1 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4 * config/i386/i386.md (*ashl<mode>3_mask_1,
5 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
6 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
9 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
12 (profile_probability): Set max_probability
13 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
15 * profile-count.c (profile_probability::differs_from_p): Do not
16 rely on max_probaiblity == 10000
18 2017-10-12 Jeff Law <law@redhat.com>
20 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
23 2017-10-12 Martin Sebor <msebor@redhat.com>
27 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
28 (handle_alias_pairs): Call it.
29 * common.opt (-Wattribute-alias): New option.
30 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
31 * doc/invoke.texi (-Wattribute-alias): Document.
33 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
36 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
38 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
40 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
41 (make_hard_regno_born, make_hard_regno_dead): Update
42 bb_killed_pseudos and bb_gen_pseudos.
44 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
46 * x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
48 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
50 * config/alpha/alpha.c (alpha_split_conditional_move):
51 Use std::swap instead of manually swapping.
52 (alpha_stdarg_optimize_hook): Ditto.
53 (alpha_canonicalize_comparison): Ditto.
55 2017-10-12 Bin Cheng <bin.cheng@arm.com>
57 * tree-loop-distribution.c (struct builtin_info): New struct.
58 (struct partition): Refactor fields into struct builtin_info.
59 (partition_free): Free struct builtin_info.
60 (build_size_arg_loc, build_addr_arg_loc): Delete.
61 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
62 information from struct builtin_info.
63 (find_single_drs): New function refactored from classify_partition.
64 Also moved builtin validity checks to this function.
65 (compute_access_range, alloc_builtin): New functions.
66 (classify_builtin_st, classify_builtin_ldst): New functions.
67 (classify_partition): Refactor code into functions find_single_drs,
68 classify_builtin_st and classify_builtin_ldst.
69 (distribute_loop): Don't do runtime alias check when distributing
71 (find_seed_stmts_for_distribution): New function.
72 (pass_loop_distribution::execute): Refactor code finding seed
73 stmts into above function. Support distribution for the innermost
74 two-level loop nest. Adjust dump information.
76 2017-10-12 Bin Cheng <bin.cheng@arm.com>
78 * tree-loop-distribution.c: Adjust the general comment.
79 (NUM_PARTITION_THRESHOLD): New macro.
80 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
81 (classify_partition): Skip builtin pattern of loop nest's inner loop.
82 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
83 in call to build_partition_graph.
84 (finalize_partitions): New parameter. Make loop distribution more
85 conservative by fusing more partitions.
86 (distribute_loop): Don't do runtime alias check in case of loop nest
88 (find_seed_stmts_for_distribution): New function.
89 (prepare_perfect_loop_nest): New function.
90 (pass_loop_distribution::execute): Refactor code finding seed stmts
91 and loop nest into above functions. Support loop nest distribution.
92 Adjust dump information accordingly.
94 2017-10-12 Bin Cheng <bin.cheng@arm.com>
96 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
97 and set PTYPE_SEQUENTIAL for merged partition.
99 2017-10-12 Richard Biener <rguenther@suse.de>
101 PR tree-optimization/69728
103 2017-09-19 Richard Biener <rguenther@suse.de>
105 PR tree-optimization/69728
106 * graphite-sese-to-poly.c (schedule_error): New global.
107 (add_loop_schedule): Handle empty domain by failing the
109 (build_original_schedule): Handle schedule_error.
111 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
112 domain by returning an unchanged schedule.
114 2017-10-12 Jakub Jelinek <jakub@redhat.com>
116 * genrecog.c (validate_pattern): For VEC_SELECT verify that
117 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
119 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
121 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
124 2017-10-12 Jakub Jelinek <jakub@redhat.com>
127 * expr.c (store_field): Don't optimize away bitsize == 0 store
128 from CALL_EXPR with addressable return type.
130 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
132 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
133 * config/rs6000/rs6000.md (sel): Delete mode attribute.
134 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
135 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
136 TARGET_ISEL instead of TARGET_ISEL<sel>.
138 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
140 * config/rs6000/rs6000.c
141 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
143 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
145 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
147 (reg_or_cint_operand): Fix comment.
148 (reg_or_zero_operand): New predicate.
