14808 * emit-rtl.h (is_leaf): Update comment about local
14809 register allocator.
14811 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14814 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
14815 for a variable to have a section before checking if the section has a
14817 Set section to.persistent if persistent attribute is set.
14818 Warn if .persistent attribute is used on an automatic variable.
14820 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
14822 PR rtl-optimization/80474
14823 * reorg.c (update_block): Do not ignore instructions in a delay slot.
14825 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
14827 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
14830 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
14832 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
14833 (casesi): Emit bounds checking as RTL.
14834 (casesi_internal_mips16_<mode>): Remove bounds checking.
14836 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
14838 * config/xtensa/xtensa.c (xtensa_option_override): Append
14839 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
14840 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
14841 xtensa_doloop_hooks): Define unconditionally.
14842 (xtensa_reorg_loops): Only call reorg_loops in the presence of
14844 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
14845 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
14846 for it in xtensa_option_override.
14847 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
14848 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
14850 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
14852 * doc/cppopts.texi: Document '-' special value to -MF.
14854 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
14856 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
14857 (cortex_a53_fconst): Likewise.
14858 (cortex_a53_fpmul): Likewise.
14859 (cortex_a53_f_load_64): Likewise.
14860 (cortex_a53_f_load_many): Likewise.
14861 (cortex_a53_advsimd_alu): Likewise.
14862 (cortex_a53_advsimd_alu_q): Likewise.
14863 (cortex_a53_advsimd_mul): Likewise.
14864 (cortex_a53_advsimd_mul_q): Likewise.
14865 (fpmac bypass): Add new bypass for fpmac-fpmac case.
14866 Add missing fmul, r2f_cvt and fconst cases.
14868 2017-06-14 Richard Biener <rguenther@suse.de>
14870 PR middle-end/81088
14871 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
14873 (fold_binary_loc): When associating do not treat pre-existing
14874 TREE_OVERFLOW on literal constants as a reason to allow
14875 TREE_OVERFLOW on associated literal constants.
14877 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14879 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
14880 (MASK_FEATURES): New macro.
14881 * config/sparc/sparc.c (sparc_option_override): Remove the special
14882 handling of -mfpu and generalize it to all MASK_FEATURES switches.
14884 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14886 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
14887 a division of 0 if non-call exceptions are enabled.
14889 2017-06-14 Andrew Pinski <apinski@cavium.com>
14890 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14893 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
14894 Improve vector initialization code gen for only variable case.
14896 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14898 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
14900 2017-06-14 Richard Biener <rguenther@suse.de>
14902 PR tree-optimization/81083
14903 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
14906 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14908 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
14909 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
14910 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
14911 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
14912 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
14913 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
14915 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14917 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
14918 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
14920 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14922 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
14924 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14926 * config/rs6000/t-rtems: Don't handle SPE.
14928 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14930 * config/rs6000/t-linux: Don't handle SPE.
14932 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14934 * config/rs6000/eabispe.h: Delete file.
14936 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14938 * config/rs6000/t-spe: Delete file.
14940 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14942 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
14943 (rs6000_legitimate_offset_address_p): Return false for anything in
14944 V2SImode or V2SFmode.
14946 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14948 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
14949 except V2SF and V2SI. Rearrange the vector modes, and add comments.
14950 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
14952 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
14953 (rs6000_legitimate_offset_address_p): Ditto.
14954 (rs6000_emit_move): Ditto.
14955 (rs6000_init_builtins): Remove V4HI_type_node.
14957 2017-06-13 Martin Liska <mliska@suse.cz>
14960 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
14961 (gate_asan): Likewise.
14962 * asan.h (asan_no_sanitize_address_p): Remove the function.
14963 (sanitize_flags_p): New function.
14964 * builtins.def: Fix coding style.
14965 * common.opt: Use renamed enum value.
14966 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
14967 * doc/extend.texi: Document no_sanitize attribute.
14968 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
14969 to SANITIZE_UNDEFINED_NONDEFAULT.
14970 * gcc.c (sanitize_spec_function): Use the renamed enum value.
14971 * gimple-fold.c (optimize_atomic_compare_exchange_p):
14972 Use sanitize_flags_p.
14973 * gimplify.c (gimplify_function_tree): Likewise.
14974 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
14975 * opts.c (parse_no_sanitize_attribute): New function.
14976 (common_handle_option): Use renamed enum value.
14977 * opts.h (parse_no_sanitize_attribute): Declare.
14978 * tree.c (sanitize_flags_p): New function.
14979 * tree.h: Declared here.
14980 * tsan.c: Use sanitize_flags_p.
14981 * ubsan.c (ubsan_expand_null_ifn): Likewise.
14982 (instrument_mem_ref): Likewise.
14983 (instrument_bool_enum_load): Likewise.
14984 (do_ubsan_in_current_function): Remove the function.
14985 (pass_ubsan::execute): Use sanitize_flags_p.
14986 * ubsan.h: Remove do_ubsan_in_current_function
14987 * tree-cfg.c (print_no_sanitize_attr_value): New function.
14988 (dump_function_to_file): Use it here.
14990 2017-06-13 Martin Jambor <mjambor@suse.cz>
14992 PR tree-optimization/80803
14993 PR tree-optimization/81063
14994 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
14995 (propagate_subaccesses_across_link): Enqueue subtree whenever
14996 necessary instead of relying on the caller.
14998 2017-06-13 Martin Jambor <mjambor@suse.cz>
15000 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
15001 that have a first_link.
15002 (sort_and_splice_var_accesses): Do not check first_link before
15004 (subtree_mark_written_and_enqueue): Likewise.
15005 (propagate_all_subaccesses): Likewise and do not stop at first
15006 parent with a first_link.
15008 2017-06-13 Martin Jambor <mjambor@suse.cz>
15010 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
15013 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15015 * match.pd: New pattern.
15017 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15019 * tree-vrp.c (is_masked_range_test): New function.
15020 (register_edge_assert_for): Determine ranges for
15023 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15025 PR tree-optimization/67328
15026 * fold-const.c (maskable_range_p): New function.
15027 (build_range_check): Generate bittests if possible.
15029 2017-06-13 Martin Liska <mliska@suse.cz>
15031 * gimple-pretty-print.c (dump_probability): Add new argument.
15032 (dump_edge_probability): Dump both probability and count.
15033 (dump_gimple_label): Likewise.
15034 (dump_gimple_bb_header): Likewise.
15036 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
15039 * config/avr/avr-devices.c: Fix indentation.
15040 * config/avr/gen-avr-mmcu-specs.c: Dito.
15042 2017-06-13 Richard Biener <rguenther@suse.de>
15044 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
15045 instead get vector type from stmt_info.
15046 (vectorizable_reduction): Adjust. Remove dead code.
15048 2017-06-13 Richard Biener <rguenther@suse.de>
15050 PR middle-end/81065
15051 * fold-const.c (extract_muldiv_1): Remove bogus distribution
15052 case of C * (x * C2 + C3).
15053 (fold_addr_of_array_ref_difference): Properly fold index difference.
15055 2017-06-12 David S. Miller <davem@davemloft.net>
15058 * config/sparc/sparc.md (return expander): Emit frame blockage if
15059 function uses alloca.
15061 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
15063 * combine.c (make_field_assignment): Check len rather than the mode
15064 precision when calling force_to_mode.
15066 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
15068 Support multilibs and devices that see flash in RAM address range.
15071 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
15072 (avr_mcu_t) <flash_pm_offset>: New field.
15073 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
15074 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
15075 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
15076 (AVR_TINY_PM_OFFSET): Remove macro.
15077 * config/avr/avr.opt (-mshort-calls): New option.
15078 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
15079 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
15080 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
15081 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
15082 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
15083 instead of avr_arch->have_jmp_call.
15084 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
15085 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
15086 avr_arch->flash_pm_offset to define.
15087 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
15088 new field flash_pm_offset. Add entry for avrxmega3.
15089 (avr_texinfo): Add entry for avrxmega3.
15090 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
15091 attiny212, attiny214,
15092 attiny412, attiny414, attiny416, attiny417,
15093 attiny814, attiny816, attiny817,
15094 attiny1614, attiny1616, attiny1617,
15095 attiny3214, attiny3216, attiny3217.
15096 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
15097 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
15098 (avr_print_operand_address) [AVR_TINY]: Same.
15099 (avr_asm_init_sections) <readonly_data_section>: Only patch
15100 callback if avr_arch->flash_pm_offset = 0.
15101 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
15102 for rodata if avr_arch->flash_pm_offset != 0.
15103 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
15104 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
15105 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
15106 (m_options): Append opt_rcall.
15107 (m_dirnames): Append dir_rcall.
15108 * config/avr/t-multilib: Regenerate.
15110 * configure.ac [target=avr]: Check whether avrxmega3 default
15111 linker description file works as needed.
15112 * configure: Regenerate.
15113 * doc/avr-mmcu.texi: Regenerate.
15114 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
15115 <__AVR_ARCH__>: Document avrxmega3 and 103.
15116 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
15117 <__AVR_SHORT_CALLS__>: Document it.
15118 <__AVR_PM_BASE_ADDRESS__>: Document it.
15119 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
15120 (AVR Variable Attributes) <progmem>: Document this is
15121 not needed for avrxmega3.
15122 (AVR Named Address Spaces) <__flash>: Dito.
15124 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
15126 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
15128 2017-06-12 Doug Rupp <rupp@adacore.com>
15130 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
15131 Append vxworks-stdint.h to the tm_file list.
15132 * config/vxworks-stdint.h: New file.
15134 2017-06-12 Martin Liska <mliska@suse.cz>
15136 PR tree-optimization/81041
15137 * tree-profile.c (gimple_gen_ic_func_profiler):
15138 Create an extra BB in profile-generate
15139 (gimple_gen_time_profiler): Likewise.
15141 2017-06-12 Jakub Jelinek <jakub@redhat.com>
15143 PR tree-optimization/81003
15144 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
15145 (update_range_test): Use it instead of force_gimple_operand_gsi.
15147 2017-06-12 Richard Biener <rguenther@suse.de>
15149 PR tree-optimization/81053
15150 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
15151 with backedge value not defined in loop. Simplify def stmt
15154 2017-06-11 Tom de Vries <tom@codesourcery.com>
15157 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
15159 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
15160 nvptx_cannot_force_const_mem.
15162 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15164 * opts.c (finish_options): Move test for flag_split_stack after
15165 it has been initialized.
15167 2017-06-11 Jason Merrill <jason@redhat.com>
15169 * tree.h (id_equal): New.
15170 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
15171 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
15172 instead of strcmp of IDENTIFIER_POINTER.
15174 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15176 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
15177 (mark_all_inlined_calls_cdtor): Fix formating.
15178 (inline_transform): Rescale profile before inlining.
15180 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15182 * cgraph.h (cgraph_edge::clone): Update prototype.
15183 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
15184 (cgraph_node::create_clone): Update.
15185 (cgraph_node::create_version_clone): Update.
15186 * tree-inline.c (copy_bb): Update.
15187 (expand_call_inline): Update.
15189 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
15191 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
15192 factored out from ...
15193 (rs6000_emit_prologue): ... here.
15195 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
15197 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
15198 factored out from ...
15199 (rs6000_emit_prologue): ... here.
15201 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15203 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
15205 (handle_missing_profiles): Fix computation of tp_first_run.
15206 (counts_to_freqs): Do not touch freqs when count is 0.
15208 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15210 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
15213 2017-06-10 Tom de Vries <tom@codesourcery.com>
15215 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
15216 attributes): Document signal effective target.
15218 2017-06-10 Tom de Vries <tom@codesourcery.com>
15220 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
15221 Document effective target stack_size.
15223 2017-06-09 David Malcolm <dmalcolm@redhat.com>
15225 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
15226 to the edit_context if they can be auto-applied.
15228 2017-06-9 Ian Lance Taylor <iant@golang.org>
15230 * opts.c (finish_options): If -fsplit-stack, disable implicit
15231 -forder-blocks-and-partition.
15232 * doc/invoke.texi (Optimize Options): Document that when using
15233 -fsplit-stack -forder-blocks-and-partition is not implicitly
15236 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15238 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
15239 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
15240 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
15241 * builtins.def (abort, trap, unreachable): Declare cold.
15242 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
15243 * tree-core.h (ECF_COLD): New.
15244 * tree.c (set_call_expr_flags): Handle ECF_COLD.
15245 (build_common_builtin_nodes): Mark unreachable and abort as cold.
15247 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15249 * predict.c (unlikely_executed_stmt_p): Cleanup.
15251 2017-06-09 Richard Biener <rguenther@suse.de>
15253 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
15254 model if the ref is always written to.
15256 2017-06-09 Tamar Christina <tamar.christina@arm.com>
15258 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
15260 2017-06-09 Tamar Christina <tamar.christina@arm.com>
15262 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
15265 2017-06-09 Tom de Vries <tom@codesourcery.com>
15268 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
15269 "target cannot support label values" when encountering LABEL_REF.
15271 2017-06-09 Martin Liska <mliska@suse.cz>
15273 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
15274 (gimple_gen_ic_func_profiler): Emit direct comparison
15275 of __gcov_indirect_call_callee with NULL.
15276 (gimple_gen_time_profiler): Change probability from
15277 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
15279 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15281 * profile.c (edge_gcov_counts): Turn to pointer.
15282 (compute_branch_probabilities, compute_branch_probabilities): Update.
15283 (branch_prob): Do not clear edge_gcov_count.
15284 * profile.h (edge_gcov_counts): Turn to pointer.
15285 (edge_gcov_count): Update.
15287 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15289 * gimple.h (gimple_check_failed): Mark cold.
15291 2017-06-09 Richard Biener <rguenther@suse.de>
15293 PR tree-optimization/66623
15294 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
15295 refactor check_reduction into two parts, properly computing
15296 whether we have to check reduction validity for outer loop
15299 2017-06-09 Richard Biener <rguenther@suse.de>
15301 PR tree-optimization/79483
15302 * graphite-scop-detection.c (order): New global.
15303 (get_order): Compute bb to order mapping that satisfies code
15304 generation constraints.
15305 (cmp_pbbs): New helper.
15306 (build_scops): Start domwalk at entry block, sort generated
15309 2017-06-09 Richard Biener <rguenther@suse.de>
15311 PR middle-end/81007
15312 * ipa-polymorphic-call.c
15313 (ipa_polymorphic_call_context::restrict_to_inner_class):
15314 Skip FIELD_DECLs with error_mark_node type.
15315 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
15318 2017-06-09 Martin Liska <mliska@suse.cz>
15320 * predict.c (struct branch_predictor): New struct.
15321 (test_prediction_value_range): New test.
15322 (predict_c_tests): New function.
15323 * selftest-run-tests.c (selftest::run_tests): Run the function.
15324 * selftest.h: Declare new tests.
15326 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
15329 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
15330 gen_add3_insn did not fail.
15331 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
15332 r0, construct that number in a temporary reg and add that reg to r0.
15333 If asked to put the result in r0 as well, fail.
15335 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
15337 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15338 for early expansion of vec_eqv.
15340 2017-06-08 Jakub Jelinek <jakub@redhat.com>
15342 PR middle-end/81005
15343 * ubsan.c (instrument_null): Avoid pointless code temporary.
15344 (pass_ubsan::execute): Instrument aggregate arguments of calls.
15346 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
15350 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
15353 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
15356 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15358 * predict.c (unlikely_executed_edge_p): Move ahead.
15359 (probably_never_executed_edge_p): Use it.
15361 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
15363 PR middle-end/79988
15364 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
15365 gimple_call_builtin_p call.
15367 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15369 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
15370 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
15371 rtl_check_failed_type2, rtl_check_failed_code1,
15372 rtl_check_failed_code2, rtl_check_failed_code_mode,
15373 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
15374 rtvec_check_failed_bounds, rtl_check_failed_flag,
15375 _fatal_insn_not_found, _fatal_insn): Likewise.
15376 * tree.h (tree_contains_struct_check_failed,
15377 tree_check_failed, tree_not_check_failed,
15378 tree_class_check_failed, tree_range_check_failed,
15379 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
15380 tree_vec_elt_check_failed, phi_node_elt_check_failed,
15381 tree_operand_check_failed, omp_clause_check_failed,
15382 omp_clause_operand_check_failed, omp_clause_range_check_failed):
15385 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15387 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
15388 flag_branch_probabilities.
15389 * ipa-inline.c (edge_badness): Likewise.
15390 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
15391 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
15392 * predict.c (maybe_hot_frequency_p): Likewise.
15393 (probably_never_executed): Likewise.
15394 * sched-ebb.c (schedule_ebbs): Likewise.
15395 * sched-rgn.c (find_single_block_region): Likewise.
15396 * tracer.c (tail_duplicate): Likewise.
15398 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15400 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
15401 longer requires x_flag_profile_use.
15403 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15405 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
15406 instead of flag_reorder_blocks_and_partition.
15407 * dbxout.c (dbxout_function_end): Likewise.
15408 * dwarf2out.c (gen_subprogram_die): Likewise.
15409 * haifa-sched.c (sched_create_recovery_edges): Likewise.
15410 * hw-doloop.c (reorg_loops): Likewise.
15411 * varasm.c (assemble_start_function,
15412 assemble_end_function): Likewise.
15413 (decide_function_section): Do not check for
15414 flag_reorder_blocks_and_partition.
15416 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
15418 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
15420 (chkp_get_hard_register_fake_addr_expr): Ditto.
15421 (chkp_build_addr_expr): Add check for hard reg case.
15422 (chkp_parse_array_and_component_ref): Ditto.
15423 (chkp_find_bounds_1): Ditto.
15424 (chkp_process_stmt): Don't generate bounds store for
15427 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15429 * predict.c (maybe_hot_bb_p): Do not check profile status.
15430 (maybe_hot_edge_p): Likewise.
15431 (probably_never_executed): Check for zero counts even if profile
15433 (unlikely_executed_edge_p): New function.
15434 (unlikely_executed_stmt_p): New function.
15435 (unlikely_executed_bb_p): New function.
15436 (set_even_probabilities): Use unlikely predicates.
15437 (combine_predictions_for_bb): Likewise.
15438 (predict_paths_for_bb): Likewise.
15439 (predict_paths_leading_to_edge): Likewise.
15440 (determine_unlikely_bbs): New function.
15441 (estimate_bb_frequencies): Use it.
15442 (compute_function_frequency): Use zero counts even if profile is
15444 * profile-count.h: Fix typo.
15446 2017-08-08 Julia Koval <julia.koval@intel.com>
15448 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
15449 _mm512_mask_cvtsepi16_storeu_epi8,
15450 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
15451 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
15452 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
15453 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
15454 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
15455 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
15456 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
15457 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
15458 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
15459 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
15460 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
15461 __builtin_ia32_pmovuswb256mem_mask,
15462 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
15463 __builtin_ia32_pmovwb512mem_mask): New builtins.
15465 2017-08-08 Julia Koval <julia.koval@intel.com>
15467 PR target/73350,80862
15468 * config/i386/subst.md (round): Fix round pattern.
15469 * config/i386/i386.c (ix86_erase_embedded_rounding):
15470 Fix erasing rounding for the fixed pattern.
15472 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15474 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
15476 2017-06-08 Martin Liska <mliska@suse.cz>
15478 PR gcov-profile/80911
15479 * gcov.c (block_info::block_info): New constructor.
15481 2017-06-07 Carl Love <cel@us.ibm.com>
15483 * config/rs6000/rs6000-c: The return type of the following
15484 built-in functions was implemented as int not long long. Fix sign
15485 of return value for the unsigned version of vec_mulo and vec_mule.
15486 vector unsigned long long vec_bperm (vector unsigned long long,
15487 vector unsigned char)
15488 vector signed long long vec_mule (vector signed int,
15490 vector unsigned long long vec_mule (vector unsigned int,
15491 vector unsigned int)
15492 vector signed long long vec_mulo (vector signed int,
15494 vector unsigned long long vec_mulo (vector unsigned int,
15495 vector unsigned int)
15496 * doc/extend.texi: Fix the documentation for the built-in
15499 2017-06-07 Carl Love <cel@us.ibm.com>
15502 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
15505 2017-06-07 Carl Love <cel@us.ibm.com>
15507 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
15508 support, Generate doublehv for signed int/float for BE case only.
15510 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
15512 * doc/invoke.texi (mcx16): Rewrite.
15514 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15516 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
15517 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
15518 *mov<mode>_softfloat, and an anonymous splitter): Use
15519 nonimmediate_operand instead of rs6000_nonimmediate_operand.
15521 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15523 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
15525 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
15526 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
15527 (rs6000_debug_reg_global): Adjust.
15528 (rs6000_init_hard_regno_mode_ok): Adjust.
15529 (rs6000_dbx_register_number): Adjust.
15530 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
15531 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
15532 Remove SPE_ACC and SPEFSCR.
15533 (REG_ALLOC_ORDER): Ditto.
15534 (FRAME_POINTER_REGNUM): Change to 111.
15535 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
15536 (REG_CLASS_NAMES): Ditto.
15537 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
15538 (REGISTER_NAMES): Ditto.
15539 (ADDITIONAL_REG_NAMES): Ditto.
15540 (rs6000_reg_names): Ditto.
15541 * config/rs6000/rs6000.md: Renumber some register number
15544 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15546 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
15548 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
15549 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
15551 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
15552 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
15553 Delete the SPE high registers.
15554 (REG_ALLOC_ORDER): Ditto.
15555 (enum reg_class): Remove SPE_HIGH_REGS.
15556 (REG_CLASS_NAMES): Ditto.
15557 (REG_CLASS_CONTENTS): Delete the SPE high registers.
15558 (REGISTER_NAMES): Ditto.
15559 (rs6000_reg_names): Ditto.
15560 * doc/tm.texi.in: Remove SPE as example.
15561 * doc/tm.texi: Regenerate.
15563 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15565 * config/rs6000/8540.md (ppc8540_brinc): Delete.
15566 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
15567 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
15568 * config/rs6000/rs6000.md (type): Remove "brinc".
15570 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15572 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
15573 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
15574 * config/rs6000/linuxspe.h: Delete file.
15575 * config/rs6000/rs6000.md: Don't include spe.md.
15576 * config/rs6000/spe.h: Delete file.
15577 * config/rs6000/spe.md: Delete file.
15578 * config/rs6000/t-rs6000: Remove spe.md.
15580 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15582 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
15583 (reg_or_none500mem_operand): Delete.
15584 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
15585 instead of reg_or_none500mem_operand.
15587 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15589 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
15590 handling of SPE flags.
15591 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
15593 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15595 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
15597 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
15598 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
15599 paired_divv2sf3): Similar.
15600 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
15601 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
15602 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
15604 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
15605 Rename the paired_* instruction patterns.
15606 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
15608 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
15609 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
15610 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
15611 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
15612 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
15613 PAIRED_VECTOR_MODE.
15614 (struct machine_function): Delete field spe_insn_chain_scanned_p.
15615 (spe_func_has_64bit_regs_p): Delete.
15616 (spe_expand_predicate_builtin): Delete.
15617 (spe_expand_evsel_builtin): Delete.
15618 (TARGET_DWARF_REGISTER_SPAN): Do not define.
15619 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
15620 (invalid_e500_subreg): Delete.
15621 (rs6000_legitimize_address): Always force_reg op2 as well, for
15622 paired single memory accesses.
15623 (rs6000_member_type_forces_blk): Delete.
15624 (rs6000_spe_function_arg): Delete.
15625 (rs6000_expand_unop_builtin): Delete SPE handling.
15626 (rs6000_expand_binop_builtin): Ditto.
15627 (spe_expand_stv_builtin): Delete.
15628 (bdesc_2arg_spe): Delete.
15629 (spe_expand_builtin): Delete.
15630 (spe_expand_predicate_builtin): Delete.
15631 (spe_expand_evsel_builtin): Delete.
15632 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
15633 (spe_init_builtins): Delete.
15634 (spe_func_has_64bit_regs_p): Delete.
15635 (savres_routine_name): Delete "info" parameter. Adjust callers.
15636 (rs6000_emit_stack_reset): Ditto.
15637 (rs6000_dwarf_register_span): Delete.
15638 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
15639 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
15640 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
15642 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
15644 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
15645 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
15646 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
15647 mulv2sf3, divv2sf3): Delete expanders.
15649 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15651 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
15653 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15655 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
15656 * config/rs6000/rs6000.c: Ditto.
15658 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15660 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
15661 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
15662 comparison_operator.
15664 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15666 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
15667 * config/rs6000/rs6000.opt: Ditto.
15668 * config/rs6000/t-rtems: Ditto.
15670 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15672 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
15673 TARGET_E500_SINGLE by 0, simplify.
15674 * config/rs6000/rs6000.c: Ditto.
15675 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
15676 (spe_build_register_parallel): Delete.
15677 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
15678 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
15679 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
15680 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
15681 (E500_CONVERT): Delete.
15682 * config/rs6000/spe.md: Remove many patterns and all define_constants.
15684 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15686 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
15687 * config/rs6000/dfp.md: Ditto.
15688 (negdd2, *negdd2_fpr): Merge.
15689 (absdd2, *absdd2_fpr): Merge.
15690 (negtd2, *negtd2_fpr): Merge.
15691 (abstd2, *abstd2_fpr): Merge.
15692 * config/rs6000/e500.h: Delete file.
15693 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
15694 TARGET_FPRS by 1 and simplify.
15695 * config/rs6000/rs6000-c.c: Ditto.
15696 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
15697 TARGET_DF_SPE by 0.
15698 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
15700 * config/rs6000/rs6000.md: Ditto.
15701 (floatdidf2, *floatdidf2_fpr): Merge.
15702 (move_from_CR_gt_bit): Delete.
15703 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
15704 (E500_CR_IOR_COMPARE): Delete.
15705 (All patterns that require !TARGET_FPRS): Delete.
15706 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
15708 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15710 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
15712 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15714 * graphds.c (add_edge): Intitialize edge's attached data.
15715 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
15716 pointer parameter. Call pointed function on each edge during
15717 graph traversing. Skip traversing the edge when the function
15719 (graphds_dfs, graphds_scc): Ditto.
15720 (for_each_edge): New parameter. Pass the new parameter to callback
15722 * graphds.h (skip_edge_callback): New function pointer type.
15723 (graphds_dfs, graphds_scc): New function pointer parameter.
15724 (graphds_edge_callback, for_each_edge): New parameter.
15726 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15728 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
15729 out code checking if runtime alias check is possible to below ...
15730 Call the new function.
15731 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
15732 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
15734 2017-06-07 Marek Polacek <polacek@redhat.com>
15737 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
15738 TYPE_OVERFLOW_WRAPS check.
15740 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15742 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
15743 if versioning is required.
15744 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
15745 peeling with the check for versioning.
15747 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15749 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
15750 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
15751 Set true to new parameter if new ssa variable is defined.
15752 (vect_gen_vector_loop_niters): Refactor. Set range information
15753 for the new vector loop bound variable.
15754 (vect_do_peeling): Ditto.
15756 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15758 * tree-affine.c (ssa.h): Include header file.
15759 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
15760 has wrapping overflow behavior.
15762 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15764 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
15766 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15768 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
15769 (tree_to_aff_combination): ... here.
15771 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15773 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
15774 reg_pressure model function.
15775 (ivopts_global_cost_for_size): Delete.
15776 (determine_set_costs, iv_ca_recount_cost): Call new model function
15777 ivopts_estimate_reg_pressure.
15779 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15781 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
15782 expensive than udiv. Remove floating point cases from mod.
15784 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15786 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
15787 Increase idiv cost.
15789 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15791 * config/aarch64/aarch64.md
15792 (copysignsf3): Fix mask generation.
15794 2017-06-07 Jakub Jelinek <jakub@redhat.com>
15796 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
15798 (class dump_manager): Add register_dumps method.
15799 * dumpfile.c: Include langhooks.h.
15800 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
15801 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
15802 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
15803 (dump_manager::dump_register): Start with 512 entries instead of 32.
15804 (dump_manager::register_dumps): New method.
15805 * toplev.c (general_init): Instead of invoking register_dumps
15806 langhook, invoke register_dumps method on the dump manager.
15807 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
15810 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
15812 * doc/md.texi: Clarify the restrictions on a define_insn condition.
15813 Say that # requires an associated define_split to exist, and that
15814 the define_split must be suitable for use after register allocation.
15816 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15818 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
15819 (compute_outgoing_frequencies): Also initialize zero counts.
15820 (find_many_sub_basic_blocks): Do not produce uninitialized profile
15821 around loops; preserve more of profile when nothing changes.
15823 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
15825 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
15827 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
15828 * config/arm/arm-cpu-cdata.h: Regenerate.
15829 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
15830 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
15831 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
15832 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
15833 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
15835 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
15837 * config/arm/t-rmprofile: Likewise.
15838 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
15840 2017-06-06 David S. Miller <davem@davemloft.net>
15843 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
15844 blockage if function uses alloca.
15846 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15848 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
15849 New "uid" fields to hold pretty-print IDs of group and ref.
15850 Memory references are now identified as <group_id>:<ref_id>
15851 instead of using [random] addresses.
15852 (dump_mem_details): Simplify, no functional change.
15853 (dump_mem_ref): Simplify and make output more concise.
15854 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
15855 (find_or_create_group): Initialize group uid.
15856 (record_ref): Initialize ref uid. Improve debug output.
15857 (prune_group_by_reuse, should_issue_prefetch_p,)
15858 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
15859 (mark_nontemporal_store, determine_loop_nest_reuse):
15860 Improve debug output.
15862 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15864 * dbgcnt.def (prefetch): New debug counter.
15865 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
15866 (schedule_prefetches): Stop issueing prefetches if debug counter
15869 2017-06-06 Tom de Vries <tom@codesourcery.com>
15871 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
15872 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
15874 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15876 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
15877 Use aarch64_reg_or_zero predicate for operand 4.
15878 (aarch64_compare_and_swap<mode> define_insn_and_split):
15879 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
15880 (aarch64_store_exclusive<mode>): Likewise for operand 2.
15882 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
15884 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
15885 (arm_compute_save_core_reg_mask): This.
15886 (thumb1_compute_save_reg_mask): Rename into ...
15887 (thumb1_compute_save_core_reg_mask): This.
15888 (arm_compute_save_reg0_reg12_mask): Adapt comment.
15889 (arm_compute_frame_layout): Likewise.
15891 2017-06-06 Richard Biener <rguenther@suse.de>
15893 PR tree-optimization/80974
15894 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
15895 keep or clear leaders SSA info.
15897 2017-06-06 Tom de Vries <tom@codesourcery.com>
15899 * config/nvptx/nvptx.c (split_mode_p): New function.
15900 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
15902 2017-06-06 Tom de Vries <tom@codesourcery.com>
15904 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
15906 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15909 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
15912 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15914 * shrink-wrap.c (handle_simple_exit): Update profile.
15915 (try_shrink_wrapping): Upate profile.
15917 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15919 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
15920 (tree_guess_outgoing_edge_probabilities): New.
15921 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
15922 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
15924 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15926 * ipa-split.c (split_function): Initialize return bb profile.
15928 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15930 * profile.c (compute_branch_probabilities): Also initialize
15931 EXIT_BLOCK profile.
15933 2017-06-06 Richard Biener <rguenther@suse.de>
15935 PR tree-optimization/80928
15936 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
15937 (vect_analyze_loop_operations): Properly guard analysis for
15939 (vect_transform_loop): Likewise.
15940 (vect_analyze_loop_2): Also reset SLP type on PHIs.
15941 (vect_model_induction_cost): Do not cost for pure SLP.
15942 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
15943 of induction in inner loop vectorization.
15944 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
15945 (vect_get_and_check_slp_defs): Handle vect_induction_def.
15946 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
15948 (vect_analyze_slp_cost_1): Cost induction.
15949 (vect_detect_hybrid_slp_stmts): Handle PHIs.
15950 (vect_get_slp_vect_defs): Likewise.
15951 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
15952 (vect_transform_stmt): Handle SLP reductions.
15953 * tree-vectorizer.h (vectorizable_induction): Adjust.
15955 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15957 * config/rs6000/rs6000.c (make_resolver_func): Update
15958 init_lowered_empty_function call.
15960 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
15962 * doc/invoke.texi: Document the -fprofile-abs-path option.
15963 * common.opt (fprofile-abs-path): New option.
15964 * gcov-io.h (gcov_write_filename): Declare.
15965 * gcov-io.c (gcov_write_filename): New function.
15966 * coverage.c (coverage_begin_function): Use gcov_write_filename.
15967 * profile.c (output_location): Likewise.
15969 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15971 * shring-wrap.c: Revert accidental commit.
15973 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
15975 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
15977 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15979 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
15981 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
15982 profile in callgraph edge.
15983 * profile-count.h (apply_probability): If THIS is 0, then result is 0
15984 (apply_scale): Likewise.
15985 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
15986 Also scale profile when inlining function with zero profile.
15987 (initialize_cfun): Update exit block profile even when it is zero.
15988 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
15989 when profile is read.
15991 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15993 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
15994 (CLONE_*): New constants to define the processors we can generate
15995 code for with the target_clone attribute.
15996 (rs6000_clone_map): New array to identify which clone processors
15997 the current program is running on.
15998 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
15999 target_clone attribute.
16000 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
16001 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
16002 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
16003 (cpu_expand_builtin): Add support for target_clone attribute.
16004 (rs6000_valid_attribute_p): Allow "default" attribute.
16005 (get_decl_name): New debug function to simplify printing the
16006 current function name in debugging statements.
16007 (rs6000_clone_priority): New functions to support the target_clone
16008 attribute, and be able to generate code to switch between ISA 2.05
16009 through ISA 3.0 (power6 through power9).
16010 (rs6000_compare_version_priority): Likewise.
16011 (rs6000_get_function_versions_dispatcher): Likewise.
16012 (make_resolver_func): Likewise.
16013 (add_condition_to_bb): Likewise.
16014 (dispatch_function_versions): Likewise.
16015 (rs6000_generate_version_dispatcher_body): Likewise.
16016 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
16017 (fusion_gpr_load_p): Fix a spacing issue.
16018 * doc/extend.texi (Common Function Attributes): Document that the
16019 PowerPC supports the target_clone attribute.
16021 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
16023 * config/arm/arm.h: explain F symbol found in description of ARM
16024 register allocation in its legend.
16026 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16028 * config/mips/frame-header-opt.c: Include profile-count.h.
16029 * config/riscv/riscv.c: Include profile-count.h
16031 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16033 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
16035 (sm_set_flag_if_changed): Add bbs field.
16036 (execute_sm_if_changed_flag_set): Pass BBS.
16037 (execute_sm): Update.
16039 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16041 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
16044 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16046 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
16047 (peephole2): New peephole2 to emit the above.
16048 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
16050 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16052 * config/aarch64/aarch64.c (define_peephole2 above
16053 *sub_<shift>_<mode>): New peephole.
16055 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16057 * config/i386/i386.c (make_resolver_func): Update.
16058 * Makefile.in: Add profile-count.h and profile-count.o
16059 * auto-profile.c (afdo_indirect_call): Update to new API.
16060 (afdo_set_bb_count): Update.
16061 (afdo_propagate_edge): Update.
16062 (afdo_propagate_circuit): Update.
16063 (afdo_calculate_branch_prob): Update.
16064 (afdo_annotate_cfg): Update.
16065 * basic-block.h: Include profile-count.h
16066 (struct edge_def): Turn count to profile_count.
16067 (struct basic_block_def): Likewie.
16068 (REG_BR_PROB_BASE): Move to profile-count.h
16069 (RDIV): Move to profile-count.h
16070 * bb-reorder.c (max_entry_count): Turn to profile_count.
16071 (find_traces): Update.
16072 (rotate_loop):Update.
16073 (connect_traces):Update.
16074 (sanitize_hot_paths):Update.
16075 * bt-load.c (migrate_btr_defs): Update.
16076 * cfg.c (RDIV): Remove.
16077 (init_flow): Use alloc_block.
16078 (alloc_block): Uninitialize count.
16079 (unchecked_make_edge): Uninitialize count.
16080 (check_bb_profile): Update.
16081 (dump_edge_info): Update.
16082 (dump_bb_info): Update.
16083 (update_bb_profile_for_threading): Update.
16084 (scale_bbs_frequencies_int): Update.
16085 (scale_bbs_frequencies_gcov_type): Update.
16086 (scale_bbs_frequencies_profile_count): New.
16087 * cfg.h (update_bb_profile_for_threading): Update.
16088 (scale_bbs_frequencies_profile_count): Declare.
16089 * cfgbuild.c (compute_outgoing_frequencies): Update.
16090 (find_many_sub_basic_blocks): Update.
16091 * cfgcleanup.c (try_forward_edges): Update.
16092 (try_crossjump_to_edge): Update.
16093 * cfgexpand.c (expand_gimple_tailcall): Update.
16094 (construct_exit_block): Update.
16095 * cfghooks.c (verify_flow_info): Update.
16096 (dump_bb_for_graph): Update.
16097 (split_edge): Update.
16098 (make_forwarder_block): Update.
16099 (duplicate_block): Update.
16100 (account_profile_record): Update.
16101 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
16102 (get_estimated_loop_iterations): Update.
16103 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
16104 (single_likely_exit): Update.
16105 * cfgloopmanip.c (scale_loop_profile): Update.
16107 (set_zero_probability): Update.
16108 (lv_adjust_loop_entry_edge): Update.
16109 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16110 (purge_dead_edges): Update.
16111 (rtl_account_profile_record): Update.
16112 * cgraph.c (cgraph_node::create): Uninitialize count.
16113 (symbol_table::create_edge): Uninitialize count.
16114 (cgraph_update_edges_for_call_stmt_node): Update.
16115 (cgraph_edge::dump_edge_flags): Update.
16116 (cgraph_node::dump): Update.
16117 (cgraph_edge::maybe_hot_p): Update.
16118 * cgraph.h: Include profile-count.h
16119 (create_clone), create_edge, create_indirect_edge): Update.
16120 (cgraph_node): Turn count to profile_count.
16121 (cgraph_edge0: Likewise.
16122 (make_speculative, clone): Update.
16123 (create_edge): Update.
16124 (init_lowered_empty_function): Update.
16125 * cgraphclones.c (cgraph_edge::clone): Update.
16126 (duplicate_thunk_for_node): Update.
16127 (cgraph_node::create_clone): Update.
16128 * cgraphunit.c (cgraph_node::analyze): Update.
16129 (cgraph_node::expand_thunk): Update.
16130 * final.c (dump_basic_block_info): Update.
16131 * gimple-streamer-in.c (input_bb): Update.
16132 * gimple-streamer-out.c (output_bb): Update.
16133 * graphite.c (print_global_statistics): Update.
16134 (print_graphite_scop_statistics): Update.
16135 * hsa-brig.c: Include basic-block.h.
16136 * hsa-dump.c: Include basic-block.h.
16137 * hsa-gen.c (T sum_slice): Update.
16138 (convert_switch_statements):Update.
16139 * hsa-regalloc.c: Include basic-block.h.
16140 * ipa-chkp.c (chkp_produce_thunks): Update.
16141 * ipa-cp.c (struct caller_statistics): Update.
16142 (init_caller_stats): Update.
16143 (gather_caller_stats): Update.
16144 (ipcp_cloning_candidate_p): Update.
16145 (good_cloning_opportunity_p): Update.
16146 (get_info_about_necessary_edges): Update.
16147 (dump_profile_updates): Update.
16148 (update_profiling_info): Update.
16149 (update_specialized_profile): Update.
16150 (perhaps_add_new_callers): Update.
16151 (decide_about_value): Update.
16152 (ipa_cp_c_finalize): Update.
16153 * ipa-devirt.c (struct odr_type_warn_count): Update.
16154 (struct decl_warn_count): Update.
16155 (struct final_warning_record): Update.
16156 (possible_polymorphic_call_targets): Update.
16157 (ipa_devirt): Update.
16158 * ipa-fnsummary.c (redirect_to_unreachable): Update.
16159 * ipa-icf.c (sem_function::merge): Update.
16160 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
16161 * ipa-inline.c (compute_uninlined_call_time): Update.
16162 (compute_inlined_call_time): Update.
16163 (want_inline_small_function_p): Update.
16164 (want_inline_self_recursive_call_p): Update.
16165 (edge_badness): Update.
16166 (lookup_recursive_calls): Update.
16167 (recursive_inlining): Update.
16168 (inline_small_functions): Update.
16169 (dump_overall_stats): Update.
16170 (dump_inline_stats): Update.
16171 * ipa-profile.c (ipa_profile_generate_summary): Update.
16172 (ipa_propagate_frequency): Update.
16173 (ipa_profile): Update.
16174 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
16175 * ipa-utils.c (ipa_merge_profiles): Update.
16176 * loop-doloop.c (doloop_modify): Update.
16177 * loop-unroll.c (report_unroll): Update.
16178 (unroll_loop_runtime_iterations): Update.
16179 * lto-cgraph.c (lto_output_edge): Update.
16180 (lto_output_node): Update.
16181 (input_node): Update.
16182 (input_edge): Update.
16183 (merge_profile_summaries): Update.
16184 * lto-streamer-in.c (input_cfg): Update.
16185 * lto-streamer-out.c (output_cfg): Update.
16186 * mcf.c (create_fixup_graph): Update.
16187 (adjust_cfg_counts): Update.
16188 (sum_edge_counts): Update.
16189 * modulo-sched.c (sms_schedule): Update.
16190 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
16191 * predict.c (maybe_hot_count_p): Update.
16192 (probably_never_executed): Update.
16193 (dump_prediction): Update.
16194 (combine_predictions_for_bb): Update.
16195 (propagate_freq): Update.
16196 (handle_missing_profiles): Update.
16197 (counts_to_freqs): Update.
16198 (rebuild_frequencies): Update.
16199 (force_edge_cold): Update.
16200 * predict.h: Include profile-count.h
16201 (maybe_hot_count_p, counts_to_freqs): UPdate.
16202 * print-rtl-function.c: Do not include cfg.h
16203 * print-rtl.c: Include basic-block.h
16204 * profile-count.c: New file.
16205 * profile-count.h: New file.
16206 * profile.c (is_edge_inconsistent): Update.
16207 (correct_negative_edge_counts): Update.
16208 (is_inconsistent): Update.
16209 (set_bb_counts): Update.
16210 (read_profile_edge_counts): Update.
16211 (compute_frequency_overlap): Update.
16212 (compute_branch_probabilities): Update; Initialize and deinitialize
16214 (branch_prob): Update.
16215 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
16216 (edge_gcov_count): New.
16217 (bb_gcov_count): New.
16218 * shrink-wrap.c (try_shrink_wrapping): Update.
16219 * tracer.c (better_p): Update.
16220 * trans-mem.c (expand_transaction): Update.
16221 (ipa_tm_insert_irr_call): Update.
16222 (ipa_tm_insert_gettmclone_call): Update.
16223 * tree-call-cdce.c: Update.
16224 * tree-cfg.c (gimple_duplicate_sese_region): Update.
16225 (gimple_duplicate_sese_tail): Update.
16226 (gimple_account_profile_record): Update.
16227 (execute_fixup_cfg): Update.
16228 * tree-inline.c (copy_bb): Update.
16229 (copy_edges_for_bb): Update.
16230 (initialize_cfun): Update.
16231 (freqs_to_counts): Update.
16232 (copy_cfg_body): Update.
16233 (expand_call_inline): Update.
16234 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
16235 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
16236 (try_unroll_loop_completely): Update.
16237 (try_peel_loop): Update.
16238 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
16239 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
16240 * tree-ssa-loop-split.c (connect_loops): Update.
16241 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
16242 * tree-ssa-reassoc.c (branch_fixup): Update.
16243 * tree-ssa-tail-merge.c (replace_block_by): Update.
16244 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
16245 (compute_path_counts): Update.
16246 (update_profile): Update.
16247 (recompute_probabilities): Update.
16248 (update_joiner_offpath_counts): Update.
16249 (estimated_freqs_path): Update.
16250 (freqs_to_counts_path): Update.
16251 (clear_counts_path): Update.
16252 (ssa_fix_duplicate_block_edges): Update.
16253 (duplicate_thread_path): Update.
16254 * tree-switch-conversion.c (case_bit_test_cmp): Update.
16255 (struct switch_conv_info): Update.
16256 * tree-tailcall.c (decrease_profile): Update.
16257 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
16258 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
16259 * value-prof.c (check_counter): Update.
16260 (gimple_divmod_fixed_value): Update.
16261 (gimple_mod_pow2): Update.
16262 (gimple_mod_subtract): Update.
16263 (gimple_ic_transform): Update.
16264 (gimple_stringop_fixed_value): Update.
16265 * value-prof.h (gimple_ic): Update.
16267 2017-06-02 Carl Love <cel@us.ibm.com>
16269 * config/rs6000/rs6000-c: Add support for built-in functions
16270 vector double vec_doublee (vector signed int);
16271 vector double vec_doublee (vector unsigned int);
16272 vector double vec_doublee (vector float);
16273 vector double vec_doubleh (vector signed int);
16274 vector double vec_doubleh (vector unsigned int);
16275 vector double vec_doubleh (vector float);
16276 vector double vec_doublel (vector signed int);
16277 vector double vec_doublel (vector unsigned int);
16278 vector double vec_doublel (vector float);
16279 vector double vec_doubleo (vector signed int);
16280 vector double vec_doubleo (vector unsigned int);
16281 vector double vec_doubleo (vector float);.
16282 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
16283 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
16285 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
16286 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
16287 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
16289 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
16290 vec_doublel, vec_doubleh.
16291 * doc/extend.texi: Update the built-in documentation file for the
16292 new built-in functions.
16294 2017-06-02 David Malcolm <dmalcolm@redhat.com>
16297 * ipa-inline-analysis.c (free_growth_caches): Set
16298 edge_removal_hook_holder to NULL after removing it.
16300 2017-06-02 Sudakshina Das <sudi.das@arm.com>
16302 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
16303 comparision with zero.
16305 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
16306 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16307 for early expansion of vec_min and vec_max builtins.
16308 (builtin_function_type): Add min/max unsigned variants to those
16309 identified as having unsigned arguments.
16311 2017-06-02 Olivier Hainque <hainque@adacore.com>
16313 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
16315 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16317 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
16318 Use VALL_F16 iterator rather than VALL.
16320 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16322 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
16323 Emit CBNZ inside loop when doing a strong exchange and comparing
16324 against zero. Generate the CC flags after the loop.
16326 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
16328 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
16329 (dl_section_ref): New.
16330 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
16331 On AIX, append an expression to subtract the size of the
16332 section length to dl_section_ref.
16334 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
16336 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16337 for early expansion of vector absolute builtins.
16339 2017-06-02 Richard Biener <rguenther@suse.de>
16341 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
16342 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
16344 2017-06-02 Richard Biener <rguenther@suse.de>
16346 PR tree-optimization/80948
16347 * tree-tailcall.c (find_tail_calls): Track stmts to move in
16348 stmt order as well.
16350 2017-06-02 Richard Biener <rguenther@suse.de>
16352 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
16354 * tree-vect-stmts.c (process_use): Do not mark backedge defs
16355 for inductions as relevant.
16357 2017-06-02 Richard Biener <rguenther@suse.de>
16359 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
16360 (vectorizable_induction): ... this. Remove dead code.
16362 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
16364 * builtins. (expand_builtin_alloca): Remove second parameter and
16365 infer its value from the first parameter instead.
16366 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
16368 2017-06-02 Jakub Jelinek <jakub@redhat.com>
16370 PR rtl-optimization/80903
16371 * loop-doloop.c (add_test): Unshare sequence.
16373 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
16375 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
16377 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
16379 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
16381 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
16382 xlogue_layout::get_instance, logue_layout::xlogue_layout,
16383 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
16384 (xlogue_layout::get_stub_rtx): Make static.
16385 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
16386 (xlogue_layout::compute_stub_managed_regs): Rename to...
16387 (xlogue_layout::count_stub_managed_regs): ...this.
16388 (xlogue_layout::is_stub_managed_reg): New function.
16389 (xlogue_layout::m_stub_names): Rename to...
16390 (xlogue_layout::s_stub_names): ...this, make static.
16391 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
16392 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
16393 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
16394 xlogue_layout::s_stub_names): Instantiate statics.
16395 (stub_managed_regs): Remove.
16396 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
16397 (disable_call_ms2sysv_xlogues): Rename to...
16398 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
16399 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
16401 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
16402 change after reload_completed.
16403 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
16405 (ix86_expand_prologue): Likewise.
16406 (ix86_expand_epilogue): Likewise.
16407 (ix86_expand_split_stack_prologue): Likewise.
16408 (ix86_compute_frame_layout): Remove frame parameter ...
16409 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
16410 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
16412 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
16413 (ix86_frame): Move from here ...
16414 * config/i386/i386.h (ix86_frame): ... to here.
16415 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
16416 complete ix86_frame data structure instead. Remove some_ld_name.
16418 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
16420 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
16421 symbols that hold a DECL_VALUE_EXPR.
16423 2017-06-01 Martin Jambor <mjambor@suse.cz>
16425 PR tree-optimization/80898
16426 * tree-sra.c (process_subtree_disqualification): Removed.
16427 (disqualify_candidate): Do not acll
16428 process_subtree_disqualification.
16429 (subtree_mark_written_and_enqueue): New function.
16430 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
16431 RHS has been disqualified and re-queue LHS if necessary. Apart
16432 from that, ignore disqualified RHS.
16434 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16436 * config/s390/s390.c (s390_emit_epilogue): Disable early return
16437 address fetch for z10 or later.
16439 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16441 * config/arc/arc.md (tst_movb): Add guard when splitting.
16443 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16445 * config/arc/arc.c (arc_can_eliminate): Test against
16446 arc_frame_pointer_needed.
16448 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16450 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
16451 to prevent store reordering.
16452 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
16453 (type): Add block type.
16454 (stack_tie): Define special instruction to be used in
16457 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16459 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
16460 constraint. It is not valid for the pattern.
16461 (noncommutative_binary_comparison): Likewise.
16463 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16465 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
16468 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16470 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
16471 be used by the reg-alloc.
16473 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16475 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
16476 reg-alloc when having mul64 or mul32x16 instructions.
16477 (mulsidi3): Likewise.
16478 (umulsidi3): Likewise.
16479 (mulsi32x16): New pattern.
16480 (mulsi64): Likewise.
16481 (mulsidi64): Likewise.
16482 (umulsidi64): Likewise.
16483 (MUL32x16_REG): Define.
16484 (mul64_600): Use MUL32x16_REG.
16485 (mac64_600): Likewise.
16486 (umul64_600): Likewise.
16487 (umac64_600): Likewise.
16489 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16491 * config/arc/arc.md (mulsi3_700): Make it commutative.
16493 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
16495 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
16497 (*sign_extendsidi2_insn): Likewise for movstosw.
16499 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
16501 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
16502 the type of the input discriminant value. Convert the
16503 discriminant value of signedness vary.
16505 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
16507 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
16508 Add to -Wall section.
16510 2017-06-01 Richard Biener <rguenther@suse.de>
16512 PR middle-end/66313
16513 * fold-const.c (fold_plusminus_mult_expr): If the factored
16514 factor may be zero use a wrapping type for the inner operation.
16515 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
16516 and handle moved defs.
16517 (process_assignment): Properly guard the unary op case. Return a
16518 tri-state indicating that moving the stmt before the call may allow
16519 to continue. Pass through to_move.
16520 (find_tail_calls): Handle moving unrelated defs before
16523 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
16526 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
16527 splitter result in the canonical way.
16529 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
16531 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
16532 also for 32bit target. Update insn attributes.
16533 (zero-extendsidi2 splitter): Allow all registers for operand 1.
16535 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
16537 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
16538 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
16539 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
16540 (_mm_maskz_min_ss): New intrinsics.
16542 2017-05-31 Martin Liska <mliska@suse.cz>
16544 * tree-vect-loop.c (vect_create_epilog_for_reduction):
16545 Change comment style to one we normally use.
16546 (vectorizable_reduction): Likewise.
16547 (vectorizable_induction): Likewise.
16548 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
16549 (vectorizable_call): Likewise.
16550 (vectorizable_simd_clone_call): Likewise.
16551 (vectorizable_conversion): Likewise.
16552 (vectorizable_assignment): Likewise.
16553 (vectorizable_shift): Likewise.
16554 (vectorizable_operation): Likewise.
16555 (vectorizable_store): Likewise.
16556 (vectorizable_load): Likewise.
16557 * tree-vectorizer.h: Likewise.
16559 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
16561 * passes.c (emergency_dump_function): New.
16562 * tree-pass.h (emergency_dump_function): Declare.
16563 * plugin.c (plugins_internal_error_function): Remove.
16564 * plugin.h (plugins_internal_error_function): Remove declaration.
16565 * toplev.c (internal_error_function): New static function. Use it...
16566 (general_init): ...here.
16568 2017-05-31 Graham Markall <graham.markall@embecosm.com>
16570 * config/arc/arc.c (arc_print_operand): Handle constant operands.
16571 (arc_rtx_costs): Add costs for new patterns.
16572 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
16573 * config/arc/predicates.md: Add _1_2_3_operand predicate.
16575 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
16577 * tree-ssa-strlen.c (get_next_strinfo): New function.
16578 (get_stridx_plus_constant): Use it.
16579 (zero_length_string): Likewise.
16580 (adjust_related_strinfos): Likewise.
16581 (adjust_last_stmt): Likewise.
16583 2017-05-31 Richard Biener <rguenther@suse.de>
16586 * config/i386/i386.c (ix86_expand_builtin): Remove assert
16587 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
16589 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
16591 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
16592 loop_vinfo argument and use of dependence distance vectors.
16593 Check instead whether the two references differ only in their
16594 initial value and assume that they have the same alignment if the
16595 difference is a multiple of the vector alignment.
16596 (vect_analyze_data_refs_alignment): Update call accordingly.
16598 2017-05-31 Martin Liska <mliska@suse.cz>
16601 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
16603 2017-05-31 Bin Cheng <bin.cheng@arm.com>
16605 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
16606 (create_intersect_range_checks): Move from ...
16607 * tree-data-ref.c (create_intersect_range_checks_index)
16608 (create_intersect_range_checks): ... to here.
16609 (create_runtime_alias_checks): New function factored from ...
16610 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
16611 here. Call above function.
16612 * tree-data-ref.h (create_runtime_alias_checks): New function.
16614 2017-05-31 Bin Cheng <bin.cheng@arm.com>
16616 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
16617 segment length for dr_b and compute it in wide_int.
16619 2017-05-31 Richard Biener <rguenther@suse.de>
16621 PR tree-optimization/80906
16622 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
16623 and pass through iv_map.
16624 (copy_bb_and_scalar_dependences): Adjust.
16625 (translate_pending_phi_nodes): Likewise.
16626 (copy_loop_close_phi_args): Handle code-generating IVs instead
16629 2017-05-30 David Malcolm <dmalcolm@redhat.com>
16631 * diagnostic-color.c (color_dict): Add "type-diff".
16632 (parse_gcc_colors): Update comment.
16633 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
16634 -fdiagnostics-show-template-tree and -fno-elide-type.
16635 (GCC_COLORS): Add type-diff to example.
16637 (-fdiagnostics-show-template-tree): New.
16638 (-fno-elide-type): New.
16639 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
16640 the pp_format_decoder callback. Call any m_format_postprocessor's
16642 (pretty_printer::pretty_printer): Initialize
16643 m_format_postprocessor.
16644 (pretty_printer::~pretty_printer): Delete any
16645 m_format_postprocessor.
16646 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
16647 (class format_postprocessor): New class.
16648 (struct pretty_printer::format_decoder): Document the new parameters.
16649 (struct pretty_printer::m_format_postprocessor): New field.
16650 * tree-diagnostic.c (default_tree_printer): Update for new
16651 bool and const char ** params.
16652 * tree-diagnostic.h (default_tree_printer): Likewise.
16654 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
16656 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
16657 (lwa_operand): Delete rs6000_gen_cell_microcode test.
16658 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
16659 rs6000_gen_cell_microcode code.
16660 (rs6000_final_prescan_insn): Delete.
16661 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
16662 "warn-cell-microcode" entries.
16663 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
16664 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
16665 throughout. Change cc_reg_not_micro_cr0_operand to
16666 cc_reg_not_cr0_operand throughout.
16667 (*extendhi<mode>2_noload): Delete.
16668 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
16669 (mwarn-cell-microcode): Delete.
16670 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
16671 -mgen-cell-microcode and -mwarn-cell-microcode.
16673 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
16676 * config/i386/constraints.md (Yd): New constraint.
16678 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
16679 and (?Yd, r) alternatives. Update insn attributes.
16680 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
16681 and (?*Yd, r) alternatives. Update insn attributes.
16682 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
16684 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
16686 * gimplify.c (gimplify_modify_expr): Don't create a
16687 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
16690 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
16692 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
16694 2017-05-30 Richard Biener <rguenther@suse.de>
16696 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
16697 and reduc_def fields.
16698 (STMT_VINFO_REDUC_TYPE): New define.
16699 (STMT_VINFO_REDUC_DEF): Likewise.
16700 (vect_force_simple_reduction): Adjust prototype.
16701 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
16702 (vect_is_simple_reduction): Remove check_reduction argument.
16703 (vect_force_simple_reduction): Adjust and set
16704 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
16705 (vectorizable_reduction): Do not re-do reduction analysis
16706 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
16707 * tree-parloops.c (gather_scalar_reductions): Adjust.
16709 2017-05-30 Richard Biener <rguenther@suse.de>
16711 PR middle-end/80901
16712 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
16715 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
16717 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
16718 Introduce unknown_misalignment parameter and remove vf.
16719 (vect_peeling_hash_get_lowest_cost):
16720 Pass unknown_misalignment parameter.
16721 (vect_enhance_data_refs_alignment):
16722 Fix unsupportable data ref treatment.
16724 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16726 * tree-vect-data-refs.c (vect_get_data_access_cost):
16727 Workaround for SLP handling.
16728 (vect_enhance_data_refs_alignment):
16729 Compute costs for doing no peeling at all, compare to the best
16730 peeling costs so far and avoid peeling if cheaper.
16732 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16734 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
16735 Return peeling info and set costs to zero for unlimited cost
16737 (vect_enhance_data_refs_alignment): Also inspect all datarefs
16738 with unknown misalignment. Compute and costs for unknown
16739 misalignment, compare them to the costs for known misalignment
16740 and choose the cheapest for peeling.
16742 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16744 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
16745 (vect_get_peeling_costs_all_drs): Create function.
16746 (vect_peeling_hash_get_lowest_cost):
16747 Use vect_get_peeling_costs_all_drs.
16748 (vect_peeling_supportable): Create function.
16749 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
16751 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16753 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
16754 DR_HAS_NEGATIVE_STEP.
16755 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
16756 (vect_enhance_data_refs_alignment): Use.
16757 (vect_duplicate_ssa_name_ptr_info): Use.
16758 * tree-vectorizer.h (dr_misalignment): Use.
16759 (known_alignment_for_access_p): Use.
16761 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16764 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
16766 (has_section_name): New function.
16768 2017-05-30 Martin Liska <mliska@suse.cz>
16771 * auto-profile.c (get_function_decl_from_block): Fix
16774 2017-05-30 Richard Biener <rguenther@suse.de>
16776 PR middle-end/80876
16777 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
16779 2017-05-30 Martin Liska <mliska@suse.cz>
16781 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
16782 * dumpfile.h (struct dump_file_info): Remove ctors.
16784 2017-05-30 Martin Liska <mliska@suse.cz>
16786 * predict.def: Fix GNU coding style.
16788 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
16790 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
16791 Mark 'to' argument with ATTRIBUTE_UNUSED.
16793 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
16795 * config/xtensa/xtensa.c (xtensa_emit_call): Use
16796 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
16797 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
16800 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
16802 * doc/install.texi (Options specification): Restore entry of
16803 --enable-sjlj-exceptions.
16805 2017-05-27 Michael Eager <eager@eagercon.com>
16808 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
16810 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
16812 * config/microblaze/microblaze.h
16813 (FIXED_REGISTERS): Update in macro.
16814 (CALL_USED_REGISTERS): Update in macro.
16816 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
16818 * doc/install.texi: Add links to macOS binary distributions.
16820 2017-05-27 Jakub Jelinek <jakub@redhat.com>
16824 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16826 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
16828 2017-05-26 Martin Liska <mliska@suse.cz>
16830 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
16832 2017-05-26 Martin Liska <mliska@suse.cz>
16834 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
16835 always leading ';; '.
16836 (dump_bb_info): Likewise.
16837 (brief_dump_cfg): Likewise.
16838 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
16839 * dumpfile.c: Remove usage of TDF_VERBOSE.
16840 * dumpfile.h (enum dump_kind): Likewise.
16841 (dump_gimple_bb_header): Do not use TDF_COMMENT.
16842 * print-tree.c (debug_verbose): Remove.
16843 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
16844 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
16845 * tree-diagnostic.c (default_tree_printer): Replace
16846 TDF_DIAGNOSTIC with TDF_SLIM.
16848 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16850 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
16851 in parameter loop, rather than loop_vinfo.
16852 (create_intersect_range_checks): Ditto.
16853 (vect_create_cond_for_alias_checks): Update call to above functions.
16855 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16857 PR tree-optimization/80815
16858 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
16859 for merging runtime alias checks. Handle negative DR_STEPs.
16861 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16863 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
16865 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
16866 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
16867 out code pruning runtime alias checks.
16868 * tree-data-ref.c (prune_runtime_alias_test_list): New function
16869 factored out from above.
16870 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
16872 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
16874 (prune_runtime_alias_test_list): New decalaration.
16876 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16878 * tree-vect-data-refs.c (compare_tree): Rename and move ...
16879 * tree-data-ref.c (data_ref_compare_tree): ... to here.
16880 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
16881 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
16882 (operator==, comp_dr_with_seg_len_pair): Ditto.
16883 (vect_prune_runtime_alias_test_list): Ditto.
16885 2017-05-26 Martin Liska <mliska@suse.cz>
16888 * params.def: Bound partial-inlining-entry-probability param.
16890 2017-05-26 Marek Polacek <polacek@redhat.com>
16893 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
16896 2017-05-26 Richard Biener <rguenther@suse.de>
16898 PR tree-optimization/80842
16899 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
16902 2017-05-26 Richard Biener <rguenther@suse.de>
16904 PR tree-optimization/80844
16905 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
16907 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
16909 * doc/md.texi (Machine Constraints): Update x86 family
16910 machine constraints section to match 'config/i386/constraints.md'.
16912 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
16914 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
16916 2017-05-25 Nathan Sidwell <nathan@acm.org>
16918 * doc/invoke.texi (--enable-languages): Update documentation.
16920 2017-05-25 Martin Liska <mliska@suse.cz>
16922 * dumpfile.c: Add TDF_FOLDING.
16923 * dumpfile.h (enum dump_kind): Likewise.
16924 * genmatch.c (dt_simplify::gen_1): Use it.
16926 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16928 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
16930 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16932 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
16933 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
16935 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16937 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
16938 * match.pd (X == C): Rewrite it here.
16939 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
16940 with_certain_nonzero_bits2): New predicates.
16941 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
16943 2017-05-24 Nathan Sidwell <nathan@acm.org>
16945 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
16948 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
16951 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
16953 * config/powerpcspe: New port. Files are copied from the rs6000
16954 port, with "rs6000" in filenames replaced by "powerpcspe".
16956 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
16958 PR rtl-optimization/80754
16959 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
16961 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
16963 * config/sparc/sparc.md (length): Return the correct value for -mflat
16964 sibcalls to match output_sibcall.
16966 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
16970 * config/rs6000/rs6000.c (struct machine_function): Add new field
16972 (rs6000_get_separate_components): Init that field, use it.
16973 (rs6000_components_for_bb): Use the field.
16975 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
16977 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
16979 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
16981 PR middle-end/80823
16982 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
16984 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16987 * config/s390/s390.c (s390_check_qrst_address): Check incoming
16988 address against address_operand predicate.
16989 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
16991 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
16993 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
16994 parameters passed indirectly.
16996 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
16998 * config/i386/i386.md (*movdi_internal): Remove SSE4
16999 alternative 18 (?r, *v). Update insn attributes.
17000 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
17001 Update insn attributes.
17002 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
17003 Update insn attributes.
17004 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
17005 alternative 1 (r, v). Remove isa attribute.
17006 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
17007 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
17008 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
17010 2017-05-23 Tom de Vries <tom@codesourcery.com>
17012 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
17015 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17017 * cgraphunit.c (symbol_table::process_new_functions): Update.
17018 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
17019 (inline_generate_summary): Rename to ...
17020 (ipa_fn_summary_generate): ... this one.
17021 (inline_read_summary): Rename to ...
17022 (ipa_fn_summary_read): ... this one.
17023 (inline_write_summary): Rename to ...
17024 (ipa_fn_summary_write): ... this one.
17025 (inline_free_summary): Rename to ...
17026 (ipa_free_fn_summary): ... this one.
17027 (pass_data_local_fn_summary, pass_local_fn_summary,
17028 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
17029 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
17030 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
17031 make_pass_ipa_fn_summary): New.
17032 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
17033 inline_write_summary, inline_free_summary): Remove.
17034 (ipa_free_fn_summary) : New.
17035 * ipa-inline.c (ipa_inline): Update.
17036 (pass_ipa_inline): Do not generate summaries.
17037 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
17039 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
17040 and add pass_ipa_fn_summary.
17041 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
17043 (make_pass_inline_parameters): Remove.
17045 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
17047 * omp-low.c (struct omp_context): Remove "default_kind" member.
17050 * omp-offload.c (execute_oacc_device_lower): Remove the
17051 parallelism dimensions function attributes for unparallelized
17052 OpenACC kernels constructs.
17054 2017-05-23 Martin Liska <mliska@suse.cz>
17056 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
17058 (cgraph_edge::make_speculative): Likewise.
17059 (cgraph_edge::resolve_speculation): Likewise.
17060 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
17061 (cgraph_node::dump): Likewise.
17062 * cgraph.h: Likewise.
17063 * cgraphunit.c (analyze_functions): Likewise.
17064 (symbol_table::compile): Likewise.
17065 * ipa-cp.c (print_all_lattices): Likewise.
17066 (determine_versionability): Likewise.
17067 (initialize_node_lattices): Likewise.
17068 (ipcp_verify_propagated_values): Likewise.
17069 (estimate_local_effects): Likewise.
17070 (update_profiling_info): Likewise.
17071 (create_specialized_node): Likewise.
17072 (perhaps_add_new_callers): Likewise.
17073 (decide_about_value): Likewise.
17074 (decide_whether_version_node): Likewise.
17075 (identify_dead_nodes): Likewise.
17076 (ipcp_store_bits_results): Likewise.
17077 * ipa-devirt.c (dump_targets): Likewise.
17078 (ipa_devirt): Likewise.
17079 * ipa-icf.c (sem_item::dump): Likewise.
17080 (sem_function::equals): Likewise.
17081 (sem_variable::equals): Likewise.
17082 (sem_item_optimizer::read_section): Likewise.
17083 (sem_item_optimizer::execute): Likewise.
17084 (congruence_class::dump): Likewise.
17085 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
17086 (dump_inline_summary): Likewise.
17087 (estimate_node_size_and_time): Likewise.
17088 (inline_analyze_function): Likewise.
17089 * ipa-inline-transform.c (inline_call): Likewise.
17090 * ipa-inline.c (report_inline_failed_reason): Likewise.
17091 (want_early_inline_function_p): Likewise.
17092 (edge_badness): Likewise.
17093 (update_edge_key): Likewise.
17094 (inline_small_functions): Likewise.
17095 * ipa-profile.c (ipa_profile): Likewise.
17096 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
17097 (ipa_make_edge_direct_to_target): Likewise.
17098 (remove_described_reference): Likewise.
17099 (ipa_impossible_devirt_target): Likewise.
17100 (propagate_controlled_uses): Likewise.
17101 (ipa_print_node_params): Likewise.
17102 (ipcp_transform_function): Likewise.
17103 * ipa-pure-const.c (pure_const_read_summary): Likewise.
17104 (propagate_pure_const): Likewise.
17105 * ipa-reference.c (generate_summary): Likewise.
17106 (read_write_all_from_decl): Likewise.
17107 (propagate): Likewise.
17108 (ipa_reference_read_optimization_summary): Likewise.
17109 * ipa-utils.c (ipa_merge_profiles): Likewise.
17110 * ipa.c (walk_polymorphic_call_targets): Likewise.
17111 (symbol_table::remove_unreachable_nodes): Likewise.
17112 (ipa_single_use): Likewise.
17113 * passes.c (execute_todo): Likewise.
17114 * predict.c (drop_profile): Likewise.
17115 * symtab.c (symtab_node::get_dump_name): New function.
17116 (symtab_node::dump_name): Likewise.
17117 (symtab_node::dump_asm_name): Likewise.
17118 (symtab_node::dump_references): Likewise.
17119 (symtab_node::dump_referring): Likewise.
17120 (symtab_node::dump_base): Likewise.
17121 (symtab_node::debug_symtab): Likewise.
17122 * tree-sra.c (convert_callers_for_node): Likewise.
17123 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
17124 * value-prof.c (init_node_map): Likewise.
17126 2017-05-23 Martin Liska <mliska@suse.cz>
17128 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
17129 and symtab_node::debug_symtab to symbol_table::debug.
17130 * cgraphunit.c (analyze_functions): Use the renamed function.
17131 (symbol_table::compile): Likewise.
17132 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
17133 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
17134 * passes.c (execute_todo): Likewise.
17135 * symtab.c (symbol_table::dump): New function.
17136 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
17138 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17140 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
17141 that nonconst implies exec.
17143 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17145 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
17146 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
17147 (inline_edge_summary_vec): Turn into ...
17148 (ipa_call_summaries): ... this one.
17149 (redirect_to_unreachable, edge_set_predicate,
17150 evaluate_properties_for_edge, inline_summary_alloc,
17151 reset_ipa_call_summary, reset_inline_summary,
17152 inline_summary_t::duplicate): Update.
17153 (inline_edge_duplication_hook): Turn to ...
17154 (ipa_call_summary_t::duplicate): ... this one.
17155 (inline_edge_removal_hook): Turn to ...
17156 (ipa_call_summary_t::remove): ... this one.
17157 (dump_inline_edge_summary): Turn to ...
17158 (dump_ipa_call_summary): ... this one.
17159 (estimate_function_body_sizes): Update.
17160 (inline_update_callee_summaries): Update.
17161 (remap_edge_change_prob): Update.
17162 (remap_edge_summaries): Update.
17163 (inline_merge_summary): Update.
17164 (do_estimate_edge_time): Update.
17165 (inline_generate_summary): Update.
17166 (inline_read_section): Update.
17167 (inline_read_summary): Update.
17168 (inline_free_summary): Update.
17169 * ipa-inline.c (can_inline_edge_p): Update.
17170 (compute_inlined_call_time): Update.
17171 (want_inline_small_function_p): Update.
17172 (edge_badness): Update.
17173 (early_inliner): Update.
17174 * ipa-inline.h (inline_edge_summary): Turn to ...
17175 (ipa_call_summary): ... this one.
17176 (ipa_call_summary_t): New class.
17177 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
17178 (ipa_call_summaries): New.
17179 (inline_edge_summary): Remove.
17180 (estimate_edge_growth): Update.
17181 * ipa-profile.c (ipa_propagate_frequency_1): Update.
17182 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
17183 * ipa-split.c (execute_split_functions): Update.
17184 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
17186 2017-05-23 Tom de Vries <tom@codesourcery.com>
17188 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
17189 attributes): Document rdrand effective target.
17191 2017-05-23 Tom de Vries <tom@codesourcery.com>
17193 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
17194 attributes): Sort alphabetically.
17196 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
17198 * config/avr/genmultilib.awk: Use gsub instead of gensub.
17200 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
17203 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
17204 V2DF/V2DI splat into two separate patterns, one that handles
17205 registers, and the other that only handles memory. Drop support
17206 for splatting from a GPR on ISA 2.07 and then splitting the
17207 splat into direct move and splat.
17208 (vsx_splat_<mode>_reg): Likewise.
17209 (vsx_splat_<mode>_mem): Likewise.
17211 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
17213 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
17215 2017-05-22 Jakub Jelinek <jakub@redhat.com>
17217 PR middle-end/80809
17218 * omp-low.c (finish_taskreg_remap): New function.
17219 (finish_taskreg_scan): If unit size of ctx->record_type
17220 is non-constant, unshare the size expression and replace
17221 decls in it with possible outer var refs.
17223 PR middle-end/80809
17224 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
17225 GOVD_SHARED rather than GOVD_PRIVATE with it.
17226 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
17227 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
17229 PR middle-end/80853
17230 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
17231 as last argument to build_outer_var_ref for pointer bases of array
17232 section reductions.
17234 2017-05-19 Martin Sebor <msebor@redhat.com>
17236 * print-tree.c (print_node): Print DECL_READ_P flag.
17238 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17240 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
17241 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
17242 * cgraph.c: Likewise.
17243 * cgraphunit.c: Likewise.
17244 * gengtype.c: Likewise.
17245 * ipa-cp.c: Likewise.
17246 * ipa-devirt.c: Likewise.
17247 * ipa-icf.c: Likewise.
17248 * ipa-predicate.c: Likewise.
17249 * ipa-profile.c: Likewise.
17250 * ipa-prop.c: Likewise.
17251 * ipa-split.c: Likewise.
17253 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
17254 edge_predicate_pool, dump_inline_hints,
17255 inline_summary::account_size_time, redirect_to_unreachable,
17256 edge_set_predicate, set_hint_predicate,
17257 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
17258 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
17259 inline_summary_t::remove, remap_hint_predicate_after_duplication,
17260 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
17261 ipa_call_summary_t::remove, initialize_growth_caches,
17262 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
17263 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
17264 mark_modified, unmodified_parm_1, unmodified_parm,
17265 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
17266 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
17267 compute_bb_predicates, will_be_nonconstant_expr_predicate,
17268 will_be_nonconstant_predicate, record_modified_bb_info,
17269 get_minimal_bb, record_modified, param_change_prob,
17270 phi_result_unknown_predicate, predicate_for_phi_result,
17271 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
17272 estimate_function_body_sizes, compute_inline_parameters,
17273 compute_inline_parameters_for_curren, pass_data_inline_parameters,
17274 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
17275 inline_update_callee_summaries, remap_edge_change_prob,
17276 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
17277 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
17278 inline_analyze_function, inline_summary_t::insert,
17279 inline_generate_summary, read_ipa_call_summary, inline_read_section,
17280 inline_read_summary, write_ipa_call_summary, inline_write_summary,
17281 inline_free_summary): Move to ipa-fnsummary.h
17282 (predicate_t): Remove.
17283 * ipa-fnsummary.c: New file.
17284 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
17285 (enum inline_hints_vals, inline_hints, agg_position_info,
17286 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
17287 inline_summaries, ipa_call_summary, ipa_call_summary_t,
17288 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
17289 dump_inline_summary, dump_inline_hints, inline_generate_summary,
17290 inline_read_summary, inline_write_summary, inline_free_summary,
17291 inline_analyze_function, initialize_inline_failed,
17292 inline_merge_summary, inline_update_overall_summary,
17293 compute_inline_parameters): Move to ipa-fnsummary.h
17294 * ipa-fnsummary.h: New file.
17295 * ipa-inline-transform.h: Include ipa-inline.h.
17296 * ipa-inline.c: LIkewise.
17298 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17300 * ipa-inline.c (edge_badness): Use inlined_time instead of
17301 inline_summaries->get.
17303 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17305 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
17307 2017-05-22 Nathan Sidwell <nathan@acm.org>
17309 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
17310 (fdump-lang): Document 'raw' option.
17311 * dumpfile.h (TDI_tu): Delete.
17312 * dumpfile.c (dump_files): Remove translation-unit.
17313 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
17315 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
17317 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
17318 command option from $(AWK) call.
17319 * config/avr/genmultilib.awk: Simplify and rewrite so that it
17320 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
17321 [FORMAT]: Remove handling of variable.
17322 * config/avr/t-multilib: Regenerate.
17324 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17326 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
17328 (dump_inline_summary): Do not print self_time.
17329 (estimate_function_body_sizes): Do not set self_time.
17330 (compute_inline_parameters): Likewise.
17331 (inline_read_section, inline_write_summary): Do not stream self_time.
17332 * ipa-inline.h (inline_summary): Drop self_time.
17334 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17336 * ipa-inline-analysis.c (account_size_time): Rename to ...
17337 (inline_summary::account_size_time): ... this one.
17338 (reset_ipa_call_summary): Turn to ...
17339 (ipa_call_summary::reset): ... this one.
17340 (reset_inline_summary): Turn to ...
17341 (inline_summary::reset): ... this one.
17342 (inline_summary_t::remove): Update.
17343 (inline_summary_t::duplicate): Update.
17344 (ipa_call_summary_t::remove): Update.
17345 (dump_inline_summary): Update.
17346 (estimate_function_body_sizes): Update.
17347 (compute_inline_parameters): Update.
17348 (estimate_node_size_and_time): Update.
17349 (inline_merge_summary): Update.
17350 (inline_update_overall_summary): Update.
17351 (inline_read_section): Update.
17352 (inline_write_summary): Update.
17353 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
17354 add account_size_time and reset member functions.
17355 (ipa_call_summary): Add reset function.
17356 * ipa-predicate.h (predicate::operator &): Constify.
17358 2017-05-22 Richard Biener <rguenther@suse.de>
17360 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
17362 2017-05-19 Jason Merrill <jason@redhat.com>
17364 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
17366 2017-05-19 Marek Polacek <polacek@redhat.com>
17369 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
17370 TYPE_OVERFLOW_WRAPS checks.
17372 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
17374 * tree-core.h (enum omp_clause_default_kind): Add
17375 "OMP_CLAUSE_DEFAULT_PRESENT".
17376 * tree-pretty-print.c (dump_omp_clause): Handle it.
17377 * gimplify.c (enum gimplify_omp_var_data): Add
17378 "GOVD_MAP_FORCE_PRESENT".
17379 (gimplify_adjust_omp_clauses_1): Map it to
17380 "GOMP_MAP_FORCE_PRESENT".
17381 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
17383 * gimplify.c (oacc_default_clause): Clarify.
17385 2017-05-19 Nathan Sidwell <nathan@acm.org>
17387 LANG_HOOK_REGISTER_DUMPS
17388 * toplev.c (general_init): Call register dump lang hook.
17389 * doc/invoke.texi: Document -fdump-lang option family.
17390 * dumpfile.c (dump_files): Remove class dump here.
17391 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
17392 * dumpfile.h (tree_dump_index): Remove TDI_class.
17393 * langhooks-def.h (lhd_register_dumps): Declare.
17394 (LANG_HOOKS_REGISTER_DUMPS): Define.
17395 (LANG_HOOKS_INITIALIZER): Add it.
17396 * langhooks.c (lhd_register_dumps): Define.
17397 * langhooks.h (struct lang_hooks): Add register_dumps.
17399 2017-05-19 Nathan Sidwell <nathan@acm.org>
17401 * context.h (context::set_passes): New.
17402 * context.c (context::context): Do not create pass manager.
17403 * toplev.c (general_init): Create pass manager here.
17405 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
17407 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
17408 use this splitter if two add or or instructions would also work for
17409 the constant we want to generate.
17411 2017-05-19 Richard Biener <rguenther@suse.de>
17414 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
17415 predicate evaluation.
17417 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
17419 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
17421 * ipa-inline.c (want_inline_small_function_p): Do not cast to
17424 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
17426 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
17427 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
17428 (inline_edge_summary_vec): Turn into ...
17429 (ipa_call_summaries): ... this one.
17430 (redirect_to_unreachable, edge_set_predicate,
17431 evaluate_properties_for_edge, inline_summary_alloc,
17432 reset_ipa_call_summary, reset_inline_summary,
17433 inline_summary_t::duplicate): Update.
17434 (inline_edge_duplication_hook): Turn to ...
17435 (ipa_call_summary_t::duplicate): ... this one.
17436 (inline_edge_removal_hook): Turn to ...
17437 (ipa_call_summary_t::remove): ... this one.
17438 (dump_inline_edge_summary): Turn to ...
17439 (dump_ipa_call_summary): ... this one.
17440 (estimate_function_body_sizes): Update.
17441 (inline_update_callee_summaries): Update.
17442 (remap_edge_change_prob): Update.
17443 (remap_edge_summaries): Update.
17444 (inline_merge_summary): Update.
17445 (do_estimate_edge_time): Update.
17446 (inline_generate_summary): Update.
17447 (inline_read_section): Update.
17448 (inline_read_summary): Update.
17449 (inline_free_summary): Update.
17450 * ipa-inline.c (can_inline_edge_p): Update.
17451 (compute_inlined_call_time): Update.
17452 (want_inline_small_function_p): Update.
17453 (edge_badness): Update.
17454 (early_inliner): Update.
17455 * ipa-inline.h (inline_edge_summary): Turn to ...
17456 (ipa_call_summary): ... this one.
17457 (ipa_call_summary_t): New class.
17458 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
17459 (ipa_call_summaries): New.
17460 (inline_edge_summary): Remove.
17461 (estimate_edge_growth): Update.
17462 * ipa-profile.c (ipa_propagate_frequency_1): Update.
17463 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
17464 * ipa-split.c (execute_split_functions): Update.
17465 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
17467 2017-05-19 Richard Biener <rguenther@suse.de>
17469 PR middle-end/80764
17470 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
17472 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
17474 * config/rs6000/rs6000.c (struct machine_function): Add field
17475 fpr_is_wrapped_separately.
17476 (rs6000_get_separate_components): Use 64 components. Handle the
17477 new FPR components.
17478 (rs6000_components_for_bb): Handle the FPR components.
17479 (rs6000_emit_prologue_components): Handle the FPR components.
17480 (rs6000_emit_epilogue_components): Handle the FPR components.
17481 (rs6000_set_handled_components): Handle the FPR components.
17482 (rs6000_emit_prologue): Don't output prologue code for those FPRs
17483 that are already separately shrink-wrapped.
17484 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
17485 that are already separately shrink-wrapped.
17487 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17490 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
17493 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
17494 (define_peephole2 for Altivec d-form load): Add peepholes to catch
17495 cases where the register allocator uses a move and an offsettable
17496 memory operation to/from a FPR register on ISA 2.06/2.07.
17497 (define_peephole2 for Altivec d-form store): Likewise.
17499 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
17502 * config/i386/mmx.md (*mov<mode>_internal): Enable
17503 alternatives 11, 12, 13 and 14 also for 32bit targets.
17504 Remove alternatives 15, 16, 17 and 18.
17505 * config/i386/sse.md (vec_concatv2di): Change
17506 alternative (!x, *y) to (x, ?!*Yn).
17508 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
17510 * dumpfile.h (enum dump_kind): Remove stray comma.
17512 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
17514 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
17515 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
17516 predicate::num_conditions
17517 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
17518 (CHANGED): turn into predicate::changed.
17519 (agg_position_info): Move to ipa-predicate.h
17520 (add_condition, predicate::add_clause, predicate::operator &=,
17521 predicate::or_with, predicate::evaluate, predicate::probability,
17522 dump_condition, dump_clause, predicate::dump,
17523 predicate::remap_after_duplication, predicate::remap_after_inlining,
17524 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
17525 (evaluate_conditions_for_known_args): Update.
17526 (set_cond_stmt_execution_predicate): Update.
17527 * ipa-inline.h: Include ipa-predicate.h
17528 (condition, inline_param_summary, conditions, agg_position_info,
17529 predicate): Move to ipa-predicate.h
17530 * ipa-predicate.c: New file.
17531 * ipa-predicate.h: New file.
17533 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
17535 * final.c (leaf_function_p): Check we are not in a sequence.
17537 2017-05-18 Martin Liska <mliska@suse.cz>
17539 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
17540 * dumpfile.c (dump_register): Use new enum dump_kind.
17541 (get_dump_file_name): Likewise.
17542 (dump_enable_all): Likewise.
17543 (dump_switch_p_1): Likewise.
17544 (enable_rtl_dump_file): Remove usage of TDF_RTL.
17545 * dumpfile.h (enum dump_kind): New enum type.
17546 (struct dump_file_info): Create constructor and
17547 format fields and comments.
17548 * passes.c (pass_manager::register_one_dump_file):
17550 * statistics.c (statistics_early_init): Likewise.
17551 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
17552 TDF_TREE with TDF_SLIM.
17553 (gather_memory_references_ref): Likewise.
17555 2017-05-18 Martin Liska <mliska@suse.cz>
17557 * vec.h (struct vnull): Use it.
17559 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
17561 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
17562 (true_predicate, false_predicate, true_predicate_p,
17563 false_predicate_p): Remove.
17564 (single_cond_predicate, not_inlined_predicate): Turn to member function
17566 (add_condition): Update.
17567 (add_clause): Turn to...
17568 (predicate::add_clause): ... this one; update; allow passing NULL
17570 (and_predicates): Turn to ...
17571 (predicate::operator &=): ... this one.
17572 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
17573 (or_predicates): Turn to ...
17574 (predicate::or_with): ... this one.
17575 (evaluate_predicate): Turn to ...
17576 (predicate::evaluate): ... this one.
17577 (predicate_probability): Turn to ...
17578 (predicate::probability): ... this one.
17579 (dump_condition): Update.
17580 (dump_predicate): Turn to ...
17581 (predicate::dump): ... this one.
17582 (account_size_time): Update.
17583 (edge_set_predicate): Update.
17584 (set_hint_predicate): UPdate.
17585 (evaluate_conditions_for_known_args): Update.
17586 (evaluate_properties_for_edge): Update.
17587 (remap_predicate_after_duplication): Turn to...
17588 (predicate::remap_after_duplication): ... this one.
17589 (remap_hint_predicate_after_duplication): Update.
17590 (inline_summary_t::duplicate): UPdate.
17591 (dump_inline_edge_summary): Update.
17592 (dump_inline_summary): Update.
17593 (set_cond_stmt_execution_predicate): Update.
17594 (set_switch_stmt_execution_predicate): Update.
17595 (compute_bb_predicates): Update.
17596 (will_be_nonconstant_expr_predicate): Update.
17597 (will_be_nonconstant_predicate): Update.
17598 (phi_result_unknown_predicate): Update.
17599 (predicate_for_phi_result): Update.
17600 (array_index_predicate): Update.
17601 (estimate_function_body_sizes): Update.
17602 (estimate_node_size_and_time): Update.
17603 (estimate_ipcp_clone_size_and_time): Update.
17604 (remap_predicate): Rename to ...
17605 (predicate::remap_after_inlining): ... this one.
17606 (remap_hint_predicate): Update.
17607 (inline_merge_summary): Update.
17608 (inline_update_overall_summary): Update.
17609 (estimate_size_after_inlining): Update.
17610 (read_predicate): Rename to ...
17611 (predicate::stream_in): ... this one.
17612 (read_inline_edge_summary): Update.
17613 (write_predicate): Rename to ...
17614 (predicate::stream_out): ... this one.
17615 (write_inline_edge_summary): Update.
17616 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
17617 (clause_t): Turn to uint32_t
17618 (predicate): Turn to class; implement constructor and operators
17620 (size_time_entry): Update.
17621 (inline_summary): Update.
17622 (inline_edge_summary): Update.
17624 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
17626 * fold-const.c (fold_binary_loc): Move transformation...
17627 * match.pd (C - X CMP X): ... here.
17629 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
17631 * config/sparc/sparc.c (sparc_option_override): Set function
17632 alignment for -mcpu=niagara7 to 64 to match the I$ line.
17633 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
17635 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
17637 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
17639 2017-05-18 Marek Polacek <polacek@redhat.com>
17642 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
17643 (pass_ubsan::execute): Call gimple_assign_single_p instead of
17644 gimple_assign_load_p.
17646 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
17648 PR middle-end/80692
17649 * real.c (do_compare): Give decimal_do_compare preference over
17650 comparing just the signs.
17652 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
17654 * doc/md.texi (Canonicalization of Instructions): Describe the
17655 canonical form of instructions that inherently set a condition
17658 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
17660 PR middle-end/80775
17661 * tree-cfg.c: Move deletion of unreachable case statements to after
17662 the merging of consecutive case labels.
17664 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
17666 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
17667 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
17668 restoring of callee-saved registers.
17670 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
17672 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
17673 * config/visium/visium.c (single_set_and_flags): Likewise.
17674 * config/visium/visium.md (Substitutions): Likewise.
17676 2017-05-17 Martin Liska <mliska@suse.cz>
17678 * cfg.c: Introduce dump_flags_t type and
17679 use it instead of int type.
17681 * cfghooks.c: Likewise.
17682 * cfghooks.h (struct cfg_hooks): Likewise.
17683 * cfgrtl.c: Likewise.
17684 * cfgrtl.h: Likewise.
17685 * cgraph.c (cgraph_node::get_body): Likewise.
17686 * coretypes.h: Likewise.
17687 * domwalk.c: Likewise.
17688 * domwalk.h: Likewise.
17689 * dumpfile.c (struct dump_option_value_info): Likewise.
17690 (dump_enable_all): Likewise.
17691 (dump_switch_p_1): Likewise.
17692 (opt_info_switch_p): Likewise.
17693 * dumpfile.h (enum tree_dump_index): Likewise.
17694 (struct dump_file_info): Likewise.
17695 * genemit.c: Likewise.
17696 * generic-match-head.c: Likewise.
17697 * gengtype.c (open_base_files): Likewise.
17698 * gimple-pretty-print.c: Likewise.
17699 * gimple-pretty-print.h: Likewise.
17700 * graph.c (print_graph_cfg): Likewise.
17701 * graphite-scop-detection.c (dot_all_sese): Likewise.
17702 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
17703 * loop-unroll.c (report_unroll): Likewise.
17704 * passes.c (pass_manager::register_one_dump_file): Likewise.
17705 * print-tree.c: Likewise.
17706 * statistics.c: Likewise.
17707 * tree-cfg.c: Likewise.
17708 * tree-cfg.h: Likewise.
17709 * tree-dfa.c: Likewise.
17710 * tree-dfa.h: Likewise.
17711 * tree-dump.c (dump_function): Likewise.
17712 * tree-dump.h (struct dump_info): Likewise.
17713 * tree-pretty-print.c: Likewise.
17714 * tree-pretty-print.h: Likewise.
17715 * tree-ssa-live.c: Likewise.
17716 * tree-ssa-live.h: Likewise.
17717 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
17718 * tree-vect-loop.c: Likewise.
17719 * tree-vect-slp.c: Likewise.
17721 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
17722 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17724 PR tree-optimization/80457
17725 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
17726 of all arguments to a statement as scalar_to_vec operations.
17727 (vectorizable_call): Adjust call to vect_model_simple_cost for
17729 (vectorizable_conversion): Likewise.
17730 (vectorizable_assignment): Likewise.
17731 (vectorizable_shift): Likewise.
17732 (vectorizable_operation): Likewise.
17733 (vectorizable_comparison): Likewise.
17734 (vect_is_simple_cond): Record the def types for operands.
17735 (vectorizable_condition): Likewise, call vect_model_simple_cost.
17736 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
17737 for statement argument count.
17739 2017-05-16 Carl Love <cel@us.ibm.com>
17741 * config/rs6000/rs6000-c: Add support for built-in functions
17742 vector unsigned long long vec_bperm (vector unsigned long long,
17743 vector unsigned char)
17744 vector signed long long vec_mule (vector signed int,
17746 vector unsigned long long vec_mule (vector unsigned int,
17747 vector unsigned int)
17748 vector signed long long vec_mulo (vector signed int,
17750 vector unsigned long long vec_mulo (vector unsigned int,
17751 vector unsigned int)
17752 vector signed char vec_sldw (vector signed char,
17753 vector signed char,
17755 vector unsigned char vec_sldw (vector unsigned char,
17756 vector unsigned char,
17758 vector signed short vec_sldw (vector signed short,
17759 vector signed short,
17761 vector unsigned short vec_sldw (vector unsigned short,
17762 vector unsigned short,
17764 vector signed int vec_sldw (vector signed int,
17767 vector unsigned int vec_sldw (vector unsigned int,
17768 vector unsigned int,
17770 vector signed long long vec_sldw (vector signed long long,
17771 vector signed long long,
17773 vector unsigned long long vec_sldw (vector unsigned long long,
17774 vector unsigned long long,
17776 * config/rs6000/rs6000-c: Add support for built-in functions
17777 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
17778 * config/rs6000/altivec.h: Add defintion for vec_sldw.
17779 * doc/extend.texi: Update the built-in documentation for the
17780 new built-in functions.
17782 2017-05-16 Marek Polacek <polacek@redhat.com>
17786 * tree.c (save_expr): Don't fold the expression.
17788 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
17790 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
17791 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
17792 and (?*y,m). Update insn attributes.
17794 2017-05-16 Martin Liska <mliska@suse.cz>
17796 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
17797 flags argument of print_gimple_stmt, print_gimple_expr,
17798 print_generic_stmt and print_generic_expr.
17799 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
17800 * coretypes.h: Likewise.
17801 * except.c (dump_eh_tree): Likewise.
17802 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
17803 * gimple-pretty-print.h: Likewise.
17804 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
17805 (backprop::push_to_worklist): Likewise.
17806 (backprop::pop_from_worklist): Likewise.
17807 (backprop::process_use): Likewise.
17808 (backprop::intersect_uses): Likewise.
17809 (note_replacement): Likewise.
17810 * gimple-ssa-store-merging.c
17811 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
17812 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
17813 (pass_store_merging::execute): Likewise.
17814 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
17815 (ssa_base_cand_dump_callback): Likewise.
17816 (dump_incr_vec): Likewise.
17817 (replace_refs): Likewise.
17818 (replace_mult_candidate): Likewise.
17819 (create_add_on_incoming_edge): Likewise.
17820 (create_phi_basis): Likewise.
17821 (insert_initializers): Likewise.
17822 (all_phi_incrs_profitable): Likewise.
17823 (introduce_cast_before_cand): Likewise.
17824 (replace_one_candidate): Likewise.
17825 * gimplify.c (gimplify_expr): Likewise.
17826 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
17827 (set_rename): Likewise.
17828 (rename_uses): Likewise.
17829 (copy_loop_phi_nodes): Likewise.
17830 (add_close_phis_to_merge_points): Likewise.
17831 (copy_loop_close_phi_args): Likewise.
17832 (copy_cond_phi_args): Likewise.
17833 (graphite_copy_stmts_from_block): Likewise.
17834 (translate_pending_phi_nodes): Likewise.
17835 * graphite-poly.c (print_pdr): Likewise.
17836 (dump_gbb_cases): Likewise.
17837 (dump_gbb_conditions): Likewise.
17838 (print_scop_params): Likewise.
17839 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
17840 (build_cross_bb_scalars_use): Likewise.
17841 (gather_bbs::before_dom_children): Likewise.
17842 * hsa-dump.c (dump_hsa_immed): Likewise.
17843 * ipa-cp.c (print_ipcp_constant_value): Likewise.
17844 (get_replacement_map): Likewise.
17845 * ipa-inline-analysis.c (dump_condition): Likewise.
17846 (estimate_function_body_sizes): Likewise.
17847 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
17848 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
17849 * ipa-prop.c (ipa_dump_param): Likewise.
17850 (ipa_print_node_jump_functions_for_edge): Likewise.
17851 (ipa_modify_call_arguments): Likewise.
17852 (ipa_modify_expr): Likewise.
17853 (ipa_dump_param_adjustments): Likewise.
17854 (ipa_dump_agg_replacement_values): Likewise.
17855 (ipcp_modif_dom_walker::before_dom_children): Likewise.
17856 * ipa-pure-const.c (check_stmt): Likewise.
17857 (pass_nothrow::execute): Likewise.
17858 * ipa-split.c (execute_split_functions): Likewise.
17859 * omp-offload.c (dump_oacc_loop_part): Likewise.
17860 (dump_oacc_loop): Likewise.
17861 * trans-mem.c (tm_log_emit): Likewise.
17862 (tm_memopt_accumulate_memops): Likewise.
17863 (dump_tm_memopt_set): Likewise.
17864 (dump_tm_memopt_transform): Likewise.
17865 * tree-cfg.c (gimple_verify_flow_info): Likewise.
17866 (print_loop): Likewise.
17867 * tree-chkp-opt.c (chkp_print_addr): Likewise.
17868 (chkp_gather_checks_info): Likewise.
17869 (chkp_get_check_result): Likewise.
17870 (chkp_remove_check_if_pass): Likewise.
17871 (chkp_use_outer_bounds_if_possible): Likewise.
17872 (chkp_reduce_bounds_lifetime): Likewise.
17873 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
17874 (chkp_mark_completed_bounds): Likewise.
17875 (chkp_register_incomplete_bounds): Likewise.
17876 (chkp_mark_invalid_bounds): Likewise.
17877 (chkp_maybe_copy_and_register_bounds): Likewise.
17878 (chkp_build_returned_bound): Likewise.
17879 (chkp_get_bound_for_parm): Likewise.
17880 (chkp_build_bndldx): Likewise.
17881 (chkp_get_bounds_by_definition): Likewise.
17882 (chkp_generate_extern_var_bounds): Likewise.
17883 (chkp_get_bounds_for_decl_addr): Likewise.
17884 * tree-chrec.c (chrec_apply): Likewise.
17885 * tree-data-ref.c (dump_data_reference): Likewise.
17886 (dump_subscript): Likewise.
17887 (dump_data_dependence_relation): Likewise.
17888 (analyze_overlapping_iterations): Likewise.
17889 * tree-inline.c (expand_call_inline): Likewise.
17890 (tree_function_versioning): Likewise.
17891 * tree-into-ssa.c (dump_defs_stack): Likewise.
17892 (dump_currdefs): Likewise.
17893 (dump_names_replaced_by): Likewise.
17894 (dump_update_ssa): Likewise.
17895 (update_ssa): Likewise.
17896 * tree-object-size.c (pass_object_sizes::execute): Likewise.
17897 * tree-parloops.c (build_new_reduction): Likewise.
17898 (try_create_reduction_list): Likewise.
17899 (ref_conflicts_with_region): Likewise.
17900 (oacc_entry_exit_ok_1): Likewise.
17901 (oacc_entry_exit_single_gang): Likewise.
17902 * tree-pretty-print.h: Likewise.
17903 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
17904 (get_scalar_evolution): Likewise.
17905 (add_to_evolution): Likewise.
17906 (get_loop_exit_condition): Likewise.
17907 (analyze_evolution_in_loop): Likewise.
17908 (analyze_initial_condition): Likewise.
17909 (analyze_scalar_evolution): Likewise.
17910 (instantiate_scev): Likewise.
17911 (number_of_latch_executions): Likewise.
17912 (gather_chrec_stats): Likewise.
17913 (final_value_replacement_loop): Likewise.
17914 (scev_const_prop): Likewise.
17915 * tree-sra.c (dump_access): Likewise.
17916 (disqualify_candidate): Likewise.
17917 (create_access): Likewise.
17918 (reject): Likewise.
17919 (maybe_add_sra_candidate): Likewise.
17920 (create_access_replacement): Likewise.
17921 (analyze_access_subtree): Likewise.
17922 (analyze_all_variable_accesses): Likewise.
17923 (sra_modify_assign): Likewise.
17924 (initialize_constant_pool_replacements): Likewise.
17925 (find_param_candidates): Likewise.
17926 (decide_one_param_reduction): Likewise.
17927 (replace_removed_params_ssa_names): Likewise.
17928 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
17929 * tree-ssa-copy.c (dump_copy_of): Likewise.
17930 (copy_prop_visit_cond_stmt): Likewise.
17931 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
17932 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
17933 (record_equivalences_from_stmt): Likewise.
17934 * tree-ssa-dse.c (compute_trims): Likewise.
17935 (delete_dead_call): Likewise.
17936 (delete_dead_assignment): Likewise.
17937 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
17938 (forward_propagate_into_cond): Likewise.
17939 (pass_forwprop::execute): Likewise.
17940 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
17941 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
17943 (move_computations_worker): Likewise.
17944 (execute_sm): Likewise.
17945 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
17946 (remove_exits_and_undefined_stmts): Likewise.
17947 (remove_redundant_iv_tests): Likewise.
17948 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
17949 (adjust_iv_update_pos): Likewise.
17950 * tree-ssa-math-opts.c (bswap_replace): Likewise.
17951 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
17952 (value_replacement): Likewise.
17953 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
17954 * tree-ssa-pre.c (print_pre_expr): Likewise.
17955 (get_representative_for): Likewise.
17956 (create_expression_by_pieces): Likewise.
17957 (insert_into_preds_of_block): Likewise.
17958 (eliminate_insert): Likewise.
17959 (eliminate_dom_walker::before_dom_children): Likewise.
17960 (eliminate): Likewise.
17961 (remove_dead_inserted_code): Likewise.
17962 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
17963 * tree-ssa-reassoc.c (get_rank): Likewise.
17964 (eliminate_duplicate_pair): Likewise.
17965 (eliminate_plus_minus_pair): Likewise.
17966 (eliminate_not_pairs): Likewise.
17967 (undistribute_ops_list): Likewise.
17968 (eliminate_redundant_comparison): Likewise.
17969 (update_range_test): Likewise.
17970 (optimize_range_tests_var_bound): Likewise.
17971 (optimize_vec_cond_expr): Likewise.
17972 (rewrite_expr_tree): Likewise.
17973 (rewrite_expr_tree_parallel): Likewise.
17974 (linearize_expr): Likewise.
17975 (break_up_subtract): Likewise.
17976 (linearize_expr_tree): Likewise.
17977 (attempt_builtin_powi): Likewise.
17978 (attempt_builtin_copysign): Likewise.
17979 (transform_stmt_to_copy): Likewise.
17980 (transform_stmt_to_multiply): Likewise.
17981 (dump_ops_vector): Likewise.
17982 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
17983 (print_scc): Likewise.
17984 (set_ssa_val_to): Likewise.
17985 (visit_reference_op_store): Likewise.
17986 (visit_use): Likewise.
17987 (sccvn_dom_walker::before_dom_children): Likewise.
17988 (run_scc_vn): Likewise.
17989 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
17991 (expr_hash_elt::print): Likewise.
17992 (const_and_copies::pop_to_marker): Likewise.
17993 (const_and_copies::record_const_or_copy_raw): Likewise.
17994 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
17995 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
17996 (dump_predicates): Likewise.
17997 (find_uninit_use): Likewise.
17998 (warn_uninitialized_phi): Likewise.
17999 (pass_late_warn_uninitialized::execute): Likewise.
18000 * tree-ssa.c (verify_vssa): Likewise.
18001 (verify_ssa): Likewise.
18002 (maybe_optimize_var): Likewise.
18003 * tree-vrp.c (dump_value_range): Likewise.
18004 (dump_all_value_ranges): Likewise.
18005 (dump_asserts_for): Likewise.
18006 (register_edge_assert_for_2): Likewise.
18007 (vrp_visit_cond_stmt): Likewise.
18008 (vrp_visit_switch_stmt): Likewise.
18009 (vrp_visit_stmt): Likewise.
18010 (vrp_visit_phi_node): Likewise.
18011 (simplify_cond_using_ranges_1): Likewise.
18012 (fold_predicate_in): Likewise.
18013 (evrp_dom_walker::before_dom_children): Likewise.
18014 (evrp_dom_walker::push_value_range): Likewise.
18015 (evrp_dom_walker::pop_value_range): Likewise.
18016 (execute_early_vrp): Likewise.