149 * config/rs6000/rs6000-protos.h (output_isel): Delete.
150 * config/rs6000/rs6000.c (output_isel): Delete.
151 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
152 instead of reg_or_cint_operand. Output instruction directly (not via
154 (isel_unsigned_<mode>): Ditto.
155 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
156 gpc_reg_operand. Add an instruction alternative for this. Output
157 instruction directly.
158 (*isel_reversed_unsigned_<mode>): Ditto.
160 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
162 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
163 (TARGET_CANONICALIZE_COMPARISON): Define.
165 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
168 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
169 Check whether the dest is REG before adding REG_EQUIV note.
171 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
174 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
176 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
177 (make_hard_regno_born, make_hard_regno_dead): Update
178 bb_killed_pseudos and bb_gen_pseudos.
180 2017-10-11 Nathan Sidwell <nathan@acm.org>
182 * incpath.h (enum incpath_kind): Name enum, prefix values.
183 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
184 * incpath.c (heads, tails): Use INC_MAX.
185 (add_env_var_paths, add_standard_paths): Use incpath_kind.
186 (merge_include_chains, split_quote_chain,
187 register_include_chains): Update incpath_kind names.
188 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
189 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
191 (add_framework_path, darwin_register_objc_includes): Likewise.
192 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
194 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
196 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
197 Do not use float_operator operator predicate.
198 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
199 * config/i386/predicates.md (float_operator): Remove predicate.
201 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
203 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
204 (*jccxf_i387): Ditto.
205 (*jcc<mode>_i387): Ditto.
206 (*jccu<mode>_i387): Ditto.
207 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
208 (*jcc_*_i387 splitters): Remove.
209 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
210 * config/i386/i386.c (ix86_split_fp_branch): Remove.
211 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
214 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
216 * profile-count.h (slow_safe_scale_64bit): New function.
217 (safe_scale_64bit): New inline.
218 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
219 * profile-count.c: Include wide-int.h
220 (slow_safe_scale_64bit): New.
222 2017-10-11 Nathan Sidwell <nathan@acm.org>
224 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
225 HAS_DECL_ASSEMBLER_NAME_P.
226 * gimple-expr.c (gimple_decl_printable_name: Check
227 HAS_DECL_ASSEMBLER_NAME_P too.
228 * ipa-utils.h (type_in_anonymous_namespace_p): Check
229 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
230 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
231 * passes.c (rest_of_decl_compilation): Check
232 HAS_DECL_ASSEMBLER_NAME_P too.
233 * recog.c (verify_changes): Likewise.
234 * tree-pretty-print.c (dump_decl_name): Likewise.
235 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
237 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
238 (SET_DECL_ASSEMBLER_NAME): Use it.
239 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
240 (COPY_DECL_ASSEMBLER_NAME): Likewise.
241 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
243 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
245 * config.gcc (i386, x86_64): Add extra objects.
246 * i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
247 (ix86_min_insn_size): Declare.
248 (ix86_issue_rate): Declare.
249 (ix86_adjust_cost): Declare.
250 (ia32_multipass_dfa_lookahead): Declare.
251 (ix86_macro_fusion_p): Declare.
252 (ix86_macro_fusion_pair_p): Declare.
253 (ix86_bd_has_dispatch): Declare.
254 (ix86_bd_do_dispatch): Declare.
255 (ix86_core2i7_init_hooks): Declare.
256 (ix86_atom_sched_reorder): Declare.
257 * i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
258 (COSTS_N_BYTES): Move to x86-tune-costs.h.
259 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
260 (rip_relative_addr_p): Rename to ...
261 (ix86_rip_relative_addr_p): ... this one; export.
262 (memory_address_length): Update.
263 (ix86_issue_rate): Move to x86-tune-sched.c.
264 (ix86_flags_dependent): Move to x86-tune-sched.c.
265 (ix86_agi_dependent): Move to x86-tune-sched.c.
266 (exact_dependency_1): Move to x86-tune-sched.c.
267 (exact_store_load_dependency): Move to x86-tune-sched.c.
268 (ix86_adjust_cost): Move to x86-tune-sched.c.
269 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
270 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
271 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
272 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
273 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
274 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
275 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
276 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
277 (min_insn_size): Rename to ...