18018 2017-05-16 Richard Biener <rguenther@suse.de>
18020 * dwarf2out.c (loc_list_from_tree_1): Do not create
18021 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
18023 2017-05-16 Richard Biener <rguenther@suse.de>
18025 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
18027 (note_variable_value_in_expr): If we resolved the decl ref
18028 do not push to the stack.
18030 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
18032 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
18033 operations in fast-math mode.
18034 (vaddq_f16): Likewise.
18035 (vmul_f16): Likewise.
18036 (vmulq_f16): Likewise.
18037 (vsub_f16): Likewise.
18038 (vsubq_f16): Likewise.
18039 * config/arm/neon.md (add<mode>3): New.
18041 (fma:<VH:mode>3): New. Also remove outdated comment.
18044 2017-05-16 Martin Liska <mliska@suse.cz>
18048 * ipa-devirt.c (warn_types_mismatch): Fix typo.
18049 (odr_types_equivalent_p): Likewise.
18051 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
18053 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
18055 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
18058 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
18059 non-interunit SSE move alternatives with '?'.
18060 (zero-extendsidi peephole2): New peephole to skip intermediate
18061 general register in SSE zero-extend sequence.
18063 2017-05-15 Jeff Law <law@redhat.com>
18065 * reorg.c (relax_delay_slots): Create a new variable to hold
18066 the temporary target rather than clobbering TARGET_LABEL.
18068 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
18069 missing argument to extract_bit_field call.
18070 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
18072 2017-05-15 Martin Liska <mliska@suse.cz>
18075 * gcc.c (process_command): Do not allow empty argument of -o option.
18077 2017-05-15 Renlin Li <renlin.li@arm.com>
18079 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
18080 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
18081 * config/aarch64/constraints.md (Usf): Add long call check.
18082 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
18083 (call_value): Likewise.
18084 (sibcall): Likewise.
18085 (sibcall_value): Likewise.
18087 (call_value_insn): New.
18088 (sibcall_insn): Update rtx pattern.
18089 (sibcall_value_insn): Likewise.
18090 (call_internal): Remove.
18091 (call_value_internal): Likewise.
18092 (sibcall_internal): Likewise.
18093 (sibcall_value_internal): Likewise.
18094 (call_reg): Likewise.
18095 (call_symbol): Likewise.
18096 (call_value_reg): Likewise.
18097 (call_value_symbol): Likewise.
18099 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
18102 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
18104 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
18106 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
18107 compatible with CCGOCmode and with CCZmode.
18109 2017-05-14 Martin Sebor <msebor@redhat.com>
18111 PR middle-end/77671
18112 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
18113 (gimple_fold_builtin_snprintf): Same.
18114 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
18115 (gimple_fold_builtin_snprintf): Same.
18116 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
18117 of character types.
18118 (is_call_safe): New function.
18119 (try_substitute_return_value): Call it.
18120 (try_simplify_call): New function.
18121 (pass_sprintf_length::handle_gimple_call): Call it.
18123 2017-05-14 Martin Sebor <msebor@redhat.com>
18125 PR middle-end/80669
18126 * builtins.c (expand_builtin_stpncpy): Simplify.
18128 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
18130 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
18131 * config/i386/i386.h
18132 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
18133 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
18134 (struct machine_function): Add new members call_ms2sysv,
18135 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
18136 (struct machine_frame_state): New fields sp_realigned and
18137 sp_realigned_offset.
18138 * config/i386/i386.c
18139 (enum xlogue_stub): New enum.
18140 (enum xlogue_stub_sets): New enum.
18141 (class xlogue_layout): New class.
18142 (struct ix86_frame): New fields stack_realign_allocate_offset,
18143 stack_realign_offset and outlined_save_offset. Modify comments to
18144 detail stack layout when using out-of-line stubs.
18145 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
18146 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
18147 -mcall-ms2sysv-xlogues.
18148 (stub_managed_regs): New static variable.
18149 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
18150 registers managed by out-of-line stub.
18151 (disable_call_ms2sysv_xlogues): New function.
18152 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
18153 m->call_ms2sysv when appropriate and compute frame layout for
18155 (sp_valid_at, fp_valid_at): New inline functions.
18156 (choose_basereg): New function.
18157 (choose_baseaddr): Add align parameter, use choose_basereg and modify
18159 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
18160 Use align parameter of choose_baseaddr to generated aligned SSE movs
18162 (pro_epilogue_adjust_stack): Modify to track
18163 machine_frame_state::sp_realigned.
18164 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
18165 (ix86_nsaved_sseregs): Likewise.
18166 (ix86_emit_save_regs): Likewise.
18167 (ix86_emit_save_regs_using_mov): Likewise.
18168 (ix86_emit_save_sse_regs_using_mov): Likewise.
18169 (get_scratch_register_on_entry): Likewise.
18170 (gen_frame_set): New function.
18171 (gen_frame_load): Likewise.
18172 (gen_frame_store): Likewise.
18173 (emit_outlined_ms2sysv_save): Likewise.
18174 (emit_outlined_ms2sysv_restore): Likewise.
18175 (ix86_expand_prologue): Modify stack re-alignment code and call
18176 emit_outlined_ms2sysv_save when appropriate.
18177 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
18178 parameter rtx_insn *insn, which allows the function to be used to only
18179 generate the notes.
18180 (ix86_expand_epilogue): Modify validity checks of frame and stack
18181 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
18182 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
18183 * config/i386/predicates.md
18184 (save_multiple): New predicate.
18185 (restore_multiple): Likewise.
18186 * config/i386/sse.md
18187 (save_multiple<mode>): New pattern.
18188 (save_multiple_realign<mode>): Likewise.
18189 (restore_multiple<mode>): Likewise.
18190 (restore_multiple_and_return<mode>): Likewise.
18191 (restore_multiple_leave_return<mode>): Likewise.
18192 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
18194 2017-05-14 Julia Koval <julia.koval@intel.com>
18196 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
18197 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
18198 (__builtin_ia32_xsetbv): New builtins.
18199 * config/i386/i386.c (ix86_expand_special_args_builtin):
18201 (ix86_expand_builtin): Special expand for new intrinsics.
18202 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
18203 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
18204 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
18206 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18208 * cfganal.c (inverted_post_order_compute): Change argument type
18210 * cfganal.h (inverted_post_order_compute): Adjust prototype.
18211 * df-core.c (rest_of_handle_df_initialize): Adjust.
18212 (rest_of_handle_df_finish): Likewise.
18213 (df_analyze_1): Likewise.
18214 (df_analyze): Likewise.
18215 (loop_inverted_post_order_compute): Change argument to be a vec *.
18216 (df_analyze_loop): Adjust.
18217 (df_get_n_blocks): Likewise.
18218 (df_get_postorder): Likewise.
18219 * df.h (struct df_d): Change field to be a vec.
18220 * lcm.c (compute_laterin): Adjust.
18221 (compute_available): Likewise.
18222 * lra-lives.c (lra_create_live_ranges_1): Likewise.
18223 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
18224 * tree-ssa-pre.c (compute_antic): Likewise.
18226 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18228 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
18229 (depth_first_search::depth_first_search): Change structure init
18230 function to this constructor.
18231 (depth_first_search::add_bb): Rename function to this member.
18232 (depth_first_search::execute): Likewise.
18233 (flow_dfs_compute_reverse_finish): Adjust.
18235 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18237 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
18238 (longest_simple_path): Likewise.
18239 * shrink-wrap.c (spread_components): Likewise.
18240 (disqualify_problematic_components): Likewise.
18241 (emit_common_heads_for_components): Likewise.
18242 (emit_common_tails_for_components): Likewise.
18243 (insert_prologue_epilogue_for_components): Likewise.
18245 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18247 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
18250 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18252 * df-core.c (df_set_blocks): Start using auto_bitmap.
18253 (df_compact_blocks): Likewise.
18254 * df-problems.c (df_rd_confluence_n): Likewise.
18255 * df-scan.c (df_insn_rescan_all): Likewise.
18256 (df_process_deferred_rescans): Likewise.
18257 (df_update_entry_block_defs): Likewise.
18258 (df_update_exit_block_uses): Likewise.
18259 (df_entry_block_bitmap_verify): Likewise.
18260 (df_exit_block_bitmap_verify): Likewise.
18261 (df_scan_verify): Likewise.
18262 * lra-constraints.c (lra_constraints): Likewise.
18263 (undo_optional_reloads): Likewise.
18264 (lra_undo_inheritance): Likewise.
18265 * lra-remat.c (calculate_gen_cands): Likewise.
18266 (do_remat): Likewise.
18267 * lra-spills.c (assign_spill_hard_regs): Likewise.
18268 (spill_pseudos): Likewise.
18269 * tree-ssa-pre.c (bitmap_set_and): Likewise.
18270 (bitmap_set_subtract_values): Likewise.
18272 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18274 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
18275 management with auto_bitmap.
18276 (fix_inter_tick): Likewise.
18277 (fix_recovery_deps): Likewise.
18278 * ira.c (add_store_equivs): Likewise.
18279 (find_moveable_pseudos): Likewise.
18280 (split_live_ranges_for_shrink_wrap): Likewise.
18281 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
18282 (rtx_reuse_manager::seen_def_p): Likewise.
18283 (rtx_reuse_manager::set_seen_def): Likewise.
18284 * print-rtl.h (class rtx_reuse_manager): Likewise.
18286 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18288 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
18290 (migrate_btr_def): Likewise.
18291 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
18292 * df-core.c (loop_post_order_compute): Likewise.
18293 (loop_inverted_post_order_compute): Likewise.
18294 * hsa-common.h: Likewise.
18295 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
18296 * init-regs.c (initialize_uninitialized_regs): Likewise.
18297 * ipa-inline.c (resolve_noninline_speculation): Likewise.
18298 (inline_small_functions): Likewise.
18299 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
18300 * ira.c (combine_and_move_insns): Likewise.
18301 (build_insn_chain): Likewise.
18302 * loop-invariant.c (find_invariants): Likewise.
18303 * lower-subreg.c (propagate_pseudo_copies): Likewise.
18304 * predict.c (tree_predict_by_opcode): Likewise.
18305 (predict_paths_leading_to): Likewise.
18306 (predict_paths_leading_to_edge): Likewise.
18307 (estimate_loops_at_level): Likewise.
18308 (estimate_loops): Likewise.
18309 * shrink-wrap.c (try_shrink_wrapping): Likewise.
18310 (spread_components): Likewise.
18311 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
18312 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
18313 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
18314 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
18315 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
18316 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
18317 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
18318 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
18319 (mark_threaded_blocks): Likewise.
18320 (thread_through_all_blocks): Likewise.
18321 * tree-ssa.c (verify_ssa): Likewise.
18322 (execute_update_addresses_taken): Likewise.
18323 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
18325 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18327 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
18329 (post_order_compute): Likewise.
18330 (inverted_post_order_compute): Likewise.
18331 (pre_and_rev_post_order_compute_fn): Likewise.
18333 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18335 * genrecog.c (int_set::int_set): Explicitly construct our
18336 auto_vec base class.
18337 * vec.h (auto_vec::auto_vec): New constructor.
18339 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18341 * bitmap.h (class auto_bitmap): New constructor taking
18342 bitmap_obstack * argument.
18344 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18346 * bitmap.h (class auto_bitmap): Change type of m_bits to
18347 bitmap_head, and adjust ctor / dtor and member operators.
18349 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
18351 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
18352 when returned register mode doesn't match original mode.
18354 2017-05-12 Jeff Law <law@redhat.com>
18355 Jakub Jelinek <jakub@redhat.com>
18357 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
18358 we look for cc setter after the compare-elim changes.
18359 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
18360 within the vector to match what compare-elim now expects.
18361 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
18362 (xorsi3_flags, one_cmplsi2_flags): Likewise.
18364 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
18365 after the compare-elim changes.
18366 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
18367 the vector to match what compare-elim now expects.
18368 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
18369 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
18370 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
18371 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
18372 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
18374 * config/visium/visium.c (single_set_and_flags): Fix where
18375 we look for cc setter after the compare-elim changes.
18376 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
18377 with the vector to match what compare-elim now expects.
18378 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
18379 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
18380 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
18381 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
18382 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
18383 (neg<mode>2_insn_set_overflow): Likewise.
18385 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
18387 PR middle-end/79794
18388 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
18389 maybe_expand_insn call, set ops[0].target. If still set after call,
18390 set alt_rtl. Add extra arg to recursive calls.
18391 (extract_bit_field): Add alt_rtl argument. Pass to
18393 * expmed.h (extract_bit_field): Fix prototype.
18394 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
18395 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
18396 to extract_bit_field_calls.
18397 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
18398 Pass alt_rtl to extract_bit_field calls.
18399 * calls.c (store_unaligned_arguments_into_psuedos)
18400 load_register_parameters): Pass extra NULL to extract_bit_field calls.
18401 * optabs.c (maybe_legitimize_operand): Clear op->target when call
18403 * optabs.h (struct expand_operand): Add target bitfield.
18405 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
18407 * compare-elim.c (try_eliminate_compare): Canonicalize
18408 operation with embedded compare to
18409 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
18410 (set (reg) (operation)].
18412 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
18414 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
18417 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
18418 cost of adding a carry flag for ADC instruction.
18419 [case MINUS]: Ignore the cost of subtracting a carry flag
18420 for SBB instruction.
18422 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
18424 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
18426 * config/rs6000/bmiintrin.h: New file.
18427 * config/rs6000/bmi2intrin.h: New file.
18428 * config/rs6000/x86intrin.h: New file.
18430 2017-05-12 Jeff Law <law@redhat.com>
18432 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
18435 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
18437 PR middle-end/80707
18438 * tree-cfg.c: Remove cfg edges of unreachable case statements.
18440 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18442 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
18443 early expansion of vector divide builtins.
18444 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
18445 builtins identified as having unsigned arguments.
18447 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18449 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
18450 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
18451 expansion of vector logical operations (and, andc, or, xor,
18454 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18456 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
18457 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
18459 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18461 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
18462 early GIMPLE expansion of vector multiplies.
18464 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18466 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
18467 TARGET_HAVE_MOVT conditional.
18468 (movt splitter): Likewise.
18470 2017-05-12 Richard Biener <rguenther@suse.de>
18472 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
18473 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
18474 Fold all stmts not inplace.
18476 2017-05-12 Richard Biener <rguenther@suse.de>
18478 PR tree-optimization/80713
18479 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
18480 inserted_exprs bit for not removed stmts.
18482 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
18484 PR middle-end/69921
18485 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
18486 parallelized" attribute for parallelized OpenACC kernels.
18487 * omp-offload.c (execute_oacc_device_lower): Use it.
18489 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
18490 Set "oacc kernels" attribute.
18491 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
18492 parameter. Adjust all users.
18493 (oacc_fn_attrib_kernels_p): Remove function.
18494 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
18495 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
18496 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
18497 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
18498 assert "oacc kernels" attribute is set.
18500 2017-05-11 Carl Love <cel@us.ibm.com>
18502 * config/rs6000/rs6000-c: Add support for built-in functions
18503 vector unsigned char vec_popcnt (vector signed char)
18504 vector unsigned char vec_popcnt (vector unsigned char)
18505 vector unsigned short vec_popcnt (vector signed short)
18506 vector unsigned short vec_popcnt (vector unsigned short)
18507 vector unsigned int vec_popcnt (vector signed int)
18508 vector unsigned int vec_popcnt (vector unsigned int)
18509 vector unsigned long long vec_popcnt (vector signed long long)
18510 vector unsigned long long vec_popcnt (vector unsigned long long)
18511 vector signed long long vec_slo (vector signed long long,
18512 vector signed char)
18513 vector signed long long vec_slo (vector signed long long,
18514 vector unsigned char)
18515 vector unsigned long long vec_slo (vector unsigned long long,
18516 vector signed char)
18517 vector unsigned long long vec_slo (vector unsigned long long,
18518 vector unsigned char)
18519 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
18520 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
18521 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
18522 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
18523 * doc/extend.texi: Update the built-in documentation file for the
18524 new built-in functions.
18526 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
18528 * attribs.h (sorted_attr_string): Move machine independent
18529 functions for target clone support from the i386 port to common
18530 code. Rename ix86_function_versions to common_function_versions.
18531 Rename make_name to make_unique_name.
18532 (common_function_versions): Likewise.
18533 (make_unique_name): Likewise.
18534 (make_dispatcher_decl): Likewise.
18535 (is_function_default_version): Likewise.
18536 * attribs.c (attr_strcmp): Likewise.
18537 (sorted_attr_string): Likewise.
18538 (common_function_versions): Likewise.
18539 (make_unique_name): Likewise.
18540 (make_dispatcher_decl): Likewise.
18541 (is_function_default_version): Likewise.
18542 * config/i386/i386.c (attr_strcmp): Likewise.
18543 (sorted_attr_string): Likewise.
18544 (ix86_function_versions): Likewise.
18545 (make_name): Likewise.
18546 (make_dispatcher_decl): Likewise.
18547 (is_function_default_version): Likewise.
18548 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
18550 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18553 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
18554 Account for direct move costs for vec_construct of integer
18557 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
18560 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
18561 (UNSPEC_STX_ATOMIC): Ditto.
18562 (loaddi_via_sse): New insn.
18563 (storedi_via_sse): Ditto.
18564 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
18565 Update corresponding peephole2 patterns.
18566 (atomic_storedi_fpu): Ditto.
18568 2017-05-11 Julia Koval <julia.koval@intel.com>
18570 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
18571 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
18573 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
18574 (__builtin_ia32_rsqrt14ss_mask): New builtins.
18575 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
18577 2017-05-11 Nathan Sidwell <nathan@acm.org>
18579 * graphite-poly.c: Include dumpfile.h.
18581 * dumpfle.h (dump_function): Declare here ...
18582 * tree-dump.h (dump_function): ... not here.
18583 * dumpfile.c: #include tree-cfg.h.
18584 (dump_function): Move here from ...
18585 * tree-dump.c (dump_function): ... here.
18586 * gimplify.c: #include splay-tree.h, not tree-dump.h.
18587 * graphite-poly.c: Don't include tree-dump.h.
18588 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
18589 * print-tree.c: Likewise.
18590 * stor-layout.c: Likewise.
18591 * tree-nested.c: Likewise.
18593 * dumpfile.c (dump_start): Use TDF_FLAGS.
18594 (dump_enable_all): Fix TDF_KIND check thinko.
18596 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
18598 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18599 array entries to represent two legal parameterizations of the
18600 overloaded __builtin_cmpb function, as represented by the
18601 P6_OV_BUILTIN_CMPB constant.
18602 (altivec_resolve_overloaded_builtin): Add special case handling
18603 for the __builtin_cmpb function, as represented by the
18604 P6_OV_BUILTIN_CMPB constant.
18605 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
18606 (BU_P6_64BIT_2): New macro.
18607 (BU_P6_OVERLOAD_2): New macro
18608 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
18609 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
18610 (CMPB): Add overload support to represent both 32-bit and 64-bit
18611 compare-bytes function.
18612 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
18613 support for TARGET_CMPB.
18614 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
18615 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18616 documentation of the __builtin_cmpb overloaded built-in function.
18618 2017-05-11 Richard Biener <rguenther@suse.de>
18620 PR tree-optimization/80705
18621 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
18622 bases are not vectorizable.
18624 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18626 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
18627 when counting register pressure.
18629 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18631 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
18632 (iv_ca_more_deps): Renamed to ...
18633 (iv_ca_compare_deps): ... this.
18634 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
18636 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18638 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
18640 (determine_group_iv_costs): ... here.
18641 (find_inv_vars_cb): Record inv var if it's not recorded before.
18643 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18645 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
18646 (get_shiftadd_cost): Ditto.
18648 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18650 * tree-ssa-address.c: Include header file.
18651 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
18653 (add_to_parts): Refactor.
18654 (addr_to_parts): New parameter. Update use of move_hint_to_base.
18655 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
18658 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18660 PR tree-optimization/53090
18661 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
18663 (extract_cond_operands): Detect condition with IV on both sides
18664 and return COMP_IV_EXPR_2.
18665 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
18666 (rewrite_use_compare): Simplify by removing call to function
18667 extract_cond_operands.
18669 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18671 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
18672 (extract_cond_operands): Detect condition comparing against non-
18673 invariant bound and return appropriate enum value.
18674 (find_interesting_uses_cond): Update use of extract_cond_operands.
18675 Handle its return value accordingly.
18676 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
18678 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18680 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
18681 nonlinear iv_use computation in loop invariant sensitive way.
18683 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18685 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
18686 (find_iv_candidates): Call relate_compare_use_with_all_cands.
18688 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18690 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
18691 (dump_cand): Support iv_cand.inv_exprs.
18692 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
18694 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
18697 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18699 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
18701 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
18702 as local function. Include necessary header files.
18703 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
18705 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18707 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
18709 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18711 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
18712 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
18713 RSHIFT_EXPR and BIT_NOT_EXPR.
18715 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18717 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
18718 (adjust_setup_cost): New parameter supporting round up adjustment.
18719 (struct address_cost_data): Delete.
18720 (force_expr_to_var_cost): Don't bound cost with spill_cost.
18721 (split_address_cost, ptr_difference_cost): Delete.
18722 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
18723 (struct ainc_cost_data): New struct.
18724 (get_address_cost_ainc): New function.
18725 (get_address_cost, get_computation_cost): Reimplement.
18726 (determine_group_iv_cost_address): Record inv_expr for all uses of
18728 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
18729 (iv_ca_has_deps): Reimplemented to ...
18730 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
18732 (iv_ca_extend): Call iv_ca_more_deps.
18734 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18736 * tree-ssa-address.c (struct mem_address): Move to header file.
18737 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
18738 * tree-ssa-address.h (struct mem_address): Move from C file.
18739 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
18741 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18743 * tree-affine.h (aff_combination_type): New interface.
18744 (aff_combination_zero_p): Remove static.
18745 (aff_combination_const_p): New interface.
18746 (aff_combination_singleton_var_p): New interfaces.
18748 2017-05-11 Richard Biener <rguenther@suse.de>
18750 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
18751 Skip unreachable blocks and destinations.
18752 (eliminate): Move stmt removal and fixup ...
18753 (fini_eliminate): ... here. Skip inserted exprs.
18754 (pass_pre::execute): Move fini_pre after fini_eliminate.
18755 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
18756 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
18757 PRE to get rid of dead code that has invalid SSA form and
18758 split critical edges again.
18760 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18762 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18764 2017-05-11 Richard Biener <rguenther@suse.de>
18766 * passes.c (execute_function_todo): Verify loops if they are
18767 said to be up-to-date.
18768 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
18769 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
18771 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
18774 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
18775 handle calling assemble_external ourself.
18778 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
18779 modes with zero size. Enhance comment.
18781 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18783 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
18784 built-ins for vec_xl and vec_xst with short and char pointer
18787 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
18789 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
18790 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
18791 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
18792 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
18793 (_mm_maskz_min_round_ss): New intrinsics.
18794 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
18795 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
18796 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
18797 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
18798 (__builtin_ia32_minss_mask_round): New builtins.
18799 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18800 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18801 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
18803 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
18804 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
18806 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
18809 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
18811 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
18812 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
18813 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
18814 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
18815 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
18816 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
18817 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
18818 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18819 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
18820 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
18821 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
18822 (__builtin_ia32_mulss_mask_round): New builtins.
18823 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18824 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18825 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
18827 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
18828 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
18830 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
18833 2017-05-10 Julia Koval <julia.koval@intel.com>
18835 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
18836 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
18837 (_mm256_setr_m128i): New intrinsics.
18839 2017-05-10 Julia Koval <julia.koval@intel.com>
18841 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
18842 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
18843 (_mm_maskz_rcp14_ss): New intrinsics.
18844 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
18845 (__builtin_ia32_rcp14ss_mask): New builtins.
18846 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
18848 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
18850 PR tree-optimization/51513
18851 * tree-cfg.c (gimple_seq_unreachable_p): New function.
18852 (assert_unreachable_fallthru_edge_p): Use it.
18853 (group_case_labels_stmt): Likewise.
18854 * tree-cfg.h: Prototype it.
18855 * stmt.c: Include cfghooks.h and tree-cfg.h.
18856 (emit_case_dispatch_table) <gap_label>: New local variable.
18857 Use it to fill dispatch table gaps.
18858 Test for default_label before updating probabilities.
18859 (expand_case) <default_label>: Remove unneeded initialization.
18860 Test for unreachable default case statement and remove its edge.
18861 Set default_label accordingly.
18862 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
18864 2017-05-10 Carl Love <cel@us.ibm.com>
18866 * config/rs6000/rs6000-c: Add support for built-in functions
18867 vector signed char vec_neg (vector signed char)
18868 vector signed short int vec_neg (vector short int)
18869 vector signed int vec_neg (vector signed int)
18870 vector signed long long vec_neg (vector signed long long)
18871 vector float vec_neg (vector float)
18872 vector double vec_neg (vector double)
18873 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
18875 * config/rs6000/altivec.h: Add define for vec_neg
18876 * doc/extend.texi: Update the built-in documentation for the
18877 new built-in functions.
18879 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18881 PR tree-optimization/77644
18882 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
18884 2017-05-10 Nathan Sidwell <nathan@acm.org>
18886 * dumpfile.h (TDI_lang_all): New.
18887 (TDF_KIND): New. Renumber others
18888 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
18890 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
18892 (get_dump_file_name): Adjust suffix generation.
18893 (dump_enable_all): Use TDF_KIND.
18894 * doc/invoke.texi (-fdump-lang-all): Document.
18896 * dumpfile.h: Tabify.
18898 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
18901 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
18902 Move member access before delete.
18904 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
18906 * tree-inline.c (expand_call_inline): Split block at stmt
18909 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
18912 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
18913 are now unused after splitting mov{sf,sd}_hardfloat.
18914 (f32_lr2): Likewise.
18915 (f32_lm): Likewise.
18916 (f32_lm2): Likewise.
18917 (f32_li): Likewise.
18918 (f32_li2): Likewise.
18919 (f32_lv): Likewise.
18920 (f32_sr): Likewise.
18921 (f32_sr2): Likewise.
18922 (f32_sm): Likewise.
18923 (f32_sm2): Likewise.
18924 (f32_si): Likewise.
18925 (f32_si2): Likewise.
18926 (f32_sv): Likewise.
18927 (f32_dm): Likewise.
18928 (f32_vsx): Likewise.
18929 (f32_av): Likewise.
18930 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
18931 For movsf, order stores so the VSX stores occur before the GPR
18932 store which encourages the register allocator to use a traditional
18933 FPR instead of a GPR. For movsd, order the stores so that the GPR
18934 store comes before the VSX stores to allow the power6 to work.
18935 This is due to the power6 not having a 32-bit integer store
18936 instruction from a FPR.
18937 (movsf_hardfloat): Likewise.
18938 (movsd_hardfloat): Likewise.
18940 2017-05-09 Martin Sebor <msebor@redhat.com>
18942 PR translation/80280
18943 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
18946 PR translation/80280
18947 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
18948 data member added in r247778.
18949 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
18951 2017-05-09 Nathan Sidwell <nathan@acm.org>
18953 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
18955 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
18958 2017-05-09 Marek Polacek <polacek@redhat.com>
18960 * doc/invoke.texi: Fix typo.
18962 2017-05-09 Richard Biener <rguenther@suse.de>
18964 * tree-vrp.c (vrp_val_is_max): Adjust comment.
18965 (vrp_val_is_min): Likewise.
18966 (set_value_range_to_value): Likewise.
18967 (set_value_range_to_nonnegative): Likewise.
18968 (gimple_assign_nonzero_p): Likewise.
18969 (gimple_stmt_nonzero_p): Likewise.
18970 (vrp_int_const_binop): Likewise. Remove unreachable case.
18971 (adjust_range_with_scev): Adjust comments.
18972 (compare_range_with_value): Likewise.
18973 (extract_range_from_phi_node): Likewise.
18974 (test_for_singularity): Likewise.
18976 2017-05-09 Richard Biener <rguenther@suse.de>
18978 * tree-vrp.c (get_single_symbol): Add assert that we don't
18979 get overflowed constants as invariant part.
18980 (compare_values_warnv): Add comment before the TREE_NO_WARNING
18981 checks. Use wi::cmp instead of recursing for integer constants.
18982 (compare_values): Just ignore whether we assumed undefined
18983 overflow instead of failing the compare.
18984 (extract_range_for_var_from_comparison_expr): Add comment before the
18985 TREE_NO_WARNING sets.
18986 (test_for_singularity): Likewise.
18987 (extract_range_from_comparison): Do not disable optimization
18988 when we assumed undefined overflow.
18989 (extract_range_basic): Remove init of unused var.
18991 2017-05-09 Richard Biener <rguenther@suse.de>
18993 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
18994 (extract_range_from_multiplicative_op_1): Adjust.
18995 (extract_range_from_binary_expr_1): Use int_const_binop.
18997 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
19000 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
19001 rs6000_store_data_bypass_p in seven define_bypass directives and
19002 in several comments.
19003 * config/rs6000/rs6000-protos.h: Add prototype for
19004 rs6000_store_data_bypass_p function.
19005 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
19006 function implements slightly different (rs6000-specific) semantics
19007 than store_data_bypass_p, returning false rather than aborting
19008 with assertion error when arguments do not satisfy the
19009 requirements of store data bypass.
19010 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
19011 rs6000_store_data_bypass_p.
19013 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
19015 * config/xtensa/xtensa-protos.h
19016 (xtensa_initial_elimination_offset): New declaration.
19017 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
19018 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
19019 macro definition, add case for FRAME_POINTER_REGNUM when
19020 FRAME_GROWS_DOWNWARD.
19021 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
19022 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
19023 xtensa_initial_elimination_offset.
19025 2017-05-08 Nathan Sidwell <nathan@acm.org>
19027 * doc/invoke.texi: Alphabetize -fdump options.
19029 2017-05-08 Martin Sebor <msebor@redhat.com>
19031 PR translation/80280
19032 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
19034 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
19036 * target.def (compute_frame_layout): New optional target hook.
19037 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
19038 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
19039 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
19041 * reload1.c (verify_initial_elim_offsets): Likewise.
19042 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
19043 (use_simple_return_p): Call arm_compute_frame_layout if needed.
19044 (arm_get_frame_offsets): Split up into this ...
19045 (arm_compute_frame_layout): ... and this function.
19047 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
19049 * config/aarch64/constraints.md (Usa): New constraint.
19050 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
19052 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
19054 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
19055 with_multilib_list after it has been checked.
19057 2017-05-08 Richard Biener <rguenther@suse.de>
19059 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
19060 (bitmap_set_subtract_values): Likewise.
19062 2017-05-08 Richard Biener <rguenther@suse.de>
19064 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
19065 (gimple_assign_nonzero): ... this and remove strict_overflow_p
19067 (gimple_stmt_nonzero_warnv_p): Rename to ...
19068 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
19070 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
19071 (extract_range_basic): Adjust, do not disable propagation on
19072 strict overflow sensitive simplification.
19073 (vrp_visit_cond_stmt): Likewise.
19075 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
19077 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
19078 body size unconditionally.
19080 2017-05-07 Jeff Law <law@redhat.com>
19083 2017-05-06 Jeff Law <law@redhat.com>
19084 PR tree-optimization/78496
19085 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
19088 PR tree-optimization/78496
19089 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
19090 (simplify_stmt_using_ranges): Call it.
19091 (vrp_dom_walker::before_dom_children): Extract equivalences
19092 from an ASSERT_EXPR with an equality comparison against a
19095 2017-05-06 Jeff Law <law@redhat.com>
19097 PR tree-optimization/78496
19098 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
19101 PR tree-optimization/78496
19102 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
19103 (simplify_stmt_using_ranges): Call it.
19104 (vrp_dom_walker::before_dom_children): Extract equivalences
19105 from an ASSERT_EXPR with an equality comparison against a
19108 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
19110 * lra-constraints.c (lra_copy_reg_equiv): New function.
19111 (split_reg): Use it to copy equivalence information from the
19112 original register to the spill register.
19114 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
19116 PR rtl-optimization/75964
19117 * simplify-rtx.c (simplify_const_relational_operation): Remove
19118 invalid handling of comparisons of integer ABS.
19120 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
19122 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
19123 initialize to zero.
19124 (init_regs): Remove declaration.
19125 (function_arg_advance_32): Initialize error_p as boolean variable.
19127 2017-05-05 Nathan Sidwell <nathan@acm.org>
19129 * store-motion.c (remove_reachable_equiv_notes): Reformat long
19130 lines. Use for (;;).
19132 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19134 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
19135 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
19136 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
19137 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
19138 VF=2 that require versioning.
19140 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19142 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
19145 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19147 * diagnostic.h (diagnostic_override_option_index): Convert from
19148 macro to inline function.
19150 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19152 * diagnostic.c (last_module_changed_p): New function.
19153 (set_last_module): New function.
19154 (diagnostic_report_current_module): Convert macro usage to
19155 the above functions.
19156 * diagnostic.h (diagnostic_context::last_module): Strengthen
19157 from const line_map * to const line_map_ordinary *.
19158 (diagnostic_last_module_changed): Delete macro.
19159 (diagnostic_set_last_module): Delete macro.
19161 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19163 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
19164 with diagnostic_report_diagnostic.
19165 (diagnostic_n_impl_richloc): Likewise.
19166 * diagnostic.h (report_diagnostic): Delete macro.
19167 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
19168 with diagnostic_report_diagnostic.
19169 * substring-locations.c (format_warning_va): Likewise.
19171 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19173 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
19174 save/restor of format_spec. Move option-printing code to...
19175 (print_option_information): ...this new function, and
19176 reimplement by simply printing to the pretty_printer,
19177 rather than appending to the format string.
19179 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19181 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
19182 handling logic into...
19183 (update_effective_level_from_pragmas): ...this new function.
19185 2017-05-04 Andrew Waterman <andrew@sifive.com>
19187 * config/riscv/riscv.opt (mstrict-align): New option.
19188 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
19189 (SLOW_UNALIGNED_ACCESS): Define.
19190 (riscv_slow_unaligned_access): Declare.
19191 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
19193 (riscv_slow_unaligned_access): New variable.
19194 (rocket_tune_info): Set slow_unaligned_access to true.
19195 (optimize_size_tune_info): Set slow_unaligned_access to false.
19196 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
19197 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
19198 (riscv_option_override): Set riscv_slow_unaligned_access.
19199 * doc/invoke.texi: Add -mstrict-align to RISC-V.
19201 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
19203 * config/riscv/riscv.md: Unify indentation.
19205 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
19210 * config/rs6000/rs6000.md (u code attribute): Add FIX and
19212 (extendsi<mode>2): Add support for doing sign extension via
19213 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
19214 don't have ISA 3.0 instructions.
19215 (extendsi<mode>2 splitter): Likewise.
19216 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
19217 generate the normal insns since SImode can now go in vector
19218 registers. Disallow the special UNSPECs needed for previous
19219 machines to hide SImode being used. Add new insns
19220 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
19221 (fix_trunc<mode>si2_stfiwx): Likewise.
19222 (fix_trunc<mode>si2_internal): Likewise.
19223 (fixuns_trunc<mode>si2): Likewise.
19224 (fixuns_trunc<mode>si2_stfiwx): Likewise.
19225 (fctiw<u>z_<mode>_smallint): Likewise.
19226 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
19227 of floating point to 32-bit integer from doing a direct move to
19228 the GPR registers to do a store.
19229 (fctiwz_<mode>): Break long line.
19231 2017-05-05 Bin Cheng <bin.cheng@arm.com>
19233 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
19234 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
19235 (addr_list, addr_offset_valid_p): New.
19236 (split_address_groups): Check offset validity with above function.
19237 (gt-tree-ssa-loop-ivopts.h): Include header file.
19239 2017-05-05 Nathan Sidwell <nathan@acm.org>
19241 * config.gcc (arm*-*-*): Add missing 'fi'.
19243 2017-05-05 Steve Ellcey <sellcey@cavium.com>
19245 * doc/invoke.texi (-fopt-info): Explicitly say order of options
19246 included in -fopt-info does not matter.
19247 * doc/optinfo.texi (-fopt-info): Fix description of default
19248 behavour. Explicitly say order of options included in -fopt-info
19251 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
19253 * config.gcc: Allow combinations of aprofile and rmprofile values for
19254 --with-multilib-list.
19255 * config/arm/t-multilib: New file.
19256 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
19257 variables. Remove setting of ISA and floating-point ABI in
19258 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
19259 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
19260 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
19261 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
19263 * config/arm/t-rmprofile: Likewise except for the matches changes.
19264 * doc/install.texi (--with-multilib-list): Document the combination of
19265 aprofile and rmprofile values and warn about pitfalls in doing that.
19267 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
19269 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
19270 (movdi_aarch64): Likewise.
19272 2017-05-05 Jakub Jelinek <jakub@redhat.com>
19274 PR tree-optimization/80632
19275 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
19277 (build_arrays): Initialize it for virtual phis.
19278 (fix_phi_nodes): Use it for virtual phis.
19280 PR tree-optimization/80558
19281 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
19282 [x, y] op z into [x op, y op z] for op & or | if conditions
19285 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
19286 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19289 * config/arm/arm.md (use_literal_pool): Remove.
19290 (64-bit immediate split): No longer takes cost into consideration
19291 if arm_disable_literal_pool is enabled.
19292 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
19293 used when arm_disable_literal_pool is enabled.
19294 (arm_max_const_double_inline_cost): Remove use of
19295 arm_disable_literal_pool.
19296 (push_minipool_fix): Add assert.
19297 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
19298 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
19299 (no_literal_pool_sf_immediate): New.
19301 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19303 PR tree-optimization/80613
19304 * tree-ssa-dce.c (propagate_necessity): Remove cases for
19305 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
19307 2017-05-05 Richard Biener <rguenther@suse.de>
19309 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
19311 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
19313 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
19314 of this flag from insn conditions due to removal from r247495.
19316 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
19318 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
19320 (arm_early_store_addr_dep_ptr): Likewise.
19321 * config/arm/aarch-common-protos.h
19322 (arm_early_load_addr_dep_ptr): Add prototype.
19323 (arm_early_store_addr_dep_ptr): Likewise.
19324 * config/arm/cortex-a53.md: Add new bypasses.
19326 2017-05-05 Jakub Jelinek <jakub@redhat.com>
19328 * tree.c (next_type_uid): Change type to unsigned.
19329 (type_hash_canon): Decrement back next_type_uid if
19330 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
19331 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
19334 2017-05-04 Martin Sebor <msebor@redhat.com>
19336 * builtins.c: Fix a trivial typo in a comment.
19338 PR middle-end/79234
19339 * builtins.c (check_sizes): Adjust to handle reading past the end.
19340 Avoid printing excessive upper bound of ranges. Use %E to print
19341 tree nodes instead of converting them to %wu.
19342 (expand_builtin_memchr): New function.
19343 (compute_dest_size): Rename...
19344 (compute_objsize): ...to this.
19345 (expand_builtin_memcpy): Adjust.
19346 (expand_builtin_mempcpy): Adjust.
19347 (expand_builtin_strcat): Adjust.
19348 (expand_builtin_strcpy): Adjust.
19349 (check_strncat_sizes): Adjust.
19350 (expand_builtin_strncat): Adjust.
19351 (expand_builtin_strncpy): Adjust and simplify.
19352 (expand_builtin_memset): Adjust.
19353 (expand_builtin_bzero): Adjust.
19354 (expand_builtin_memcmp): Adjust.
19355 (expand_builtin): Handle memcmp.
19356 (maybe_emit_chk_warning): Check strncat just once.
19358 2017-05-04 Martin Sebor <msebor@redhat.com>
19360 PR preprocessor/79214
19361 PR middle-end/79222
19362 PR middle-end/79223
19363 * builtins.c (check_sizes): Add inlining context and issue
19364 warnings even when -Wno-system-headers is set.
19365 (check_strncat_sizes): Same.
19366 (expand_builtin_strncat): Same.
19367 (expand_builtin_memmove): New function.
19368 (expand_builtin_stpncpy): Same.
19369 (expand_builtin): Handle memmove and stpncpy.
19371 2017-05-04 Bin Cheng <bin.cheng@arm.com>
19373 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
19374 which is not used any more.
19376 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19378 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
19380 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19382 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
19383 (cortexa53_tunings): Likewise.
19384 (cortexa57_tunings): Likewise.
19385 (cortexa72_tunings): Likewise.
19386 (cortexa73_tunings): Likewise.
19388 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19390 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
19391 Set loop alignment to 8.
19393 2017-05-04 Martin Sebor <msebor@redhat.com>
19395 PR translation/80280
19396 * builtins.c (expand_builtin_object_size): Add missing quoting to
19397 %D and like directives.
19398 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
19399 (hsa_type_for_tree_type): Same.
19400 (verify_function_arguments): Same.
19401 * symtab.c (symbol_table::change_decl_assembler_name): Same.
19402 * varasm.c (get_section): Same.
19405 2017-05-04 Martin Sebor <msebor@redhat.com>
19407 PR translation/80280
19408 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
19410 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19412 * config/aarch64/aarch64.c (generic_addrcost_table):
19413 Change HI/TI mode setting.
19415 2017-05-04 Martin Jambor <mjambor@suse.cz>
19417 PR tree-optimization/80622
19418 * tree-sra.c (comes_initialized_p): New function.
19419 (build_accesses_from_assign): Only set write lazily when
19420 comes_initialized_p is false.
19421 (analyze_access_subtree): Use comes_initialized_p.
19422 (propagate_subaccesses_across_link): Assert !comes_initialized_p
19423 instead of testing for PARM_DECL.
19425 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19427 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
19428 constraint on operand 0 to allow more general addressing modes.
19429 Adjust output template.
19430 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
19432 * config/aarch64/aarch64-protos.h
19433 (aarch64_address_valid_for_prefetch_p): Declare prototype.
19434 * config/aarch64/constraints.md (Dp): New address constraint.
19435 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
19438 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
19440 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
19441 update use of estimate_ipcp_clone_size_and_time.
19442 (estimate_local_effects): Update use of
19443 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
19444 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
19445 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
19446 Return nonspecialized time.
19448 2017-05-04 Richard Biener <rguenther@suse.de>
19450 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
19451 for the last VUSE which def dominates the PHI. Directly call
19453 (get_continuation_for_phi_1): Remove.
19455 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
19457 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
19458 to explain the use of truncating division. Cap the number of
19459 iterations to the maximum given by nb_iterations_upper_bound,
19462 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19464 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
19465 * configure: Regenerate.
19466 * config.in: Regenerate.
19467 * config/i386/driver-mingw32.c: new file.
19468 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
19469 * config.host: Link driver-mingw32.o on MinGW host.
19470 * doc/install.texi: Document new --enable-mingw-wildcard configure
19473 2017-05-04 Marek Polacek <polacek@redhat.com>
19475 PR tree-optimization/80612
19476 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
19478 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19479 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
19481 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
19482 (movt splitter): Likewise.
19483 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
19484 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
19485 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
19486 block for Thumb-1 with MOVT.
19487 (thumb2_legitimate_address_p): Move code block ...
19488 (can_avoid_literal_pool_for_label_p): ... into this new function.
19489 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
19491 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
19492 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
19493 "M-profile targets with the MOVT instruction".
19495 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19497 * config/arm/arm-builtins.c (arm_init_builtins): Rename
19498 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
19499 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
19501 2017-05-04 Martin Liska <mliska@suse.cz>
19503 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
19504 variable cond_code.
19506 2017-05-04 Richard Biener <rguenther@suse.de>
19508 * tree.c (array_at_struct_end_p): Handle arrays at struct
19509 end with flexarrays more conservatively. Refactor and treat
19510 arrays of arrays or aggregates more strict. Fix
19511 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
19512 * tree.c (array_at_struct_end_p): Adjust prototype.
19513 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
19514 * gimple-fold.c (get_range_strlen): Likewise.
19515 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
19517 2017-05-04 Richard Biener <rguenther@suse.de>
19519 PR tree-optimization/31130
19520 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
19522 (supports_overflow_infinity): Likewise.
19523 (is_negative_overflow_infinity): Likewise.
19524 (is_positive_overflow_infinity): Likewise.
19525 (is_overflow_infinity): Likewise.
19526 (stmt_overflow_infinity): Likewise.
19527 (overflow_infinity_range_p): Likewise.
19528 (usable_range_p): Remove as always returning true.
19529 (make_overflow_infinity): Remove.
19530 (negative_overflow_infinity): Likewise.
19531 (positive_overflow_infinity): Likewise.
19532 (avoid_overflow_infinity): Likewise.
19533 (set_value_range): Adjust accordingly.
19534 (set_value_range_to_nonnegative): Likewise, remove now unused
19535 overflow_infinity arg.
19536 (vrp_operand_equal_p): Adjust.
19537 (update_value_range): Likewise.
19538 (range_int_cst_singleton_p): Likewise.
19539 (operand_less_p): Likewise.
19540 (compare_values_warnv): Likewise.
19541 (extract_range_for_var_from_comparison_expr): Likewise.
19542 (vrp_int_const_binop): Likewise.
19543 (zero_nonzero_bits_from_vr): Likewise.
19544 (extract_range_from_multiplicative_op_1): Likewise.
19545 (extract_range_from_binary_expr_1): Likewise.
19546 (extract_range_from_unary_expr): Likewise.
19547 (extract_range_from_comparison): Likewise.
19548 (extract_range_basic): Likewise.
19549 (adjust_range_with_scev): Likewise.
19550 (compare_ranges): Likewise.
19551 (compare_range_with_value): Likewise.
19552 (dump_value_range): Likewise.
19553 (test_for_singularity): Likewise, remove strict_overflow_p parameter
19555 (simplify_cond_using_ranges): Adjust.
19557 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
19559 * brig-builtins.def: Added a builtin for class_f64.
19560 * builtin-types.def: Added a builtin type needed by class_f64.
19562 2017-05-03 Jason Merrill <jason@redhat.com>
19564 * timevar.def: Add TV_CONSTEXPR.
19566 2017-05-03 David Malcolm <dmalcolm@redhat.com>
19568 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
19570 2017-05-03 Martin Jambor <mjambor@suse.cz>
19572 * ipa-prop.c (ipa_update_after_lto_read): Removed.
19573 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
19574 * ipa-cp.c (ipcp_propagate_stage): Do not call
19575 ipa_update_after_lto_read.
19576 * ipa-inline.c (ipa_inline): Likewise.
19578 2017-05-03 Martin Jambor <mjambor@suse.cz>
19580 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
19581 tag. Added a default constructor and a destructor.
19582 (ipa_edge_args_sum_t): New class;
19583 (ipa_edge_args_sum): Declare.
19584 (ipa_edge_args_vector): Remove declaration.
19585 (IPA_EDGE_REF): Use ipa_edge_args_sum.
19586 (ipa_free_edge_args_substructures): Remove declaration.
19587 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
19588 (ipa_edge_args_info_available_for_edge_p): Likewise.
19589 * ipa-prop.c (ipa_edge_args_vector): Removed.
19590 (edge_removal_hook_holder): Likewise.
19591 (edge_duplication_hook_holder): Likewise.
19592 (ipa_edge_args_sum): New variable.
19593 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
19594 ipa_edge_args_vector.
19595 (ipa_free_edge_args_substructures): Likewise.
19596 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
19597 ipa_edge_args_vector.
19598 (ipa_edge_removal_hook): Turned into method
19599 ipa_edge_args_sum_t::remove.
19600 (ipa_edge_duplication_hook): Turned into method
19601 ipa_edge_args_sum_t::duplicate.
19602 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
19603 registering edge hooks.
19604 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
19605 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
19606 ipa_edge_args_sum instead of ipa_edge_args_vector.
19607 * ipa-profile.c (ipa_profile): Likewise.
19609 2017-05-03 Martin Jambor <mjambor@suse.cz>
19611 * symbol-summary.h (function_summary): New method exists.
19612 (function_summary::symtab_removal): Deallocate through release.
19613 (call_summary): New class.
19614 (gt_ggc_mx): New overload.
19615 (gt_pch_nx): Likewise.
19616 (gt_pch_nx): Likewise.
19618 2017-05-03 Jeff Law <law@redhat.com>
19620 PR tree-optimization/78496
19621 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
19622 from simplify_cond_using_ranges. Split off code to walk
19623 backwards through casts into ...
19624 (simplify_cond_using_ranges_2): New function.
19625 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
19626 (execute_vrp): After identifying jump threads, call
19627 simplify_cond_using_ranges_2.
19629 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
19632 * ipa-inline.h (inline_summary): Add ctor.
19633 (create_ggc): Do not use ggc_cleared_alloc.
19635 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
19636 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19638 * gcc.c (handle_braces): Support escaping in switch matching
19640 * doc/invoke.texi (Spec Files): Document it.
19641 Remove superfluous @code markup in items.
19643 2017-05-03 David Malcolm <dmalcolm@redhat.com>
19645 * diagnostic-show-locus.c (struct column_range): New struct.
19646 (get_affected_columns): New function.
19647 (get_printed_columns): New function.
19648 (struct correction): New struct.
19649 (correction::ensure_capacity): New function.
19650 (correction::ensure_terminated): New function.
19651 (struct line_corrections): New struct.
19652 (line_corrections::~line_corrections): New dtor.
19653 (line_corrections::add_hint): New function.
19654 (layout::print_trailing_fixits): Reimplement in terms of the new
19656 (selftest::test_overlapped_fixit_printing): New function.
19657 (selftest::diagnostic_show_locus_c_tests): Call it.
19659 2017-05-03 Nathan Sidwell <nathan@acm.org>
19661 Canonicalize canonical type hashing
19662 * tree.h (type_hash_canon_hash): Declare.
19663 * tree.c (type_hash_list, attribute_hash_list): Move into
19664 type_hash_canon_hash.
19665 (build_type_attribute_qual_variant): Break out hash code calc into
19666 type_hash_canon_hash.
19667 (type_hash_canon_hash): New. Generic type hash computation.
19668 (build_range_type_1, build_array_type_1, build_function_type,
19669 build_method_type_directly, build_offset_type, build_complex_type,
19670 make_vector_type): Call it.
19672 2017-05-03 Richard Biener <rguenther@suse.de>
19674 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
19675 When all DRs have unknown misaligned do not always peel
19676 when there is a store but apply the same costing model as if
19677 there were only loads.
19679 2017-05-03 Richard Biener <rguenther@suse.de>
19682 PR tree-optimization/80492
19683 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
19684 compare_base_decls returning dont-know properly.
19686 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
19688 * config/arm/iterators.md (CCSI): New mode iterator.
19689 (arch): New mode attribute.
19690 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
19691 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
19692 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
19693 code iterator for success result mode.
19694 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
19695 the corresponding new insn generators.
19697 2017-05-03 Bin Cheng <bin.cheng@arm.com>
19700 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19701 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19703 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
19705 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
19706 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
19707 (DDR_A): Wrap DDR argument in brackets.
19708 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
19709 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
19710 (DDR_REVERSED_P): Likewise.
19712 2017-05-03 Jakub Jelinek <jakub@redhat.com>
19714 PR tree-optimization/79472
19715 * tree-switch-conversion.c (struct switch_conv_info): Add
19716 contiguous_range and default_case_nonstandard fields.
19717 (collect_switch_conv_info): Compute contiguous_range and
19718 default_case_nonstandard fields, don't clear final_bb if
19719 contiguous_range and only the default case doesn't have the required
19721 (check_all_empty_except_final): Set default_case_nonstandard instead
19722 of failing if contiguous_range and the default case doesn't have empty
19724 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
19725 and only the default case doesn't have the required constants. Skip
19727 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
19728 if default_case_nonstandard.
19729 (build_constructors): Build constant 1 just once. Assert that default
19730 values aren't inserted in between cases if contiguous_range. Skip
19732 (build_arrays): Skip virtual phis.