278 (ix86_min_insn_size): ... this one; export.
279 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
280 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
281 (core2i7_first_cycle_multipass_backtrack): Move to
282 x86-tune-sched-core.c.
283 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
284 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
285 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
286 (ix86_avoid_jump_mispredicts): Update.
287 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
288 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
289 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
290 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
291 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
292 (MAX_INSN): Move to ix86-tune-sched-bd.c.
293 (MAX_IMM): Move to ix86-tune-sched-bd.c.
294 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
295 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
296 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
297 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
298 (MAX_STORE): Move to ix86-tune-sched-bd.c.
299 (BIG): Move to ix86-tune-sched-bd.c.
300 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
301 (enum insn_path): Move to ix86-tune-sched-bd.c.
302 (get_mem_group): Move to ix86-tune-sched-bd.c.
303 (is_cmp): Move to ix86-tune-sched-bd.c.
304 (dispatch_violation): Move to ix86-tune-sched-bd.c.
305 (is_branch): Move to ix86-tune-sched-bd.c.
306 (is_prefetch): Move to ix86-tune-sched-bd.c.
307 (init_window): Move to ix86-tune-sched-bd.c.
308 (allocate_window): Move to ix86-tune-sched-bd.c.
309 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
310 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
311 (process_end_window): Move to ix86-tune-sched-bd.c.
312 (allocate_next_window): Move to ix86-tune-sched-bd.c.
313 (find_constant): Move to ix86-tune-sched-bd.c.
314 (get_num_immediates): Move to ix86-tune-sched-bd.c.
315 (has_immediate): Move to ix86-tune-sched-bd.c.
316 (get_insn_path): Move to ix86-tune-sched-bd.c.
317 (get_insn_group): Move to ix86-tune-sched-bd.c.
318 (count_num_restricted): Move to ix86-tune-sched-bd.c.
319 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
320 (add_insn_window): Move to ix86-tune-sched-bd.c.
321 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
322 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
323 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
324 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
325 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
326 (do_dispatch): Move to ix86-tune-sched-bd.c.
327 (has_dispatch): Move to ix86-tune-sched-bd.c.
328 * i386/t-i386: Add new object files.
329 * i386/x86-tune-costs.h: New file.
330 * i386/x86-tune-sched-atom.c: New file.
331 * i386/x86-tune-sched-bd.c: New file.
332 * i386/x86-tune-sched-core.c: New file.
333 * i386/x86-tune-sched.c: New file.
335 2017-10-11 Liu Hao <lh_mouse@126.com>
337 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
338 the generic version below instead.
339 (should_colorize): Recognize Windows consoles as terminals
341 * pretty-print.c [__MINGW32__] (write_all): New function.
342 [__MINGW32__] (find_esc_head): Likewise.
343 [__MINGW32__] (find_esc_terminator): Likewise.
344 [__MINGW32__] (eat_esc_sequence): Likewise.
345 [__MINGW32__] (mingw_ansi_fputs): New function that handles
347 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
350 2017-10-11 Richard Biener <rguenther@suse.de>
352 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
353 Properly call analyze_scalar_evolution with the loop of the stmt.
355 2017-10-11 Richard Biener <rguenther@suse.de>
357 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
358 * tree-core.h (tree_base): Add chrec_var union member.
359 * tree.h (CHREC_VAR): Remove.
360 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
361 * tree-chrec.h (build_polynomial_chrec): Adjust.
362 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
363 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
365 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
367 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
368 * match.pd: ... here.
369 ((T) X == (T) Y): Relax condition.
371 2017-10-11 Bin Cheng <bin.cheng@arm.com>
373 PR tree-optimization/82472
374 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
376 (break_alias_scc_partitions): Update postorder number.
378 2017-10-11 Martin Liska <mliska@suse.cz>
381 * opts.c (parse_no_sanitize_attribute): Do not use error_value
383 * opts.h (parse_no_sanitize_attribute): Remove last argument.
385 2017-10-11 Martin Liska <mliska@suse.cz>
387 * print-rtl.c (print_insn): Move declaration of idbuf
388 to same scope as name.