19733 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
19734 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
19735 Handle virtual phis.
19736 (gen_inbound_check): Handle default_case_nonstandard case.
19737 (process_switch): Adjust check_final_bb caller. Call
19738 gather_default_values with the first non-default case instead of
19739 default case if default_case_nonstandard.
19741 2017-05-02 Nathan Sidwell <nathan@acm.org>
19743 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
19744 check. Fix formatting.
19746 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
19748 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
19749 errors when comparing specialized and unspecialized times.
19751 2017-05-02 David Malcolm <dmalcolm@redhat.com>
19753 * diagnostic-show-locus.c
19754 (layout::should_print_annotation_line_p): Make private.
19755 (layout::print_annotation_line): Make private.
19756 (layout::annotation_line_showed_range_p): Make private.
19757 (layout::show_ruler): Make private.
19758 (layout::print_source_line): Make private. Pass in line and
19759 line_width, rather than calling location_get_source_line. Drop
19761 (layout::print_leading_fixits): New method.
19762 (layout::print_any_fixits): Rename to...
19763 (layout::print_trailing_fixits): ...this, and make private.
19764 Don't print newline fixits.
19765 (diagnostic_show_locus): Move logic for printing one row into...
19766 (layout::print_line): ...this new function. Move the
19767 location_get_source_line call and error-handling from
19768 print_source_line to here. Call print_leading_fixits, and rename
19769 print_any_fixits to print_trailing_fixits.
19770 (selftest::test_fixit_insert_containing_newline): Update now that
19771 newlines are partially supported.
19772 (selftest::test_fixit_insert_containing_newline_2): New test.
19773 (selftest::test_fixit_replace_containing_newline): Update comments.
19774 (selftest::diagnostic_show_locus_c_tests): Call the new test.
19775 * edit-context.c (class added_line): New class.
19776 (class edited_line): Describe newline handling in comment.
19777 (edited_line::actually_edited_p): New method.
19778 (edited_line::print_content): Delete redundant decl.
19779 (edited_line::m_predecessors): New field.
19780 (edited_file::print_content): Call edited_line::print_content.
19781 (edited_file::print_diff): Update to support newlines.
19782 (edited_file::print_diff_hunk): Likewise.
19783 (edited_file::print_run_of_changed_lines): New function.
19784 (edited_file::print_diff_line): Convert to...
19785 (print_diff_line): ...this.
19786 (edited_file::get_effective_line_count): New function.
19787 (edited_line::edited_line): Initialize new field m_predecessors.
19788 (edited_line::~edited_line): Clean up m_predecessors.
19789 (edited_line::apply_fixit): Handle newlines.
19790 (edited_line::get_effective_line_count): New function.
19791 (edited_line::print_content): New function.
19792 (edited_line::print_diff_lines): New function.
19793 (selftest::test_applying_fixits_insert_containing_newline): New
19795 (selftest::test_applying_fixits_replace_containing_newline): New
19797 (selftest::insert_line): New function.
19798 (selftest::test_applying_fixits_multiple_lines): Add example of
19800 (selftest::edit_context_c_tests): Call the new tests.
19802 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19804 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
19805 parameter cand. Update dump information.
19806 (get_computation_cost): Update uses.
19808 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19810 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
19811 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
19812 (get_computation_at, rewrite_use_address): Update use of
19813 get_computation_aff.
19815 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19817 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
19818 (get_computation): Delete.
19819 (get_computation_cost): Implement like get_computation_cost_at.
19820 Use get_computation_at.
19821 (get_computation_cost_at): Delete.
19822 (rewrite_use_nonlinear_expr): Use get_computation_at.
19823 (rewrite_use_compare, remove_unused_ivs): Ditto.
19825 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19827 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
19829 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19831 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
19832 (ivopts_global_cost_for_size): Rename parameter and update uses.
19833 (iv_ca_recount_cost): Update uses.
19834 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
19835 candidates seperately in n_invs and n_cands.
19836 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
19838 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19840 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
19841 (find_inv_vars_cb): New.
19842 (find_depends): Renamed to ...
19843 (find_inv_vars): ... this.
19844 (add_candidate_1, force_var_cost): Call find_inv_vars.
19845 (split_address_cost, determine_group_iv_cost_cond): Ditto.
19847 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19849 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
19850 inv_vars. Add inv_exprs.
19851 (struct iv_cand): Rename depends_on to inv_vars.
19852 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
19853 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
19854 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
19855 (dump_cand): Dump inv_vars.
19856 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
19857 (record_invariant, find_depends, add_candidate_1): Ditto.
19858 (set_group_iv_cost, force_var_cost): Ditto.
19859 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
19860 (get_computation_cost_at, get_computation_cost): Ditto.
19861 (determine_group_iv_cost_generic): Ditto.
19862 (determine_group_iv_cost_address): Ditto.
19863 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
19864 (determine_group_iv_costs): Ditto.
19865 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
19866 (iv_ca_set_remove_invariants): Renamed to ...
19867 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
19868 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
19869 (iv_ca_set_add_invariants): Renamed to ...
19870 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
19871 (iv_ca_set_cp): Use iv_ca_set_add_invs.
19872 (iv_ca_has_deps): Support inv_vars and inv_exprs.
19873 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
19874 (create_new_ivs): Remove useless dump.
19876 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19878 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
19880 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
19881 (iv_ca_set_no_cp, create_new_iv): Ditto.
19883 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19885 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19887 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
19889 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
19890 function tree_check2.
19892 2017-05-02 Martin Liska <mliska@suse.cz>
19894 * doc/gcov.texi: Add missing preposition.
19895 * gcov.c (function_info::function_info): Properly fill up
19896 all member variables.
19898 2017-05-02 Tamar Christina <tamar.christina@arm.com>
19900 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
19902 2017-05-02 Tamar Christina <tamar.christina@arm.com>
19904 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
19906 2017-05-02 Martin Liska <mliska@suse.cz>
19909 * lto-streamer-in.c (lto_read_tree_1): Remove
19910 LTO_STREAMER_DEBUG.
19911 * lto-streamer.c (struct tree_hash_entry): Likewise.
19912 (struct tree_entry_hasher): Likewise.
19913 (tree_entry_hasher::hash): Likewise.
19914 (tree_entry_hasher::equal): Likewise.
19915 (lto_streamer_init): Likewise.
19916 (lto_orig_address_map): Likewise.
19917 (lto_orig_address_get): Likewise.
19918 (lto_orig_address_remove): Likewise.
19919 * lto-streamer.h: Likewise.
19920 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
19921 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
19923 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
19925 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
19926 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
19927 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
19928 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
19929 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
19930 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
19931 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
19932 (mm_maskz_sub_ss): New intrinsics.
19933 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19934 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
19935 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
19936 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
19937 (__builtin_ia32_subss_mask_round): New builtins.
19938 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19939 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
19940 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
19942 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
19943 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
19945 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
19948 2017-05-02 Martin Jambor <mjambor@suse.cz>
19950 PR tree-optimization/78687
19951 * tree-sra.c (access): New field parent.
19952 (process_subtree_disqualification): New function.
19953 (disqualify_candidate): Call it.
19954 (build_accesses_from_assign): Reset write flag if creating an
19956 (build_access_subtree): Fill in parent field and also prpagate
19957 down grp_write flag.
19958 (create_artificial_child_access): New parameter set_grp_write, set
19959 grp_write to its value.
19960 (propagate_subaccesses_across_link): Also propagate grp_write flag
19962 (propagate_all_subaccesses): Push the closest parent back to work
19963 queue if add_access_to_work_queue returned true.
19965 2017-05-02 Richard Biener <rguenther@suse.de>
19967 * common.opt (fstrict-overflow): Alias negative to fwrapv.
19968 * doc/invoke.texi (fstrict-overflow): Remove all traces of
19969 -fstrict-overflow documentation.
19970 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
19971 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
19972 flag_strict_overflow.
19973 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
19974 * lto-opts.c (lto_write_options): Do not stream it.
19975 * lto-wrapper.c (merge_and_complain): Do not handle it.
19976 * opts.c (default_options_table): Do not set -fstrict-overflow.
19977 (finish_options): Likewise do not clear it when sanitizing.
19978 * simplify-rtx.c (simplify_const_relational_operation): Do not
19979 test flag_strict_overflow.
19981 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
19983 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
19984 using enabled attribute.
19985 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
19986 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
19987 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
19988 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
19989 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
19990 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
19991 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
19992 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
19993 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
19994 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
19996 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
19998 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
20000 2017-05-02 Richard Biener <rguenther@suse.de>
20002 PR tree-optimization/80591
20004 2017-04-10 Richard Biener <rguenther@suse.de>
20006 * tree-ssa-structalias.c (find_func_aliases): Properly handle
20009 2017-05-02 Richard Biener <rguenther@suse.de>
20011 PR tree-optimization/80549
20012 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
20013 (cleanup_tree_cfg_noloop): Create forwarders to known loop
20014 headers if they do not have a preheader.
20016 2017-05-02 Martin Liska <mliska@suse.cz>
20019 * common.opt: Fix typo.
20020 * doc/invoke.texi: Likewise.
20022 2017-05-01 Jan Beulich <jbeulich@suse.com>
20024 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
20025 swapping, add (x,x,m,x,n) alternative.
20027 2017-05-01 Nathan Sidwell <nathan@acm.org>
20029 * calls.c (combine_pending_stack_adjustment_and_call): Remove
20030 unnecessary unadjusted_alignment check.
20032 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
20035 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
20037 * gimplify.c (gimplify_cilk_detach): New function.
20038 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
20039 * tree-core.h: Document EXPR_CILK_SPAWN.
20040 * tree.h (EXPR_CILK_SPAWN): Define.
20042 2017-05-01 David Malcolm <dmalcolm@redhat.com>
20044 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
20045 to use new fixit_hint representation, using the "replace" logic.
20046 (get_line_span_for_fixit_hint): Likewise.
20047 (layout::print_any_fixits): Likewise.
20048 (selftest::test_one_liner_many_fixits): Rename to...
20049 (selftest::test_one_liner_many_fixits_1): ...this, and update
20050 comment and expected output to reflect that the multiple fix-it
20051 hints are now consolidated into one insertion.
20052 (selftest::test_one_liner_many_fixits_2): New test.
20053 (selftest::test_diagnostic_show_locus_one_liner): Update for
20055 (selftest::test_fixit_consolidation): Update for fix-it API
20057 * diagnostic.c (print_parseable_fixits): Likewise.
20058 * edit-context.c (edited_line::m_line_events): Convert from
20059 auto_vec <line_event *> to auto_vec <line_event>.
20060 (class line_event): Convert from abstract base class to a concrete
20061 class, taking over the role of replace_event.
20062 (class insert_event): Delete.
20063 (class replace_event): Rename to class line_event. Convert to
20065 (edit_context::add_fixits): Reimplement.
20066 (edit_context::apply_insert): Delete.
20067 (edit_context::apply_replace): Rename to...
20068 (edit_context::apply_fixit): ...this. Convert to half-open range.
20069 (edited_file::apply_insert): Delete.
20070 (edited_file::apply_replace): Rename to...
20071 (edited_file::apply_fixit): ...this.
20072 (edited_line::~edited_line): Drop deletion of events.
20073 (edited_line::apply_insert): Delete.
20074 (edited_line::apply_replace): Rename to...
20075 (edited_line::apply_fixit): ...this. Convert to half-open range.
20076 Update for change to type of m_line_events.
20077 * edit-context.h (edit_context::apply_insert): Delete.
20078 (edit_context::apply_replace): Rename to...
20079 (edit_context::apply_fixit): ...this.
20081 2017-05-01 Martin Sebor <msebor@redhat.com>
20083 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
20086 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
20089 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
20090 __get_cpuid_max returns 0.
20091 (__get_cpuid_count): Ditto.
20093 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
20095 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
20096 replacement expression is another instance of one of its arguments.
20098 2017-05-01 Jakub Jelinek <jakub@redhat.com>
20101 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
20102 check for stack push/pop autoinc.
20103 * config/i386/i386.c (ix86_agi_dependent): Return false
20104 if the only reason why modified_in_p returned true is that
20105 addr is SP based and set_insn is a push or pop.
20107 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
20109 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
20112 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
20115 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
20116 (account_size_time): Use two predicates - exec_pred and
20118 (evaluate_conditions_for_known_args): Compute both clause and
20120 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
20121 (inline_summary_t::duplicate): Update.
20122 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
20124 (compute_inline_parameters): Likewise.
20125 (estimate_edge_size_and_time): Update caluclation of time.
20126 (estimate_node_size_and_time): Compute both time and nonspecialized
20128 (estimate_ipcp_clone_size_and_time): Update.
20129 (inline_merge_summary): Update.
20130 (do_estimate_edge_time): Update.
20131 (do_estimate_edge_size): Update.
20132 (do_estimate_edge_hints): Update.
20133 (inline_read_section, inline_write_summary): Stream both new predicates.
20134 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
20136 (compute_inlined_call_time): Cleanup.
20137 (big_speedup_p): Update.
20138 (edge_badness): Update.
20139 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
20140 (size_time_entry): Replace predicate by exec_predicate and
20141 nonconst_predicate.
20142 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
20143 (estimate_edge_time): Return also nonspec_time.
20144 (reset_edge_growth_cache): Update.
20146 2017-04-29 Jakub Jelinek <jakub@redhat.com>
20148 PR rtl-optimization/80491
20149 * ifcvt.c (noce_process_if_block): When looking for x setter
20150 with missing else_bb, don't check only the insn right before
20151 cond_earliest, but look for the last insn that x is modified in
20152 within the same bb.
20154 PR rtl-optimization/80491
20155 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
20157 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
20159 PR tree-optimization/80487
20160 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
20162 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20164 PR tree-optimization/79697
20165 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
20166 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
20167 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
20169 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
20170 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
20172 2017-04-28 Martin Sebor <msebor@redhat.com>
20174 PR tree-optimization/80523
20175 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
20176 (init_target_to_host_charmap, target_to_host, target_strtol10): New
20178 (maybe_warn, format_directive, parse_directive): Use new functions.
20179 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
20181 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
20183 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
20185 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
20187 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
20188 target_header_dir): Set correctly.
20189 * configure: Regenerated.
20190 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
20191 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
20192 instead of SYSTEM_HEADER_DIR.
20194 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
20196 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
20197 (estimate_local_effects): Likewise.
20198 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
20199 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
20200 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
20201 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
20202 do_estimate_edge_time, estimate_edge_time): Likewise.
20203 * ipa-inline-analysis.c (estimate_node_size_and_time,
20204 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
20205 (estimate_time_after_inlining): Remove.
20207 2017-04-28 Martin Liska <mliska@suse.cz>
20209 * doc/gcov.texi: Enhance documentation of gcov.
20211 2017-04-28 Martin Liska <mliska@suse.cz>
20213 * doc/gcov.texi: Sort options in alphabetic order.
20214 * doc/gcov-dump.texi: Likewise.
20215 * doc/gcov-tool.texi: Likewise.
20216 * gcov.c (print_usage): Likewise.
20217 * gcov-dump.c (print_usage): Likewise.
20218 * gcov-tool.c (print_merge_usage_message): Likewise.
20219 (print_rewrite_usage_message): Likewise.
20220 (print_overlap_usage_message): Likewise.
20222 2017-04-28 Martin Liska <mliska@suse.cz>
20224 PR gcov-profile/53915
20225 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
20227 2017-04-28 Martin Liska <mliska@suse.cz>
20229 PR gcov-profile/79891
20230 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
20231 is marked by compiler as living on a line.
20232 (get_cycles_count): Remove usage of the union.
20233 (output_intermediate_file): Likewise.
20234 (find_source): Fix GNU coding style.
20235 (accumulate_line_counts): Remove old non-all block mode.
20236 (output_lines): Remove usage of the union.
20237 * profile.c (output_location): Include all BBs, even if
20238 belonging to a same line (and file) as a previous BB.
20240 2017-04-28 Martin Liska <mliska@suse.cz>
20242 * gcov.c (process_args): Handle new argument 'w'.
20243 (read_graph_file): Assign ID to BBs.
20244 (output_branch_count): Display BB # if verbose flag is set.
20245 (output_lines): Likewise for arcs.
20246 (print_usage): Add '--verbose' option help.
20247 * doc/gcov.texi: Document --verbose (-w) option.
20249 2017-04-28 Martin Liska <mliska@suse.cz>
20251 * gcov.c (struct block_location_info): New struct.
20252 (process_file): Fill up the new structure.
20253 (read_graph_file): Replace usage of encoding by the newly added
20255 (add_line_counts): Likewise.
20256 (accumulate_line_counts): Remove usage of the union.
20257 (function_info::function_info): New function.
20258 (function_info::~function_info): Likewise.
20259 (process_file): Call delete instead of release_function.
20260 (release_function): Release the function.
20261 (release_structures): Call delete instead of release_function.
20262 (solve_flow_graph): Replace usage of num_blocks.
20263 (find_exception_blocks): Likewise.
20264 (output_lines): Fix GNU coding style.
20266 2017-04-28 Martin Liska <mliska@suse.cz>
20269 * coverage.c (coverage_remove_note_file): New function.
20270 * coverage.h: Declare the function.
20271 * toplev.c (finalize): Clean if an error has been seen.
20273 2017-04-28 Martin Liska <mliska@suse.cz>
20275 PR gcov-profile/80031
20276 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
20277 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
20278 * gcov.c (read_graph_file): Read just number of blocks.
20279 * profile.c (branch_prob): Do not stream 0 flags per a basic
20282 2017-04-28 Martin Liska <mliska@suse.cz>
20284 * gcov-dump.c (tag_*): Add new argument to declarations.
20285 (dump_gcov_file): Likewise.
20286 (tag_blocks): Add and use new argument depth.
20287 (tag_arcs): Likewise.
20288 (tag_lines): Likewise.
20289 (tag_counters): Likewise.
20290 (tag_summary): Likewise.
20291 (dump_working_sets): Use depth to do a proper indentation.
20293 2017-04-28 Jakub Jelinek <jakub@redhat.com>
20296 * cgraph.h (symtab_node::debug_symtab): No longer inline.
20297 * symtab.c (symtab_node::debug_symtab): Move definition here.
20299 2017-04-28 Richard Biener <rguenther@suse.de>
20301 * lto-streamer.h (LTO_major_version): Bump to 7.
20303 2017-04-28 Richard Biener <rguenther@suse.de>
20305 * tree-vrp.c (assert_info): New struct.
20306 (add_assert_info): New helper.
20307 (register_edge_assert_for_2): Refactor to add asserts to a vector
20309 (register_edge_assert_for_1): Likewise.
20310 (register_edge_assert_for): Likewise.
20311 (finish_register_edge_assert_for): New helper actually registering
20312 asserts where live on edge.
20313 (find_conditional_asserts): Adjust.
20314 (find_switch_asserts): Likewise.
20315 (evrp_dom_walker::try_find_new_range): Generalize.
20316 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
20318 2017-04-27 Marek Polacek <polacek@redhat.com>
20321 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
20322 arg10 and arg11 to itype.
20324 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
20326 * doc/extend.texi (Object Size Checking): Improve grammar.
20328 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
20331 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
20332 that the logic for permitting reciprocal estimates matches that
20335 2017-04-27 Jakub Jelinek <jakub@redhat.com>
20338 * tree.c (type_cache_hasher::equal): Only compare
20339 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
20340 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
20341 non-aggregate element types.
20342 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
20343 about the flag on ARRAY_TYPEs in the comment, formatting fix.
20345 2017-04-27 Richard Biener <rguenther@suse.de>
20347 PR middle-end/80533
20348 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
20349 stripping ARRAY_REFs from MEM_EXPR make sure we're not
20350 keeping a reference to a trailing array.
20352 2017-04-27 Richard Biener <rguenther@suse.de>
20354 PR middle-end/80539
20355 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
20356 being in loop-closed SSA form conservatively.
20357 (chrec_fold_multiply_poly_poly): Likewise.
20359 2017-04-27 Tamar Christina <tamar.christina@arm.com>
20361 PR middle-end/79665
20362 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
20363 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
20365 2017-04-27 Jakub Jelinek <jakub@redhat.com>
20368 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
20369 (aarch64_function_arg_alignment): Return unsigned int again, but still
20370 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
20371 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
20372 Don't emit -Wpsabi note.
20373 (aarch64_function_arg_boundary): Likewise.
20374 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
20377 2017-04-26 Nathan Sidwell <nathan@acm.org>
20379 * tree.h (crc32_unsigned_n): Declare.
20380 (crc32_unsigned, crc32_unsigned): Make inline.
20381 * tree.c (crc32_unsigned_bits): Replace with ...
20382 (crc32_unsigned_n): ... this.
20383 (crc32_unsigned, crc32_byte): Remove.
20384 (crc32_string): Remove unnecessary braces.
20386 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
20388 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
20389 * ipa-inline-analysis.c (MAX_TIME): Remove.
20390 (account_size_time): Use sreal for time.
20391 (dump_inline_summary): Update.
20392 (estimate_function_body_sizes): Update.
20393 (estimate_edge_size_and_time): Update.
20394 (estimate_calls_size_and_time): Update.
20395 (estimate_node_size_and_time): Update.
20396 (inline_merge_summary): Update.
20397 (inline_update_overall_summary): Update.
20398 (estimate_time_after_inlining): Update.
20399 (inline_read_section): Update.
20400 (inline_write_summary): Update.
20401 * ipa-inline.c (compute_uninlined_call_time): Update.
20402 (compute_inlined_call_time): Update.
20403 (recursive_inlining): Update.
20404 (inline_small_functions): Update.
20405 (dump_overall_stats): Update.
20406 * ipa-inline.h: Include sreal.h.
20407 (size_time_entry): Turn time to sreal.
20408 (inline_summary): Turn self_time nad time to sreal.
20410 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
20412 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
20414 (sreal::stream_out, sreal::stream_in): New.
20415 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
20417 2017-04-25 Jakub Jelinek <jakub@redhat.com>
20419 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
20422 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
20425 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
20427 (dimode_scalar_chain::compute_convert_gain): Ditto.
20428 (dimode_scalar_chain::make_vector_copies): Ditto.
20429 (dimode_scalar_chain::convert_reg): Ditto.
20430 (dimode_scalar_chain::convert_insn): Ditto.
20431 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
20432 (VI248_AVX512BW_1): New mode iterator.
20433 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
20434 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
20437 2017-04-25 Martin Sebor <msebor@redhat.com>
20439 PR tree-optimization/80497
20440 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
20441 constants are representable in HOST_WIDE_INT.
20442 (parse_directive): Ditto.
20444 2017-04-25 Martin Sebor <msebor@redhat.com>
20447 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
20448 (new_zero_array): Adjust signature.
20449 (dom_info::dom_init): Used unsigned rather that size_t.
20450 (dom_info::dom_info): Same.
20452 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20453 Jakub Jelinek <jakub@redhat.com>
20456 * config/arm/arm.c: Include gimple.h.
20457 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
20458 returns negative, increment ncrn only if it returned positive.
20459 (arm_needs_doubleword_align): Return int instead of bool,
20460 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
20461 members, but if there is any such non-FIELD_DECL
20462 > PARM_BOUNDARY aligned decl, return -1 instead of false.
20463 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
20464 returns negative, increment nregs only if it returned positive.
20465 (arm_setup_incoming_varargs): Likewise.
20466 (arm_function_arg_boundary): Emit -Wpsabi note if
20467 arm_needs_doubleword_align returns negative, return
20468 DOUBLEWORD_ALIGNMENT only if it returned positive.
20470 2017-04-25 Marek Polacek <polacek@redhat.com>
20473 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
20474 first argument to type.
20476 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
20479 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
20480 type checks to test for compatibility instead of equality.
20482 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20483 Jakub Jelinek <jakub@redhat.com>
20486 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
20488 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
20489 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
20490 the alignment computation, but return their maximum in warn_alignment.
20491 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
20492 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
20494 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
20495 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
20498 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20500 * config/arc/simdext.md (dmpyh): Fix typo.
20502 2017-04-25 Richard Biener <rguenther@suse.de>
20504 PR tree-optimization/80492
20505 * alias.c (compare_base_decls): Handle registers with asm
20506 specification conservatively.
20507 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
20508 compare_base_decls returning dont-know properly.
20510 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20512 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
20513 (legitimate_offset_address_p): New function.
20514 (arc_legitimate_address_p): Use above function.
20516 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20518 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
20520 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20522 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
20523 ACCH registers whenever they are available.
20525 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20527 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
20528 double regs fix when not used.
20530 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20532 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
20534 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
20535 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
20537 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20539 * config/arc/arc.c (arc_output_addsi): Check for h-register class
20540 when emitting short ADD instructions.
20542 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20544 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
20546 (cmpsi_cc_c_insn): Likewise.
20547 (cbranchsi4_scratch): Compute proper instruction length using
20548 compact_hreg_operand.
20549 * config/arc/predicates.md (compact_hreg_operand): New predicate.
20551 2017-04-25 Richard Biener <rguenther@suse.de>
20553 PR middle-end/80509
20554 * passes.c (pass_manager::pass_manager): Initialize
20555 m_name_to_pass_map.
20557 2017-04-25 Richard Biener <rguenther@suse.de>
20559 PR tree-optimization/79201
20560 * tree-ssa-sink.c (statement_sink_location): Handle calls.
20562 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20565 * config/s390/vector.md: Split MEM->GPR vector moves for
20566 non-s_operand addresses.
20568 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20571 * config/s390/predicates.md (reload_const_wide_int_operand): New
20573 * config/s390/s390.md ("movti"): Remove d/P alternative.
20574 ("movti_bigconst"): New pattern definition.
20576 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20579 * s390-protos.h (s390_expand_cs_hqi): Removed.
20580 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
20581 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
20582 modes as well as CCZ1mode and CCZmode.
20583 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
20584 signature of s390_emit_compare_and_swap.
20585 (s390_expand_cs_hqi): Likewise, make static.
20586 (s390_expand_cs_tdsi): Generate an explicit compare before trying
20587 compare-and-swap, in some cases.
20588 (s390_expand_cs): Wrapper function.
20589 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
20591 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
20592 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
20593 patterns for small and large integers. Forbid symref memory operands.
20594 Move expander to s390.c. Require cc register.
20595 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
20596 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
20597 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
20598 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
20599 symref memory operands. Remove CC mode and call s390_match_ccmode
20601 ("atomic_exchange<mode>"): Allow and implement all integer modes.
20603 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20605 * config/s390/s390.md (define_peephole2): New peephole to help
20606 combining the load-and-test pattern with volatile memory.
20608 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20610 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
20611 with CCZmode for TARGET_Z196.
20613 2017-04-25 Jakub Jelinek <jakub@redhat.com>
20615 PR rtl-optimization/80501
20616 * combine.c (make_compound_operation_int): Set subreg_code to SET
20617 even for AND with mask of the sign bit of mode.
20619 PR rtl-optimization/80500
20620 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
20621 sum's initial value.
20623 2017-04-25 Julian Brown <julian@codesourcery.com>
20624 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20626 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
20628 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
20630 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
20632 2017-04-25 Julian Brown <julian@codesourcery.com>
20633 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20635 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
20636 (thunderx2t99_sha): New Reservation.
20638 2017-04-25 Julian Brown <julian@codesourcery.com>
20639 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20641 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
20642 type for 1-element load.
20644 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
20646 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
20648 2017-04-24 Martin Jambor <mjambor@suse.cz>
20650 PR tree-optimization/80293
20651 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
20652 char arrays not totally scalarizable if it is false.
20653 (analyze_all_variable_accesses): Pass correct value in the new
20654 parameter. Add a statistics counter.
20656 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
20658 PR middle-end/79931
20659 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
20661 2017-04-24 Richard Biener <rguenther@suse.de>
20663 PR tree-optimization/80494
20664 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
20665 out for complex types.
20667 2017-04-24 Richard Biener <rguenther@suse.de>
20669 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
20670 * tree-ssa-sccvn.c (print_scc): Print SCC size.
20671 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
20672 (DFS): Adjust and never fail.
20673 (sccvn_dom_walker::fail): Remove.
20674 (sccvn_dom_walker::before_dom_children): Adjust.
20675 (run_scc_vn): Likewise and never fail.
20676 * tree-ssa-pre.c (pass_pre::execute): Adjust.
20677 (pass_fre::execute): Likewise.
20679 2017-04-24 Richard Biener <rguenther@suse.de>
20681 PR tree-optimization/79725
20682 * tree-ssa-sink.c (statement_sink_location): Return whether
20683 failure reason was zero uses. Move that check later.
20684 (sink_code_in_bb): Deal with zero uses by removing the stmt
20687 2017-04-24 Richard Biener <rguenther@suse.de>
20690 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
20691 pointer-based references.
20693 2017-04-24 Richard Biener <rguenther@suse.de>
20696 * pass_manager.h (pass_manager::operator new): Remove.
20697 (pass_manager::operator delete): Likewise.
20698 * passes.c (pass_manager::operator new): Remove.
20699 (pass_manager::operator delete): Likewise.
20700 (pass_manager::pass_manager): Zero individual pass members.
20702 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
20705 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
20706 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
20707 Check "XEXP (src, 1)" operand here.
20708 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
20709 Check "XEXP (src, 1)" operand here.
20710 (dimode_scalar_chain::make_vector_copies): Detect count register
20711 of a shift instruction. Zero extend count register from QImode
20712 to DImode to satisfy vector shift pattern count operand predicate.
20713 Substitute vector shift count operand with a DImode copy.
20714 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
20717 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
20719 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
20720 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
20721 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
20722 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
20723 (UNSPEC_NOREX_MEM): Remove definition.
20725 2017-04-21 Richard Biener <rguenther@suse.de>
20727 PR tree-optimization/79547
20728 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
20729 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
20730 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
20731 without any constraints.
20733 2017-04-21 Richard Biener <rguenther@suse.de>
20735 PR tree-optimization/78847
20736 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
20738 2017-04-21 Richard Biener <rguenther@suse.de>
20740 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
20741 (build_distinct_type_copy): Likewise.
20742 (build_variant_type_copy): Likewise.
20743 * tree.c (build_qualified_type): Pass down mem-stat info.
20744 (build_distinct_type_copy): Likewise.
20745 (build_variant_type_copy): Likewise.
20747 2017-04-21 Richard Biener <rguenther@suse.de>
20749 PR tree-optimization/80237
20750 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
20752 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
20753 for a simplified result.
20755 2016-04-21 Richard Biener <rguenther@suse.de>
20757 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
20758 sth as strict as a simple_iv but a chrec without symbols and an
20759 operand defined in the loop we are peeling (and not some subloop).
20760 (propagate_constants_for_unrolling): Propagate all constants.
20762 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
20765 * config/i386/i386.c (print_reg): Remove assert for disalowed
20766 regno values, call output_operand_lossage instead.
20768 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
20771 * config/i386/constraints.md (Yc): New register constraint.
20772 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
20773 Use Yc constraint for alternative 2 of operand 0. Remove
20774 preferred_for_speed attribute.
20776 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
20778 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
20779 lastprivate clauses in SIMT case.
20781 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
20783 * doc/invoke.texi (-Wextra-semi): Document new warning option.
20785 2017-04-20 Richard Biener <rguenther@suse.de>
20787 PR tree-optimization/57796
20788 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
20789 as N scalar stores.
20790 (vect_model_load_cost): Cost gathers as N scalar loads.
20792 2017-04-20 Richard Biener <rguenther@suse.de>
20794 * ggc-page.c (ggc_allocated_p): Rename to ...
20795 (safe_lookup_page_table_entry): ... this and return the lookup
20797 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
20799 2017-04-20 Richard Biener <rguenther@suse.de>
20801 PR tree-optimization/80453
20802 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
20803 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
20804 from the conditions.
20805 (vn_phi_eq): Pass them down.
20806 (vn_phi_lookup): Record them.
20807 (vn_phi_insert): Likewise.
20809 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
20811 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
20812 uninitialized variable warning to avoid buffer overrun.
20814 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
20817 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
20818 is suppressed for '{ 0 }' in C.
20820 2017-04-20 Jakub Jelinek <jakub@redhat.com>
20822 * BASE-VER: Set to 8.0.0.
20824 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
20826 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
20827 priority .init_array and .fini_array section with SECTION_NOTYPE
20830 2017-04-20 Jakub Jelinek <jakub@redhat.com>
20832 PR middle-end/80423
20833 * tree.h (build_array_type): Add typeless_storage default argument.
20834 * tree.c (type_cache_hasher::equal): Also compare
20835 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
20836 (build_array_type): Add typeless_storage argument, set
20837 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
20839 (build_nonshared_array_type): Adjust build_array_type_1 caller.
20840 (build_array_type): Likewise. Add typeless_storage argument.
20842 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
20843 Jakub Jelinek <jakub@redhat.com>
20845 PR tree-optimization/80426
20846 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
20847 operation on symbolic operands, also compute the overflow for the
20848 invariant part when the operation degenerates into a negation.
20850 2017-04-19 Jakub Jelinek <jakub@redhat.com>
20853 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
20854 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
20857 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
20859 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
20862 * config/avr/avr.c (tree.h): Include it.
20863 (cgraph.h): Include it.
20864 (avr_encode_section_info): Don't warn for uninitialized progmem
20865 variable if it's just an alias.
20867 2017-04-19 Richard Biener <rguenther@suse.de>
20870 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
20871 when needed by AutoPGO.
20873 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
20876 * doc/lto.texi: Remove an extra 'that'.
20878 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
20880 PR rtl-optimization/80429
20881 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
20882 are only used in debug insns.
20884 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
20885 Vladimir Makarov <vmakarov@redhat.com>
20887 * config/sparc/predicates.md (input_operand): Add comment. Return
20888 true for any memory operand when LRA is in progress.
20889 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
20891 2017-04-18 Jeff Law <law@redhat.com>
20894 * mips.md ({return,simple_return}_internal): Do not overwrite
20897 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20899 PR tree-optimization/80443
20900 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
20901 instead of adding 1, subtract -1 and similarly instead of subtracting
20904 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
20906 PR rtl-optimization/80357
20907 * haifa-sched.c (tmp_bitmap): New variable.
20908 (model_recompute): Handle duplicate use records.
20909 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
20910 (free_global_sched_pressure_data): Free it.
20912 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
20915 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
20916 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
20917 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
20918 instead of SYSTEM_HEADER_DIR.
20920 2017-04-18 Jeff Law <law@redhat.com>
20922 PR middle-end/80422
20923 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
20924 predecessors after walking up the insn chain.
20926 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20929 * dwarf2out.c (modified_type_die): Try harder not to emit internal
20930 sizetype type into debug info.
20932 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
20935 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
20936 unneeded test for TARGET_UPPER_REGS_SF.
20937 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
20939 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20942 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
20943 instead of gsi_after_labels.
20945 2017-04-18 Jeff Law <law@redhat.com>
20947 * regcprop.c (maybe_mode_change): Avoid creating copies of the
20951 2017-04-13 Jeff Law <law@redhat.com>
20952 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
20953 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
20955 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
20958 * config/avr/avr.c (intl.h): Include it.
20959 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
20961 2017-04-18 Martin Liska <mliska@suse.cz>
20963 PR gcov-profile/78783
20964 * gcov-tool.c (gcov_output_files): Validate that destination
20965 file is either removed by the tool or by a user.
20967 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
20968 Guy Benyei <guybe@mellanox.com>
20970 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
20971 block, and do not negate it, the stored id is already negative.
20973 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
20975 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
20977 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
20980 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
20981 masks of options that should be turned off if the VSX vector
20982 options are turned off.
20983 (OTHER_P8_VECTOR_MASKS): Likewise.
20984 (OTHER_VSX_VECTOR_MASKS): Likewise.
20985 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
20986 rs6000_disable_incompatible_switches to validate no type switches
20988 (rs6000_incompatible_switch): New function to disallow turning on
20989 other vector options if -mno-vsx, -mno-power8-vector, or
20990 -mno-power9-vector are specified.
20992 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
20994 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
20996 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
20998 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
20999 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
21000 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
21001 (ARG_POINTER_CFA_OFFSET): Likewise.
21003 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21005 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
21006 conditions to take advantage of various optimizations.
21008 2017-04-13 Jeff Law <law@redhat.com>
21010 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
21011 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
21012 (zero_extendsidi2_dext): Likewise.
21014 2017-04-13 Jakub Jelinek <jakub@redhat.com>
21017 * fold-const.c (fold_ternary_loc): Revert
21018 use op0 instead of fold_convert_loc (loc, type, arg0) part of
21021 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
21023 PR rtl-optimization/80343
21024 * lra-remat.c (update_scratch_ops): Assign original hard reg to
21025 new scratch pseudo.
21027 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
21030 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
21031 to ubsan_encode_value.
21033 2017-04-13 Jeff Law <law@redhat.com>
21035 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
21036 appearing in DEBUG_INSNs.
21038 2017-04-13 Martin Liska <mliska@suse.cz>
21040 PR gcov-profile/80413
21041 * gcov-io.c (gcov_write_string): Copy to buffer just when
21042 allocated size is greater than zero.
21044 2017-04-13 Jakub Jelinek <jakub@redhat.com>
21047 * dwarf2out.c (decls_for_scope): Ignore declarations of
21048 current_function_decl in BLOCK_NONLOCALIZED_VARS.
21050 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
21053 * ipa-visibility.c (non_local_p): Fix typos.
21054 (localize_node): When localizing symbol in same comdat group,
21055 dissolve the group only when we know external symbols are going
21057 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
21059 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21061 PR tree-optimization/79390
21062 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
21063 order does not result in usable sequence, retry with reversed operand
21069 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
21070 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
21071 op0 instead of fold_convert_loc (loc, type, arg0).
21073 2017-04-12 Jeff Law <law@redhat.com>
21075 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
21076 has a delay slot in the generated code.
21078 * config/cris/cris.md (cris_preferred_reload_class): Return
21079 GENNONACR_REGS rather than GENERAL_REGS.
21081 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21084 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
21085 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
21086 signedness of the result type.
21088 2017-04-12 Richard Biener <rguenther@suse.de>
21089 Jeff Law <law@redhat.com>
21091 PR tree-optimization/80359
21092 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
21093 trim stores to TARGET_MEM_REFs.
21095 2017-04-12 Richard Biener <rguenther@suse.de>
21097 PR tree-optimization/79390
21098 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
21099 threading case even more.
21101 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
21104 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
21105 for quad_address_p for TImode, instead of just not indexed_address.
21107 2017-04-12 Richard Biener <rguenther@suse.de>
21108 Bernd Edlinger <bernd.edlinger@hotmail.de>
21110 PR middle-end/79671
21111 * alias.c (component_uses_parent_alias_set_from): Handle
21112 TYPE_TYPELESS_STORAGE.
21113 (get_alias_set): Likewise.
21114 * tree-core.h (tree_type_common): Add typeless_storage flag.
21115 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
21116 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
21117 for types containing members with TYPE_TYPELESS_STORAGE.
21118 (place_field): Likewise.
21119 (layout_type): Likewise for ARRAY_TYPE.
21120 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
21121 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
21122 TYPE_TYPELESS_STORAGE.
21123 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
21125 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21128 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
21129 first argument to type.
21131 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21135 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
21136 CONST0_RTX (mode) rather than const0_rtx where appropriate.
21137 (rs6000_expand_binop_builtin): Likewise.
21138 (rs6000_expand_ternop_builtin): Likewise; also add missing
21139 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
21140 vshasigma built-ins.
21141 * doc/extend.texi: Document that vec_xxpermdi's third argument
21142 must be a constant.
21144 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
21146 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
21147 Use shift_const cost parameter when calculating gain of STV shifts.
21149 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
21151 PR rtl-optimization/70478
21152 * lra-constraints.c (process_alt_operands): Check memory for
21153 disfavoring memory insn operand.
21155 2017-04-11 Jakub Jelinek <jakub@redhat.com>
21157 PR middle-end/80100
21158 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
21159 left shift in unsigned HOST_WIDE_INT type.
21161 PR rtl-optimization/80385
21162 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
21163 (not (neg X)) into (plus X -1) for complex or non-integral modes.
21166 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
21167 if they have any depend clauses.
21169 2017-04-11 Martin Liska <mliska@suse.cz>
21172 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
21173 * ipa-split.c (split_function): Create a local comdat symbol
21174 if caller is in a comdat group.
21176 2017-04-11 Martin Liska <mliska@suse.cz>
21179 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
21182 2017-04-11 Martin Sebor <msebor@redhat.com>
21184 PR middle-end/80364
21185 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
21186 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
21188 (directive::set_width, directive::set_precision, format_character):
21190 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
21193 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
21196 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
21197 conflict, set target->arch_name instead of target->cpu_name.
21199 2017-04-11 Richard Biener <rguenther@suse.de>
21201 PR tree-optimization/80374
21202 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
21203 build_zero_cst, remove fold_convertible_p check again.
21205 2017-04-11 Martin Liska <mliska@suse.cz>
21208 * ubsan.c (instrument_object_size): Do not instrument register
21211 2017-04-11 Jakub Jelinek <jakub@redhat.com>
21214 * config/i386/i386-builtin-types.def
21215 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
21216 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
21217 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
21218 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
21219 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
21220 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
21221 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
21222 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
21223 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
21224 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
21225 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
21226 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
21227 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
21228 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
21229 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
21230 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
21231 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
21232 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
21233 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
21234 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
21235 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
21236 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
21237 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
21238 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
21239 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
21240 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
21241 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
21242 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
21243 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
21244 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
21245 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
21246 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
21247 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
21248 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
21249 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
21250 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
21251 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
21252 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
21253 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
21254 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
21255 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
21256 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
21257 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
21258 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
21259 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
21260 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
21262 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
21263 flag to second_arg_count, handle 4 argument function type _COUNT
21264 aliases, handle second_arg_count on second argument rather than last.
21266 2017-04-10 Jeff Law <law@redhat.com>
21268 PR tree-optimization/80374
21269 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
21270 record anything if we can not convert integer_zero_node to the
21273 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21276 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21277 Enhance special handling given to the TARGET_P9_MINMAX option in
21278 relation to certain other options.
21280 2017-04-10 Bin Cheng <bin.cheng@arm.com>
21282 PR tree-optimization/80153
21283 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
21284 remove POINTER_PLUS_EXPR's base part directly, rather than through
21287 2017-04-10 Richard Biener <rguenther@suse.de>
21288 Bin Cheng <bin.cheng@arm.com>
21290 PR tree-optimization/80153
21291 * tree-affine.c (aff_combination_to_tree): Get base pointer from
21292 the first element of pointer type aff_tree. Build result expr in
21294 (add_elt_to_tree): Convert to type unconditionally. Remove other
21295 fold_convert calls.
21296 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
21297 (rewrite_use_nonlinear_expr): Check invariant using iv information.
21299 2017-04-10 Richard Biener <rguenther@suse.de>
21301 * tree-ssa-structalias.c (find_func_aliases): Properly handle
21304 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
21306 PR rtl-optimization/70478
21307 * lra-constraints.c (curr_small_class_check): New.
21308 (update_and_check_small_class_inputs): New.
21309 (process_alt_operands): Update curr_small_class_check. Disfavor
21310 alternative insn memory operands. Check available regs for small
21313 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
21316 * config/mips/mips.opt (-mvirt): Update description.
21317 * doc/invoke.texi (-mvirt): Likewise.
21319 2017-04-10 Richard Biener <rguenther@suse.de>
21321 PR middle-end/80362
21322 * fold-const.c (fold_binary_loc): Look at unstripped ops when
21323 looking for NEGATE_EXPR in -A / -B to A / B folding.
21325 2017-04-10 Martin Liska <mliska@suse.cz>
21327 PR gcov-profile/80224
21328 * gcov.c (print_usage): Fix usage string.
21329 (get_gcov_intermediate_filename): Remove.
21330 (output_gcov_file): Use both for normal and intermediate format.
21331 (generate_results): Do not initialize special file for
21332 intermediate format.
21334 2017-04-10 Richard Biener <rguenther@suse.de>
21336 PR tree-optimization/80304
21337 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
21340 2017-04-10 Nathan Sidwell <nathan@acm.org>
21343 * config/rs6000/rs6000.c (rs6000_vector_type): New.
21344 (rs6000_init_builtins): Use it.
21346 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21348 * config/arm/arm.md (<mrc>): Add mode to SET source.
21349 (<mrrc>): Likewise.
21351 2017-04-10 Richard Biener <rguenther@suse.de>
21353 PR middle-end/80344
21354 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
21356 2017-04-10 Jakub Jelinek <jakub@redhat.com>
21359 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
21360 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
21361 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
21362 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
21363 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
21364 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
21365 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
21366 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
21367 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
21368 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
21369 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
21370 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
21371 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
21372 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
21373 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
21374 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
21375 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
21376 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
21377 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
21378 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
21379 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
21380 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
21381 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
21383 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
21385 PR rtl-optimization/70478
21386 * lra-constraints.c: Reverse the last patch.
21388 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
21390 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
21391 Add comment for WCHAR_T.
21393 2017-04-08 Martin Liska <mliska@suse.cz>
21396 2017-04-07 Martin Liska <mliska@suse.cz>
21399 * ipa-split.c (split_function): Add function part to a same comdat
21402 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21405 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
21407 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
21409 * rs6000/rs6000.c (vec_load_pendulum): Rename...
21410 (vec_pairing): ...to this.
21411 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
21412 (rs6000_sched_init): Adjust for name change.
21413 (struct rs6000_sched_context): Likewise.
21414 (rs6000_init_sched_context): Likewise.
21415 (rs6000_set_sched_context): Likewise.
21417 2017-04-07 Jakub Jelinek <jakub@redhat.com>
21423 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
21425 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
21426 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
21427 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
21429 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
21431 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
21433 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
21435 PR rtl-optimization/70703
21436 * ira-color.c (update_conflict_hard_regno_costs): Use
21437 int64_t instead of HOST_WIDE_INT.
21439 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
21441 PR rtl-optimization/70478
21442 * lra-constraints.c (process_alt_operands): Disfavor alternative
21443 insn memory operands.
21445 2017-04-07 Jeff Law <law@redhat.com>
21447 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
21448 CALL and NOTE_INSN_CALL_ARG_LOCATION.
21450 2017-04-07 Martin Liska <mliska@suse.cz>
21453 * config/aarch64/aarch64.c (aarch64_process_target_attr):
21454 Show error message instead of an ICE.
21456 2017-04-07 Martin Liska <mliska@suse.cz>
21459 * ipa-split.c (split_function): Add function part to a same comdat
21462 2017-04-07 Richard Biener <rguenther@suse.de>
21464 PR middle-end/80341
21465 * tree.c (get_unwidened): Also handle ! for_type case for
21467 * convert.c (do_narrow): Split out from ...
21468 (convert_to_integer_1): ... here. Do not pass final truncation
21469 type to get_unwidened for TRUNC_DIV_EXPR.
21471 2017-04-07 Richard Biener <rguenther@suse.de>
21473 * tree-affine.c (wide_int_ext_for_comb): Take type rather
21475 (aff_combination_const): Adjust.
21476 (aff_combination_scale): Likewise.
21477 (aff_combination_add_elt): Likewise.
21478 (aff_combination_add_cst): Likewise.
21479 (aff_combination_convert): Likewise.
21480 (add_elt_to_tree): Likewise. Remove unused argument.
21481 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
21483 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
21485 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
21487 * config/arm/arm.c (arm_default_short_enums): Use
21488 ARM_DEFAULT_SHORT_ENUMS.
21489 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
21491 2017-04-06 Jakub Jelinek <jakub@redhat.com>
21494 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
21495 members with redundant out-of-class redeclaration.
21497 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
21500 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
21501 * config/i386/i386.md (*zero_extendsidi2):
21502 Add (?*x,*x) and (?*v,*v) alternatives.
21504 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
21507 * config/i386/i386.c (ix86_expand_builtin)
21508 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
21509 mode from insn data. Convert operands to insn operand mode.
21510 Copy operands that don't satisfy insn predicate to a register.
21512 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
21514 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
21517 2017-04-06 Richard Biener <rguenther@suse.de>
21519 PR tree-optimization/80334
21520 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
21521 preserve alignment of accesses.
21523 2017-04-06 Richard Biener <rguenther@suse.de>
21525 PR tree-optimization/80262
21526 * tree-sra.c (build_ref_for_offset): Preserve address-space
21528 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
21529 Drop useless address-space information on MEM_REF offsets.
21531 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
21533 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
21535 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
21537 PR rtl-optimization/70703
21538 * ira-color.c (update_conflict_hard_regno_costs): Use
21539 HOST_WIDE_INT instead of long.
21541 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
21544 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
21545 not defined for x86_64 target. Add -mmmx target option when __SSE2__
21547 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
21548 for x86_64 target. Handle -m3dnowa option.
21550 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
21552 PR rtl-optimization/70703
21553 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
21554 (update_conflict_hard_regno_costs): Use long instead of unsigned
21555 arithmetic for cost calculation.
21557 2017-04-05 Jakub Jelinek <jakub@redhat.com>
21558 Bernd Edlinger <bernd.edlinger@hotmail.de>
21561 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
21564 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
21567 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
21568 ptr_mode with Pmode throughout.
21569 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
21570 into probe_stack_range and use DImode.
21572 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
21575 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
21576 call_eh_return is true.
21578 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21580 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
21581 Initialize last_match_fntype_index.
21583 2017-04-05 Jakub Jelinek <jakub@redhat.com>
21586 * tree-nvr.c: Include internal-fn.h.
21587 (pass_return_slot::execute): Ignore internal calls without
21590 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21591 Richard Biener <rguenther@suse.de>
21594 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
21595 captures used multiple times, except for the last use.
21596 * generic-match-head.c: Include gimplify.h.
21598 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21600 PR tree-optimization/79390
21601 * target.h (struct noce_if_info): Declare.
21602 * targhooks.h (default_noce_conversion_profitable_p): Declare.
21603 * target.def (noce_conversion_profitable_p): New target hook.
21604 * ifcvt.h (struct noce_if_info): New type, moved from ...
21605 * ifcvt.c (struct noce_if_info): ... here.
21606 (noce_conversion_profitable_p): Renamed to ...
21607 (default_noce_conversion_profitable_p): ... this. No longer
21609 (noce_try_store_flag_constants, noce_try_addcc,
21610 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
21611 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
21612 instead of noce_conversion_profitable_p.
21613 * config/i386/i386.c: Include ifcvt.h.
21614 (ix86_option_override_internal): Don't override
21615 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
21616 (ix86_noce_conversion_profitable_p): New function.
21617 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
21618 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
21619 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
21620 * doc/tm.texi: Regenerated.
21622 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21624 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
21627 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21630 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
21631 instructions for small multiply cores.
21633 2017-04-04 Jeff Law <law@redhat.com>
21635 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
21637 (mips_expand_vec_perm_const): Initialize elements in orig_perm
21638 that are not set by the loop over the elements.
21640 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21643 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
21644 int mode, convert_modes it to mode as unsigned, otherwise use
21645 lowpart_subreg to mode rather than SImode.
21646 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
21647 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
21648 Use DImode instead of SImode for the shift count operand.
21649 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
21652 2017-04-04 Richard Biener <rguenther@suse.de>
21654 PR middle-end/80281
21655 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
21656 arithmetic done for the negate or the plus. Simplify.
21657 (A - (-B) -> A + B): Likewise.
21658 * fold-const.c (split_tree): Make sure to not negate pointers.
21660 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
21662 PR rtl-optimization/60818
21663 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
21664 a compare of comparisons with the thing compared if this results
21665 in a different machine mode.
21667 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
21669 * alias.c (base_alias_check): Fix typo in comment.
21670 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
21671 * cgraphunit.c (symbol_table::compile): Likewise.
21672 * collect2.c (maybe_run_lto_and_relink): Likewise.
21673 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
21674 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
21675 * config/avr/avr.c (avr_map_op_t): Likewise.
21676 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
21677 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
21678 * config/epiphany/epiphany.md (movcc): Likewise.
21679 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
21680 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
21682 * config/mips/mips.c (mips_save_restore_reg): Likewise.
21683 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
21684 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
21685 * config/sh/sh.c (sh_rtx_costs): Likewise.
21686 * fold-const.c (fold_truth_andor): Likewise.
21687 * genautomata.c (collapse_flag): Likewise.
21688 * gengtype.h (struct type::u::s): Likewise.
21689 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
21690 * input.c (FORMAT_AMOUNT): Likewise.
21691 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
21692 (known_aggs_to_agg_replacement_list): Likewise.
21693 * ipa-inline-analysis.c: Likewise.
21694 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
21695 * ipa-polymorphic-call.c
21696 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
21697 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
21698 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
21700 * modulo-sched.c (apply_reg_moves): Likewise.
21701 * omp-expand.c (build_omp_regions_1): Likewise.
21702 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
21703 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
21704 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
21705 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
21706 * value-prof.c: Likewise.
21707 * var-tracking.c (val_reset): Likewise.
21709 2017-04-03 Richard Biener <rguenther@suse.de>
21711 PR tree-optimization/80275
21712 * fold-const.c (split_address_to_core_and_offset): Handle
21715 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
21717 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
21718 descriptors is at least equal to that of functions.
21720 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
21722 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
21724 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
21727 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
21728 (mov<IMOD4:mode>): New expander.
21729 (*mov<IMOD4:mode>_internal): New insn and split pattern.
21731 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
21733 PR rtl-optimization/79405
21734 * fwprop.c (propagations_left): New variable.
21735 (forward_propagate_into): Decrement it.
21736 (fwprop_init): Initialize it.
21737 (fw_prop): If the variable has reached zero, stop propagating.
21738 (fwprop_addr): Ditto.
21740 2017-03-31 Jakub Jelinek <jakub@redhat.com>
21743 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
21744 a FUNCTION_DECL, pass it as decl instead of origin to
21747 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
21749 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
21752 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
21755 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
21756 TARGET_VSX_SMALL_INTEGER.
21758 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21760 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
21761 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
21763 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
21765 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
21766 extraction from odd-numbered MSA register.
21768 2017-03-31 Jakub Jelinek <jakub@redhat.com>
21770 PR middle-end/80173
21771 * expmed.c (store_bit_field_1): Don't attempt to create
21772 a word subreg out of hard registers wider than word if they
21773 have HARD_REGNO_NREGS of 1 for their mode.
21775 PR middle-end/80163
21776 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
21777 conversions to integer types wider than word and pointer.
21780 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
21781 (rtx_equal_for_cselib_p): Pass 0 to it.
21782 * cselib.c (cselib_hasher::equal): Likewise.
21783 (rtx_equal_for_cselib_1): Add depth argument. If depth
21784 is 128, don't look up VALUE locs and punt. Increment
21785 depth in recursive calls when walking VALUE locs.
21787 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
21789 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
21790 (make_gcov_file_name): Use the canonical path name for generating
21792 (read_line): Fix handling of files with ascii null bytes.
21794 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
21796 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
21797 to initialise a vector register instead
21798 of using a const_int.
21800 2017-03-30 Jakub Jelinek <jakub@redhat.com>
21802 PR translation/80189
21803 * gimplify.c (omp_default_clause): Use %qs instead of %s in
21804 diagnostic messages.
21806 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
21809 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
21810 (dfp_diex_<mode>): Update mode of operand 1.
21811 * doc/extend.texi (dxex, dxexq): Document change to return type.
21812 (diex, diexq): Document change to argument type.
21814 2017-03-30 Martin Jambor <mjambor@suse.cz>
21817 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
21818 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
21819 it reflects the signature changes performed at the callee side.
21820 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
21821 to cgraph_build_function_type_skip_args.
21822 (build_function_decl_skip_args): Adjust call to the above function.
21824 2017-03-30 Jakub Jelinek <jakub@redhat.com>
21827 * config/i386/sse.md
21828 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
21829 register as dest whenever it is a MEM not rtx_equal_p to the
21830 corresponding dup operand, and when forcing into reg move the
21831 reg into the memory afterwards.
21832 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
21833 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
21834 for the force_reg mode.
21835 (avx512vl_vextractf128<mode>): Use register as dest either
21836 always when a MEM, or when it is a MEM not rtx_equal_p to the
21837 corresponding dup operand, or even not when it is a CONST_VECTOR
21838 depending on the mode and lo vs. hi.
21839 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
21841 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
21842 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
21843 Likewise. Require that operands[2] is even.
21844 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
21845 Remove extraneous parens. Require that operands[2] is a multiple
21847 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
21848 operands[0] is a MEM if <mask_applied>, the predicates/constraints
21849 disallow memory then.
21851 2017-03-30 Richard Biener <rguenther@suse.de>
21853 PR tree-optimization/77498
21854 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
21855 to non-constants over backedges.
21857 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
21859 PR rtl-optimization/80233
21860 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
21861 as last_combined_insn. Do not test for BARRIER_P separately.
21863 2017-03-29 Andreas Schwab <schwab@suse.de>
21866 * calls.c (prepare_call_address): Convert funexp to Pmode before
21867 copying to temp reg.
21869 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21871 PR tree-optimization/80158
21872 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
21873 Handle possible future case of more than one alternate
21875 (replace_rhs_if_not_dup): Likewise.
21876 (replace_one_candidate): Likewise.
21878 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
21880 PR rtl-optimization/80193
21881 * ira.c (ira): Do not check allocation for LRA.
21883 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
21885 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
21886 (nvptx_output_simt_exit): Declare.
21887 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
21888 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
21889 (init_softstack_frame): Move initialization of crtl->is_leaf to...
21890 (nvptx_declare_function_name): ...here. Emit declaration of local
21891 memory space buffer for omp_simt_enter insn.
21892 (nvptx_output_unisimt_switch): New.
21893 (nvptx_output_softstack_switch): New.
21894 (nvptx_output_simt_enter): New.
21895 (nvptx_output_simt_exit): New.
21896 * config/nvptx/nvptx.h (struct machine_function): New fields
21897 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
21898 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
21899 (UNSPECV_SIMT_EXIT): Ditto.
21900 (omp_simt_enter_insn): New insn.
21901 (omp_simt_enter): New expansion.
21902 (omp_simt_exit): New insn.
21903 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
21905 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
21906 (expand_GOMP_SIMT_ENTER_ALLOC): New.
21907 (expand_GOMP_SIMT_EXIT): New.
21908 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
21909 (GOMP_SIMT_ENTER_ALLOC): Ditto.
21910 (GOMP_SIMT_EXIT): Ditto.
21911 * target-insns.def (omp_simt_enter): New insn.
21912 (omp_simt_exit): Ditto.
21913 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
21915 (lower_rec_simd_input_clauses): Implement SIMT privatization.
21916 (lower_rec_input_clauses): Likewise.
21917 (lower_lastprivate_clauses): Handle SIMT privatization.
21919 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
21920 (ompdevlow_adjust_simt_enter): New.
21921 (find_simtpriv_var_op): New.
21922 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
21923 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
21925 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
21926 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
21927 (copy_decl_for_dup_finish): Ditto.
21929 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
21931 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
21934 * config/i386/i386.c (ix86_option_override_internal): Always
21935 allow -mpreferred-stack-boundary=3 for 64-bit targets.
21937 2017-03-28 Bin Cheng <bin.cheng@arm.com>
21939 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
21941 2017-03-28 Bin Cheng <bin.cheng@arm.com>
21943 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
21944 mark new edge's irreducible flag accordign to it.
21945 (vect_do_peeling): Check loop preheader edge's irreducible flag
21946 and pass it to function slpeel_add_loop_guard.
21948 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
21950 PR tree-optimization/80218
21951 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
21952 Update block frequencies and counts.
21954 2017-03-28 Richard Biener <rguenther@suse.de>
21956 PR tree-optimization/78644
21957 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
21958 of a simplification result we may not use it at all.
21960 2017-03-28 Richard Biener <rguenther@suse.de>
21963 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
21964 without arguments, generate default definition of a SSA name.
21966 2017-03-28 Richard Biener <rguenther@suse.de>
21968 PR middle-end/80222
21969 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
21970 TYPE_REF_CAN_ALIAS_ALL references.
21971 * fold-const.c (fold_indirect_ref_1): Likewise.
21973 2017-03-28 Martin Liska <mliska@suse.cz>
21976 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
21977 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
21979 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
21980 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
21982 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
21983 (EXTRA_SPECS): Define.
21984 (SUBTARGET_EXTRA_SPECS): Likewise.
21985 (SUBTARGET_CPP_SPEC): Likewise.
21986 * config/arc/elf.h (EXTRA_SPECS): Renamed to
21987 SUBTARGET_EXTRA_SPECS.
21988 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
21990 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
21992 * config/arc/simdext.md (vst64_insn): Update pattern.
21993 (vld32wh_insn): Likewise.
21994 (vld32wl_insn): Likewise.
21995 (vld64_insn): Likewise.
21996 (vld32_insn): Likewise.
21998 2017-03-28 Marek Polacek <polacek@redhat.com>
22001 * fold-const.c (fold_comparison): Use protected_set_expr_location
22002 instead of SET_EXPR_LOCATION.
22004 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
22006 * tree.c (add_expr): Avoid name lookup warning.
22008 2017-03-27 Jeff Law <law@redhat.com>
22010 PR tree-optimization/80216
22011 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
22012 function name. Limit recursion depth.
22013 (record_temporary_equivalences): Corresponding changes.
22015 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
22017 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
22020 2017-03-27 Jakub Jelinek <jakub@redhat.com>
22023 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
22025 * cfgcleanup.c (reg_note_cfa_p): New array.
22026 (insns_have_identical_cfa_notes): New function.
22027 (old_insns_match_p): Don't cross-jump in between /f
22028 and non-/f instructions. If both i1 and i2 are frame related,
22029 verify all CFA notes, their order and content.
22031 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
22034 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
22035 HImode and SImode with zero extend to DImode to one insn.
22036 (bswap<mode>2_extenddi): Likewise.
22037 (bswapsi2_extenddi): Likewise.
22038 (bswaphi2_extendsi): Likewise.
22039 (bswaphi2): Combine bswap HImode and SImode into one insn.
22040 Separate memory insns from swapping register.
22041 (bswapsi2): Likewise.
22042 (bswap<mode>2): Likewise.
22043 (bswaphi2_internal): Delete, no longer used.
22044 (bswapsi2_internal): Likewise.
22045 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
22046 store, and gpr<-gpr swap insns.
22047 (bswap<mode>2_store): Likewise.
22048 (bswaphi2_reg): Register only splitter, combine with the splitter.
22049 (bswaphi2 splitter): Likewise.
22050 (bswapsi2_reg): Likewise.
22051 (bswapsi2 splitter): Likewise.
22052 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
22053 the insns into load, store, and register/register insns.
22054 (bswapdi2_ldbrx): Likewise.
22055 (bswapdi2_load): Likewise.
22056 (bswapdi2_store): Likewise.
22057 (bswapdi2_reg): Likewise.
22059 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
22061 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
22062 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
22064 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
22067 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
22069 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
22070 special handling for target option conflicts between dform
22071 options (-mpower9-dform, -mpower9-dform-vector,
22072 -mpower9-dform-scalar) and -mno-direct-move.
22074 2017-03-27 Richard Biener <rguenther@suse.de>
22076 PR tree-optimization/80181
22077 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
22079 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22081 * config/arc/predicates.md (move_double_src_operand): Replace the
22082 call to move_double_src_operand with a call to address_operand.
22084 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22086 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
22087 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
22088 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
22090 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22092 * config/arc/predicates.md (long_immediate_loadstore_operand):
22093 Consider scaled addresses cases.
22095 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22097 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
22098 restored when in interrupt.
22099 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
22100 doesn't have delay slot.
22102 2017-03-27 Richard Biener <rguenther@suse.de>
22105 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
22106 inlined thunk clones.
22108 2017-03-27 Jakub Jelinek <jakub@redhat.com>
22111 * asan.c (instrument_derefs): Copy over last operand from
22112 original COMPONENT_REF to the new COMPONENT_REF with
22113 DECL_BIT_FIELD_REPRESENTATIVE.
22114 * ubsan.c (instrument_object_size): Likewise.
22116 2017-03-27 Richard Biener <rguenther@suse.de>
22118 PR tree-optimization/80170
22119 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
22120 sure DR/SCEV didnt fold in constants we do not see when looking
22121 at the reference base alignment.
22123 2017-03-27 Richard Biener <rguenther@suse.de>
22125 PR middle-end/80171
22126 * gimple-fold.c (fold_ctor_reference): Properly guard against
22127 NULL return value from canonicalize_constructor_val.
22129 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
22132 * config/i386/i386.c (ix86_expand_builtin)
22133 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
22134 flags reg setting and flags reg using instructions.
22135 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
22136 clobbering instructions to zero extend op2.
22138 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
22140 * doc/install.texi (Configuration) <--with-aix-soname>:
22141 Update link to AIX ld.
22143 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
22145 PR rtl-optimization/80160
22146 PR rtl-optimization/80159
22147 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
22148 reg_alternate_class into account.
22150 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
22153 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
22154 to consider in curr_insn_transform.
22156 2017-03-24 Jakub Jelinek <jakub@redhat.com>
22158 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
22159 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
22160 and emit_mode_inner.
22162 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22164 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
22165 argument to the overloaded builtin variants. Use the new flag to
22166 deprecate certain builtin variants.
22167 * config/s390/s390-builtin-types.def: Add new builtin types.
22168 * config/s390/s390-builtins.h: Support new flags field for
22169 overloaded builtins.
22170 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
22171 (s390_macro_to_expand): Enable vector float data type.
22172 (s390_cpu_cpp_builtins_internal): Indicate support of the new
22173 builtins by incrementing the __VEC__ version number.
22174 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
22176 (s390_resolve_overloaded_builtin): Emit error messages depending
22177 on the builtin flags.
22178 * config/s390/s390.c (s390_expand_builtin): Support additional
22179 flags argument. Change error message to match the messages
22180 emitted in s390-c.c.
22181 * config/s390/s390.md: New UNSPEC_* constants.
22182 (op_type): Add new instruction types.
22183 * config/s390/vecintrin.h: Add new builtins and test data class
22185 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
22186 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
22187 (VEC_INEXACT, VEC_NOINEXACT): New constants.
22188 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
22189 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
22190 ("vec_mergel<mode>"): V_HW -> VEC_HW.
22192 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
22193 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
22194 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
22195 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
22197 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
22198 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
22199 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
22200 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
22202 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
22203 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
22204 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
22205 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
22206 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
22207 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
22208 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
22210 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
22211 ("vec_scatter_element<V_HW_4:mode>_DI")
22212 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
22213 ("vec_fpint<mode>", "vflls")
22214 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
22215 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
22216 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
22217 ("*vec_cmphe<mode>_cc"): ... these.
22219 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
22220 mode constant instead of magic value.
22222 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22224 * config/s390/s390.c (s390_expand_vec_compare): Support other
22225 vector floating point modes than just V2DF.
22226 (s390_expand_vcond): Likewise.
22227 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
22228 (s390_cannot_change_mode_class): Prevent mode changes between TF
22229 and V1TF in vector registers.
22230 * config/s390/s390.md (DF, SF): New mode attributes.
22231 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
22232 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
22233 SFmode support for VRs.
22234 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
22236 (VFT, VF_HW): New mode iterators.
22237 (vw, sdx): New mode attributes.
22238 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
22239 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
22240 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
22241 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
22242 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
22243 also the new vector floating point modes. Renaming to ...
22245 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
22246 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
22247 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
22248 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
22249 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
22250 ("vec_unordered<mode>"): ... these.
22252 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
22253 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
22254 ("*vec_extendv2df"): New insn definitions.
22256 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22258 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
22259 ("mulditi3_2", "*muldi3_sign"): New patterns.
22260 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
22261 rename the pattern definition.
22263 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22265 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
22267 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
22269 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22271 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
22272 instruction if possible.
22273 * config/s390/vector.md (vec_halfnumelts): New mode
22275 ("*vec_vllezlf<mode>"): New pattern.
22277 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22279 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
22280 ("popcountv4si2", "popcountv2di2"): Rename to ...
22281 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
22282 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
22284 ("popcount<mode>2_vxe"): New pattern.
22286 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22288 * common/config/s390/s390-common.c (processor_flags_table): Add
22290 * config.gcc: Add arch12.
22291 * config/s390/driver-native.c (s390_host_detect_local_cpu):
22292 Default to arch12 for unknown CPU model numbers.
22293 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
22294 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
22295 PROCESSOR_max sanity check.
22296 * config/s390/s390-opts.h (enum processor_type): Add
22298 * config/s390/s390.c (processor_table): Add arch12.
22299 (s390_expand_builtin): Add check for B_VXE flag.
22300 (s390_issue_rate): Add PROCESSOR_ARCH12.
22301 (s390_get_sched_attrmask): Likewise.
22302 (s390_get_unit_mask): Likewise.
22303 (s390_sched_score): Enable z13 scheduling for arch12.
22304 (s390_sched_reorder): Likewise.
22305 (s390_sched_variable_issue): Likewise.
22306 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
22308 (s390_tune_attr): Use z13 scheduling also for arch12.
22309 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
22310 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
22311 (TARGET_VXE_P): New macros.
22312 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
22313 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
22314 * config/s390/s390.opt: Add arch12 as processor_type.
22316 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22318 * config/s390/s390.md
22319 ("fixuns_truncdddi2", "fixuns_trunctddi2")
22320 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
22321 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
22323 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
22324 Rename expanders to ...
22326 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
22327 ("fixuns_truncdddi2_emu"): ... these.
22329 ("fixuns_trunc<mode>si2_emu"): New expander.
22331 ("*fixuns_truncdfdi2_z13"): Rename to ...
22332 ("*fixuns_truncdfdi2_vx"): ... this.
22334 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22336 * config/s390/2964.md: Remove the single element vector compare
22337 instructions which are no longer used.
22338 * config/s390/s390.c (s390_select_ccmode): Remove handling of
22340 (s390_canonicalize_comparison): Remove handling of DFmode
22342 (s390_expand_vec_compare_scalar): Remove function.
22343 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
22344 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
22346 ("*cmp<mode>_ccs"): Add wfcdb instruction.
22348 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22350 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
22352 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
22353 will anyway by matched by mov<mode>_64dfp.
22355 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22357 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
22358 vlef/vstef. Add missing operand to vleif.
22360 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22362 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
22363 pair for all vector types with 64 bit elements.
22364 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
22365 * config/s390/vector.md (V_HW_64): ... here.
22366 (V_128_NOSINGLE): New mode iterator.
22367 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
22368 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
22369 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
22370 ("*vec_load_pairv2di"): Change to ...
22371 ("*vec_load_pair<mode>"): ... this one.
22373 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22375 * config/s390/constraints.md: Add comments.
22376 (jKK): Reject element sizes > 8 bytes.
22377 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
22379 * config/s390/s390.md: Add the s_operand checks formerly in
22380 s390_split_ok_p to various splitters where they are still
22382 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
22383 for 128 bit vectors. Plus two splitters.
22385 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22387 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
22390 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22393 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
22394 error if the boundary argument is not constant.
22396 2017-03-24 Jakub Jelinek <jakub@redhat.com>
22398 PR rtl-optimization/80112
22399 * loop-doloop.c (doloop_condition_get): Don't check condition
22400 if cmp isn't SET with IF_THEN_ELSE src.
22402 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22404 PR tree-optimization/80158
22405 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
22406 replacing a candidate statement, also replace it for the
22407 candidate's alternate interpretation.
22408 (replace_rhs_if_not_dup): Likewise.
22409 (replace_one_candidate): Likewise.
22411 2017-03-24 Richard Biener <rguenther@suse.de>
22413 PR tree-optimization/80167
22414 * graphite-isl-ast-to-gimple.c
22415 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
22417 (translate_isl_ast_to_gimple::get_rename): Likewise.
22419 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
22421 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
22422 handling of certain combinations of target options, including the
22423 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
22424 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
22426 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22429 * config/arm/arm.md (*load_multiple): Add reload_completed to
22430 matching condition.
22432 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22433 Richard Biener <rguenth@suse.de>
22435 PR tree-optimization/79908
22436 PR tree-optimization/80136
22437 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
22438 been cast away, gimplify_and_add suffices.
22440 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
22442 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
22444 2017-03-23 Richard Biener <rguenther@suse.de>
22446 PR tree-optimization/80032
22447 * gimplify.c (gimple_push_cleanup): Forced unconditional
22448 cleanups still have to go to the conditional_cleanups
22451 2017-03-22 Jakub Jelinek <jakub@redhat.com>
22453 PR tree-optimization/80072
22454 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
22456 (next_operand_entry_id): Change type to unsigned int.
22457 (sort_by_operand_rank): Make sure to return the right return value
22458 even if unsigned fields are bigger than INT_MAX.
22459 (struct oecount): Change cnt and id type to unsigned int.
22460 (oecount_hasher::equal): Formatting fix.
22461 (oecount_cmp): Make sure to return the right return value
22462 even if unsigned fields are bigger than INT_MAX.
22463 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
22466 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
22467 TREE_READONLY on result if writing it more than once.
22470 * doc/invoke.texi (-fsanitize=thread): Document that with
22471 -fnon-call-exceptions atomics are not able to throw
22475 * tsan.c: Include tree-eh.h.
22476 (instrument_builtin_call): Call maybe_clean_eh_stmt or
22477 maybe_clean_or_replace_eh_stmt where needed.
22478 (instrument_memory_accesses): Add cfg_changed argument.
22479 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
22480 if it returned true.
22481 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
22483 PR rtl-optimization/63191
22484 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
22485 wrapper function, moved the whole old content into ...
22486 (ix86_delegitimize_address_1): ... this. New inline function.
22487 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
22488 true as last argument instead of ix86_delegitimize_address.
22490 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
22492 * config/aarch64/aarch64.c (generic_branch_cost): Copy
22493 cortexa57_branch_cost.
22495 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
22497 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
22499 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22502 * doc/md.texi (Constraints): Document wA constraint.
22503 * config/rs6000/constraints.md (wA): New.
22504 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
22505 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
22506 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
22507 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
22509 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
22512 * gimplify.c (is_oacc_declared): New function.
22513 (oacc_default_clause): Use it to set default flags for acc declared
22514 variables inside parallel regions.
22515 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
22516 declared variables.
22517 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
22518 declare attribute to any decl as necessary.
22520 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
22523 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
22524 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
22525 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
22526 (arm_arch_lpae): This.
22527 * config/arm/arm.c (arm_arch7ve): Rename into ...
22528 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
22529 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
22532 2017-03-22 Martin Liska <mliska@suse.cz>
22535 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
22536 error message instead of an ICE.
22538 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22540 * doc/extend.texi (6.11 Additional Floating Types): Revise.
22542 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
22544 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
22546 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
22549 2017-03-21 Martin Sebor <msebor@redhat.com>
22551 * doc/extend.texi: Use "cannot" instead of "can't."
22552 * doc/hostconfig.texi: Same.
22553 * doc/install.texi: Same.
22554 * doc/invoke.texi: Same.
22555 * doc/loop.texi: Same.
22556 * doc/md.texi: Same.
22557 * doc/objc.texi: Same.
22558 * doc/rtl.texi: Same.
22559 * doc/tm.texi: Same.
22560 * doc/tm.texi.in: Same.
22561 * doc/trouble.texi: Same.
22563 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
22566 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
22567 (collect_checksum_attributes): Set it.
22568 (die_checksum_ordered): Use it.
22570 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22572 PR tree-optimization/79908
22573 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
22574 change: For a VA_ARG whose LHS has been cast away, use
22575 force_gimple_operand to construct the side effects.
22577 2017-03-21 David Malcolm <dmalcolm@redhat.com>
22579 PR translation/80001
22580 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
22581 more amenable to translation.
22582 (oacc_loop_auto_partitions): Likewise.
22584 2017-03-21 Marek Polacek <polacek@redhat.com>
22585 Martin Sebor <msebor@redhat.com>
22587 PR tree-optimization/80109
22588 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
22589 on INTEGRAL_TYPE_P.
22591 2017-03-21 Jakub Jelinek <jakub@redhat.com>
22592 Segher Boessenkool <segher@kernel.crashing.org>
22595 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
22596 check reg_used_between_p between insn and one of succ or succ2
22597 depending on if succ is artificial insn not inserted into insn
22600 2017-03-21 Martin Liska <mliska@suse.cz>
22602 PR gcov-profile/80081
22603 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
22604 * doc/gcc.texi: Include gcov-dump stuff.
22605 * doc/gcov-dump.texi: New file.
22607 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
22609 PR rtl-optimization/79150
22610 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
22611 conditional jump, if the jump is the last insn of the loop.
22613 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22614 Richard Biener <rguenth@suse.de>
22616 PR tree-optimization/79908
22617 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
22618 been cast away, use force_gimple_operand to construct the side
22621 2017-03-21 Martin Liska <mliska@suse.cz>
22623 PR libfortran/79956
22624 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
22627 2017-03-21 Brad Spengler <spender@grsecurity.net>
22630 * plugin.c (htab_hash_plugin): New function.
22631 (add_new_plugin): Use it and adjust.
22632 (parse_plugin_arg_opt): Adjust.
22633 (init_one_plugin): Likewise.
22635 2017-03-21 Richard Biener <rguenther@suse.de>
22637 PR tree-optimization/80032
22638 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
22639 if set force the cleanup to happen unconditionally.
22640 (gimplify_target_expr): Push inserted clobbers with force_uncond
22641 to avoid them being removed by control-dependent DCE.
22643 2017-03-21 Richard Biener <rguenther@suse.de>
22645 PR tree-optimization/80122
22646 * tree-inline.c (copy_bb): Do not expans va-arg packs or
22647 va_arg_pack_len when the inlined call stmt requires pack
22649 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
22651 2017-03-21 Jakub Jelinek <jakub@redhat.com>
22654 * tsan.c (instrument_builtin_call): If the memory model argument
22655 is not a constant, assume it is valid.
22658 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
22661 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
22663 PR rtl-optimization/79910
22664 * combine.c (can_combine_p): Do not allow combining an I0 or I1
22665 if its dest is used by an insn before I2 (other than the combined
22666 insns themselves, which are properly handled already).
22668 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
22671 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
22673 * combine.c (record_used_regs): New static function.
22674 (try_combine): Handle situations where there is an additional
22675 instruction between I2 and I3 which needs to have a LOG_LINK
22679 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
22681 * combine.c (try_combine): Delete redundant i1 test. Call
22682 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
22684 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22687 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
22688 alternatives 13/14.
22690 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22692 PR tree-optimization/80054
22693 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
22694 the optimization if a PHI or any of its arguments is not dominated
22695 by the candidate's basis. Use gphi* rather than gimple* as
22697 (replace_profitable_candidates): Clean up a gimple* variable that
22698 should be a gphi* variable.
22700 2017-03-20 Martin Sebor <msebor@redhat.com>
22703 * doc/extend.texi (attribute constructor): Document present limitation.
22705 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
22708 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
22709 __POWER9_VECTOR__ #ifdef control, change template definition to
22710 use Power9-specific built-in function.
22711 (vec_any_eq): Likewise.
22712 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
22713 to control outcomes from this test.
22714 (vector_ae_<mode>p): For VEC_F modes, likewise.
22716 2017-03-20 Ian Lance Taylor <iant@google.com>
22718 * config/i386/i386.c (ix86_function_regparm): Save an extra
22719 register for -fsplit-stack with DECL_STATIC_CHAIN.
22721 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
22724 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
22725 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
22727 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
22729 * config/riscv/riscv.c (riscv_print_operand): Use "fence
22731 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
22734 2017-03-20 Marek Polacek <polacek@redhat.com>
22737 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
22739 2017-03-20 Richard Biener <rguenther@suse.de>
22741 PR tree-optimization/80113
22742 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
22743 allocate extra SSA name for PHI def.
22744 (add_close_phis_to_outer_loops): Likewise.
22745 (add_close_phis_to_merge_points): Likewise.
22746 (copy_loop_close_phi_args): Likewise.
22747 (copy_cond_phi_nodes): Likewise.
22749 2017-03-20 Martin Liska <mliska@suse.cz>
22751 PR middle-end/79753
22752 * tree-chkp.c (chkp_build_returned_bound): Do not build
22753 returned bounds for a LHS that's not a BOUNDED_P type.
22755 2017-03-20 Martin Liska <mliska@suse.cz>
22759 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
22760 COMPLEX_CST and VECTOR_CST.
22762 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22765 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
22766 target operand. A new splitter adds the clobber statement in case
22767 the target operand is dead anyway.
22769 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
22771 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
22772 to age-old versions of binutils and glibc.
22774 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
22776 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
22778 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
22780 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
22782 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
22784 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
22785 requirement for binutils 2.13.
22787 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
22789 * combine.c (try_combine): Delete redundant i1 test. Call
22790 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
22792 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
22794 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
22795 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
22797 <riscv64-*-elf>: Re-arrange section
22798 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
22799 <riscv32-*-linux>: Likewise.
22800 <riscv64-*-elf>: Likewise
22801 <riscv64-*-linux>: Likewise.
22803 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
22806 * aarch64.opt(verbose-cost-dump): Fix typo.
22808 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
22811 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
22812 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
22814 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
22816 * reload.c (find_reloads): When reloading a nonoffsettable address,
22817 use RELOAD_OTHER for it and its address reloads.
22819 PR rtl-optimization/79910
22820 * combine.c (record_used_regs): New static function.
22821 (try_combine): Handle situations where there is an additional
22822 instruction between I2 and I3 which needs to have a LOG_LINK
22825 2017-03-17 Jeff Law <law@redhat.com>
22827 PR tree-optimization/71437
22828 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
22829 conditional in the hash table first.
22830 (vrp_dom_walker::before_dom_children): Extract condition from
22831 ASSERT_EXPR. Record condition, its inverion and any implied
22832 conditions as well.
22834 2017-03-17 Marek Polacek <polacek@redhat.com>
22835 Markus Trippelsdorf <markus@trippelsdorf.de>
22837 PR tree-optimization/80079
22838 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
22841 2017-03-17 Richard Biener <rguenther@suse.de>
22843 PR middle-end/80075
22844 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
22845 Properly verify the LHS before the RHS possibly claims to be
22847 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
22850 2017-03-17 Martin Jambor <mjambor@suse.cz>
22852 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
22853 (List of -O2 options): Likewise.
22854 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
22857 2017-03-17 Tom de Vries <tom@codesourcery.com>
22859 * gcov-dump.c (print_usage): Print bug_report_url.
22861 2017-03-17 Richard Biener <rguenther@suse.de>
22863 PR middle-end/80050
22864 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
22865 (parser::peek): Likewise.
22867 2017-03-17 Richard Biener <rguenther@suse.de>
22869 PR tree-optimization/80048
22870 * sese.c (free_sese_info): Properly release rename_map and
22871 copied_bb_map elements.
22873 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
22875 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
22876 Add linked-list forward and backlinks. Insert on
22877 construction, remove on destruction.
22878 (class pass_store_merging): Add m_stores_head field.
22879 (pass_store_merging::terminate_and_process_all_chains):
22880 Iterate over m_stores_head list.
22881 (pass_store_merging::terminate_all_aliasing_chains):
22883 (pass_store_merging::execute): Check for debug stmts first.
22884 Push new chains onto the m_stores_head stack.
22886 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
22889 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
22890 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
22891 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
22893 2017-03-16 Jeff Law <law@redhat.com>
22895 PR tree-optimization/71437
22896 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
22897 member function. Implementation moved into after_dom_children
22898 member function and into the threader's thread_outgoing_edges
22900 (dom_opt_dom_walker::after_dom_children): Simplify by moving
22901 some code into new thread_outgoing_edges.
22902 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
22903 definition. Simplify marker handling (do it here). Assume we always
22904 have the available expression and the const/copies tables.
22905 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
22907 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
22908 * tree-vrp.c (equiv_stack): No longer file scoped.
22909 (vrp_dom_walker): New class.
22910 (vrp_dom_walker::before_dom_children): New member function.
22911 (vrp_dom_walker::after_dom_children): Likewise.
22912 (identify_jump_threads): Setup domwalker. Use it rather than
22913 walking edges in a random order by hand. Simplify setup/finalization.
22914 (finalize_jump_threads): Remove.
22915 (vrp_finalize): Do not call identify_jump_threads here.
22916 (execute_vrp): Do it here instead and call thread_through_all_blocks
22919 PR tree-optimization/71437
22920 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
22922 (simplify_stmt_for_jump_threading): Add basic_block argument. All
22924 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
22925 (dom_opt_dom_walker::thread_across_edge): Remove
22926 handle_dominating_asserts argument. All callers changed.
22927 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
22928 changes. Remove calls to lhs_of_dominating_assert. Other
22929 uses of handle_dominating_asserts turn into unconditional code
22930 (simplify_control_stmt_condition_1): Likewise.
22931 (simplify_control_stmt_condition): Likewise.
22932 (thread_through_normal_block, thread_across_edge): Likewise.
22933 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
22934 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
22935 object if it is not an SSA_NAME.
22936 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
22937 before calling into the VRP specific simplifiers.
22938 (identify_jump_threads): Remove handle_dominating_asserts
22941 2017-03-16 Jakub Jelinek <jakub@redhat.com>
22944 * tree-diagnostic.c (default_tree_printer): No longer static.
22945 * tree-diagnostic.h (default_tree_printer): New prototype.
22947 2017-03-16 Tamar Christina <tamar.christina@arm.com>
22949 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
22950 Change ins into fmov.
22952 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22954 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
22955 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
22956 Use h_con constraint for operand 1.
22957 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
22958 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
22960 2017-03-15 Jeff Law <law@redhat.com>
22962 PR tree-optimization/71437
22963 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
22964 (record_temporary_equivalences): Use it.
22966 PR tree-optimization/71437
22967 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
22968 tree-ssa-scopedtables.
22969 (lookup_avail_expr, build_and_record_new_cond): Likewise.
22970 (record_conditions, record_cond, vuse_eq): Likewise.
22971 (record_edge_info): Adjust to API tweak of record_conditions.
22972 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
22973 (record_temporary_equivalences, optimize_stmt): Likewise.
22974 (eliminate_redundant_computations): Likewise.
22975 (record_equivalences_from_stmt): Likewise.
22976 * tree-ssa-scopedtables.c: Include options.h and params.h.
22977 (vuse_eq): New function, moved from tree-ssa-dom.c
22978 (build_and_record_new_cond): Likewise.
22979 (record_conditions): Likewise. Accept vector of conditions rather
22980 than edge_equivalence structure for first argument.
22981 for the first argument.
22982 (avail_exprs_stack::lookup_avail_expr): New member function, moved
22983 from tree-ssa-dom.c.
22984 (avail_exprs_stack::record_cond): Likewise.
22985 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
22986 from tree-ssa-dom.c.
22987 (avail_exprs_stack): Add new member functions lookup_avail_expr
22989 (record_conditions): Declare.
22991 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
22994 * lra-constraints.c (process_alt_operands): Increase reject for
22995 reloading an input/output operand.
22997 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
23000 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
23001 insns to convert from signed/unsigned char/short to IEEE 128-bit
23003 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
23005 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
23008 * config/i386/i386.c (ix86_vector_duplicate_value): Create
23009 subreg of inner mode for values already in registers.
23011 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
23013 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
23014 iteration reg is used after the loop.
23016 2017-03-14 Martin Sebor <msebor@redhat.com>
23018 PR tree-optimization/79800
23019 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
23020 precision in negative-positive range.
23021 (format_floating): Call non-const overload with adjusted precision.
23023 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
23026 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
23029 2017-03-14 Martin Sebor <msebor@redhat.com>
23031 PR middle-end/80020
23032 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
23033 * builtins.def (aligned_alloc): Use it.
23036 * Makefile.in (GTFILES): Add calls.c.
23037 * calls.c: Include "gt-calls.h".
23039 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
23041 PR rtl-optimization/79728
23042 * regs.h (struct target_regs): New field
23043 x_contains_allocatable_regs_of_mode.
23044 (contains_allocatable_regs_of_mode): New macro.
23045 * reginfo.c (init_reg_sets_1): Initialize it, and change
23046 contains_reg_of_mode so it includes global regs as well.
23047 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
23048 rather than contains_regs_of_mode.
23050 2017-03-14 Martin Liska <mliska@suse.cz>
23052 * doc/invoke.texi: Document options that can't be combined with
23053 -fcheck-pointer-bounds.
23055 2017-03-14 Martin Liska <mliska@suse.cz>
23057 PR middle-end/79831
23058 * doc/invoke.texi (-Wchkp): Document the option.
23060 2017-03-14 Martin Liska <mliska@suse.cz>
23062 * Makefile.in: Install gcov-dump.
23064 2017-03-14 Martin Liska <mliska@suse.cz>
23066 * multiple_target.c (expand_target_clones): Bail out for
23067 an invalid attribute.
23069 2017-03-14 Richard Biener <rguenther@suse.de>
23071 * alias.c (struct alias_set_entry): Pack properly.
23072 * cfgloop.h (struct loop): Likewise.
23073 * cse.c (struct set): Likewise.
23074 * ipa-utils.c (struct searchc_env): Likewise.
23075 * loop-invariant.c (struct invariant): Likewise.
23076 * lra-remat.c (struct cand): Likewise.
23077 * recog.c (struct change_t): Likewise.
23078 * rtl.h (struct address_info): Likewise.
23079 * symbol-summary.h (function_summary): Likewise.
23080 * tree-loop-distribution.c (struct partition): Likewise.
23081 * tree-object-size.c (struct object_size_info): Likewise.
23082 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
23083 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
23084 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
23085 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
23086 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
23087 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
23088 (struct _stmt_vec_info): Likewise.
23090 2017-03-14 Martin Liska <mliska@suse.cz>
23093 * multiple_target.c (create_dispatcher_calls): Check that
23094 a target can create a function dispatcher.
23096 2017-03-14 Martin Liska <mliska@suse.cz>
23099 * multiple_target.c (expand_target_clones): Drop local.local
23100 flag for default implementation.
23102 2017-03-14 Richard Biener <rguenther@suse.de>
23104 PR tree-optimization/80030
23105 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
23107 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
23109 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
23110 gcc_fallthrough() instead of __attribute__((fallthrough));
23112 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
23114 * doc/gcc.texi: Remove "up" link to (DIR).
23115 * doc/gccint.texi: Ditto.
23117 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
23119 * doc/install.texi (Specific) <avr>: Remove reference to
23122 2017-03-13 Jeff Law <law@redhat.com>
23124 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
23125 attribute rather than comments.
23127 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
23128 match_scratch operand is highest.
23130 2017-03-13 Martin Liska <mliska@suse.cz>
23132 PR middle-end/78339
23133 * ipa-pure-const.c (warn_function_noreturn): If the declarations
23134 is a CHKP clone, use original declaration.
23136 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23138 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
23139 (arc_conditional_register_usage): Use a different allocation order
23140 when optimizing for size.
23141 * common/config/arc/arc-common.c (arc_option_optimization_table):
23142 Section anchors default on when optimizing for size.
23144 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23146 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
23148 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23150 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
23151 * config/arc/arc.md (cpu_facility): Add cd variant.
23152 (*movqi_insn): Add code density variant.
23153 (*movhi_insn): Likewise.
23154 (*movqi_insn): Likewise.
23155 (*addsi3_mixed): Likewise.
23156 (subsi3_insn): Likewise.
23158 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23160 * config/arc/arc.md (movsi_cond_exec): Update constraint.
23162 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23164 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
23165 expressions with MINUS and UNARY ops.
23167 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23170 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
23172 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
23173 between vec_select and vector argument.
23174 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
23175 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
23176 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
23177 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
23178 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
23179 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
23181 2017-03-13 Richard Biener <rguenther@suse.de>
23184 * params.def (vect-max-peeling-for-alignment): Fix typo.
23186 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23188 * doc/install.texi (Specific) <mips-*-*>: Remove description of
23189 issue that only occurred with binutils below 2.18.
23191 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23193 * doc/install.texi (Specific) <cris-axis-elf>: No longer
23194 refer to binutils 2.11/2.12 minimum.
23196 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23198 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
23199 ftp.kernel.org and simplify binutils requirement.
23201 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
23203 * doc/invoke.texi (Warning Options): Fix spelling of link-time
23205 (Optimize Options): Ditto. Also remove redundancy.
23207 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23209 PR translation/79848
23210 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
23212 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
23213 to G_ to avoid double translation.
23215 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23217 PR translation/79923
23218 * auto-profile.c (get_combined_location): Convert leading
23219 character of diagnostics to lower case and remove trailing period.
23220 (read_profile): Likewise for various diagnostics.
23221 * config/arm/arm.c (arm_option_override): Remove trailing period
23222 from various diagnostics.
23223 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
23224 (msp430_expand_delay_cycles): Likewise.
23226 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23229 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
23230 full command-line argument, rather than just "str".
23231 (aarch64_validate_march): Likewise.
23232 (aarch64_validate_mtune): Likewise.
23234 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
23236 PR rtl-optimization/78911
23237 * lra-assigns.c (must_not_spill_p): New function.
23238 (spill_for): Use it.
23240 2017-03-10 Jakub Jelinek <jakub@redhat.com>
23242 PR tree-optimization/79981
23243 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
23244 ATOMIC_COMPARE_EXCHANGE ifn result.
23245 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
23246 IFN_ATOMIC_COMPARE_EXCHANGE.
23248 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23251 * opts.c (parse_sanitizer_options): Add missing question mark to
23252 "did you mean" message.
23254 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23256 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
23258 (VMULEUH_UNS): Likewise.
23259 (VMULOUB_UNS): Likewise.
23260 (VMULOUH_UNS): Likewise.
23261 * config/rs6000/rs6000.c (builtin_function_type): Remove
23262 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
23264 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23267 * read-rtl-function.c (function_reader::read_rtx_operand): Update
23268 x with result of extra_parsing_for_operand_code_0.
23269 (function_reader::extra_parsing_for_operand_code_0): Convert
23270 return type from void to rtx, returning x. When reading
23271 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
23272 larger size containing struct block_symbol.
23274 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
23276 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
23277 -mfloat128-hardware without -m64.
23279 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
23282 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
23283 entries to the case statement that marks unsigned arguments to
23284 overloaded functions.
23286 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
23288 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
23289 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
23291 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
23294 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
23295 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
23297 2017-03-10 Martin Liska <mliska@suse.cz>
23301 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
23302 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
23305 2017-03-10 Olivier Hainque <hainque@adacore.com>
23307 * tree-switch-conversion (array_value_type): Start by resetting
23308 candidate type to it's main variant.
23310 2017-03-10 Jakub Jelinek <jakub@redhat.com>
23312 PR rtl-optimization/79909
23313 * combine.c (try_combine): Use simplify_replace_rtx on individual
23314 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
23315 of the whole CALL_INSN_FUNCTION_USAGE.
23317 PR tree-optimization/79972
23318 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
23319 get_range_info on SSA_NAMEs. Formatting fixes.
23321 2017-03-10 Richard Biener <rguenther@suse.de>
23322 Jakub Jelinek <jakub@redhat.com>
23324 PR tree-optimization/77975
23325 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
23326 edge to be constant.
23327 (get_val_for): For constant x return it. Formatting fix.
23328 (loop_niter_by_eval): Avoid pointless looping if the next iteration
23329 would use the same bases as the current one.
23331 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23333 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
23334 instead of vec_select for V1TImode.
23335 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
23337 (VSX_LE_128): Add V1TI to this mode iterator.
23338 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
23339 (*vsx_le_perm_store_<mode>): Likewise.
23340 (pre-reload splitter for VSX stores): Likewise.
23341 (post-reload splitter for VSX stores): Likewise.
23342 (*vsx_xxpermdi2_le_<mode>): Likewise.
23343 (*vsx_lxvd2x2_le_<mode>): Likewise.
23344 (*vsx_stxvd2x2_le_<mode>): Likewise.
23346 2017-03-09 Michael Eager <eager@eagercon.com>
23348 Correct failures with --enable-checking=yes,rtl.
23350 * config/microblaze/microblaze.c (microblaze_expand_shift):
23351 Replace GET_CODE test with CONST_INT_P and INTVAL test with
23352 test for const0_rtx.
23353 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
23354 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
23356 2017-03-09 Richard Biener <rguenther@suse.de>
23358 PR tree-optimization/79977
23359 * graphite-scop-detection.c (scop_detection::merge_sese):
23360 Handle the case of extra exits to blocks dominating the entry.
23362 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
23364 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
23367 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
23369 PR rtl-optimization/79949
23370 * lra-constraints.c (process_alt_operands): Check memory when
23371 trying to predict a cycle. Print about the overall increase.
23373 2017-03-09 Richard Biener <rguenther@suse.de>
23375 PR middle-end/79971
23376 * gimple-expr.c (useless_type_conversion_p): Preserve
23377 TYPE_SATURATING for fixed-point types.
23379 2017-03-09 Richard Biener <rguenther@suse.de>
23382 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
23383 alignment of BLKmode params.
23385 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23388 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
23389 (VALL_NO_V2Q): Likewise.
23391 * config/aarch64/aarch64-simd.md
23392 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
23394 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
23395 VALL_NO_V2Q mode iterator.
23396 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
23398 2017-03-09 Martin Liska <mliska@suse.cz>
23400 PR tree-optimization/79631
23401 * tree-chkp-opt.c (chkp_is_constant_addr): Call
23402 tree_int_cst_sign_bit just for INTEGER constants.
23404 2017-03-09 Martin Liska <mliska@suse.cz>
23408 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
23411 2017-03-09 Marek Polacek <polacek@redhat.com>
23414 * tree.c (inchash::add_expr): Handle TREE_VEC.
23416 2017-03-09 Martin Liska <mliska@suse.cz>
23419 (chkp_narrow_size_and_offset): New function.
23420 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
23421 (void chkp_parse_bit_field_ref): New function.
23422 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
23423 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
23425 2017-03-09 Martin Liska <mliska@suse.cz>
23428 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
23429 (chkp_find_bounds_1): Remove gcc_unreachable.
23431 2017-03-09 Jakub Jelinek <jakub@redhat.com>
23434 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
23435 BUILT_IN_SYNC*, determine the access type from the size suffix and
23436 always build a MEM_REF with that type. Handle forgotten
23437 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
23440 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
23441 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
23442 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
23443 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
23444 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
23445 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
23446 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
23447 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
23448 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
23449 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
23450 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
23451 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
23452 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
23453 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
23454 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
23455 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
23456 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
23457 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
23458 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
23459 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
23460 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
23461 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
23462 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
23463 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
23464 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
23465 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
23466 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
23467 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
23468 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
23469 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
23470 definitions outside of __OPTIMIZE__ guarded section.
23473 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
23474 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
23475 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
23476 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
23479 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23481 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
23482 ("vfenez<mode>"): Add missing constraints.
23484 2017-03-08 Martin Sebor <msebor@redhat.com>
23487 * config/nds32/nds32.c (nds32_option_override):
23488 Fix misspelled diagnostic.
23490 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23493 * gimplify.c (gimplify_omp_for): Replace index var in outer
23494 taskloop statement with an artificial variable and add
23495 OMP_CLAUSE_PRIVATE clause for it.
23497 2017-03-08 Richard Biener <rguenther@suse.de>
23499 PR tree-optimization/79955
23500 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
23501 for accesses that are completely outside of the variable.
23503 2017-03-08 Andrew Haley <aph@redhat.com>
23505 PR tree-optimization/79943
23506 * tree-ssa-loop-split.c (compute_new_first_bound): When
23507 calculating the new upper bound, (END-BEG) should be added, not
23510 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23512 * config/avr/avr.md (setmemhi): Make sure match_dup
23513 operand number comes before match_scratch.
23515 2017-03-08 Richard Biener <rguenther@suse.de>
23517 PR tree-optimization/79920
23518 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
23519 with ncopies == 1 to ...
23520 (vect_transform_slp_perm_load): ... here. Properly compute
23521 all element loads by iterating VF times over the group. Do
23522 not handle ncopies (computed in a broken way) in
23523 vect_create_mask_and_perm.
23525 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23528 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
23529 is a uniform vector, use uniform_vector_p return value instead of
23530 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
23532 2017-03-07 Marek Polacek <polacek@redhat.com>
23534 PR middle-end/79809
23535 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
23536 (alloca_call_type): Likewise.
23538 2017-03-07 Martin Liska <mliska@suse.cz>
23540 * gcov.c (process_args): Put comment to correct location.
23542 2017-03-07 Martin Liska <mliska@suse.cz>
23544 PR middle-end/68270
23545 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
23546 Use array_at_struct_end_p instead of DECL_CHAIN (field).
23547 (chkp_narrow_bounds_for_field): Likewise.
23548 (chkp_parse_array_and_component_ref): Pass one more argument to
23551 2017-03-07 Richard Biener <rguenther@suse.de>
23553 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
23556 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
23558 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
23559 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
23561 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23564 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
23565 to end of description.
23566 (PARAM_MAX_STORES_TO_MERGE): Likewise.
23568 2017-03-07 Jakub Jelinek <jakub@redhat.com>
23570 PR rtl-optimization/79901
23571 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
23573 (*avx512f_<code><mode>3<mask_name>): ... this.
23574 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
23575 iterator instead of VI8_AVX2_AVX512BW.
23577 PR rtl-optimization/79901
23578 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
23579 min/max expander, expand it using expand_vec_cond_expr.
23582 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
23585 2017-03-06 Jakub Jelinek <jakub@redhat.com>
23588 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
23589 to void * for PCH reasons.
23590 * dwarf2out.c (output_loc_operands, output_die): Cast
23591 v.val_vec.array to unsigned char *.
23593 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
23596 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
23599 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
23601 PR rtl-optimization/79571
23602 * lra-constraints.c (process_alt_operands): Calculate static
23603 reject and subtract it from overall when only addresses will be
23606 2017-03-06 Julia Koval <julia.koval@intel.com>
23609 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
23610 incoming stack boundary to 128 for 64-bit targets.
23612 2017-03-06 Richard Biener <rguenther@suse.de>
23614 PR tree-optimization/79894
23615 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
23616 to NULL after folding it.
23618 2017-03-06 Richard Biener <rguenther@suse.de>
23620 PR tree-optimization/79824
23621 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
23622 check disabling peeling for gaps.
23624 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
23626 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
23627 attributes): Document gettimeofday.
23629 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
23631 * config/s390/s390.c (s390_option_override_internal): Set
23632 PARAM_MIN_VECT_LOOP_BOUND
23634 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
23636 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
23637 * config/s390/s390.md: Likewise.
23639 2017-03-06 Jakub Jelinek <jakub@redhat.com>
23642 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
23643 (<avx2_avx512>_perm<mode>): Rename to ...
23644 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
23646 (<avx512>_perm<mode>_mask): Rename to ...
23647 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
23649 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
23650 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
23651 instead of VI8F_256_512.
23652 (avx512f_perm<mode>): New define_expand.
23653 (avx512f_perm<mode>_mask): Likewise.
23654 (avx512f_perm<mode>_1<mask_name>): New define_insn.
23655 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
23657 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23659 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
23660 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
23662 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
23664 2017-03-06 Martin Liska <mliska@suse.cz>
23667 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
23668 when having a SSA NAME w/o VAR_DECL assigned to it.
23670 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23672 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
23673 msa_dpsub_<su>_d): Fix MODE for vec_select.
23675 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23677 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
23679 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
23681 2017-03-06 Richard Biener <rguenther@suse.de>
23683 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
23684 * plugin.c (register_plugin_info): Likewise.
23685 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
23687 2017-03-05 Jakub Jelinek <jakub@redhat.com>
23689 * config/i386/sse.md (sse_storehps, sse_storelps,
23690 avx_<castmode><avxsizesuffix>_<castmode>,
23691 avx512f_<castmode><avxsizesuffix>_<castmode>,
23692 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
23693 in condition that at least one operand is not a MEM.
23695 2017-03-03 Jakub Jelinek <jakub@redhat.com>
23697 PR middle-end/79805
23698 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
23699 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
23701 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
23702 gimple_call_nothrow_p flag based on whether original builtin can throw.
23703 If it can, emit following stmts on the fallthrough edge.
23704 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
23705 don't create new bb if inserting just debug stmts on the edge, try to
23706 insert them on the fallthru bb or just reset debug stmts.
23708 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
23711 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
23712 restore recog_data (including the operand rtxes inside it) around
23713 the call to get_insn_template.
23715 2017-03-03 Martin Sebor <msebor@redhat.com>
23717 PR tree-optimization/79699
23718 * context.c (context::~context): Free MPFR caches to avoid
23719 a memory leak on program exit.
23721 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23723 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
23724 Use wide_int::ulow () instead of .elt (0).
23726 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
23728 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
23729 (*pushxf): Limit oF constraint to 32bit targets and add oC
23730 constraint for 64bit targets.
23731 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
23732 (*pushdf): Change rmF constraint to rmC.
23734 2017-03-03 Martin Liska <mliska@suse.cz>
23736 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
23737 Remove unused variable.
23739 2017-03-03 Jakub Jelinek <jakub@redhat.com>
23742 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
23743 is a memory operand, increase num_memory.
23744 (ix86_expand_args_builtin): Likewise.
23746 2017-03-03 Jan Hubicka <jh@suse.cz>
23749 * ipa-devirt.c (maybe_record_node): Properly handle
23750 __cxa_pure_virtual visibility.
23752 2017-03-03 Martin Liska <mliska@suse.cz>
23754 PR tree-optimization/79803
23755 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
23757 (pass_loop_prefetch::execute): Disabled optimization if an
23758 assumption about L1 cache size is not met.
23760 2017-03-03 Martin Liska <mliska@suse.cz>
23762 PR rtl-optimization/79574
23763 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
23764 (hash_scan_set): Likewise.
23765 (dump_hash_table): Likewise.
23766 (hoist_code): Likewise.
23768 2017-03-03 Richard Biener <rguenther@suse.de>
23770 * fixed-value.c (fixed_from_string): Restore use of elt (1)
23771 in place of uhigh ().
23772 (fixed_convert_from_real): Likewise.
23774 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
23777 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
23779 2017-03-03 Richard Biener <rguenther@suse.de>
23781 PR middle-end/79818
23782 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
23783 TYPE_OVERFLOW_UNDEFINED check.
23785 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23787 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
23789 (vector_ae_<mode>_p): Likewise.
23790 (vector_nez_<mode>_p): Likewise.
23791 (vector_ne_v2di_p): Likewise.
23792 (vector_ae_v2di_p): Likewise.
23793 (vector_ne_<mode>_p): Likewise.
23794 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
23796 (vsx_tsqrt<mode>2_fe): Likewise.
23798 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
23801 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
23803 2017-03-02 Jakub Jelinek <jakub@redhat.com>
23805 PR rtl-optimization/79780
23806 * cprop.c (one_cprop_pass): When second and further conditional trap
23807 in a single basic block is turned into an unconditional trap, turn it
23808 into a deleted note to avoid RTL verification failures.
23810 2017-03-02 Richard Biener <rguenther@suse.de>
23812 * fold-const.c (const_binop): Use ulow () instead of elt (0).
23814 2017-03-02 Richard Biener <rguenther@suse.de>
23816 PR tree-optimization/79345
23818 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
23819 param and abort the walk, returning -1 if it is hit.
23820 (walk_aliased_vdefs): Take a limit param and pass it on.
23821 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
23822 defaulting to 0 and return a signed int.
23823 * tree-ssa-uninit.c (struct check_defs_data): New struct.
23824 (check_defs): New helper.
23825 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
23826 about uninitialized memory.
23827 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
23828 bogus uninitialized warning.
23829 (fixed_convert_from_real): Likewise.
23831 2017-03-02 Bin Cheng <bin.cheng@arm.com>
23833 PR tree-optimization/66768
23834 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
23835 iv_use if base object can't be determined.
23837 2017-03-02 Jakub Jelinek <jakub@redhat.com>
23839 PR tree-optimization/79345
23840 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
23841 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
23842 (get_pattern_stats): Initialize it.
23843 * genemit.c (gen_expand): Verify match_scratch numbers come after
23844 match_operand/match_dup numbers.
23845 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
23846 match_scratch numbers.
23847 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
23849 * config/s390/s390.md (trunctdsd2): Likewise.
23851 2017-03-02 Richard Biener <rguenther@suse.de>
23853 * wide-int.h (wide_int_storage::operator=): Implement in terms
23856 2017-03-02 Richard Biener <rguenther@suse.de>
23858 PR tree-optimization/79777
23859 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
23860 the to insert expression to sth existing.
23862 2017-03-01 Martin Sebor <msebor@redhat.com>
23864 PR middle-end/79692
23865 * gimple-ssa-sprintf.c
23866 (directive::known_width_and_precision): New function.
23867 (format_integer): Use it.
23868 (get_mpfr_format_length): Consider the full range of precision
23869 when computing %g output with the # flag. Set the likely byte
23870 count to 3 rather than 1 when precision is indeterminate.
23871 (format_floating): Correct the lower bound of precision.
23873 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23875 * doc/invoke.texi: Document default code model for 64-bit Linux.
23877 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23880 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
23881 udiv rather than div since input pattern is unsigned.
23883 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
23885 * config/i386/i386.c (print_reg): Warn for values of
23886 unsupported size in integer register.
23888 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
23891 * config/rs6000/predicates.md (current_file_function_operand): Do
23892 not allow self calls to be local if the function is replaceable.
23894 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
23897 * config/rs6000/altivec.h (vec_ctz and others): Change the
23898 preprocessor macro that controls conditional compilation from
23899 _ARCH_PWR9 to __POWER9_VECTOR__.
23900 (vec_all_ne): Change parameterization of __altivec_scalar_pred
23901 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
23902 control (instead of _ARCH_PWR9 control) so that template
23903 definition uses power9-specific function.
23904 (vec_any_eq): Likewise.
23905 (vec_all_ne): Change macro definition to use a power9-specific
23906 expansion under #ifdef __POWER9_VECTOR__ control (instead of
23907 _ARCH_PWR9 control).
23908 (vec_any_eq) Likewise.
23909 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
23910 expansion for CMPNEF to remove support for xvcmpnesp instruction.
23911 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
23912 support for xvcmpnedp instruction.
23913 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
23914 macro expansion so that Power9 implementation of vec_all_ne does
23915 not use the AltiVec predicate framework.
23916 (VCMPNEH_P): Likewise.
23917 (VCMPNEW_P): Likewise.
23918 (VCMPNED_P): Likewise.
23919 (VCMPNEFP_P): Likewise.
23920 (VCMPNEDP_P): Likewise.
23921 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
23922 implementation of vec_any_eq to not use AltiVec predicate
23924 (VCMPAEH_P): Likewise.
23925 (VCMPAEW_P): Likewise.
23926 (VCMPAED_P): Likewise.
23927 (VCMPAEFP_P): Likewise.
23928 (VCMPAEDP_P): Likewise.
23929 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
23930 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
23931 not use the AltiVec predicate framework.
23932 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
23933 of vec_any_eq to not use AltiVec predicate framework.
23934 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
23935 support for predefined __POWER9_VECTOR__ macro to indicate that
23936 Power9 instruction selection is enabled.
23937 (altivec_overloaded_builtins): Remove extraneous
23938 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
23939 function argument types RS6000_BTI_bool_V16QI and
23940 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
23941 entry for overloaded function argument types RS6000_BTI_bool_V4SI
23942 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
23943 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
23944 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
23945 Power9 for implementations of vec_cmpne. Change the signature for
23946 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
23947 (representing vec_all_ne) to remove the previously described first
23948 argument of type RS6000_BTI_INTSI, as this was an artifact of
23949 reliance on the AltiVec predicate framework, which is no longer
23950 used in the implementation of these functions. Add
23951 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
23952 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
23953 since, unlike the AltiVec predicate framework implementation, we
23954 do not share function descriptors between vec_alle and vec_anyeq.
23955 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
23956 set of modes that receive special treatment even when
23957 TARGET_P9_VECTOR is true. The special treatment emits code that
23958 does not depend on Power9 instructions.
23959 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
23960 define_expand to not rely on AltiVec predicate framework.
23961 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
23963 (vector_ne_v2di_p): Change this define_expand to not rely on
23964 AltiVec predicate framework.
23965 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
23967 (vector_ne_<mode>_p): Change this define_expand to not rely on
23968 AltiVec predicate framework.
23969 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
23971 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
23972 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
23973 define_insn pattern.
23974 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
23975 define_insn pattern because the xvcmpne<VSs>. instruction is not
23977 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
23978 instruction is not supported.
23980 2017-03-01 Jakub Jelinek <jakub@redhat.com>
23982 * config/nvptx/nvptx.c: Include intl.h.
23984 2017-03-01 Martin Jambor <mjambor@suse.cz>
23987 * ipa-prop.h (ipa_bits): Removed field known.
23988 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
23989 to pointers. Adjusted their comments to warn about their sharing.
23990 (ipcp_transformation_summary): Change bits to a vector of pointers.
23991 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
23992 (ipa_get_ipa_bits_for_value): Declare.
23993 * tree-vrp.h (value_range): Mark as GTY((for_user)).
23994 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
23995 (ipa_bits_hash_table): Likewise.
23996 (ipa_vr_ggc_hash_traits): Likewise.
23997 (ipa_vr_hash_table): Likewise.
23998 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
23999 being pointers and vr_known being removed.
24000 (ipa_set_jf_unknown): Likewise.
24001 (ipa_get_ipa_bits_for_value): New function.
24002 (ipa_set_jfunc_bits): Likewise.
24003 (ipa_get_value_range): New overloaded functions.
24004 (ipa_set_jfunc_vr): Likewise.
24005 (ipa_compute_jump_functions_for_edge): Use the above functions to
24006 construct bits and vr parts of jump functions.
24007 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
24008 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
24010 (ipcp_grow_transformations_if_necessary): Also allocate
24011 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
24013 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
24014 them. Fix too long lines.
24015 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
24016 vr_known being removed.
24017 (ipa_read_jump_function): Use new setter functions to construct bits
24018 and vr parts of jump functions or set them to NULL.
24019 (write_ipcp_transformation_info): Adjust for bits being pointers.
24020 (read_ipcp_transformation_info): Likewise.
24021 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
24023 Include gt-ipa-prop.h.
24024 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
24026 (ipcp_store_bits_results): Likewise.
24027 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
24028 Do not write to existing jump functions but use a temporary instead.
24030 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24033 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
24034 attempt to use its first operand as BIT_FIELD_REF base.
24036 2017-03-01 Richard Biener <rguenther@suse.de>
24038 PR middle-end/79721
24039 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
24040 interpolating formula in wrapping arithmetic.
24041 (chrec_apply): Convert chrec_evaluate return value to wanted type.
24043 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24045 PR tree-optimization/79734
24046 * tree-vect-generic.c (expand_vector_condition): Optimize
24047 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
24048 Handle VEC_COND_EXPR where comparison has different inner width from
24049 type's inner width.
24051 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
24053 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
24054 markup, and similar issues. Remove @opindex entries for things
24055 that aren't options. Add missing -mmpy-option entries.
24057 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24059 PR tree-optimization/79737
24060 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
24061 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
24062 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
24063 instead of byte_size. Formatting fix.
24064 (shift_bytes_in_array_right): Formatting fix.
24066 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
24069 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
24070 condition on optimize for the leaf function test.
24072 2017-02-28 Martin Liska <mliska@suse.cz>
24075 * read-rtl-function.c (function_reader::handle_unknown_directive):
24076 Bail out when one uses -flto.
24078 2017-02-28 Martin Liska <mliska@suse.cz>
24080 * common.opt: Replace space with tabular for options of <number>
24082 * config/i386/i386.opt: Show <number> value for
24083 -mlarge-data-threshold.
24084 * opts.c (print_filtered_help): Do not display number in hexadecimal
24087 2017-02-28 Martin Liska <mliska@suse.cz>
24089 * common.opt: Fix --help=option -Q for options which are of
24092 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
24094 * config/i386/i386.c (print_reg): Error out for values
24095 of 8-bit size in invalid integer register.
24097 2017-02-28 Martin Sebor <msebor@redhat.com>
24099 PR tree-optimization/79691
24100 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
24102 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24105 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
24106 gcc_unreachable with output_operand_lossage.
24108 2017-02-28 Richard Biener <rguenther@suse.de>
24110 PR tree-optimization/79740
24111 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
24113 (visit_nary_op): Insert the nary into the hashtable if we
24114 pattern-matched sth.
24115 * tree-ssa-pre.c (eliminate_insert): Robustify.
24117 2017-02-28 Richard Biener <rguenther@suse.de>
24119 PR middle-end/79731
24120 * fold-const.c (decode_field_reference): Reject out-of-bound
24123 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24125 * config/i386/i386.c: Include intl.h.
24126 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
24127 instead of just cond ? "..." : "...".
24128 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
24129 * coverage.c (read_counts_file): Likewise.
24130 * omp-offload.c: Include intl.h.
24131 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
24132 of just cond ? "..." : "...".
24133 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
24134 of just cond ? "..." : "...".
24136 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
24139 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
24141 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
24142 'tune for' CPU name.
24143 * config/arm/arm-cpu-data.h: Regenerated.
24145 2017-02-28 Richard Biener <rguenther@suse.de>
24147 PR tree-optimization/79732
24148 * tree-inline.c (expand_call_inline): Do not shadow var.
24150 2017-02-28 Richard Biener <rguenther@suse.de>
24152 PR tree-optimization/79723
24153 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
24154 address-space properly.
24156 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
24158 * doc/optinfo.texi (Optimization groups): Fix option used for
24160 * doc/invoke.texi (-fopt-info): Document "omp".
24161 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
24162 (OPTGROUP_ALL): Add OPTGROUP_OMP.
24163 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
24164 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
24165 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
24167 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
24169 * dumpfile.c (optgroup_options): Instead of "openmp", associate
24170 OPTGROUP_OMP with "omp".
24172 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
24175 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
24176 for arithmetic shift of unsigned V2DI.
24178 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
24180 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
24181 arc/linux.h headers.
24182 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
24183 (LINK_SPEC): Likewise.
24184 (ARC_TLS_EXTRA_START_SPEC): Likewise.
24185 (EXTRA_SPECS): Likewise.
24186 (STARTFILE_SPEC): Likewise.
24187 (ENDFILE_SPEC): Likewise.
24188 (LIB_SPEC): Likewise.
24189 (TARGET_SDATA_DEFAULT): Likewise.
24190 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
24191 (MULTILIB_DEFAULTS): Likewise.
24192 (DWARF2_UNWIND_INFO): Likewise.
24193 * config/arc/big.h: New file.
24194 * config/arc/elf.h: Likewise.
24195 * config/arc/linux.h: Likewise.
24196 * config/arc/t-uClibc: Remove.
24198 2017-02-27 Bin Cheng <bin.cheng@arm.com>
24200 PR tree-optimization/77536
24201 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
24202 (tree_transform_and_unroll_loop): Use above function to compute the
24203 estimated niter of unrolled loop and use it when scaling profile.
24204 Also use count info rather than frequency if it's non-zero.
24205 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
24206 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
24207 (vect_transform_loop): Call above function.
24209 2017-02-27 Richard Biener <rguenther@suse.de>
24211 PR tree-optimization/45397
24212 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
24213 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
24214 (visit_nary_op): Add pattern matching for CSEing sign-changed
24215 or truncated operations with wider ones.
24217 2017-02-27 Richard Biener <rguenther@suse.de>
24219 PR tree-optimization/79690
24220 * tree-vect-stmts.c (vectorizable_store): Use vector type
24221 built from the DR with address-space.
24223 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
24225 * doc/invoke.texi (Optimize Options): Refine the description
24226 of asan-use-after-return.
24228 2017-02-25 Alan Modra <amodra@gmail.com>
24230 PR rtl-optimization/79584
24231 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
24232 base, not ad->base_term, the reg within base. Remove assertion
24233 that ad->base == ad->base_term. Replace gen_int_mode using
24234 bogus mode with const0_rtx.
24236 2017-02-25 Jakub Jelinek <jakub@redhat.com>
24238 PR middle-end/79396
24239 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
24240 FMA_EXPR like tcc_binary or tcc_unary.
24242 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
24245 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
24247 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
24248 (output_loc_operands): Handle DW_OP_call_ref and
24249 DW_OP_GNU_variable_value.
24250 (struct variable_value_struct): New type.
24251 (struct variable_value_hasher): Likewise.
24252 (variable_value_hash): New variable.
24253 (string_types): Remove.
24254 (copy_loc_descr): New function.
24255 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
24256 (prepend_loc_descr_to_each): New function.
24257 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
24258 instead of add_loc_descr_to_each if the first argument is single
24259 location list and the second has multiple.
24260 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
24261 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
24262 when looking for variable value which doesn't have other location info.
24263 (loc_list_from_tree): Formatting fix.
24264 (gen_array_type_die): Simplify DW_AT_string_length handling.
24265 (adjust_string_types): Remove.
24266 (gen_subprogram_die): Don't call adjust_string_types nor test/set
24267 string_types. Call resolve_variable_values.
24268 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
24269 (resolve_addr_in_expr): Likewise. Add A argument.
24270 (copy_deref_exprloc): Remove deref argument. Adjust for the
24271 original expression being DW_OP_GNU_variable_value with optionally
24272 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
24273 optionally after it.
24274 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
24275 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
24276 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
24277 (variable_value_hasher::hash, variable_value_hasher::equal): New
24279 (resolve_variable_value_in_expr, resolve_variable_value,
24280 resolve_variable_values, note_variable_value_in_expr,
24281 note_variable_value): New functions.
24282 (dwarf2out_early_finish): Call note_variable_value on all toplevel
24285 2017-02-24 Jakub Jelinek <jakub@redhat.com>
24288 * opts.h (handle_generated_option): Add GENERATED_P argument.
24289 * opts-common.c (handle_option): Adjust function comment.
24290 (handle_generated_option): Add GENERATED_P argument, pass it to
24292 (control_warning_option): Pass false to handle_generated_option
24294 * opts.c (maybe_default_option): Pass true to handle_generated_option
24296 * optc-gen.awk: Likewise.
24298 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24300 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
24301 a REG, look at the REG it is a SUBREG of.
24302 (splitter for cmpeqsi_t): Ditto.
24304 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24306 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
24307 the special USEs with the pattern of the insn, not the insn itself.
24309 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
24312 * doc/invoke.texi: Document -mload-store-pairs.
24314 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24315 Sandra Loosemore <sandra@codesourcery.com>
24317 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
24318 argument isn't a CONST_INT.
24319 (nios2_alternate_compare_const): Assert op is a CONST_INT.
24320 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
24321 (nios2_validate_compare): Bypass alternate compare logic if *op2
24322 is not a CONST_INT.
24323 (ldstwm_operation_p): Return false if first_base is not a REG or
24324 if first_offset is not a CONST_INT.
24326 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24328 * config/cris/cris.md: Use correct operand in a define_peephole2.
24330 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24332 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
24334 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24336 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
24337 this_insn if it is an INSN or JUMP_INSN.
24338 (force_offsettable): Look at base, not at addr.
24339 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
24340 on things that aren't necessarily CONST_INTs.
24342 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
24344 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
24345 -mfpmath=sse is the default also for x86-32 targets with SSE2
24346 instruction set when @option{-ffast-math} is enabled
24348 2017-02-24 Jeff Law <law@redhat.com>
24350 PR rtl-optimizatoin/79286
24351 * ira.c (update_equiv_regs): Drop may_trap_p exception to
24354 2017-02-24 Richard Biener <rguenther@suse.de>
24356 PR tree-optimization/79389
24357 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
24360 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
24362 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
24363 function comment to reflect reality.
24364 (loop_exits_before_overflow): Fix typo in function description.
24366 2017-02-24 Richard Biener <rguenther@suse.de>
24368 PR tree-optimization/79389
24369 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
24370 properly that a threading opportunity exists. Detect conditional
24371 copy/constant propagation opportunities.
24373 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
24375 * config/visium/visium.md (type): Add trap.
24376 (b): New mode attribute.
24377 (*btst): Rename into...
24378 (*btst<mode>): ...this and adjust.
24379 (*cbranchsi4_btst_insn): Rename into...
24380 (*cbranch<mode>4_btst_insn): ...this and adjust.
24381 (trap): New define_insn.
24383 2017-02-23 Jakub Jelinek <jakub@redhat.com>
24385 PR tree-optimization/79389
24386 * ifcvt.c (struct noce_if_info): Add rev_cond field.
24387 (noce_reversed_cond_code): New function.
24388 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
24389 reversed_comparison_code. Formatting fix.
24390 (noce_try_store_flag): Test rev_cond != NULL in addition to
24391 reversed_comparison_code.
24392 (noce_try_store_flag_constants): Likewise.
24393 (noce_try_store_flag_mask): Likewise.
24394 (noce_try_addcc): Use rev_cond if non-NULL instead of
24395 reversed_comparison_code.
24396 (noce_try_cmove_arith): Likewise. Formatting fixes.
24397 (noce_try_minmax, noce_try_abs): Clear rev_cond.
24398 (noce_find_if_block): Initialize rev_cond.
24399 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
24400 instead of false as last argument never attempt to reverse it
24403 2017-02-23 Bin Cheng <bin.cheng@arm.com>
24405 PR tree-optimization/79663
24406 * tree-predcom.c (combine_chains): Process refs in reverse order
24407 only for ZERO length chains, and add explaining comment.
24409 2017-02-23 Jeff Law <law@redhat.com>
24411 PR tree-optimization/79578
24412 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
24413 in call to operand_equal_p.
24415 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
24418 * config/i386/cpuid.h: Fix another undefined behavior.
24420 2017-02-23 Richard Biener <rguenther@suse.de>
24422 PR tree-optimization/79683
24423 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
24424 vector types for data-refs.
24426 2017-02-23 Martin Liska <mliska@suse.cz>
24428 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
24430 2017-02-23 Jakub Jelinek <jakub@redhat.com>
24432 PR middle-end/79665
24433 * internal-fn.c (get_range_pos_neg): Moved to ...
24434 * tree.c (get_range_pos_neg): ... here. No longer static.
24435 * tree.h (get_range_pos_neg): New prototype.
24436 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
24437 are known to be in between 0 and signed maximum inclusive, try to
24438 expand both unsigned and signed divmod and use the cheaper one from
24441 2017-02-22 Jeff Law <law@redhat.com>
24443 PR tree-optimization/79578
24444 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
24445 to compare base operands.
24447 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
24450 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
24451 gpc_reg_operand instead of fpr_reg_operand.
24453 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
24455 * config/mips/mips.c (mips_return_in_memory): Force FP
24456 vector types to be returned in memory for o32 ABI.
24458 2017-02-22 Jakub Jelinek <jakub@redhat.com>
24460 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
24461 instead of DW_TAG_member for static data member declarations and don't
24462 set no_linkage_name for static inline data members.
24463 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
24466 2017-02-22 Martin Liska <mliska@suse.cz>
24468 * doc/invoke.texi: Replace inequality signs with square brackets
24471 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24473 PR tree-optimization/68644
24474 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
24476 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
24479 * lra-constraints.c (simplify_operand_subreg): Handle
24480 WORD_REGISTER_OPERATIONS targets.
24482 2017-02-22 Jakub Jelinek <jakub@redhat.com>
24485 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
24486 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
24487 elimination by swapping fld*.
24489 2017-02-22 Richard Biener <rguenther@suse.de>
24491 PR tree-optimization/79673
24492 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
24493 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
24494 irrelevant address-space qualifiers and avoiding a
24495 ADDR_SPACE_CONVERT_EXPR from fold_convert.
24497 2017-02-22 Richard Biener <rguenther@suse.de>
24499 PR tree-optimization/79666
24500 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
24501 to not symbolically negate if that may introduce undefined
24504 2017-02-22 Martin Liska <mliska@suse.cz>
24507 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
24508 * data-streamer-out.c (streamer_write_gcov_count_stream):
24510 * value-prof.c (stream_out_histogram_value): Make assert more
24511 precise based on type of counter.
24513 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
24516 * config/i386/i386.md (standard_x87sse_constant_load splitter):
24517 Use nonimmediate_operand instead of memory_operand for operand 1.
24518 (float-extend standard_x87sse_constant_load splitter): Ditto.
24520 2017-02-21 Jeff Law <law@redhat.com>
24522 PR tree-optimization/79621
24523 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
24524 blocks with edges to themselves.
24526 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24529 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
24530 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
24531 Use gimple_call_builtin_p.
24534 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
24535 on temporarily removed DEBUG_INSNs.
24537 PR tree-optimization/79649
24538 * tree-loop-distribution.c (classify_partition): Give up on
24539 non-generic address space loads/stores.
24541 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
24543 * doc/loop.texi (Loop manipulation): Remove nonexistent
24544 tree_ssa_loop_version from the documentation.
24545 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
24547 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24550 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
24551 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
24552 * config/rs6000/rs6000.c: Include except.h.
24553 (rs6000_expand_split_stack_prologue): Call
24554 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
24556 2017-02-21 Martin Jambor <mjambor@suse.cz>
24559 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
24560 have been analyzed.
24562 2017-02-21 Martin Jambor <mjambor@suse.cz>
24564 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
24565 for backward compatibility only.
24566 * doc/invoke.texi (Option Summary): Remove all references to
24567 -fipa-cp-alignment.
24569 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
24573 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24575 * lra-constraints.c (curr_insn_transform): Handle
24576 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
24578 2017-02-21 Martin Liska <mliska@suse.cz>
24580 * config/i386/i386.opt: Replace -masm-dialect with -masm.
24582 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
24584 PR translation/79638
24585 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
24587 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
24590 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
24591 (arm_function_ok_for_sibcall): Return false for an indirect call by
24592 descriptor if all the argument registers are used.
24593 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
24594 alignment of the function.
24596 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24598 PR tree-optimization/61441
24599 * simplify-rtx.c (simplify_const_unary_operation): For
24600 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
24601 the sNaN unmodified.
24603 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
24605 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
24606 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
24607 instead of SYSTEM_HEADER_DIR.
24609 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
24610 Martin Liška <mliska@suse.cz>
24612 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
24613 Fix typos and grammar, use active voice, and clarify.
24615 2017-02-20 Marek Polacek <polacek@redhat.com>
24617 PR middle-end/79537
24618 * gimplify.c (gimplify_expr): Handle unused *&&L;.
24621 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
24623 2017-02-20 Jakub Jelinek <jakub@redhat.com>
24626 * config/i386/i386.c (ix86_expand_builtin): Handle
24627 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
24628 ix86_builtins_isa[fcode].isa as a requirement of those
24629 flags and any other flag in the bitmask.
24630 (ix86_init_mmx_sse_builtins): Use 0 instead of
24631 ~OPTION_MASK_ISA_64BIT as mask.
24632 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
24633 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
24634 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
24635 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
24637 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24640 * lra-constraints.c (split_reg): Check requested split mode
24641 is supported by the register.
24643 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24645 * lra-constraints.c (simplify_operand_subreg): Remove early
24648 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24651 * lra-constraints.c (curr_insn_transform): Tighten condition
24652 for converting SUBREG reloads from OP_OUT to OP_INOUT.
24654 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24657 * lra-constraints.c (curr_insn_transform): Handle
24658 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
24660 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
24663 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
24665 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
24667 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
24670 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
24673 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24675 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
24678 2017-02-18 Jakub Jelinek <jakub@redhat.com>
24680 * final.c (last_columnnum, override_columnnum): New variables.
24681 (final_start_function): Set last_columnnum, pass it to begin_prologue
24682 hook and pass 0 to dwarf2out_begin_prologue.
24683 (final_scan_insn): Update override_columnnum. Pass last_columnnum
24684 to source_line debug hook.
24685 (notice_source_line): Compute last_columnnum and for debug_column_info
24686 return true on column changes.
24687 * debug.h (struct gcc_debug_hooks): Add column argument to
24688 source_line and begin_prologue hooks.
24689 (debug_nothing_int_charstar_int_bool): Remove prototype.
24690 (debug_nothing_int_int_charstar,
24691 debug_nothing_int_int_charstar_int_bool): New prototypes.
24692 (dwarf2out_begin_prologue): Add column argument.
24693 * debug.c (do_nothing_debug_hooks): Adjust source_line and
24694 begin_prologue hooks.
24695 (debug_nothing_int_charstar_int_bool): Remove.
24696 (debug_nothing_int_int_charstar,
24697 debug_nothing_int_int_charstar_int_bool): New functions.
24698 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
24699 through to dwarf2out_source_line.
24700 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
24701 (dwarf2out_source_line): Add column argument, emit it if requested.
24702 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
24704 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
24705 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
24706 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
24707 through to dwarf2out_begin_prologue.
24708 (vmsdbgout_source_line): Add column argument, pass it through to
24709 dwarf2out_source_line.
24710 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
24711 dbxout_source_line caller.
24712 (dbxout_source_line): Add column argument.
24714 * common.opt (gno-column-info, gcolumn-info): New options.
24715 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
24716 (check_die): Also test for multiple DW_AT_decl_column attributes.
24717 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
24718 DW_AT_decl_column if requested.
24719 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
24721 (gen_variable_die): Likewise.
24722 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
24723 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
24726 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
24727 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
24728 (ix86_handle_option): Handle OPT_m3dnowa.
24729 * doc/invoke.texi (-m3dnowa): Document.
24730 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
24731 -m3dnowa instead of -m3dnow -march=athlon.
24734 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
24735 instead of gcc_assert for K, r and R code checks. Formatting fixes.
24737 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24740 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
24741 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
24742 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
24743 generator for vsx_xxpermdi_<mode>_be.
24744 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
24745 force big-endian semantics.
24746 (vsx_xxpermdi_<mode>_be): New define_expand with same
24747 implementation as previous version of vsx_xxpermdi_<mode>.
24749 2017-02-17 Jakub Jelinek <jakub@redhat.com>
24751 PR tree-optimization/79327
24752 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
24753 variable, its initialization and use.
24755 2017-02-17 Julia Koval <julia.koval@intel.com>
24757 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
24758 (OPTION_MASK_ISA_PKU_UNSET): New.
24759 (ix86_handle_option): Handle -mrdpid.
24760 * config/i386/cpuid.h (bit_RDPID): New.
24761 * config/i386/driver-i386.c (host_detect_local_cpu):
24762 Detect RDPID feature.
24763 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
24764 * config/i386/i386-c.c (ix86_target_macros_internal):
24766 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
24767 (ix86_valid_target_attribute_inner_p): Add "rdpid".
24768 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
24769 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
24770 * config/i386/i386.md (define_insn "rdpid"): New.
24771 * config/i386/i386.opt Add -mrdpid.
24772 * config/i386/immintrin.h (_rdpid_u32): New.
24774 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
24776 PR rtl-optimization/79541
24777 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
24778 instead of transforming it into USE.
24780 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
24782 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
24783 If HONOR_SNANS (SFmode) force the input to a register.
24784 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
24785 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
24786 an frsp or similar insn.
24788 2017-02-17 Martin Liska <mliska@suse.cz>
24790 PR rtl-optimization/79577
24791 * params.def (selsched-max-sched-times): Increase minimum to 1.
24793 2017-02-17 Martin Liska <mliska@suse.cz>
24795 PR rtl-optimization/79574
24796 * gcse.c (want_to_gcse_p): Prevent integer overflow.
24798 2017-02-17 Martin Liska <mliska@suse.cz>
24800 PR tree-optimization/79529
24801 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
24802 ssa_defined_default_def_p to handle cases which are implicitly
24804 * tree-ssa.c (ssa_defined_default_def_p): New function.
24805 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
24806 which are implicitly defined.
24807 * tree-ssa.h (ssa_defined_default_def_p): Declare.
24809 2017-02-17 Richard Biener <rguenther@suse.de>
24811 PR middle-end/79576
24812 * params.def (max-ssa-name-query-depth): Limit to 10.
24814 2017-02-17 Richard Biener <rguenther@suse.de>
24816 PR tree-optimization/79552
24817 * tree-ssa-structalias.c (visit_loadstore): Properly verify
24820 2017-02-17 Richard Biener <rguenther@suse.de>
24823 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
24825 2017-02-17 Marek Polacek <polacek@redhat.com>
24827 PR middle-end/79536
24828 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
24829 (fold_negate_expr): New wrapper.
24831 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
24833 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
24834 Correct terminology and de-emphasize pre-standard behavior.
24836 2017-02-16 Alan Modra <amodra@gmail.com>
24838 PR rtl-optimization/79286
24839 * ira.c (def_dominates_uses): New function.
24840 (update_equiv_regs): Don't create an equivalence for insns that
24841 may trap where the register def does not dominate the use.
24843 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
24845 PR rtl-optimization/78127
24846 * lra.c (lra): Call lra_eliminate before finish the loop after
24849 2017-02-16 Richard Biener <rguenther@suse.de>
24851 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
24853 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
24854 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
24855 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
24856 (isl_val_int_from_wi): New function.
24857 (extract_affine_gmp): Rename to ...
24858 (extract_affine_wi): ... this, take a widest_int.
24859 (extract_affine_int): Just wrap extract_affine_wi.
24860 (add_param_constraints): Use isl_val_int_from_wi.
24861 (add_loop_constraints): Likewise, and extract_affine_wi.
24863 2017-02-15 Jeff Law <law@redhat.com>
24865 PR middle-end/79521
24866 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
24867 ira_init_register_move_cost_if_necessary.
24869 2017-02-15 Martin Sebor <msebor@redhat.com>
24871 PR middle-end/32003
24872 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
24873 removed in a prior commit.
24875 2017-02-15 Bin Cheng <bin.cheng@arm.com>
24877 PR tree-optimization/79347
24878 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
24879 counters during peeling.
24881 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
24883 * Makefile.in (site.exp): Remove "set ISLVER".
24885 2017-02-15 Jakub Jelinek <jakub@redhat.com>
24888 * real.c (real_from_integer): Call real_convert even for decimal.
24890 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
24893 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
24895 2017-02-14 Andrew Pinski <apinski@cavium.com>
24897 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
24898 cores and change the partno/implementer to be correct.
24899 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
24900 the 'B" as the implementer.
24901 * config/aarch64/aarch64-tune.md: Regenerate.
24903 2017-02-14 Carl Love <cel@us.ibm.com>
24905 * config/rs6000/rs6000.c: Add case statement entry to make the
24906 xvcvuxdsp built-in argument unsigned.
24907 * config/rs6000/vsx.md: Fix the source and return operand types so they
24908 match the instruction definitions from the ISA document. Fix typo
24909 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
24912 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
24915 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
24916 member early_clobber_alts.
24917 * lra-lives.c (reg_early_clobber_p): New.
24918 (process_bb_lives): Use it.
24919 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
24920 (debug_operand_data): Initialize early_clobber_alts.
24921 (setup_operand_alternative): Set up early_clobber_alts.
24922 (collect_non_operand_hard_regs): Ditto. Pass early clobber
24923 alternatives to new_insn_reg.
24924 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
24926 (lra_update_insn_regno_info): Pass the new arg.
24928 2017-02-14 Jakub Jelinek <jakub@redhat.com>
24930 PR middle-end/79505
24931 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
24932 (new_oacc_loop_raw): Don't clear already cleared fields.
24935 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
24936 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
24937 _mm512_prefetch_i64gather_ps): New inline functions and macros.
24939 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
24942 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
24944 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
24947 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
24948 the extra instruction to the right place to store 128-bit constant
24951 2017-02-14 Martin Sebor <msebor@redhat.com>
24953 PR middle-end/79448
24954 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
24955 warning for strings of unknown length.
24957 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
24959 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
24961 2017-02-14 Jeff Law <law@redhat.com>
24964 * ira-costs.c (scan_one_insn): Initialize register move costs
24965 for pseudos seen in USE/CLOBBER insns.
24967 PR tree-optimization/79095
24968 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
24969 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
24970 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
24971 if the operands are known to be not equal, then the resulting range
24973 (intersect_ranges): If the new range is ~[0,0] and the old range is
24974 wide, then prefer ~[0,0].
24975 * tree-vrp.c (overflow_comparison_p_1): New function.
24976 (overflow_comparison_p): New function.
24977 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
24978 if NAME is used in an overflow test.
24979 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
24980 overflow check that can be expressed as an equality test, then adjust
24981 ops to be that equality test.
24983 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24985 * config/s390/s390-builtin-types.def: Remove flags argument.
24986 * config/s390/s390.c (s390_init_builtins): Likewise.
24988 2017-02-14 Martin Liska <mliska@suse.cz>
24990 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
24991 vector. Fix trailing white spaces.
24993 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
24995 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
24998 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25000 PR rtl-optimization/68664
25001 * config/arm/arm.c (arm_sched_can_speculate_insn):
25002 New function. Declare prototype.
25003 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
25005 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25007 PR rtl-optimization/68664
25008 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
25010 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
25012 2017-02-14 Amit Pawar <amit.pawar@amd.com>
25014 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
25015 max skip bytes for function, loop and jump.
25017 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25019 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
25020 ABS_EXPR for gimple dump.
25022 2017-02-14 Jakub Jelinek <jakub@redhat.com>
25025 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
25027 PR tree-optimization/79408
25028 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
25029 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
25030 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
25031 also if rhs1 is INTEGER_CST.
25033 2017-02-14 Richard Biener <rguenther@suse.de>
25035 PR middle-end/79432
25036 * tree-into-ssa.c (insert_phi_nodes): When the function can
25037 have abnormal edges rewrite SSA names with broken use-def
25038 dominance out of SSA and register them for PHI insertion.
25040 2017-02-13 Martin Sebor <msebor@redhat.com>
25042 PR middle-end/79496
25043 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
25044 clearing info.nowrite flag when snprintf size argument is a range.
25046 2017-02-13 Jakub Jelinek <jakub@redhat.com>
25048 * cprop.c (cprop_jump): Add missing space in string literal.
25049 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
25050 (get_constraint_for_component_ref): Likewise.
25051 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
25052 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
25053 * lra-constraints.c (process_alt_operands): Likewise.
25054 * ipa-inline.c (inline_small_functions): Likewise.
25055 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
25056 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
25057 * trans-mem.c (diagnose_tm_1_op): Likewise.
25058 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
25059 (grid_parallel_clauses_gridifiable): Likewise.
25061 * config/nvptx/mkoffload.c (process): Add space in between
25064 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
25065 "MOD4_SSE_REGS" and "ALL_REGS".
25067 * spellcheck.c (test_data): Add , in between "foo" and "food".
25069 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25072 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
25073 boundary crossing check and subsequent code generation agree.
25075 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25077 * config/aarch64/aarch64.c (has_memory_op): Delete.
25078 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
25081 2017-02-13 Jakub Jelinek <jakub@redhat.com>
25083 PR rtl-optimization/79388
25084 PR rtl-optimization/79450
25085 * combine.c (distribute_notes): When removing TEM_INSN for which
25086 corresponding dest has last value recorded, invalidate that last
25089 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25091 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
25092 of explicit '@'. Add missing assembly comment marker on branch costs
25095 2017-02-13 Nathan Sidwell <nathan@acm.org>
25097 * gengtype-lex.l (<in_struct>): Add '/'.
25099 2017-02-13 Martin Liska <mliska@suse.cz>
25102 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
25104 2017-02-13 Richard Biener <rguenther@suse.de>
25106 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
25108 * configure: Re-generate.
25109 * config.in: Likewise.
25110 * graphite-dependences.c: Simplify as if
25111 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
25112 * graphite-isl-ast-to-gimple.c: Likewise.
25113 * graphite-optimize-isl.c: Likewise.
25114 * graphite-poly.c: Likewise.
25115 * graphite-sese-to-poly.c: Likewise.
25116 * graphite.h: Likewise.
25117 * toplev.c: Include isl/version.h and use isl_version () for
25118 printing the ISL version.
25119 * doc/install.texi: Update ISL requirement.
25121 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
25123 * doc/standards.texi (Standards): Update reference to
25126 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
25128 * doc/extend.texi (Named Address Spaces): sourceware.org now
25130 * doc/install.texi (Binaries): Ditto.
25133 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
25135 * doc/cpp.texi: Replace "stringify"/"stringification" with C
25136 standard terminology "stringize"/"stringizing" throughout.
25137 * doc/cppinternals.texi: Likewise.
25139 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
25141 * doc/extend.texi: Fix some spelling mistakes and typos.
25142 * doc/invoke.texi: Likewise.
25144 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
25147 * params.def (inline-min-speedup) Change from 10 to 8.
25149 2017-02-11 Jakub Jelinek <jakub@redhat.com>
25151 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
25154 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
25157 * ipa-inline-analysis.c (get_minimal_bb): New function.
25158 (record_modified): Use it.
25159 (remap_edge_change_prob): Handle also ancestor functions.
25161 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
25163 * doc/contrib.texi (Contributors): Remove broken link into
25164 the Mauve CVS repository.
25166 2017-02-11 Jakub Jelinek <jakub@redhat.com>
25168 PR middle-end/79454
25169 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
25170 result computation whenever lhs doesn't have vector mode, not
25171 just when it has BLKmode.
25173 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
25175 * doc/makefile.texi (profiledbootstrap): Refer to the
25176 installation instructions only in textual form.
25178 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25181 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
25183 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
25185 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
25186 (Specific): Update mingw-w64 reference.
25188 (Specific): Remove broken link to Renesas RX processor.
25190 2017-02-10 Richard Biener <rguenther@suse.de>
25192 * toplev.c (process_options): Do not mention obsolete graphite
25193 options when printing sorry message about missing graphite support.
25194 Mention -floop-nest-optimize.
25196 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
25198 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
25199 (vtst_p16): Likewise.
25200 (vtstq_p8): Likewise.
25201 (vtstq_p16): Likewise.
25203 (vtstq_p64): Likewise.
25204 * config/arm/arm_neon.h (vgetq_lane_p64): New.
25205 (vset_lane_p64): New.
25206 (vsetq_lane_p64): New.
25208 2017-02-10 Jakub Jelinek <jakub@redhat.com>
25210 PR tree-optimization/79411
25211 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
25212 stmt operands are SSA_NAMEs used in abnormal phis.
25213 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
25216 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
25219 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
25222 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
25224 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
25226 2017-02-09 Jakub Jelinek <jakub@redhat.com>
25228 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
25232 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
25233 not arbitrary TREE_CONSTANT.
25236 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
25237 "omp declare target link" attribute unless is_global_var.
25238 * omp-offload.c (find_link_var_op): Likewise.
25240 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
25241 Chung-Lin Tang <cltang@codesourcery.com>
25243 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
25245 (gimplify_adjust_omp_clauses): Don't delete TILE.
25246 (gimplify_omp_for): Deal with TILE.
25247 * internal-fn.c (expand_GOACC_TILE): New function.
25248 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
25250 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
25251 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
25253 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
25254 avoid DIV for outermost collapse var.
25255 (expand_oacc_for): Insert tile element loop as needed. Adjust.
25256 Remove out of date comments, fix whitespace.
25257 * omp-general.c (omp_extract_for_data): Deal with tiling.
25258 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
25259 adjust OLF_DIM_BASE value.
25260 (struct omp_for_data): Add tiling field.
25261 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
25262 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
25263 for auto loops. Remove default auto determining, moved to
25264 oacc_loop_fixed_partitions.
25265 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
25266 stmts, add e_mask field.
25267 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
25268 (oacc_thread_numbers): Use oacc_dim_call.
25269 (oacc_xform_tile): New.
25270 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
25271 (finish_oacc_loop): Adjust for ifns vector.
25272 (oacc_loop_discover_walk): Append loop abstraction sites to list,
25273 add case for GOACC_TILE fns.
25274 (oacc_loop_xform_loop): Delete.
25275 (oacc_loop_process): Iterate over call list directly, and add
25276 handling for GOACC_TILE fns.
25277 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
25279 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
25280 vector partitioning to outer loops. Assign 2 partitions to loops
25281 when available. Add TILE handling.
25282 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
25283 (execite_oacc_device_lower): Process GOACC_TILE fns,
25284 ignore unknown specs.
25285 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
25286 * tree.c (omp_clause_num_ops): Adjust TILE ops.
25287 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
25289 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
25291 * configure.ac (ACX_BUGURL): Update.
25292 * configure: Regenerate.
25294 2017-02-09 Richard Biener <rguenther@suse.de>
25296 PR tree-optimization/69823
25297 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
25298 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
25300 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
25302 * config/arc/arc-c.def: Add __NPS400__ definition.
25303 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
25304 (TARGET_NPS400): Define.
25306 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
25308 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
25310 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
25311 pointer, arch_info.
25312 (arc_cpu_types): Fill the arch_info field with a pointer into the
25313 arc_arch_types table.
25314 (arc_selected_cpu): Declare.
25315 * config/arc/arc.c (arc_selected_cpu): Make global.
25316 (arc_selected_arch): Delete.
25317 (arc_base_cpu): Delete.
25318 (arc_override_options): Remove references to deleted variables,
25319 update access to arch information.
25320 (ARC_OPT): Update access to arch information.
25321 (ARC_OPTX): Likewise.
25322 * config/arc/arc.h (arc_base_cpu): Remove declaration.
25323 (TARGET_ARC600): Update access to arch information.
25324 (TARGET_ARC601): Likewise.
25325 (TARGET_ARC700): Likewise.
25326 (TARGET_EM): Likewise.
25327 (TARGET_HS): Likewise.
25328 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
25331 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
25334 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
25335 condition/operands for integer GE/LE/GEU/LEU operations.
25337 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
25339 PR translation/79397
25340 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
25343 2017-02-08 Martin Jambor <mjambor@suse.cz>
25346 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
25347 whether allocation happened.
25348 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
25349 nothing was allocated.
25351 2017-02-08 Jakub Jelinek <jakub@redhat.com>
25353 PR tree-optimization/79408
25354 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
25355 constant, but SSA_NAME with a known integer range, use the minimum
25356 of that range instead of op1 to determine if modulo can be replaced
25357 with its first operand.
25359 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25361 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
25363 2017-02-08 Richard Biener <rguenther@suse.de>
25365 PR tree-optimization/71824
25366 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
25367 Check all loops contained in the merged region.
25369 2017-02-07 Andrew Pinski <apinski@cavium.com>
25371 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
25373 2017-02-07 Andrew Pinski <apinski@cavium.com>
25375 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
25376 (thunderxt88): Likewise.
25377 (thunderxt81): Disable LSE and change v8.1 to v8.
25378 (thunderxt83): Likewise.
25380 2017-02-07 Jakub Jelinek <jakub@redhat.com>
25381 Richard Biener <rguenther@suse.de>
25383 PR middle-end/79399
25384 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
25385 type from int to size_t.
25386 * ira-costs.c (struct_costs_size): Change type from int to size_t.
25388 2017-02-07 Jakub Jelinek <jakub@redhat.com>
25390 PR rtl-optimization/79386
25391 * cprop.c (bypass_conditional_jumps): Initialize
25392 bypass_last_basic_block already before splitting bbs after
25393 unconditional traps...
25394 (bypass_conditional_jumps): ... rather than here.
25397 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
25398 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
25399 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
25400 fix -masm=intel patterns.
25402 2017-02-07 Richard Biener <rguenther@suse.de>
25404 PR tree-optimization/79256
25405 PR middle-end/79278
25406 * builtins.c (get_object_alignment_2): Use min_align_of_type
25407 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
25408 and ADJUST_FIELD_ALIGN.
25410 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
25412 * doc/tm.texi: Regenerate.
25413 * stor-layout.c (layout_decl): Adjust.
25414 (update_alignment_for_field): Likewise.
25415 (place_field): Likewise.
25416 (min_align_of_type): Likewise.
25417 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
25418 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
25419 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
25420 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
25421 * config/frv/frv.c (frv_adjust_field_align): Likewise.
25422 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
25423 * config/i386/i386.c (x86_field_alignment): Likewise.
25424 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
25425 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
25426 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
25427 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
25428 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
25429 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
25433 2017-01-30 Richard Biener <rguenther@suse.de>
25435 PR tree-optimization/79256
25436 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
25437 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
25440 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
25442 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
25443 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
25444 builtins to SImode and emit a zero-extend, if necessary.
25446 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25448 * docs/invoke.texi (RISC-V Options): Alphabetize.
25450 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25452 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
25455 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25457 * config/riscv/riscv.c: New file.
25458 * common/config/riscv/riscv-common.c: Likewise.
25459 * config.gcc: Likewise.
25460 * config/riscv/constraints.md: Likewise.
25461 * config/riscv/elf.h: Likewise.
25462 * config/riscv/generic.md: Likewise.
25463 * config/riscv/linux.h: Likewise.
25464 * config/riscv/multilib-generator: Likewise.
25465 * config/riscv/peephole.md: Likewise.
25466 * config/riscv/pic.md: Likewise.
25467 * config/riscv/predicates.md: Likewise.
25468 * config/riscv/riscv-builtins.c: Likewise.
25469 * config/riscv/riscv-c.c: Likewise.
25470 * config/riscv/riscv-ftypes.def: Likewise.
25471 * config/riscv/riscv-modes.def: Likewise.
25472 * config/riscv/riscv-opts.h: Likewise.
25473 * config/riscv/riscv-protos.h: Likewise.
25474 * config/riscv/riscv.h: Likewise.
25475 * config/riscv/riscv.md: Likewise.
25476 * config/riscv/riscv.opt: Likewise.
25477 * config/riscv/sync.md: Likewise.
25478 * config/riscv/t-elf-multilib: Likewise.
25479 * config/riscv/t-linux: Likewise.
25480 * config/riscv/t-linux-multilib: Likewise.
25481 * config/riscv/t-riscv: Likewise.
25482 * configure.ac: Likewise.
25483 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
25484 Waterman as RISC-V maintainers.
25485 * doc/install.texi: Add RISC-V entries.
25486 * doc/invoke.texi: Add RISC-V options section.
25487 * doc/md.texi: Add RISC-V constraints section.
25488 * configure: Regenerated.
25490 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
25493 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
25494 false values to be constant vectors with all 0 or all 1 bits set.
25495 (vcondu<mode><mode>): Likewise.
25496 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
25498 (fpmask_comparison_operator): Update comment.
25499 (vecint_comparison_operator): New predicate.
25500 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
25501 vector conditionals when the true and false values are constant
25502 vectors with all 0 bits or all 1 bits set.
25504 2017-02-06 Martin Sebor <msebor@redhat.com>
25506 PR tree-optimization/79376
25507 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
25509 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
25511 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
25512 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
25513 to simplify split condition.
25515 2017-02-06 Jakub Jelinek <jakub@redhat.com>
25517 * omp-expand.c (oxpand_omp_atomic_fetch_op,
25518 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
25521 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
25523 PR rtl-optimization/68664
25524 * target.def (can_speculate_insn): New hook.
25525 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
25526 * doc/tm.texi: Regenerate.
25527 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
25528 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
25529 (rs6000_sched_can_speculate_insn): New function.
25531 2017-02-06 Jakub Jelinek <jakub@redhat.com>
25533 PR tree-optimization/79284
25534 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
25535 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
25536 vectorizable_mask_load_store, vectorizable_operation,
25537 vect_is_simple_cond, get_same_sized_vectype): Use it instead
25538 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
25539 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
25540 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
25541 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
25542 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
25543 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
25544 is_gimple_assign (stmt). Replace another such test with
25545 is_gimple_assign (stmt).
25547 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
25550 * config/avr/avr.c (rtl-iter.h): Include it.
25551 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
25552 (avr_legitimate_combined_insn): ...and implementation.
25554 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
25556 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
25557 * config/s390/s390.c (s390_const_operand_ok)
25558 (s390_canonicalize_comparison, s390_extract_part)
25559 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
25560 (s390_contiguous_bitmask_p, s390_rtx_costs)
25561 (legitimize_pic_address): Likewise.
25562 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
25563 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
25564 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
25565 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
25566 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
25568 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
25570 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
25571 REGNO($0) == REGNO($1).
25573 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25575 * config/s390/linux.h(SIZE_TYPE): Add comment.
25577 2017-02-06 Julian Brown <julian@codesourcery.com>
25578 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25579 Virendra Pathak <virendra.pathak@broadcom.com>
25581 * config/aarch64/aarch64-cores.def: Change the scheduler
25583 * config/aarch64/aarch64.md: Include thunderx2t99.md.
25584 * config/aarch64/thunderx2t99.md: New file.
25586 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25588 * doc/standards.texi (Go Language): Update link to language
25591 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
25593 * tree-eh.c (lower_resx): Sanitize profile.
25594 (cleanup_empty_eh_move_lp): Likewise.
25596 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
25599 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
25601 * cfgloopmanip.h (loop_version): Update prototype.
25602 * modulo-sched.c (sms_schedule): Update call of loop_version.
25603 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
25604 * tree-parloops.c (gen_parallel_loop): Likewise.
25605 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
25606 * tree-ssa-loop-split.c (split_loop): Likewise.
25607 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
25608 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25610 2017-02-05 Martin Liska <mliska@suse.cz>
25613 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
25615 (print_operand_address): Initialize a struct to zero.
25617 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25619 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
25620 garbage collector only in textual form.
25622 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25624 * doc/extend.texi (x86 specific memory model extensions for
25625 transactional memory): Simplify a phrase.
25627 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
25630 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
25631 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
25632 (atomic_storedi_1): Likewise.
25634 2017-02-04 Jakub Jelinek <jakub@redhat.com>
25636 PR tree-optimization/79338
25637 * tree-parloops.c (gather_scalar_reductions): Don't call
25638 vect_analyze_loop_form for loop->inner before destroying loop's
25641 2017-02-03 Martin Sebor <msebor@redhat.com>
25643 PR tree-optimization/79327
25644 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
25645 when precision has resulted in leading zeros.
25646 (format_integer): Adjust the likely counter to assume an unknown
25647 argument that may be zero is non-zero.
25649 2017-02-03 Jason Merrill <jason@redhat.com>
25652 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
25653 avoid copying non-taken branch.
25655 2017-02-03 Jakub Jelinek <jakub@redhat.com>
25657 PR tree-optimization/79340
25658 * tree-vect-loop.c (vectorizable_reduction): Release
25659 vec_defs elements after safe_splicing them into other vectors.
25662 PR tree-optimization/79327
25663 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
25664 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
25666 (format_integer): Use wide_int_to_tree instead of build_int_cst
25667 + to_?hwi. If argmin is NULL, just set argmin and argmax to
25668 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
25669 of shortest and longest sequence.
25671 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
25673 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
25674 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
25676 2017-02-03 Walter Lee <walt@tilera.com>
25679 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
25680 after initial stackframe link reg save.
25681 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
25683 2017-02-03 Jakub Jelinek <jakub@redhat.com>
25686 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
25687 wu for stxssp alternative.
25689 2017-02-03 Martin Sebor <msebor@redhat.com>
25691 PR tree-optimization/79352
25692 * gimple-fold.c (get_range_strlen): Add argument.
25693 (get_range_strlen): Change return type to bool.
25694 (get_maxval_strlen): Pass in a dummy argument.
25695 * gimple-fold.h (get_range_strlen): Change return type to bool.
25696 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
25697 * tree.h (array_at_struct_end_p): Add argument.
25698 * tree.c (array_at_struct_end_p): Handle it.
25700 2017-02-03 Martin Liska <mliska@suse.cz>
25703 * multiple_target.c (create_dispatcher_calls): Redirect edge
25704 from a caller of a dispatcher.
25705 (expand_target_clones): Make the clones local.
25706 (ipa_target_clone): Do both target clones and resolvers.
25707 (ipa_dispatcher_calls): Remove the pass.
25708 (pass_dispatcher_calls::gate): Likewise.
25709 (make_pass_dispatcher_calls): Likewise.
25710 * passes.def (pass_target_clone): Put as very first IPA early
25713 2017-02-03 Martin Liska <mliska@suse.cz>
25715 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
25716 in case of a function with ifunc attribute.
25718 2017-02-03 Martin Liska <mliska@suse.cz>
25720 * cgraph.c (cgraph_node::dump): Dump function version info.
25721 * symtab.c (symtab_node::dump_base): Add missing new line.
25723 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
25725 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
25726 (ifcombine_ifandif): Use it.
25728 2017-02-03 Martin Liska <mliska@suse.cz>
25730 * doc/invoke.texi: Document default value for
25731 use-after-scope-direct-emission-threshold.
25733 2017-02-03 Martin Liska <mliska@suse.cz>
25735 PR tree-optimization/79339
25736 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
25737 (format_floating): Likewise.
25739 2017-02-03 Martin Liska <mliska@suse.cz>
25742 * ipa-prop.c (ipa_node_params_t::insert): Remove current
25744 (ipa_node_params_t::remove): Likewise.
25745 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
25746 initialization from removed ipa_node_params_t::insert.
25747 (ipa_node_params::~ipa_node_params): Move from removed
25748 ipa_node_params_t::release.
25749 * symbol-summary.h (symbol_summary::m_released): New member.
25750 Do not release a summary twice. Do not allow to call finalizer
25751 for types of a summary that live in GGC memory.
25753 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25755 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
25758 2017-02-02 Martin Sebor <msebor@redhat.com>
25760 PR middle-end/79275
25761 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
25762 (format_string): Tighten up the range of output for non-constant
25763 strings and correct the expected range for wide non-constant strings.
25765 2017-02-02 Martin Sebor <msebor@redhat.com>
25767 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
25769 PR middle-end/32003
25770 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
25772 (-fdump-tree-@var): Add to index and document how to come up
25773 with pass-specific option and dump file names.
25774 (-fdump-passes): Clarify where to look for output.
25776 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
25778 PR middle-end/77445
25779 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
25780 statistics of the analyzed path; allow threading for speed when
25781 any of BBs along the path are optimized for speed.
25783 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
25785 PR middle-end/78468
25786 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
25787 settings of the virtual registers.
25790 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
25792 * explow.c (get_dynamic_stack_size): Take known alignment of stack
25793 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
25796 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25798 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
25799 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
25801 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25803 * config/s390/s390.md: Add missing comments with the expanded
25805 * config/s390/vector.md: Likewise.
25806 * config/s390/vx-builtins.md: Likewise.
25808 2017-02-02 Jakub Jelinek <jakub@redhat.com>
25811 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
25812 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
25813 conditions on a single line.
25815 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25817 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
25818 __S390_VX__ to __VX__.
25820 2017-02-01 Andrew Pinski <apinski@cavium.com>
25822 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
25823 stmt_info to record_stmt_cost.
25824 (vect_get_known_peeling_cost): Pass stmt_info if known to
25826 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
25827 cpu_vector_cost field into
25828 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
25829 field into vec_int_stmt_cost and vec_fp_stmt_cost.
25830 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
25831 splitting of scalar_stmt_cost and vec_stmt_cost.
25832 (thunderx_vector_cost): Likewise.
25833 (cortexa57_vector_cost): LIkewise.
25834 (exynosm1_vector_cost): Likewise.
25835 (xgene1_vector_cost): Likewise.
25836 (thunderx2t99_vector_cost): Improve after the splitting of the two
25838 (aarch64_builtin_vectorization_cost): Update for the splitting of
25839 scalar_stmt_cost and vec_stmt_cost.
25841 2017-02-01 Torvald Riegel <triegel@redhat.com>
25842 Richard Henderson <rth@redhat.com>
25844 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
25845 conditional on existance of a fast atomic load.
25846 * optabs-query.c (can_atomic_load_p): New function.
25847 * optabs-query.h (can_atomic_load_p): Declare it.
25848 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
25849 no fast atomic load is available for the particular size of access.
25850 (expand_atomic_compare_and_swap): Likewise.
25851 (expand_atomic_load): Likewise.
25852 (expand_atomic_store): Likewise.
25853 (expand_atomic_fetch_op): Likewise.
25854 * testsuite/lib/target-supports.exp
25855 (check_effective_target_sync_int_128): Remove x86 because it provides
25856 no fast atomic load.
25857 (check_effective_target_sync_int_128_runtime): Likewise.
25859 2017-02-01 Richard Biener <rguenther@suse.de>
25861 * graphite.c: Include tree-vectorizer.h for find_loop_location.
25862 (graphite_transform_loops): Provide opt-info for optimized nests.
25863 * tree-parloop.c (parallelize_loops): Provide opt-info for
25864 parallelized loops.
25866 2017-02-01 Richard Biener <rguenther@suse.de>
25868 PR middle-end/79315
25869 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
25870 was not set before.
25872 2017-02-01 Richard Biener <rguenther@suse.de>
25874 PR tree-optimization/71824
25875 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
25876 Verify the loops are valid in the merged SESE region.
25877 (scop_detection::can_represent_loop_1): Check analyzing the
25878 evolution of the number of iterations in the region succeeds.
25880 2017-01-31 Ian Lance Taylor <iant@golang.org>
25882 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
25883 REG_ARGS_SIZE note to 32-bit push insns and call insn.
25885 2017-01-31 David Malcolm <dmalcolm@redhat.com>
25887 PR preprocessor/79210
25888 * input.c (get_substring_ranges_for_loc): Replace line_width
25889 assertion with error-handling.
25891 2017-01-31 Richard Biener <rguenther@suse.de>
25893 PR tree-optimization/77318
25894 * graphite-sese-to-poly.c (extract_affine): Fix assert.
25895 (create_pw_aff_from_tree): Take loop parameter.
25896 (add_condition_to_pbb): Pass loop of the condition to
25897 create_pw_aff_from_tree.
25899 2017-01-31 Jakub Jelinek <jakub@redhat.com>
25901 * config/s390/s390.c (s390_asan_shadow_offset): New function.
25902 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
25904 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
25908 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
25910 (convert_int_to_float128): Likewise.
25911 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
25912 (convert_int_to_float128): Likewise.
25913 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
25914 (UNSPEC_IEEE128_CONVERT): Likewise.
25915 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
25916 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
25917 Use local variables for IBM extended format.
25918 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
25919 (fix_trunc<mode>si2_fprs): Likewise.
25920 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
25921 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
25922 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
25923 to know that we can now have integers of all sizes in vector
25925 (fix<uns>_<mode>di2_hw): Likewise.
25926 (float<uns>_<mode>si2_hw): Likewise.
25927 (fix_<mode>si2_hw): Likewise.
25928 (fixuns_<mode>si2_hw): Likewise.
25929 (float<uns>_<mode>di2_hw): Likewise.
25930 (float_<mode>di2_hw): Likewise.
25931 (float_<mode>si2_hw): Likewise.
25932 (floatuns_<mode>di2_hw): Likewise.
25933 (floatuns_<mode>si2_hw): Likewise.
25934 (xscvqp<su>wz_<mode>): Delete, no longer used.
25935 (xscvqp<su>dz_<mode>): Likewise.
25936 (xscv<su>dqp_<mode>): Likewise.
25937 (ieee128_mfvsrd_64bit): Likewise.
25938 (ieee128_mfvsrd_32bit): Likewise.
25939 (ieee128_mfvsrwz): Likewise.
25940 (ieee128_mtvsrw): Likewise.
25941 (ieee128_mtvsrd_64bit): Likewise.
25942 (ieee128_mtvsrd_32bit): Likewise.
25944 2017-01-31 Martin Liska <mliska@suse.cz>
25947 * ipa-prop.c (ipa_free_all_node_params): Call release method
25948 instead of ~sumbol_summary to not to trigger double times
25951 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
25953 PR tree-optimization/71691
25954 * bitmap.h (class auto_bitmap): New.
25955 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
25956 is_maybe_undefined instead of ssa_undefined_value_p.
25958 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25960 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
25961 __S390_ARCH_LEVEL__ to __ARCH__.
25963 2017-01-31 Jakub Jelinek <jakub@redhat.com>
25965 PR tree-optimization/79267
25966 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
25967 if should_remove_lhs_p is true.
25969 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
25972 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
25973 (add_alignment_attribute): New.
25974 (base_type_die): Add alignment attribute.
25975 (subrange_type_die): Likewise.
25976 (modified_type_die): Likewise.
25977 (gen_array_type_die): Likewise.
25978 (gen_descr_array_type_die: Likewise.
25979 (gen_enumeration_type_die): Likewise.
25980 (gen_subprogram_die): Likewise.
25981 (gen_variable_die): Likewise.
25982 (gen_field_die): Likewise.
25983 (gen_ptr_to_mbr_type_die): Likewise.
25984 (gen_struct_or_union_type_die): Likewise.
25985 (gen_subroutine_type_die): Likewise.
25986 (gen_typedef_die): Likewise.
25987 (base_type_cmp): Compare alignment attribute.
25989 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25992 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
25993 (setb_unsigned) New pattern for setb with CCUNS.
25994 * config/rs6000/rs6000.c (expand_block_compare): Use a different
25995 subfc./subfe sequence to avoid overflow problems. Generate a
25996 shorter sequence with cmpld/setb for power9.
25997 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
25998 for generating subfc. instruction.
25999 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
26000 now uses this instruction.
26002 2017-01-30 Ian Lance Taylor <iant@google.com>
26005 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
26006 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
26008 2017-01-30 Martin Sebor <msebor@redhat.com>
26010 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
26011 Move constant to the right of a relational operator.
26012 (get_mpfr_format_length, format_character, format_string): Ditto.
26013 (should_warn_p, maybe_warn): Same.
26015 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
26017 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
26020 * asan.c (get_translation_unit_decl): Remove function.
26021 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
26023 2017-01-30 Martin Liska <mliska@suse.cz>
26025 PR gcov-profile/79259
26026 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
26027 -fprofile-generate.
26029 2017-01-30 Martin Liska <mliska@suse.cz>
26032 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
26033 Initialize variables with NULL value.
26035 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
26038 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
26040 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
26042 2017-01-30 Richard Biener <rguenther@suse.de>
26044 PR tree-optimization/79276
26045 * tree-vrp.c (process_assert_insertions): Properly adjust common
26046 when removing a duplicate.
26048 * gcc.dg/torture/pr79276.c: New testcase.
26050 2017-01-30 Richard Biener <rguenther@suse.de>
26052 PR tree-optimization/79256
26053 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
26054 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
26056 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
26058 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
26061 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
26062 ("*r<noxa>sbg_<mode>_sll_bitmask")
26063 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
26064 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
26065 Use contiguous_bitmask_nowrap_operand.
26067 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26070 * config/rs6000/altivec.h (vec_xl): Revise #define.
26071 (vec_xst): Likewise.
26073 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
26075 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
26077 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
26079 PR rtl-optimization/79194
26080 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
26081 traps before call to bypass_conditional_jumps.
26083 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
26085 PR tree-optimization/71374
26086 * lra-constraints.c (check_conflict_input_operands): New.
26087 (match_reload): Use it.
26089 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
26092 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
26093 account to calculate conflict_set.
26095 2017-01-27 Bin Cheng <bin.cheng@arm.com>
26097 PR rtl-optimization/78559
26098 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
26099 other_insn in combine.
26101 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26103 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
26104 uint16_type_node for BT_UINT16.
26106 2017-01-27 David Malcolm <dmalcolm@redhat.com>
26108 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
26109 "RTL Tests" to menu.
26110 (GIMPLE Tests): New node.
26111 (RTL Tests): New node.
26113 2017-01-27 Richard Biener <rguenther@suse.de>
26115 PR tree-optimization/79245
26116 * tree-loop-distribution.c (distribute_loop): Apply cost
26117 modeling also to detected patterns.
26119 2017-01-27 Richard Biener <rguenther@suse.de>
26121 PR tree-optimization/71433
26122 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
26123 (compare_assert_loc): New function.
26124 (process_assert_insertions): Sort and optimize assert locations
26125 to remove duplicates and push down identical assertions on
26126 edges to their destination block.
26128 2017-01-27 Richard Biener <rguenther@suse.de>
26130 PR tree-optimization/79244
26131 * tree-vrp.c (remove_range_assertions): Forcefully propagate
26132 out SSA names even if abnormal.
26134 2017-01-27 Jakub Jelinek <jakub@redhat.com>
26136 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
26137 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
26138 instead of MPFR_RNDN.
26140 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
26143 * arm.c (arm_option_override): Don't call build_target_option_node
26144 until after doing all option overrides.
26145 (arm_valid_target_attribute_tree): Likewise.
26147 2017-01-27 Martin Liska <mliska@suse.cz>
26149 * doc/invoke.texi (-fprofile-arcs): Document profiling support
26150 for {cd}tors and C++ {cd}tors.
26152 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
26154 * config/s390/s390.md ("*setmem_long_and")
26155 ("*setmem_long_and_31z"): Use zero_extend instead of and.
26157 2017-01-26 Martin Sebor <msebor@redhat.com>
26159 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
26162 2017-01-26 Martin Sebor <msebor@redhat.com>
26164 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
26165 HAVE_DFmode before using XFmode or DFmode.
26166 (parse_directive): Avoid using the z length modifier to avoid
26167 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
26169 PR middle-end/78703
26170 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
26171 to accept adjustment as an array.
26172 (get_int_range): New function.
26173 (struct directive): Make width and prec arrays.
26174 (directive::set_width, directive::set_precision): Call get_int_range.
26175 (format_integer, format_floating): Handle width and precision ranges.
26176 (format_string, parse_directive): Same.
26178 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26181 * dwarf2out.c (generate_skeleton_bottom_up): For children with
26182 comdat_type_p set, just clone them, but keep the children in the
26186 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
26187 which have direct callers with -fvar-tracking-assignments enabled
26189 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
26190 inside of type units.
26192 2017-01-26 Martin Sebor <msebor@redhat.com>
26194 PR middle-end/78703
26195 * gimple-ssa-sprintf.c (struct result_range): Add likely and
26197 (struct format_result): Replace number_chars, number_chars_min,
26198 and number_chars_max with a single member of struct result_range.
26200 (format_result::operator+=): Adjust.
26201 (struct fmtresult): Remove bounded. Handle likely and unlikely
26203 (fmtresult::adjust_for_width_or_precision): New function.
26204 (fmtresult:type_max_digits): New function.
26205 (bytes_remaining): Handle likely and unlikely counters.
26206 (min_bytes_remaining): Remove.
26207 (format_percent): Simplify.
26208 (format_integer, format_floating): Set likely and unlikely counters.
26209 (get_string_length, format_character, format_string): Same.
26210 (format_plain, should_warn_p): New function.
26211 (maybe_warn): Call should_warn_p. Update diagnostic messages
26212 and handle those for all directives, including plain strings.
26213 (format_directive): Handle likely and unlikely counters.
26214 Remove unnecessary quoting from diagnostics. Add an informational
26216 (add_bytes): Remove.
26217 (pass_sprintf_length::compute_format_length): Simplify.
26218 (try_substitute_return_value): Handle likely and unlikely counters.
26220 2017-01-26 Carl Love <cel@us.ibm.com>
26222 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
26223 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
26225 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
26228 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
26229 endianess for subregs into account.
26230 * lra-constraints.c (lra_constraints): Do risky transformations
26231 always on the first iteration.
26232 * lra-lives.c (check_pseudos_live_through_calls): Add arg
26233 last_call_used_reg_set.
26234 (process_bb_lives): Define and use last_call_used_reg_set.
26235 * lra.c (lra): Always continue after lra_constraints on the first
26238 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
26240 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
26242 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
26244 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26246 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
26247 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
26248 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
26249 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
26250 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
26251 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
26252 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
26253 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
26254 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
26256 2017-01-26 Marek Polacek <polacek@redhat.com>
26259 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
26260 for the third operand.
26262 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26264 PR middle-end/79236
26265 * omp-low.c (struct omp_context): Add simt_stmt field.
26266 (scan_omp_for): Return omp_context *.
26267 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
26268 context to the _simt_ SIMD stmt.
26269 (lower_omp_for): For combined SIMD with sibling _simt_
26270 SIMD, make sure to use the same decls in _looptemp_
26271 clauses as in the sibling.
26273 2017-01-26 David Sherwood <david.sherwood@arm.com>
26275 PR middle-end/79212
26276 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
26279 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26282 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
26283 emit fld b; fld a; if possible.
26285 * brig-builtins.def: Update copyright years.
26286 * config/arm/arm_acle_builtins.def: Update copyright years.
26288 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
26291 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
26292 constraint instead of o for the stxsd instruction.
26294 2017-01-25 Carl Love <cel@us.ibm.com>
26296 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
26297 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
26299 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
26301 * doc/invoke.texi (C++ Dialect Options): Fix typo.
26303 2017-01-25 Richard Biener <rguenther@suse.de>
26305 PR tree-optimization/69264
26306 * target.def (vector_alignment_reachable): Improve documentation.
26307 * doc/tm.texi: Regenerate.
26308 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
26310 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
26311 earlier changes with respect to TYPE_USER_ALIGN.
26312 (vector_alignment_reachable_p): Likewise. Improve dumping.
26314 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26317 * config/arm/arm.md (xordi3): Force constant operand into a register
26320 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26322 * doc/invoke.texi (-fstore-merging): Correct default optimization
26323 levels at which it is enabled.
26324 (-O): Move -fstore-merging from list to...
26327 2017-01-25 Richard Biener <rguenther@suse.de>
26330 * omp-expand.c: Include debug.h.
26331 (expand_omp_taskreg): Make sure to generate early debug before
26332 outlining anything from a function.
26333 (expand_omp_target): Likewise.
26334 (grid_expand_target_grid_body): Likewise.
26336 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
26339 * asan.c (get_translation_unit_decl): New function.
26340 (asan_add_global): Extract modules file name from globals
26341 TRANSLATION_UNIT_DECL name.
26343 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
26346 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
26347 for long calls with APCS frame and VFP.
26349 2017-01-24 David Malcolm <dmalcolm@redhat.com>
26351 * cfg.c (original_copy_tables_initialized_p): New function.
26352 * cfg.h (original_copy_tables_initialized_p): New decl.
26353 * cfgrtl.c (relink_block_chain): Guard the call to
26354 free_original_copy_tables with a call to
26355 original_copy_tables_initialized_p.
26356 * cgraph.h (symtab_node::native_rtl_p): New decl.
26357 * cgraphunit.c (symtab_node::native_rtl_p): New function.
26358 (symtab_node::needed_p): Don't assert for early assembly output
26359 for __RTL functions.
26360 (cgraph_node::finalize_function): Set "force_output" for __RTL
26362 (cgraph_node::analyze): Bail out early for __RTL functions.
26363 (analyze_functions): Update assertion to support __RTL functions.
26364 (cgraph_node::expand): Bail out early for __RTL functions.
26365 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
26367 * function.h (struct function): Update comment for field
26369 * gimple-expr.c: Include "tree-pass.h".
26370 (gimple_has_body_p): Return false for __RTL functions.
26371 * Makefile.in (OBJS): Add run-rtl-passes.o.
26372 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
26374 (gcc::pass_manager::get_clean_slate): New accessor.
26375 * passes.c: Include "insn-addr.h".
26376 (should_skip_pass_p): Add logging. Update logic for running
26377 "expand" to be compatible with both __GIMPLE and __RTL. Guard
26378 property-provider override so it is only done for gimple passes.
26380 (skip_pass): New function.
26381 (execute_one_pass): Call skip_pass when skipping passes.
26382 * read-md.c (md_reader::read_char): Support filtering
26383 the input to a subset of line numbers.
26384 (md_reader::md_reader): Initialize fields
26385 m_first_line and m_last_line.
26386 (md_reader::read_file_fragment): New function.
26387 * read-md.h (md_reader::read_file_fragment): New decl.
26388 (md_reader::m_first_line): New field.
26389 (md_reader::m_last_line): New field.
26390 * read-rtl-function.c (function_reader::create_function): Only
26391 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
26392 curr_properties. Set DECL_INITIAL to a dummy block.
26393 (read_rtl_function_body_from_file_range): New function.
26394 * read-rtl-function.h (read_rtl_function_body_from_file_range):
26396 * run-rtl-passes.c: New file.
26397 * run-rtl-passes.h: New file.
26399 2017-01-24 Jeff Law <law@redhat.com>
26401 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
26404 2017-01-24 Bin Cheng <bin.cheng@arm.com>
26406 PR tree-optimization/79159
26407 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
26408 (record_nonwrapping_iv): Improve boundary using above function if no
26409 value range information.
26411 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
26412 Martin Jambor <mjambor@suse.cz>
26414 * brig-builtins.def: New file.
26415 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
26416 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
26417 (DEF_HSAIL_SAT_BUILTIN): Likewise.
26418 (DEF_HSAIL_INTR_BUILTIN): Likewise.
26419 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
26420 * builtin-types.def (BT_INT8): New.
26421 (BT_INT16): Likewise.
26422 (BT_UINT8): Likewise.
26423 (BT_UINT16): Likewise.
26424 (BT_FN_ULONG): Likewise.
26425 (BT_FN_UINT_INT): Likewise.
26426 (BT_FN_UINT_ULONG): Likewise.
26427 (BT_FN_UINT_LONG): Likewise.
26428 (BT_FN_UINT_PTR): Likewise.
26429 (BT_FN_ULONG_PTR): Likewise.
26430 (BT_FN_INT8_FLOAT): Likewise.
26431 (BT_FN_INT16_FLOAT): Likewise.
26432 (BT_FN_UINT32_FLOAT): Likewise.
26433 (BT_FN_UINT16_FLOAT): Likewise.
26434 (BT_FN_UINT8_FLOAT): Likewise.
26435 (BT_FN_UINT64_FLOAT): Likewise.
26436 (BT_FN_UINT16_UINT32): Likewise.
26437 (BT_FN_UINT32_UINT16): Likewise.
26438 (BT_FN_UINT16_UINT16_UINT16): Likewise.
26439 (BT_FN_INT_PTR_INT): Likewise.
26440 (BT_FN_UINT_PTR_UINT): Likewise.
26441 (BT_FN_LONG_PTR_LONG): Likewise.
26442 (BT_FN_ULONG_PTR_ULONG): Likewise.
26443 (BT_FN_VOID_UINT64_UINT64): Likewise.
26444 (BT_FN_UINT8_UINT8_UINT8): Likewise.
26445 (BT_FN_INT8_INT8_INT8): Likewise.
26446 (BT_FN_INT16_INT16_INT16): Likewise.
26447 (BT_FN_INT_INT_INT): Likewise.
26448 (BT_FN_UINT_FLOAT_UINT): Likewise.
26449 (BT_FN_FLOAT_UINT_UINT): Likewise.
26450 (BT_FN_ULONG_UINT_UINT): Likewise.
26451 (BT_FN_ULONG_UINT_PTR): Likewise.
26452 (BT_FN_ULONG_ULONG_ULONG): Likewise.
26453 (BT_FN_UINT_UINT_UINT): Likewise.
26454 (BT_FN_VOID_UINT_PTR): Likewise.
26455 (BT_FN_UINT_UINT_PTR: Likewise.
26456 (BT_FN_UINT32_UINT64_PTR): Likewise.
26457 (BT_FN_INT_INT_UINT_UINT): Likewise.
26458 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
26459 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
26460 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
26461 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
26462 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
26463 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
26464 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
26465 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
26466 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
26467 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
26468 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
26469 * doc/frontends.texi: List BRIG FE.
26470 * doc/install.texi (Testing): Add BRIG tesring requirements.
26471 * doc/invoke.texi (Overall Options): Mention BRIG.
26472 * doc/standards.texi (Standards): Doucment BRIG HSA version.
26474 2017-01-24 Richard Biener <rguenther@suse.de>
26476 PR translation/79208
26477 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
26479 2017-01-24 Martin Jambor <mjambor@suse.cz>
26482 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
26483 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
26484 and known_contexts.
26486 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
26488 PR middle-end/79123
26489 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
26490 casts from signed to unsigned really don't have a range.
26492 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
26494 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
26495 GMP_RNDx for compatiblity.
26497 2017-01-24 Martin Liska <mliska@suse.cz>
26500 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
26501 that would prevent us to call alloca with -1 as argument.
26503 2017-01-24 Jakub Jelinek <jakub@redhat.com>
26505 * dwarf2out.c (output_compilation_unit_header, output_file_names):
26506 Avoid -Wformat-security warning.
26508 2017-01-23 Andrew Pinski <apinski@cavium.com>
26510 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
26513 2017-01-23 Martin Sebor <msebor@redhat.com>
26515 PR middle-end/78703
26516 * gimple-ssa-sprintf.c (warn_level): New global.
26517 (format_integer): Use it here and throughout the rest of the file.
26518 Use the same switch to compute sign as base.
26519 (maybe_warn): New function.
26520 (format_directive): Factor out warnings into maybe_warn.
26521 Add debugging output. Use warn_level.
26522 (add_bytes): Use warn_level.
26523 (pass_sprintf_length::compute_format_length): Add debugging output.
26524 (try_substitute_return_value): Same.
26525 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
26527 PR middle-end/78703
26528 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
26529 (struct fmtresult, format_integer, format_floating): Adjust.
26530 (fmtresult::fmtresult): Set max correctly in two argument ctor.
26531 (get_string_length, format_string,format_directive): Same.
26532 (pass_sprintf_length::compute_format_length): Same.
26533 (try_substitute_return_value): Simplify slightly.
26535 PR middle-end/78703
26536 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
26537 (fmtresult::operator+=): Outlined.
26538 (struct fmtresult): Add ctors.
26539 (struct conversion_spec): Rename...
26540 (struct directive): ...to this. Add and remove data members.
26541 (directive::set_width, directive::set_precision): New functions.
26542 (format_percent): Use fmtresult ctor.
26543 (get_width_and_precision): Remove.
26544 (format_integer): Make naming changes. Avoid computing width and
26546 (format_floating): Same. Adjust indentation.
26547 (format_character, format_none): New functions.
26548 (format_string): Moved character handling to format_character.
26549 (format_directive): Remove arguments, change return type.
26550 (parse_directive): New function.
26551 (pass_sprintf_length::compute_format_length): Move directive
26552 parsing to parse_directive.
26554 2017-01-23 Jakub Jelinek <jakub@redhat.com>
26556 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
26557 (assign_assembler_name_if_needed): ... this.
26558 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
26559 (assign_assembler_name_if_needed): ... this.
26560 (free_lang_data_in_cgraph): Adjust callers.
26561 * cgraphunit.c (cgraph_node::analyze): Likewise.
26562 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
26565 2017-01-23 Richard Biener <rguenther@suse.de>
26567 PR tree-optimization/79088
26568 PR tree-optimization/79188
26569 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
26570 resetting loop bounds after last path deletion. Reset loop
26571 bounds of the target loop, make code match the comments.
26572 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
26573 Make sure loops need no fixups.
26575 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
26577 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
26578 exponent support with double type for first argument.
26579 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
26580 type returned by __builtin_vec_extract_sig,
26581 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
26582 functions from "vector int" to "vector unsigned int" or from
26583 "vector long long int" to "vector unsigned long long int".
26584 Changed type returned by __builtin_vec_extract_exp,
26585 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
26586 functions from "vector int" to "vector unsigned int" or from
26587 "vector long long int" to "vector unsigned long long int".
26588 Changed return type of __builtin_vec_test_data_class,
26589 __builtin_vec_test_data_class_sp, and
26590 __builtin_vec_test_data_class_dp from "vector int" to
26591 "vector bool int" or from "vector long long int" to "vector bool
26592 long long int" and changed second argument type from "unsigned
26593 int" to "int". Added new overloaded function forms "vector float
26594 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
26595 "vector float __builtin_vec_insert_exp_sp (vector float, vector
26596 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
26597 double, vector unsigned long long int)" and "vector double
26598 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
26599 long int)". Changed return type of
26600 __builtin_scalar_test_data_class and
26601 __builtin_scalar_test_data_class_sp and
26602 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
26603 int" and changed second argument from "unsigned int" to "int".
26604 Changed type returned by __builtin_scalar_test_neg,
26605 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
26606 from "int" to "bool int". Added new overloaded function form
26607 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
26608 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
26609 exponent double-precision with floating point first argument.
26610 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
26611 documentation of scalar_test_data_class, scalar_test_neg,
26612 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
26613 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
26614 vec_test_data_class built-in functions to reflect refinements in
26615 their type signatures.
26617 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
26619 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
26621 (aarch64_elf_asm_destructor): Likewise.
26623 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
26625 PR rtl-optimization/78634
26626 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
26627 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
26628 * ifcvt.c (noce_try_cmove): Add missing cost check.
26630 PR rtl-optimization/71724
26631 * combine.c (if_then_else_cond): Look for situations where it is
26632 beneficial to undo the work of one of the recursive calls.
26634 2017-01-23 Bin Cheng <bin.cheng@arm.com>
26636 PR tree-optimization/70754
26637 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
26638 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
26639 combined stmt before it if not NULL.
26640 (combine_chains): Process refs reversely and compute dominance point
26643 2017-01-23 Martin Liska <mliska@suse.cz>
26645 PR tree-optimization/79196
26646 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
26647 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
26649 (strlen_optimize_stmt): Call the renamed function.
26651 2017-01-23 Michael Matz <matz@suse.de>
26653 PR tree-optimization/78384
26654 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
26656 2017-01-23 Richard Biener <rguenther@suse.de>
26658 PR tree-optimization/79186
26659 * tree-vrp.c (register_new_assert_for): Make sure we've seen
26660 both incoming edges before moving an assert.
26662 2017-01-23 Martin Jambor <mjambor@suse.cz>
26664 * ipa-prop.c (load_from_param_1): Removed.
26665 (load_from_unmodified_param): Bits from load_from_param_1 put back
26667 (load_from_param): Removed.
26668 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
26669 with stmt. Reverted back to use of load_from_unmodified_param.
26671 2017-01-23 Martin Jambor <mjambor@suse.cz>
26674 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
26675 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
26676 field a pointer to garbage collected vector, mark lattices and
26677 ipcp_orig_node with GTY((skip)).
26678 (ipa_get_param_count): Adjust to descriptors being a pointer.
26679 (ipa_get_param): Likewise.
26680 (ipa_get_type): Likewise.
26681 (ipa_get_param_move_cost): Likewise.
26682 (ipa_set_param_used): Likewise.
26683 (ipa_get_controlled_uses): Likewise.
26684 (ipa_set_controlled_uses): Likewise.
26685 (ipa_is_param_used): Likewise.
26686 (ipa_node_params_t): Move into garbage collector. New methods insert
26688 (ipa_node_params_sum): Annotate wth GTY(()).
26689 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
26691 (ipa_load_from_parm_agg): Adjust declaration.
26692 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
26693 * ipa-profile.c (ipa_profile): Likewise.
26694 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
26695 (ipa_populate_param_decls): Make descriptors parameter garbage
26697 (ipa_dump_param): Adjust to descriptors being a pointer.
26698 (ipa_alloc_node_params): Likewise.
26699 (ipa_initialize_node_params): Likewise.
26700 (load_from_param_1): Make descriptors parameter garbage collected.
26701 (load_from_unmodified_param): Likewise.
26702 (load_from_param): Likewise.
26703 (ipa_load_from_parm_agg): Likewise.
26704 (ipa_node_params::~ipa_node_params): Removed.
26705 (ipa_free_all_node_params): Remove call to delete operator.
26706 (ipa_node_params_t::insert): New.
26707 (ipa_node_params_t::remove): Likewise.
26708 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
26709 copy known_csts and known_contexts vectors.
26710 (ipa_read_node_info): Adjust to descriptors being a pointer.
26711 (ipcp_modif_dom_walker): Make m_descriptors field garbage
26713 (ipcp_transform_function): Make descriptors variable garbage
26716 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
26718 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
26719 * config/i386/avx512dqintrin.h: Ditto.
26720 * config/i386/avx512fintrin.h: Ditto.
26721 * config/i386/i386.c: Handle new builtins.
26722 * config/i386/i386-builtin.def: Add new builtins.
26723 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
26724 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
26726 2017-01-23 Jakub Jelinek <jakub@redhat.com>
26727 Martin Liska <mliska@suse.cz>
26729 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
26730 * asan.c (asan_expand_poison_ifn): Support stores and use
26731 appropriate ASAN report function.
26732 * internal-fn.c (expand_ASAN_POISON_USE): New function.
26733 * internal-fn.def (ASAN_POISON_USE): Declare.
26734 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
26735 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
26736 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
26737 ASAN_POISON calls w/o LHS.
26738 * tree-ssa.c (execute_update_addresses_taken): Create clobber
26739 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
26740 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
26741 * gimplify.c (asan_poison_variables): Add attribute
26742 use_after_scope_memory to variables that really needs to live
26744 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
26745 having the attribute.
26747 2017-01-23 Martin Liska <mliska@suse.cz>
26749 * asan.c (create_asan_shadow_var): New function.
26750 (asan_expand_poison_ifn): Likewise.
26751 * asan.h (asan_expand_poison_ifn): New declaration.
26752 * internal-fn.c (expand_ASAN_POISON): Likewise.
26753 * internal-fn.def (ASAN_POISON): New builtin.
26754 * sanopt.c (pass_sanopt::execute): Expand
26755 asan_expand_poison_ifn.
26756 * tree-inline.c (copy_decl_for_dup_finish): Make function
26758 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
26759 * tree-ssa.c (is_asan_mark_p): New function.
26760 (execute_update_addresses_taken): Rewrite local variables
26761 (identified just by use-after-scope as addressable) into SSA.
26763 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
26765 * doc/install.texi (Specific): opensource.apple.com uses https
26766 now. Remove trailing slash.
26768 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
26770 * README.Portability: Remove note on an Irix compatibility issue.
26772 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
26774 * gcov.c (INCLUDE_ALGORITHM): Define.
26775 (INCLUDE_VECTOR): Define.
26776 No longer include <vector> and <algorithm> directly.
26778 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
26780 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
26782 * doc/invoke.texi (Code Gen Options): Ditto.
26784 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
26787 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
26789 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
26791 rtl-optimization/79125
26792 * cprop.c (local_cprop_pass): Handle cases where we make an
26793 unconditional trap.
26795 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
26799 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
26800 read from, for big endian.
26802 2017-01-20 Jiong Wang <jiong.wang@arm.com>
26804 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
26805 register pauth builtins for LP64 only.
26807 2017-01-20 Marek Polacek <polacek@redhat.com>
26810 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
26813 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
26815 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
26817 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
26818 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
26819 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
26821 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26824 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
26825 in big-endian mode when they are not a single duplicated value.
26827 2017-01-20 Richard Biener <rguenther@suse.de>
26829 * BASE-VER: Bump to 7.0.1.
26831 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
26833 * omp-low.c (omplow_simd_context): New struct. Use it...
26834 (lower_rec_simd_input_clauses): ...here and...
26835 (lower_rec_input_clauses): ...here to hold common data. Adjust all
26836 references to idx, lane, max_vf, is_simt.
26838 2017-01-20 Graham Markall <graham.markall@embecosm.com>
26840 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
26843 2017-01-20 Martin Jambor <mjambor@suse.cz>
26845 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
26846 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
26848 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
26849 (GTFILES): Rename hsa.c to hsa-common.c.
26850 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
26851 * hsa-dump.c: Likewise.
26852 * hsa-gen.c: Likewise.
26853 * hsa-regalloc.c: Likewise.
26854 * ipa-hsa.c: Likewise.
26855 * omp-expand.c: Likewise.
26856 * omp-low.c: Likewise.
26857 * toplev.c: Likewise.
26859 2017-01-20 Marek Polacek <polacek@redhat.com>
26862 * doc/invoke.texi: Document -Wduplicated-branches.
26863 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
26864 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
26865 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
26866 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
26867 return 0 only when not OEP_LEXICOGRAPHIC.
26868 (fold_build_cleanup_point_expr): Use the expression
26869 location when building CLEANUP_POINT_EXPR.
26870 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
26871 * tree.c (add_expr): Handle error_mark_node.
26873 2017-01-20 Martin Liska <mliska@suse.cz>
26876 * tree-profile.c (init_ic_make_global_vars): Do not call
26878 (gimple_init_gcov_profiler): Likewise.
26880 2017-01-20 Martin Liska <mliska@suse.cz>
26883 * cgraph.h (maybe_create_reference): Remove argument and
26885 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
26887 * ipa-cp.c (create_specialized_node): Likewise.
26888 * symtab.c (symtab_node::maybe_create_reference): Handle
26889 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
26891 2017-01-20 Martin Liska <mliska@suse.cz>
26893 * read-rtl-function.c (function_reader::create_function): Use
26894 build_decl instread of build_decl_stat.
26896 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
26898 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
26899 * config/i386/avx512dqintrin.h: Ditto.
26900 * config/i386/avx512fintrin.h: Ditto.
26901 * config/i386/i386-builtin-types.def: Add new types.
26902 * config/i386/i386.c: Handle new types.
26903 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
26904 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
26905 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
26906 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
26907 (__builtin_ia32_kshiftridi): New.
26908 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
26910 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
26914 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
26915 define to rs6000_init_stack_protect_guard.
26916 (rs6000_init_stack_protect_guard): New function.
26918 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
26919 Yunqiang Su <yunqiang.su@imgtec.com>
26921 * config.gcc (supported_defaults): Add madd4.
26922 (with_madd4): Add validation.
26923 (all_defaults): Add madd4.
26924 * config/mips/mips.opt (mmadd4): New option.
26925 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
26927 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
26929 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
26930 (ISA_HAS_FUSED_MADD4): Likewise.
26931 * doc/invoke.texi (-mmadd4): Document the new option.
26932 * doc/install.texi (--with-madd4): Document the new option.
26934 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26936 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
26937 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
26938 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
26939 (aarch64_init_pauth_hint_builtins): New.
26940 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
26941 (aarch64_expand_builtin): Expand new builtins.
26943 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26945 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
26946 * combine-stack-adj.c (no_unhandled_cfa): Handle
26947 REG_CFA_TOGGLE_RA_MANGLE.
26948 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
26949 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
26950 info for return address signing.
26951 (aarch64_expand_epilogue): Likewise.
26953 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26955 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
26956 * config/aarch64/aarch64-protos.h
26957 (aarch64_return_address_signing_enabled): New declaration.
26958 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
26960 (aarch64_expand_prologue): Sign return address before it's pushed onto
26962 (aarch64_expand_epilogue): Authenticate return address fetched from
26964 (aarch64_override_options): Sanity check for ILP32 and ISA level.
26965 (aarch64_attributes): New function attributes for "sign-return-address".
26966 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
26967 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
26968 ("*do_return"): Generate combined instructions according to key index.
26969 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
26970 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
26972 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
26973 * config/aarch64/aarch64.opt (msign-return-address=): New.
26974 * doc/extend.texi (AArch64 Function Attributes): Documents
26975 "sign-return-address=".
26976 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
26978 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
26980 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
26981 overall option summary.
26983 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26985 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
26986 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
26987 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
26988 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
26990 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
26992 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
26993 -mpower9-minmax by default for -mcpu=power9.
26994 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
26995 128-bit floating point.
26997 2017-01-20 Alan Modra <amodra@gmail.com>
26999 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
27000 optimizing for size.
27002 2017-01-20 Alan Modra <amodra@gmail.com>
27005 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
27006 for strcmp and strncmp from corresponding builtin decl.
27008 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
27010 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
27011 instead of i386/rtems-64.h.
27012 * config/i386/rtems-64.h: Remove.
27014 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
27018 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
27020 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
27022 2017-01-19 Tamar Christina <tamar.christina@arm.com>
27024 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
27025 Change int to HOST_WIDE_INT.
27026 * config/aarch64/aarch64-protos.h
27027 (aarch64_simd_gen_const_vector_dup): Likewise.
27028 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
27030 2017-01-19 David Malcolm <dmalcolm@redhat.com>
27032 * langhooks-def.h (lhd_type_for_size): New decl.
27033 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
27034 * langhooks.c (lhd_type_for_size): New function, taken from
27037 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
27039 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
27040 define_bypass for CR latency.
27041 (power9-cracked-alu): Update bypass latency and remove power9-branch.
27042 (power9-alu2): Add define_bypass for CR latency.
27044 (power9-mul): Update insn latency.
27045 (power9-mul-compare): Update insn latency, bypass latency and remove
27048 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27050 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
27052 * config/aarch64/aarch64.md
27053 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
27054 aarch64_nopcrelative_literal_loads.
27055 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
27057 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
27059 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
27060 TARGET_LOONGSON_3A.
27061 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
27063 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
27066 * config.gcc (supported_defaults): Add lxc1-sxc1.
27067 (with_lxc1_sxc1): Add validation.
27068 (all_defaults): Add lxc1-sxc1.
27069 * config/mips/mips.opt (mlxc1-sxc1): New option.
27070 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
27072 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
27073 __mips_no_lxc1_sxc1.
27074 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
27075 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
27076 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
27078 2017-01-19 Richard Biener <rguenther@suse.de>
27080 PR tree-optimization/72488
27081 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
27082 sure to restore SSA info.
27083 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
27085 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
27087 PR rtl-optimization/79121
27088 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
27089 of the inner type when shifting an extended value.
27091 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
27094 * symtab.c (symtab_node::equal_address_to): Fix comparing of
27095 interposable aliases.
27097 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
27100 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
27101 Use the evmergelohi instruction.
27102 (mov_si<mode>_e500_subreg4_2_le): Likewise.
27103 (mov_sitf_e500_subreg8_2_be): Likewise.
27104 (mov_sitf_e500_subreg12_2_le): Likewise.
27105 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
27106 (mov_si<mode>_e500_subreg4_2_be): Likewise.
27107 (mov_sitf_e500_subreg8_2_le): Likewise.
27108 (mov_sitf_e500_subreg12_2_be): Likewise.
27110 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27112 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
27113 attribute from vecsimple to vecperm.
27114 (altivec_vbpermq2): Likewise.
27116 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27119 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
27121 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27122 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
27123 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
27124 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
27125 case where N arg is SIZE_MAX.
27126 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
27127 (cmpstrsi): Add pattern.
27129 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27131 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27132 __builtin_vec_revb builtins.
27133 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
27134 built-in functions to support generation of the ISA 3.0 XXBR<x>
27135 vector byte reverse instructions.
27136 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
27137 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
27138 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
27139 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
27140 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
27141 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
27142 (P9V_BUILTIN_VEC_REVB): Likewise.
27143 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
27144 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
27145 (p9_xxbrq_v16qi): Likewise.
27146 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
27147 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
27148 (p9_xxbrh_v8hi): Likewise.
27149 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
27150 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
27151 vec_revb built-in functions.
27153 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
27155 PR rtl-optimization/78952
27156 * config/i386/i386.md (any_extract): New code iterator.
27157 (*insvqi_2): Use any_extract for source operand.
27158 (*insvqi_3): Use any_shiftrt for source operand.
27160 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
27162 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
27164 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
27166 2017-01-18 Matthias Klose <doko@ubuntu.com>
27168 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
27170 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27172 * config/rs6000/altivec.h (vec_bperm): Change #define.
27173 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
27174 (altivec_vbpermq2): New define_insn.
27175 (altivec_vbpermd): Likewise.
27176 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
27177 function interface.
27178 (VBPERMD): Likewise.
27179 (VBPERM): New polymorphic function interface.
27180 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
27181 Add entries for P9V_BUILTIN_VEC_VBPERM.
27182 * doc/extend.texi: Add interfaces for vec_bperm.
27184 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27186 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
27187 first letter of error messages.
27188 (s390_resolve_overloaded_builtin): Likewise.
27189 * config/s390/s390.c (s390_expand_builtin): Likewise.
27190 (s390_invalid_arg_for_unprototyped_fn): Likewise.
27191 (s390_valid_target_attribute_inner_p): Likewise.
27192 * config/s390/s390.md ("tabort"): Likewise.
27194 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
27196 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
27197 (ISA_AVOID_DIV_HILO): New macro.
27198 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
27199 (ISA_HAS_DDIV): Likewise.
27201 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27203 * doc/invoke.texi (fabi-version): Correct number of occurrences.
27205 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27207 * doc/invoke.texi (fabi-version): Spelling fix.
27209 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27212 * doc/invoke.texi (fabi-version): Mention mangling fix for
27215 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27218 * doc/invoke.texi (fabi-version): Document discriminator mangling.
27220 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
27223 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
27224 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
27226 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
27228 (stack_protect_test): Ditto.
27229 * config/rs6000/rs6000.opt (mstack-protector-guard=,
27230 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
27232 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
27233 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
27234 -mstack-protector-guard-offset=.
27235 (RS/6000 and PowerPC Options): Ditto.
27237 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
27239 * config/i386/i386.h (MASK_CLASS_P): New define.
27240 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
27241 there are no registers from different register sets also when
27242 mask registers are used. Update function comment.
27243 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
27244 to (*k/*r) and (*k/*km) alternatives.
27246 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
27248 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
27249 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
27250 (EH_RETURN_HANDLER_RTX): New define.
27251 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
27252 Force frame pointer in EH return functions.
27253 (aarch64_expand_epilogue): Add barrier for eh_return.
27254 (aarch64_final_eh_return_addr): Remove.
27255 (aarch64_eh_return_handler_rtx): New function.
27256 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
27258 (aarch64_eh_return_handler_rtx): New prototype.
27260 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27262 * config/rs6000/altivec.h (vec_rlmi): New #define.
27263 (vec_vrlnm): Likewise.
27264 (vec_rlnm): Likewise.
27265 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
27266 (UNSPEC_VRLNM): Likewise.
27267 (VIlong): New mode iterator.
27268 (altivec_vrl<VI_char>mi): New define_insn.
27269 (altivec_vrl<VI_char>nm): Likewise.
27270 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
27272 (VRLDNM): Likewise.
27273 (RLNM): New polymorphic function entry.
27274 (VRLWMI): New monomorphic function entry.
27275 (VRLDMI): Likewise.
27276 (RLMI): New polymorphic function entry.
27277 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
27278 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
27279 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
27282 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27285 * dwarf2out.c (field_byte_offset): Restore the
27286 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
27287 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
27288 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
27291 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
27294 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
27296 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27299 * dwarf2out.c (add_data_member_location_attribute): For constant
27300 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
27301 instead of DW_AT_data_member_location, DW_AT_bit_offset and
27302 DW_AT_byte_size attributes.
27304 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
27306 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
27307 after forcing to constant memory when the code model is medium.
27309 2017-01-17 Julia Koval <julia.koval@intel.com>
27312 * config/i386/avx512fintrin.h
27313 (_mm512_i32gather_ps): Change __addr type to void const*.
27314 (_mm512_mask_i32gather_ps): Ditto.
27315 (_mm512_i32gather_pd): Ditto.
27316 (_mm512_mask_i32gather_pd): Ditto.
27317 (_mm512_i64gather_ps): Ditto.
27318 (_mm512_mask_i64gather_ps): Ditto.
27319 (_mm512_i64gather_pd): Ditto.
27320 (_mm512_mask_i64gather_pd): Ditto.
27321 (_mm512_i32gather_epi32): Ditto.
27322 (_mm512_mask_i32gather_epi32): Ditto.
27323 (_mm512_i32gather_epi64): Ditto.
27324 (_mm512_mask_i32gather_epi64): Ditto.
27325 (_mm512_i64gather_epi32): Ditto.
27326 (_mm512_mask_i64gather_epi32): Ditto.
27327 (_mm512_i64gather_epi64): Ditto.
27328 (_mm512_mask_i64gather_epi64): Ditto.
27329 (_mm512_i32scatter_ps): Change __addr type to void*.
27330 (_mm512_mask_i32scatter_ps): Ditto.
27331 (_mm512_i32scatter_pd): Ditto.
27332 (_mm512_mask_i32scatter_pd): Ditto.
27333 (_mm512_i64scatter_ps): Ditto.
27334 (_mm512_mask_i64scatter_ps): Ditto.
27335 (_mm512_i64scatter_pd): Ditto.
27336 (_mm512_mask_i64scatter_pd): Ditto.
27337 (_mm512_i32scatter_epi32): Ditto.
27338 (_mm512_mask_i32scatter_epi32): Ditto.
27339 (_mm512_i32scatter_epi64): Ditto.
27340 (_mm512_mask_i32scatter_epi64): Ditto.
27341 (_mm512_i64scatter_epi32): Ditto.
27342 (_mm512_mask_i64scatter_epi32): Ditto.
27343 (_mm512_i64scatter_epi64): Ditto.
27344 (_mm512_mask_i64scatter_epi64): Ditto.
27345 * config/i386/avx512pfintrin.h
27346 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
27347 (_mm512_mask_prefetch_i32gather_ps): Ditto.
27348 (_mm512_mask_prefetch_i64gather_pd): Ditto.
27349 (_mm512_mask_prefetch_i64gather_ps): Ditto.
27350 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
27351 (_mm512_prefetch_i32scatter_ps): Ditto.
27352 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
27353 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
27354 (_mm512_prefetch_i64scatter_pd): Ditto.
27355 (_mm512_prefetch_i64scatter_ps): Ditto.
27356 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
27357 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
27358 * config/i386/avx512vlintrin.h
27359 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
27360 (_mm_mmask_i32gather_ps): Ditto.
27361 (_mm256_mmask_i32gather_pd): Ditto.
27362 (_mm_mmask_i32gather_pd): Ditto.
27363 (_mm256_mmask_i64gather_ps): Ditto.
27364 (_mm_mmask_i64gather_ps): Ditto.
27365 (_mm256_mmask_i64gather_pd): Ditto.
27366 (_mm_mmask_i64gather_pd): Ditto.
27367 (_mm256_mmask_i32gather_epi32): Ditto.
27368 (_mm_mmask_i32gather_epi32): Ditto.
27369 (_mm256_mmask_i32gather_epi64): Ditto.
27370 (_mm_mmask_i32gather_epi64): Ditto.
27371 (_mm256_mmask_i64gather_epi32): Ditto.
27372 (_mm_mmask_i64gather_epi32): Ditto.
27373 (_mm256_mmask_i64gather_epi64): Ditto.
27374 (_mm_mmask_i64gather_epi64): Ditto.
27375 (_mm256_i32scatter_ps): Change __addr type to void*.
27376 (_mm256_mask_i32scatter_ps): Ditto.
27377 (_mm_i32scatter_ps): Ditto.
27378 (_mm_mask_i32scatter_ps): Ditto.
27379 (_mm256_i32scatter_pd): Ditto.
27380 (_mm256_mask_i32scatter_pd): Ditto.
27381 (_mm_i32scatter_pd): Ditto.
27382 (_mm_mask_i32scatter_pd): Ditto.
27383 (_mm256_i64scatter_ps): Ditto.
27384 (_mm256_mask_i64scatter_ps): Ditto.
27385 (_mm_i64scatter_ps): Ditto.
27386 (_mm_mask_i64scatter_ps): Ditto.
27387 (_mm256_i64scatter_pd): Ditto.
27388 (_mm256_mask_i64scatter_pd): Ditto.
27389 (_mm_i64scatter_pd): Ditto.
27390 (_mm_mask_i64scatter_pd): Ditto.
27391 (_mm256_i32scatter_epi32): Ditto.
27392 (_mm256_mask_i32scatter_epi32): Ditto.
27393 (_mm_i32scatter_epi32): Ditto.
27394 (_mm_mask_i32scatter_epi32): Ditto.
27395 (_mm256_i32scatter_epi64): Ditto.
27396 (_mm256_mask_i32scatter_epi64): Ditto.
27397 (_mm_i32scatter_epi64): Ditto.
27398 (_mm_mask_i32scatter_epi64): Ditto.
27399 (_mm256_i64scatter_epi32): Ditto.
27400 (_mm256_mask_i64scatter_epi32): Ditto.
27401 (_mm_i64scatter_epi32): Ditto.
27402 (_mm_mask_i64scatter_epi32): Ditto.
27403 (_mm256_i64scatter_epi64): Ditto.
27404 (_mm256_mask_i64scatter_epi64): Ditto.
27405 (_mm_i64scatter_epi64): Ditto.
27406 (_mm_mask_i64scatter_epi64): Ditto.
27407 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
27408 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
27409 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
27410 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
27411 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
27412 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
27413 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
27414 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
27415 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
27416 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
27417 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
27418 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
27419 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
27420 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
27421 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
27422 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
27423 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
27424 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
27425 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
27426 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
27427 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
27428 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
27429 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
27430 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
27431 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
27432 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
27433 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
27434 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
27435 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
27436 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
27437 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
27438 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
27439 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
27440 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
27441 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
27442 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
27443 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
27444 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
27445 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
27446 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
27447 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
27448 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
27449 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
27450 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
27451 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
27452 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
27453 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
27454 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
27455 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
27456 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
27457 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
27458 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
27459 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
27460 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
27461 definitions accordingly.
27463 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
27464 Kuan-Lin Chen <kuanlinchentw@gmail.com>
27467 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
27470 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
27473 * ira-conflicts.c (ira_build_conflicts): Update total conflict
27474 hard regs for inner regno.
27476 2017-01-17 Martin Liska <mliska@suse.cz>
27479 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
27480 assumption and add comment.
27482 2017-01-17 Nathan Sidwell <nathan@acm.org>
27484 * ipa-visibility.c (localize_node): New function, broken out of ...
27485 (function_and_variable_visibility): ... here. Call it.
27487 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
27489 PR middle-end/77445
27490 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
27491 correctly set frequency of oudgoing edge.
27492 (duplicate_thread_path): Fix profile updating.
27494 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27497 * configure.ac: Add GCC_BASE_VER.
27498 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
27499 version from BASE-VER file.
27500 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
27501 (gcc.o): Depend on $(BASEVER).
27502 * common.opt (dumpfullversion): New option.
27503 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
27504 * doc/invoke.texi: Document -dumpfullversion.
27505 * doc/install.texi: Document --with-gcc-major-version-only.
27506 * configure: Regenerated.
27508 2017-01-17 Richard Biener <rguenther@suse.de>
27510 PR tree-optimization/71433
27511 * tree-vrp.c (register_new_assert_for): Merge same asserts
27512 on all incoming edges.
27513 (process_assert_insertions_for): Handle insertions at the
27516 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
27518 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
27519 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
27521 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
27524 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
27527 2017-01-17 Alan Modra <amodra@gmail.com>
27530 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
27531 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
27532 symbolic stack limit when pic.
27534 2017-01-16 Martin Sebor <msebor@redhat.com>
27536 PR tree-optimization/78608
27537 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
27539 2017-01-16 Jeff Law <law@redhat.com>
27542 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
27543 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
27544 for several include directories that may be relative to sysroot.
27545 * config/i386/x-mingw32 (gplus_includedir): Define.
27546 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
27547 (native_system_includedir): Likewise.
27548 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
27549 override if TARGET_SYSTEM_ROOT is defined.
27550 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
27552 PR tree-optimization/79090
27553 PR tree-optimization/33562
27554 PR tree-optimization/61912
27555 PR tree-optimization/77485
27556 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
27557 and computed trims into the dump file.
27559 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
27561 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
27563 2017-01-16 Jakub Jelinek <jakub@redhat.com>
27566 * gimplify.c (gimplify_init_constructor): If want_value and
27567 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
27571 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
27572 sequence. Formatting fixes.
27573 (doloop_optimize): Formatting fixes.
27576 * gcc.c (debug_level_greater_than_spec_func): New function.
27577 (static_spec_functions): Add debug-level-gt spec function.
27578 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
27580 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
27581 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
27582 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
27583 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
27584 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
27585 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
27587 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
27589 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
27590 QImode fixups to general and mask registers only.
27592 2017-01-16 Carl Love <cel@us.ibm.com>
27594 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
27595 for built-in functions
27596 vector signed char vec_nabs (vector signed char)
27597 vector signed short vec_nabs (vector signed short)
27598 vector signed int vec_nabs (vector signed int)
27599 vector signed long long vec_nabs (vector signed long long)
27600 vector float vec_nabs (vector float)
27601 vector double vec_nabs (vector double)
27602 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
27604 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
27605 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
27606 * doc/extend.texi: Update the documentation file for the new built-in
27609 2017-01-16 Martin Sebor <msebor@redhat.com>
27611 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
27614 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27616 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
27617 UNSPEC_VSX__XXSPLTD to require special splat handling.
27619 2017-01-16 David Malcolm <dmalcolm@redhat.com>
27622 * system.h: Poison strndup.
27624 2017-01-16 Alan Modra <amodra@gmail.com>
27627 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
27630 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
27632 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
27634 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
27636 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
27637 call recog here. Assert that INSN_CODE (insn) is non-negative.
27639 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
27642 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
27644 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
27645 in the currently scheduled RTL fragment.
27647 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
27649 PR rtl-optimization/78751
27650 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
27653 2017-01-14 Jeff Law <law@redhat.com>
27655 PR tree-optimization/79090
27656 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
27657 variable length stores.
27658 (compute_trims): Delete dead assignment to *trim_tail.
27659 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
27662 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
27664 PR rtl-optimization/78626
27665 PR rtl-optimization/78727
27666 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
27667 of a block, and split such blocks after everything else is finished.
27669 2017-01-14 Alan Modra <amodra@gmail.com>
27672 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
27673 target legitimate_combined_insn.
27674 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
27675 (rs6000_legitimate_combined_insn): New function.
27676 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
27678 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
27679 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
27680 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
27682 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
27684 * doc/frontends.texi (G++ and GCC): Remove references to Java.
27686 2017-01-13 Jeff Law <law@redhat.com>
27688 PR tree-optimization/33562
27689 PR tree-optimization/61912
27690 PR tree-optimization/77485
27691 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
27693 (delete_dead_assignment): Likewise.
27694 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
27695 statement to delete_dead_call and delete_dead_assignment.
27697 2017-01-13 David Malcolm <dmalcolm@redhat.com>
27700 * substring-locations.c (format_warning_va): Strengthen case 1 so
27701 that both endpoints of the substring must be within the format
27702 range for just the substring to be printed.
27704 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
27706 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
27707 * config/i386/i386.c (ix86_target_string): Add missing options
27708 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
27709 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
27710 flags_other and ix86_target_other to flags2_other. Display unknown
27712 (ix86_valid_target_attribute_inner_p): Add missing options and
27713 reorder options by implied ISAs, as in ix86_target_string.
27715 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
27717 * hash-table.h (hash_table::too_empty_p): New function.
27718 (hash_table::expand): Use it.
27719 (hash_table::traverse): Likewise.
27720 (hash_table::empty_slot): Use sizeof (value_type) instead of
27721 sizeof (PTR) to convert bytes to elements. Shrink the table
27722 if the current size is excessive for the current number of
27725 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
27727 * ira-costs.c (record_reg_classes): Break from the inner loop
27728 early once alt_fail is known to be true. Update outer loop
27729 handling accordingly.
27731 2017-01-13 Jeff Law <law@redhat.com>
27733 * tree-ssa-dse.c (decrement_count): New function.
27734 (increment_start_addr, maybe_trim_memstar_call): Likewise.
27735 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
27736 when we know the partially dead statement is a mem* function.
27738 PR tree-optimization/61912
27739 PR tree-optimization/77485
27740 * tree-ssa-dse.c: Include expr.h.
27741 (maybe_trim_constructor_store): New function.
27742 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
27744 PR tree-optimization/33562
27745 PR tree-optimization/61912
27746 PR tree-optimization/77485
27747 * doc/invoke.texi: Document new dse-max-object-size param.
27748 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
27749 * tree-ssa-dse.c: Include params.h.
27750 (dse_store_status): New enum.
27751 (initialize_ao_ref_for_dse): New, partially extracted from
27753 (valid_ao_ref_for_dse, normalize_ref): New.
27754 (setup_live_bytes_from_ref, compute_trims): Likewise.
27755 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
27756 (maybe_trim_partially_dead_store): Likewise.
27757 (maybe_trim_complex_store): Likewise.
27758 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
27759 Track what bytes live from the original store. Return tri-state
27760 for dead, partially dead or live.
27761 (dse_dom_walker): Add constructor, destructor and new private members.
27762 (delete_dead_call, delete_dead_assignment): New extracted from
27764 (dse_optimize_stmt): Make a member of dse_dom_walker.
27765 Use initialize_ao_ref_for_dse.
27767 PR tree-optimization/33562
27768 PR tree-optimization/61912
27769 PR tree-optimization/77485
27770 * sbitmap.h (bitmap_count_bits): Prototype.
27771 (bitmap_clear_range, bitmap_set_range): Likewise.
27772 * sbitmap.c (bitmap_clear_range): New function.
27773 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
27775 2017-01-13 Martin Liska <mliska@suse.cz>
27778 * function.c (set_cfun): Add new argument force.
27779 * function.h (set_cfun): Likewise.
27780 * ipa-inline-transform.c (inline_call): Use the function when
27781 strict alising from is dropped for function we inline to.
27783 2017-01-13 Richard Biener <rguenther@suse.de>
27785 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
27786 for dumping GIMPLE INTEGER_CSTs.
27788 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27790 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
27791 to 201112L since C++17.
27793 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
27796 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
27797 if -fsanitize=kernel-address is present.
27799 2017-01-13 Richard Biener <rguenther@suse.de>
27801 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
27802 as _Literal ( type ) number in case usual suffixes do not
27803 preserve all information.
27805 2017-01-13 Richard Biener <rguenther@suse.de>
27807 PR tree-optimization/77283
27808 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
27809 and ssa-iterators.h.
27810 (is_feasible_trace): Implement a cost model based on joiner
27813 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
27816 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
27817 char or short to __float128/_Float128 directly.
27819 2017-01-12 Martin Sebor <msebor@redhat.com>
27821 to -Wformat-overflow.
27822 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
27823 (min_bytes_remaining): Same.
27824 (get_string_length): Same.
27825 (format_string): Same.
27826 (format_directive): Same.
27828 (pass_sprintf_length::handle_gimple_call): Same.
27830 2017-01-12 Jakub Jelinek <jakub@redhat.com>
27832 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
27833 info.nowrite calls with no lhs that can't throw. Return bool
27834 whether gsi_remove has been called or not.
27835 (pass_sprintf_length::handle_gimple_call): Return bool whether
27836 try_substitute_return_value called gsi_remove. Formatting fix.
27837 (pass_sprintf_length::execute): Don't use gsi_remove if
27838 handle_gimple_call returned true.
27841 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
27842 be removed due to side-effects, don't remove following barrier nor
27843 turn the successor edge into fallthru edge.
27845 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27848 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
27849 element-reversing loads and stores as not swappable.
27851 2017-01-12 Nathan Sidwell <nathan@acm.org>
27852 Nicolai Stange <nicstange@gmail.com>
27854 * combine.c (try_combine): Don't ignore result of overlap checking
27855 loop. Combine overlap & asm check into single loop.
27857 2017-01-12 Richard Biener <rguenther@suse.de>
27859 * tree-pretty-print.c (dump_generic_node): Provide -gimple
27860 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
27862 2017-01-12 Richard Biener <rguenther@suse.de>
27864 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
27865 and TS_TARGET_OPTION directly derive from TS_BASE.
27866 * tree-core.h (tree_optimization_option): Derive from tree_base.
27867 (tree_target_option): Likewise.
27869 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
27871 * config/i386/i386.c (memory_address_length): Increase len
27872 only when rip_relative_addr_p returns false.
27874 2017-01-11 Julia Koval <julia.koval@intel.com>
27876 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
27877 (OPTION_MASK_ISA_SGX_SET): New.
27878 (ix86_handle_option): Handle OPT_msgx.
27879 * config.gcc: Added sgxintrin.h.
27880 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
27881 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
27882 * config/i386/i386.c (ix86_target_string): Add -msgx.
27884 (ix86_option_override_internal): Handle new options.
27885 (ix86_valid_target_attribute_inner_p): Add sgx.
27886 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
27887 * config/i386/i386.opt: Add msgx.
27888 * config/i386/sgxintrin.h: New file.
27889 * config/i386/x86intrin.h: Add sgxintrin.h.
27891 2017-01-11 Jakub Jelinek <jakub@redhat.com>
27894 * fold-const.c (maybe_nonzero_address): Return 1 for function
27896 (tree_single_nonzero_warnv_p): Don't handle function local objects
27900 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
27903 2017-01-11 David Malcolm <dmalcolm@redhat.com>
27906 * opts.c: Include "spellcheck.h"
27907 (struct string_fragment): New struct.
27908 (struct edit_distance_traits<const string_fragment &>): New
27910 (get_closest_sanitizer_option): New function.
27911 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
27913 2017-01-11 Jakub Jelinek <jakub@redhat.com>
27915 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
27917 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
27918 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
27919 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
27920 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
27921 for initial die_offset if dwarf_split_debug_info.
27922 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
27923 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
27925 (output_skeleton_debug_sections): Formatting fix. Use
27926 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
27927 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
27929 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
27931 * config/arm/cortex-a53.md: Add bypasses for
27932 cortex_a53_r2f_cvt.
27933 (cortex_a53_r2f): Only use for transfers.
27934 (cortex_a53_f2r): Likewise.
27935 (cortex_a53_r2f_cvt): Add reservation for conversions.
27936 (cortex_a53_f2r_cvt): Likewise.
27938 2017-01-11 Tamar Christina <tamar.christina@arm.com>
27940 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
27941 to all inlined functions, change static to extern.
27943 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
27946 * config/arm/arm.c (legitimize_pic_address): Handle reference to
27948 (arm_assemble_integer): Likewise.
27950 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
27952 * config.gcc: Use new awk script to check CPU, FPU and architecture
27953 parameters for --with-... options.
27954 * config/arm/parsecpu.awk: New file
27955 * config/arm/arm-cpus.in: New file.
27956 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
27958 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
27960 * config/arm/t-arm: Update dependency rules.
27961 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
27962 of processing .def files.
27963 * config/arm/genopt.sh: Deleted.
27964 * config/arm/gentune.sh: Deleted.
27965 * config/arm/arm-cores.def: Deleted.
27966 * config/arm/arm-arches.def: Deleted.
27967 * config/arm/arm-fpus.def: Deleted.
27968 * config/arm/arm-tune.md: Regenerated.
27969 * config/arm/arm-tables.opt: Regenerated.
27970 * config/arm/arm-cpu.h: New generated file.
27971 * config/arm/arm-cpu-data.h: New generated file.
27972 * config/arm/arm-cpu-cdata.h: New generated file.
27974 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
27977 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
27979 (input_varpool_node): Unpack dynamically_initialized bit.
27981 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
27983 PR rtl-optimization/79032
27984 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
27985 the alignment of the adjusted memory reference against that of MODE,
27986 instead of the alignment of the original memory reference.
27988 2017-01-11 Martin Jambor <mjambor@suse.cz>
27990 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
27992 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
27993 decorated functions.
27995 2017-01-11 Richard Biener <rguenther@suse.de>
27997 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
27998 set range/nonnull info for PHI results. Do not set it on
27999 stmts marked for removal.
28001 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
28003 * expr.c (store_field): In the bitfield case, fetch the return value
28004 from the registers before applying a single big-endian adjustment.
28005 Always do a final load for a BLKmode value not larger than a word.
28007 2017-01-10 David Malcolm <dmalcolm@redhat.com>
28010 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
28011 that we correctly handle column numbers greater than
28012 LINE_MAP_MAX_COLUMN_NUMBER.
28014 2017-01-10 Martin Sebor <msebor@redhat.com>
28016 PR middle-end/78245
28017 * gimple-ssa-sprintf.c (get_destination_size): Call
28018 {init,fini}object_sizes.
28019 * tree-object-size.c (addr_object_size): Adjust.
28020 (pass_through_call): Adjust.
28021 (pass_object_sizes::execute): Adjust.
28022 * tree-object-size.h (fini_object_sizes): Declare.
28024 2017-01-10 Martin Sebor <msebor@redhat.com>
28026 PR tree-optimization/78775
28027 * builtins.c (get_size_range): Move...
28028 * calls.c: ...to here.
28029 (alloc_max_size): Accept zero argument.
28030 (operand_signed_p): Remove.
28031 (maybe_warn_alloc_args_overflow): Call get_size_range.
28032 * calls.h (get_size_range): Declare.
28034 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
28036 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
28037 from TI's devices.csv file as of September 2016.
28038 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
28040 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
28042 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
28043 * doc/invoke.texi: Likewise.
28044 * doc/md.texi: Likewise.
28045 * doc/objc.texi: Likewise.
28047 2017-01-10 Joshua Conner <joshconner@google.com>
28049 * config/arm/fuchsia-elf.h: New file.
28050 * config/fuchsia.h: New file.
28051 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
28052 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
28054 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
28056 2016-01-10 Richard Biener <rguenther@suse.de>
28058 PR tree-optimization/79034
28059 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28060 Propagate out degenerate PHIs in the joiner.
28062 2017-01-10 Martin Liska <mliska@suse.cz>
28064 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
28065 (sort_congruence_classes_by_decl_uid): Likewise.
28066 (sort_congruence_class_groups_by_decl_uid): Likewise.
28067 (sem_item_optimizer::merge_classes): Sort class, groups in these
28068 classes and members in the groups by DECL_UID of declarations.
28069 This would make merge operations stable.
28071 2017-01-10 Martin Liska <mliska@suse.cz>
28073 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
28074 usage of m_classes_vec.
28075 (sem_item_optimizer::~sem_item_optimizer): Likewise.
28076 (sem_item_optimizer::get_group_by_hash): Likewise.
28077 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
28078 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
28079 (sem_item_optimizer::verify_classes): Likewise.
28080 (sem_item_optimizer::process_cong_reduction): Likewise.
28081 (sem_item_optimizer::dump_cong_classes): Likewise.
28082 (sem_item_optimizer::merge_classes): Likewise.
28083 * ipa-icf.h (congruence_class_hash): Rename from
28084 congruence_class_group_hash. Remove declaration of m_classes_vec.
28086 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
28088 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
28089 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
28090 * config.gcc: Add avx512vpopcntdqintrin.h.
28091 * config/i386/avx512vpopcntdqintrin.h: New.
28092 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
28093 * config/i386/i386-builtin-types.def: Add new types.
28094 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
28095 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
28096 __builtin_ia32_vpopcountq_v8di_mask): New.
28097 * config/i386/i386-c.c (ix86_target_macros_internal): Define
28098 __AVX512VPOPCNTDQ__.
28099 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
28100 (PTA_AVX512VPOPCNTDQ): Define.
28101 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
28102 TARGET_AVX512VPOPCNTDQ_P): Define.
28103 * config/i386/i386.opt: Add mavx512vpopcntdq.
28104 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
28105 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
28107 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28109 PR middle-end/77484
28110 * predict.def (PRED_CALL): Set to 67.
28112 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
28114 * expr.c (store_field): In the bitfield case, if the value comes from
28115 a function call and is of an aggregate type returned in registers, do
28116 not modify the field mode; extract the value in all cases if the mode
28117 is BLKmode and the size is not larger than a word.
28119 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
28122 * config/i386/cpuid.h: Fix undefined behavior.
28124 2017-01-04 Jeff Law <law@redhat.com>
28126 PR tree-optimization/79007
28127 PR tree-optimization/67955
28128 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
28129 conservative for pt.null when flag_non_call_exceptions is on.
28131 2017-01-09 Jakub Jelinek <jakub@redhat.com>
28133 PR translation/79019
28134 PR translation/79020
28135 * params.def (PARAM_INLINE_MIN_SPEEDUP,
28136 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
28137 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
28139 * config/avr/avr.opt (maccumulate-args): Likewise.
28140 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
28141 * common.opt (freport-bug): Likewise.
28142 * cif-code.def (CIF_FINAL_ERROR): Likewise.
28143 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
28144 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
28145 translatable string.
28146 * config/i386/i386.c (function_value_32): Likewise.
28147 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
28148 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
28150 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
28151 * common/config/msp430/msp430-common.c (msp430_handle_option):
28153 * symtab.c (symtab_node::verify_base): Likewise.
28154 * opts.c (set_debug_level): Likewise.
28155 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
28156 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
28157 missing whitespace to translatable strings.
28158 * config/avr/avr.md (bswapsi2): Fix typo in comment.
28159 * config/sh/superh.h: Likewise.
28160 * config/i386/xopintrin.h: Likewise.
28161 * config/i386/znver1.md: Likewise.
28162 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
28163 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
28164 * double-int.h (struct double_int): Likewise.
28165 * double-int.c (div_and_round_double): Likewise.
28166 * wide-int.cc: Likewise.
28167 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
28168 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
28169 * cfgcleanup.c (crossjumps_occured): Renamed to ...
28170 (crossjumps_occurred): ... this.
28171 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
28174 PR tree-optimization/78899
28175 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
28176 returning bool return struct loop *, NULL for failure and the new
28178 (versionable_outer_loop_p): Don't version outer loop if it has
28179 dont_vectorized bit set.
28180 (tree_if_conversion): When versioning outer loop, ensure
28181 tree_if_conversion is performed also on the inner loop of the
28182 non-vectorizable outer loop copy.
28183 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
28184 LOOP_VECTORIZED in inner loop of the scalar outer loop and
28185 prevent vectorization of it.
28186 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
28187 the outer loop vectorization of the non-scalar version is attempted
28188 before vectorization of the inner loop in scalar version. If
28189 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
28190 vectorization of its inner loop.
28191 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
28192 has 2 inner loops, rename also on edges from bb whose single pred
28193 is outer_loop->header. Fix typo in function comment.
28195 2017-01-09 Martin Sebor <msebor@redhat.com>
28198 * asan.c (asan_emit_stack_protection): Increase local buffer size
28199 to avoid snprintf truncation warning.
28201 2017-01-09 Andrew Pinski <apinski@cavium.com>
28203 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
28204 to reference thunderx2t99 for the tuning structure
28205 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
28207 (thunderx2t99_extra_costs): This.
28208 * config/aarch64/aarch64-tune.md: Regenerate.
28209 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
28210 (vulcan_addrcost_table): This.
28211 (vulcan_regmove_cost): Rename to ...
28212 (thunderx2t99_regmove_cost): This.
28213 (vulcan_vector_cost): Rename to ...
28214 (thunderx2t99_vector_cost): this.
28215 (vulcan_branch_cost): Rename to ...
28216 (thunderx2t99_branch_cost): This.
28217 (vulcan_tunings): Rename to ...
28218 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
28219 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
28221 2017-01-09 Martin Jambor <mjambor@suse.cz>
28225 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
28226 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
28227 (propagate_vr_accross_jump_function): Use the above function for all
28228 value range computations for pass-through jump functions and type
28229 converasion from explicit value range values.
28230 (ipcp_propagate_stage): Do not attempt to deduce types of formal
28231 parameters from TYPE_ARG_TYPES.
28232 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
28233 (ipa_write_node_info): Stream type of the actual argument.
28234 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
28236 2017-01-09 Martin Liska <mliska@suse.cz>
28239 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
28240 (lookup_compiler): Do not show error message with have_E.
28242 2017-01-09 Jakub Jelinek <jakub@redhat.com>
28244 PR tree-optimization/78938
28245 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
28246 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
28247 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
28248 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
28251 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28253 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
28256 2017-01-09 Richard Biener <rguenther@suse.de>
28258 PR tree-optimization/78997
28259 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
28260 name condition properly.
28262 2017-01-09 Richard Biener <rguenther@suse.de>
28265 * dwarf2out.c (is_cxx): New overload with context.
28266 (is_naming_typedef_decl): Use it.
28268 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28270 * invoke.texi (Option Summary): Correct spacing in option lists
28271 and add line breaks to fix over-long lines.
28273 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28275 PR middle-end/17660
28277 * extend.texi (Common Variable Attributes): Add xref to GCC
28278 Internals manual to explain mode attribute keywords.
28280 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28283 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
28284 and Preprocessor Options.
28285 (Options for Linking): Document -pthread here....
28286 (RS/6000 and PowerPC Options): ...not here.
28287 (Solaris 2 Options): ...or here.
28288 * doc/cppopts.texi: Document -pthread.
28290 2017-01-08 Martin Sebor <msebor@redhat.com>
28292 PR middle-end/77708
28293 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
28294 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
28295 New member functions.
28296 (format_directive): Used them.
28298 (pass_sprintf_length::handle_gimple_call): Same.
28299 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
28300 to avoid truncation for any argument.
28301 (extract_affine_mul): Same.
28302 * tree.c (get_file_function_name): Same.
28304 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28306 PR middle-end/77484
28307 * predict.def (PRED_INDIR_CALL): Set to 86.
28309 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28311 PR preprocessor/54124
28312 * doc/cppopts.texi: Reformat -d subtable to list the full name
28313 of the options. Add cross-reference to the docs for the general
28314 compiler -d options.
28315 * doc/invoke.texi (Developer Options): Add cross-reference to the
28316 preprocessor-specific -d option documentation.
28318 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28320 PR preprocessor/13498
28321 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
28322 redudant material, and reflect new command-line options.
28323 (System Headers): Likewise.
28325 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28327 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
28328 -isystem, and -idirafter. Copy-edit.
28329 * doc/cppopts.texi: Copy-edit. Remove contradiction about
28330 default for -ftrack-macro-expansion. Delete obsolete and
28331 badly-formatted implementation details about -fdebug-cpp output.
28332 * doc/cppwarnopts.texi: Copy-edit.
28334 2017-01-07 David Malcolm <dmalcolm@redhat.com>
28337 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
28338 that the transition from a max line width >= 1<<10 to narrower
28339 lines works correctly.
28341 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
28343 * doc/options.texi (PerFunction): New.
28344 * opt-functions.awk (switch_flags): Map both Optimization and
28345 PerFunction to CL_OPTIMIZATION.
28346 * opth-gen.awk: Test for PerFunction flag along with
28348 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
28349 it only when the latter is present. Skip those that don't in
28350 the hash function generator.
28351 * common.opt (fvar-tracking): Mark as PerFunction instead of
28353 (fvar-tracking-assignments): Likewise.
28354 (fvar-tracking-assignments-toggle): Likewise.
28355 (fvar-tracking-uninit): Likewise.
28357 2017-01-07 Jakub Jelinek <jakub@redhat.com>
28359 PR translation/79018
28360 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
28363 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
28366 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
28367 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
28368 TARGET_LONG_JUMP_TABLE_OFFSETS.
28369 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
28370 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
28371 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
28372 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
28373 * config/m68k/m68k.md (tablejump expander): Likewise.
28374 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
28375 TARGET_LONG_JUMP_TABLE_OFFSETS.
28376 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
28377 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
28379 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
28380 David Holsgrove <david.holsgrove@xilinx.com>
28382 * common/config/microblaze/microblaze-common.c
28383 (TARGET_EXCEPT_UNWIND_INFO): Remove.
28384 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
28386 * config/microblaze/microblaze.c (microblaze_must_save_register)
28387 (microblaze_expand_epilogue, microblaze_return_addr): Handle
28389 (microblaze_eh_return): New function.
28390 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
28391 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
28392 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
28393 * config/microblaze/microblaze.md (eh_return): New pattern.
28395 2017-01-06 Jakub Jelinek <jakub@redhat.com>
28397 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
28398 GCC_DIAGNOSTIC_STRINGIFY): Define.
28400 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
28402 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28404 * config/arm/arm.md (<mcrr>): New.
28406 * config/arm/arm.c (arm_arch5te): New.
28407 (arm_option_override): Set arm_arch5te.
28408 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
28410 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
28411 (arm_mcrr_qualifiers): ... this. New.
28412 (MRRC_QUALIFIERS): Define to...
28413 (arm_mrrc_qualifiers): ... this. New.
28414 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
28416 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
28417 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
28418 (MRRCI, mrrc, MRRC): New.
28419 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
28420 VUNSPEC_MRRC2): New.
28422 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28424 * config/arm/arm.md (<mcr>): New.
28426 * config/arm/arm.c (arm_coproc_builtin_available): Add
28427 support for mcr, mrc, mcr2 and mrc2.
28428 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
28429 (arm_mcr_qualifiers): ... this. New.
28430 (MRC_QUALIFIERS): Define to ...
28431 (arm_mrc_qualifiers): ... this. New.
28432 (MCR_QUALIFIERS): Define to ...
28433 (arm_mcr_qualifiers): ... this. New.
28434 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
28436 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
28437 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
28438 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
28439 VUNSPEC_MRC2): New.
28441 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28443 * config/arm/arm.md (*ldc): New.
28447 * config/arm/arm.c (arm_coproc_builtin_available): Add
28448 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
28449 (arm_coproc_ldc_stc_legitimate_address): New.
28450 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
28451 'qualifier_const_pointer'.
28452 (LDC_QUALIFIERS): Define to...
28453 (arm_ldc_qualifiers): ... this. New.
28454 (STC_QUALIFIERS): Define to...
28455 (arm_stc_qualifiers): ... this. New.
28456 * config/arm/arm-protos.h
28457 (arm_coproc_ldc_stc_legitimate_address): New.
28458 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
28459 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
28460 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
28461 stc2, stcl, stc2l): New.
28462 * config/arm/constraints.md (Uz): New.
28463 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
28464 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
28465 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
28466 VUNSPEC_STC2L): New.
28468 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28470 * config/arm/arm.md (<cdp>): New.
28471 * config/arm/arm.c (neon_const_bounds): Rename this ...
28472 (arm_const_bounds): ... this.
28473 (arm_coproc_builtin_available): New.
28474 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
28475 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
28476 (CDP_QUALIFIERS): Define to...
28477 (arm_cdp_qualifiers): ... this. New.
28479 (arm_expand_builtin_args): Add case for 6 arguments.
28480 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
28481 (arm_const_bounds): ... this.
28482 (arm_coproc_builtin_available): New.
28483 * config/arm/arm_acle.h (__arm_cdp): New.
28485 * config/arm/arm_acle_builtins.def (cdp): New.
28487 * config/arm/iterators.md (CDPI,CDP,cdp): New.
28488 * config/arm/neon.md: Rename all 'neon_const_bounds' to
28489 'arm_const_bounds'.
28490 * config/arm/types.md (coproc): New.
28491 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
28492 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
28493 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
28494 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
28496 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28498 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
28499 (UBINOP_QUALIFIERS): New.
28501 (acle_builtin_data): New. Change comment.
28502 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
28503 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
28504 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
28505 arm_acle_builtins.def.
28506 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
28507 (arm_init_acle_builtins): New.
28508 (CRC32_BUILTIN): Remove.
28509 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
28510 crc32cb, crc32ch and crc32cw.
28511 (arm_init_crc32_builtins): Remove.
28512 (arm_init_builtins): Use arm_init_acle_builtins rather
28513 than arm_init_crc32_builtins.
28514 (arm_expand_acle_builtin): New.
28515 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
28516 * config/arm/arm_acle_builtins.def: New.
28518 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28520 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
28521 (arm_builtin_datum): ... this.
28522 (arm_init_neon_builtin): Rename to ...
28523 (arm_init_builtin): ... this. Add a new parameters PREFIX
28524 and USE_SIG_IN_NAME.
28525 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
28526 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
28527 'arm_builtin_datum'.
28528 (arm_init_vfp_builtins): Likewise.
28529 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
28530 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
28531 (arm_expand_neon_args): Rename to ...
28532 (arm_expand_builtin_args): ... this. Rename builtin_arg
28533 enum values and differentiate between ARG_BUILTIN_MEMORY
28534 and ARG_BUILTIN_NEON_MEMORY.
28535 (arm_expand_neon_builtin_1): Rename to ...
28536 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
28537 values, arm_expand_builtin_args and add bool parameter NEON.
28538 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
28539 (arm_expand_vfp_builtin): Likewise.
28540 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
28542 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28544 PR middle-end/77484
28545 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
28546 * predict.c (tree_estimate_probability_bb): Reverse direction of
28547 polymorphic call predictor.
28549 2017-01-06 David Malcolm <dmalcolm@redhat.com>
28551 * passes.c (execute_one_pass): Split out pass-skipping logic into...
28552 (determine_pass_name_match): ...this new function and...
28553 (should_skip_pass_p): ...this new function.
28555 2017-01-06 Nathan Sidwell <nathan@acm.org>
28557 * ipa-visibility.c (function_and_variable_visibility): Reformat
28558 comments and long lines. Remove extrneous if.
28559 * symtab.c (symtab_node::make_decl_local): Fix code format.
28560 (symtab_node::set_section_for_node): Fix comment typo.
28562 2017-01-06 Martin Liska <mliska@suse.cz>
28565 * lra-constraints.c: Rename invariant to lra_invariant.
28566 * predict.c (set_even_probabilities): Initialize e to NULL.
28568 2017-01-05 Martin Sebor <msebor@redhat.com>
28570 PR tree-optimization/78910
28571 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
28572 (format_integer): Correct off-by-one error in the handling
28573 of precision with negative numbers in signed conversions..
28575 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
28577 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
28579 2017-01-05 Jakub Jelinek <jakub@redhat.com>
28581 PR tree-optimization/71016
28582 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
28583 factor_out_conditional_conversion. Formatting fix.
28584 (factor_out_conditional_conversion): Add cond_stmt argument.
28585 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
28586 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
28589 2017-01-05 David Malcolm <dmalcolm@redhat.com>
28591 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
28592 read-rtl-function.o, and selftest-rtl.o.
28593 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
28594 (selftest::aarch64_test_loading_full_dump): New function.
28595 (selftest::aarch64_run_selftests): New function.
28596 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
28597 selftest::aarch64_run_selftests.
28598 * config/i386/i386.c
28599 (selftest::ix86_test_loading_dump_fragment_1): New function.
28600 (selftest::ix86_test_loading_call_insn): New function.
28601 (selftest::ix86_test_loading_full_dump): New function.
28602 (selftest::ix86_test_loading_unspec): New function.
28603 (selftest::ix86_run_selftests): Call the new functions.
28604 * emit-rtl.c (maybe_set_max_label_num): New function.
28605 * emit-rtl.h (maybe_set_max_label_num): New decl.
28606 * function.c (instantiate_decls): Guard call to
28607 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
28608 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
28610 * gensupport.c (gen_reader::gen_reader): Pass "false"
28611 for new "compact" param of rtx_reader.
28612 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
28613 rather than an empty string for NULL strings.
28614 * read-md.c: Potentially include config.h rather than bconfig.h.
28615 Wrap include of errors.h with #ifdef GENERATOR_FILE.
28616 (have_error): New global, copied from errors.c.
28617 (md_reader::read_name): Rename to...
28618 (md_reader::read_name_1): ...this, adding "out_loc" param,
28619 and converting "missing name or number" to returning false, rather
28621 (md_reader::read_name): Reimplement in terms of read_name_1.
28622 (md_reader::read_name_or_nil): New function.
28623 (md_reader::read_string): Handle "(nil)" by returning NULL.
28624 (md_reader::md_reader): Add new param "compact".
28625 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
28626 (md_reader::read_file): New method.
28627 * read-md.h (md_reader::md_reader): Add new param "compact".
28628 (md_reader::read_file): New method.
28629 (md_reader::is_compact): New accessor.
28630 (md_reader::read_name): Convert return type from void to file_location.
28631 (md_reader::read_name_or_nil): New decl.
28632 (md_reader::read_name_1): New decl.
28633 (md_reader::m_compact): New field.
28634 (noop_reader::noop_reader): Pass "false" for new "compact" param
28636 (rtx_reader::rtx_reader): Add new "compact" param.
28637 (rtx_reader::read_rtx_operand): Make virtual and convert return
28638 type from void to rtx.
28639 (rtx_reader::read_until): New decl.
28640 (rtx_reader::handle_any_trailing_information): New virtual function.
28641 (rtx_reader::postprocess): New virtual function.
28642 (rtx_reader::finalize_string): New virtual function.
28643 (rtx_reader::m_in_call_function_usage): New field.
28644 (rtx_reader::m_reuse_rtx_by_id): New field.
28645 * read-rtl-function.c: New file.
28646 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
28647 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
28648 (selftest::verify_three_block_rtl_cfg): New decl.
28649 * read-rtl-function.h: New file.
28650 * read-rtl.c: Potentially include config.h rather than bconfig.h.
28651 For host, include function.h, memmodel.h, and emit-rtl.h.
28652 (one_time_initialization): New function.
28653 (struct compact_insn_name): New struct.
28654 (compact_insn_names): New array.
28655 (find_code): Handle insn codes in compact dumps.
28656 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
28657 (bind_subst_iter_and_attr): Likewise.
28658 (add_condition_to_string): Likewise.
28659 (add_condition_to_rtx): Likewise.
28660 (apply_attribute_uses): Likewise.
28661 (add_current_iterators): Likewise.
28662 (apply_iterators): Likewise.
28663 (initialize_iterators): Guard usage of apply_subst_iterator with
28664 #ifdef GENERATOR_FILE.
28665 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
28666 (md_reader::read_mapping): Likewise.
28667 (add_define_attr_for_define_subst): Likewise.
28668 (add_define_subst_attr): Likewise.
28669 (read_subst_mapping): Likewise.
28670 (check_code_iterator): Likewise.
28671 (rtx_reader::read_rtx): Likewise. Move one-time initialization
28673 (one_time_initialization): New function.
28674 (rtx_reader::read_until): New method.
28675 (read_flags): New function.
28676 (parse_reg_note_name): New function.
28677 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
28678 Handle reuse_rtx ids.
28679 Wrap iterator lookup within #ifdef GENERATOR_FILE.
28680 Add parsing support for RTL dumps, mirroring the special-cases in
28681 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
28682 values, and calling handle_any_trailing_information.
28683 (rtx_reader::read_rtx_operand): Convert return type from void
28684 to rtx, returning return_rtx. Handle case 'e'. Call
28685 finalize_string on XSTR and XTMPL fields.
28686 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
28687 "(nil)" values were omitted. Call the postprocess vfunc on the
28689 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
28690 class ctor. Initialize m_in_call_function_usage. Call
28691 one_time_initialization.
28692 * rtl-tests.c (selftest::test_uncond_jump): Call
28693 set_new_first_and_last_insn.
28694 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
28695 * selftest-rtl.c: New file.
28696 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
28697 (selftest::get_insn_by_uid): New decl.
28698 * selftest-run-tests.c (selftest::run_tests): Call
28699 read_rtl_function_c_tests.
28700 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
28701 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
28704 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
28706 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
28707 operands in a special way. Assert that pos+len <= mode precision.
28709 2017-01-05 Jakub Jelinek <jakub@redhat.com>
28711 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
28712 3 argument Alias with unlimited for the negative form.
28713 (fno-vect-cost-model): Removed.
28715 2017-01-05 Martin Liska <mliska@suse.cz>
28717 * hsa-gen.c (gen_hsa_divmod): New function.
28718 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
28720 2017-01-05 Martin Liska <mliska@suse.cz>
28723 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
28726 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28728 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
28729 small constant length operands.
28731 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28733 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
28734 between loop iterations.
28736 2017-01-05 Martin Liska <mliska@suse.cz>
28739 * gimplify.c (gimplify_decl_expr): Compare to
28740 asan_poisoned_variables instread of checking flags.
28741 (gimplify_target_expr): Likewise.
28742 (gimplify_expr): Likewise.
28743 (gimplify_function_tree): Conditionally initialize
28744 asan_poisoned_variables.
28746 2017-01-04 Jeff Law <law@redhat.com>
28748 PR tree-optimizatin/78812
28749 * rtl.h (contains_mem_rtx_p): Prototype.
28750 * ifcvt.c (containts_mem_rtx_p): Move from here to...
28751 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
28752 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
28753 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
28754 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
28756 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28758 * input.c (assert_char_at_range): Default-initialize actual_range.
28760 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28762 * df-scan.c (df_ref_create_structure): Make regno unsigned,
28763 to match the caller.
28765 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28767 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
28768 insns after final jump in test to emit dummy move.
28770 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28772 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
28773 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
28775 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28777 * multiple_target.c (create_dispatcher_calls): Init e_next.
28778 * tree-ssa-loop-split.c (split_loop): Init border.
28779 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
28782 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
28787 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
28788 (altivec_register_operand): Do not return true if the operand
28789 contains a SUBREG mixing SImode and SFmode.
28790 (vsx_register_operand): Likewise.
28791 (vsx_reg_sfsubreg_ok): New predicate.
28792 (vfloat_operand): Do not return true if the operand contains a
28793 SUBREG mixing SImode and SFmode.
28794 (vint_operand): Likewise.
28795 (vlogical_operand): Likewise.
28796 (gpc_reg_operand): Likewise.
28797 (int_reg_operand): Likewise.
28798 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
28799 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
28800 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
28802 (rs6000_emit_move_si_sf_subreg): New helper function.
28803 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
28804 fixup SUBREGs involving SImode and SFmode.
28805 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
28806 numbers for the new peephole2 optimization.
28807 (peephole2 for SFmode unions): New peephole2 to optimize cases in
28808 the GLIBC math library that do AND/IOR/XOR operations on single
28809 precision floating point.
28810 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
28811 target macros to say whether we need to avoid SUBREGs mixing
28813 (TARGET_ALLOW_SF_SUBREG): Likewise.
28814 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
28815 (UNSPEC_SI_FROM_SF): Likewise.
28816 (iorxor): Change spacing.
28817 (and_ior_xor): New iterator for AND, IOR, and XOR.
28818 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
28819 (movdi_from_sf_zero_ext): Likewise.
28820 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
28821 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
28822 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
28823 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
28824 (fms<mode>4): Likewise.
28825 (fnma<mode>4): Likewise.
28826 (fnms<mode>4): Likewise.
28827 (nfma<mode>4): Likewise.
28828 (nfms<mode>4): Likewise.
28830 2017-01-04 Marek Polacek <polacek@redhat.com>
28833 * doc/invoke.texi: Document -Wpointer-compare.
28835 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28837 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
28840 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
28841 descriptions for -gdwarf-5 and emit them as uleb128 instead of
28844 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
28847 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
28848 documentation of the powerpc_popcntb_ok attribute.
28849 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28850 code to issue warning messages if a requested CPU configuration is
28851 not supported by the binary (assembler and loader) toolchain.
28852 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
28853 made to define a built-in function that has been disabled.
28854 (paired_init_builtins): Add assertion to prevent ICE if attempt is
28855 made to define a built-in function that has been disabled.
28856 (altivec_init_builtins): Add comment explaining why definition
28857 of the DST built-in functions is not preceded by an assertion
28858 check. Add assertions to prevent ICE if attempts are made to
28859 define an altivec predicate or an abs* built-in function that has
28861 (htm_init_builtins): Add comment explaining why definition of the
28862 htm built-in functions is not preceded by an assertion check.
28864 2017-01-04 Jeff Law <law@redhat.com>
28866 PR tree-optimizatin/67955
28867 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
28868 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
28869 the points-to solution does not include pt_null. Use DECL_PT_UID
28872 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
28874 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
28875 Use gen_int_mode instead of gen_lopwart for const_int operands.
28877 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28879 PR tree-optimization/71563
28880 * match.pd: Simplify X << Y into X if Y is known to be 0 or
28881 out of range value - has low bits known to be zero.
28883 2017-01-04 Alan Modra <amodra@gmail.com>
28885 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
28886 * configure: Regenerate.
28887 * config.in: Regenerate.
28889 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28892 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
28893 a substring of the message, but strcmp with the whole message. Ifdef
28894 ENABLE_NLS, translate the message first using dgettext.
28896 2017-01-03 Jeff Law <law@redhat.com>
28898 PR tree-optimizatin/78856
28899 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
28900 (mark_threaded_blocks): Remove code to truncate thread paths that
28901 cross multiple loop headers. Instead invalidate the cached loop
28902 iteration information and handle case of a thread path walking
28903 into an irreducible region.
28905 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
28908 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
28909 assertions. Add support for doing the signbit if the IEEE 128-bit
28910 floating point value is in a GPR.
28911 * config/rs6000/rs6000.md (Fsignbit): Delete.
28912 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
28913 Update the length attribute if the value is in a GPR.
28914 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
28915 the sign or zero extension instruction, since the value is always 0/1.
28916 (signbit<mode>2_dm2): Delete using <Fsignbit>.
28919 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
28920 extracting SImode to a GPR register so that we can generate a
28921 store, limit the vector to be in a traditional Altivec register
28922 for the vextuwrx instruction.
28924 2017-01-03 Ian Lance Taylor <iant@google.com>
28926 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
28928 2017-01-03 Martin Sebor <msebor@redhat.com>
28930 PR tree-optimization/78696
28931 * gimple-ssa-sprintf.c (format_floating): Correct handling of
28932 precision. Use MPFR for %f for greater fidelity. Correct handling
28934 (pass_sprintf_length::compute_format_length): Set width and precision
28935 specified by asrerisk to void_node for vararg functions.
28936 (try_substitute_return_value): Adjust dump output.
28938 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
28940 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
28942 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
28944 * doc/invoke.texi (SPARC options): Document -mlra as the default.
28945 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
28946 -mlra/-mno-lra was passed to the compiler.
28948 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
28950 PR rtl-optimization/65618
28951 * emit-rtl.c (try_split): Move initialization of "before" and
28952 "after" to just before the call to emit_insn_after_setloc.
28954 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
28956 * doc/md.texi (Standard Names): Remove reference to Java frontend.
28958 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
28960 * dwarf2out.c (gen_enumeration_type_die): When
28961 -gno-strict-dwarf, add a DW_AT_encoding attribute.
28963 2017-01-03 Jakub Jelinek <jakub@redhat.com>
28965 PR tree-optimization/78965
28966 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
28967 Change first argument from const call_info & to call_info &. For %n
28968 set info.nowrite to false.
28970 PR middle-end/78901
28971 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
28972 possibly throwing calls.
28974 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
28975 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
28976 and fns handling, rather than in a separate case SSA_NAME.
28978 2017-01-02 Jeff Law <law@redhat.com>
28980 * config/darwin-driver.c (darwin_driver_init): Const-correctness
28981 fixes for first_period and second_period variables.
28983 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
28986 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
28987 (*insvqi_1): New insn pattern.
28988 (*insvqi_1_mem_rex64): Ditto.
28989 (*insvqi_2): Ditto.
28990 (*insvqi_3): Rename from *insvqi.
28992 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
28994 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
28996 * doc/cfg.texi (Edges): Remove reference to Java.
28997 (Maintaining the CFG): Ditto.
28999 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29001 PR middle-end/77674
29002 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
29003 transparent aliases.
29005 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29007 PR middle-end/77484
29008 * predict.def (PRED_CALL): Update hitrate.
29009 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
29010 * predict.c (tree_estimate_probability_bb): Split CALL predictor
29011 into direct/indirect/polymorphic variants.
29013 2017-01-01 Jakub Jelinek <jakub@redhat.com>
29015 Update copyright years.
29017 * gcc.c (process_command): Update copyright notice dates.
29018 * gcov-dump.c (print_version): Ditto.
29019 * gcov.c (print_version): Ditto.
29020 * gcov-tool.c (print_version): Ditto.
29021 * gengtype.c (create_file): Ditto.
29022 * doc/cpp.texi: Bump @copying's copyright year.
29023 * doc/cppinternals.texi: Ditto.
29024 * doc/gcc.texi: Ditto.
29025 * doc/gccint.texi: Ditto.
29026 * doc/gcov.texi: Ditto.
29027 * doc/install.texi: Ditto.
29028 * doc/invoke.texi: Ditto.
29030 Copyright (C) 2017 Free Software Foundation, Inc.
29032 Copying and distribution of this file, with or without modification,
29033 are permitted in any medium without royalty provided the copyright
29034 notice and this notice are preserved.