390 2017-10-11 Martin Liska <mliska@suse.cz>
395 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
398 2017-10-11 Martin Liska <mliska@suse.cz>
401 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
404 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
406 * config.gcc (default_gnu_indirect_function): Default to yes for
407 arm*-*-linux* with glibc.
409 2017-10-11 Richard Biener <rguenther@suse.de>
411 * tree-scalar-evolution.c (get_scalar_evolution): Handle
412 default-defs and types we do not want to analyze.
413 (interpret_loop_phi): Replace unreachable code with an assert.
414 (compute_scalar_evolution_in_loop): Remove and inline ...
415 (analyze_scalar_evolution_1): ... here, replacing condition with
416 what makes the intent clearer. Remove handling of cases
417 get_scalar_evolution now handles.
419 2017-10-10 Jim Wilson <wilson@tuliptree.org>
421 PR rtl-optimization/81434
422 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
423 comment for main loop. In sched_group_found if, also add checks for
424 pass and min_cost_group.
426 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
428 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
429 (rs6000_insn_cost): New function.
430 * config/rs6000/rs6000.md (cost): New attribute.
432 2017-10-10 Jakub Jelinek <jakub@redhat.com>
433 H.J. Lu <hongjiu.lu@intel.com>
437 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
438 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
439 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
440 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
441 to OPTION_MASK_ISA_AVX512VL - builtins that have both
442 OPTION_MASK_ISA_MMX and some other bit set require both
443 mmx and the ISAs without the mmx bit.
444 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
445 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
446 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
447 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
448 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
449 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
450 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
451 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
452 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
453 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
454 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
455 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
456 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
457 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
458 Add OPTION_MASK_ISA_MMX.
460 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
462 * config.gcc (armv7*-*-freebsd*): New target.
463 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
465 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
467 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
468 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
471 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
473 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
474 a template parameter.
475 (WIDE_INT_REF_FOR): Update accordingly.
476 * tree.h (wi::int_traits <const_tree>): Delete.
477 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
478 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
479 (wi::tree_to_wide_ref): New typedef.
480 (wi::to_wide): New function.
481 * calls.c (get_size_range): Use wi::to_wide when operating on
483 * cgraph.c (cgraph_node::create_thunk): Likewise.
484 * config/i386/i386.c (ix86_data_alignment): Likewise.
485 (ix86_local_alignment): Likewise.
486 * dbxout.c (stabstr_O): Likewise.
487 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
488 * expr.c (const_vector_from_tree): Likewise.
489 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
490 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
491 (fold_negate_expr_1, int_const_binop_1, const_binop)
492 (fold_convert_const_int_from_real, optimize_bit_field_compare)
493 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
494 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
495 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
496 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
497 (fold_not_const, round_up_loc): Likewise.
498 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
499 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
500 (alloca_call_type): Likewise.
501 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
502 * godump.c (go_output_typedef): Likewise.
503 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
504 * internal-fn.c (get_min_precision): Likewise.
505 * ipa-cp.c (ipcp_store_vr_results): Likewise.
506 * ipa-polymorphic-call.c
507 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
508 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
509 (ipa_modify_call_arguments): Likewise.
510 * match.pd: Likewise.
511 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
512 * print-tree.c (print_node_brief, print_node): Likewise.
513 * stmt.c (expand_case): Likewise.
514 * stor-layout.c (layout_type): Likewise.
515 * tree-affine.c (tree_to_aff_combination): Likewise.
516 * tree-cfg.c (group_case_labels_stmt): Likewise.
517 * tree-data-ref.c (dr_analyze_indices): Likewise.
518 (prune_runtime_alias_test_list): Likewise.
519 * tree-dump.c (dequeue_and_dump): Likewise.
520 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
521 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
522 * tree-pretty-print.c (dump_generic_node): Likewise.
523 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
524 (simple_iv_with_niters): Likewise.
525 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
526 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
527 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
528 * tree-ssa-loop-niter.c (split_to_var_and_offset)
529 (refine_value_range_using_guard, number_of_iterations_ne_max)
530 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
531 (get_cst_init_from_scev, record_nonwrapping_iv)
532 (scev_var_range_cant_overflow): Likewise.
533 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
534 * tree-ssa-pre.c (compute_avail): Likewise.
535 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
536 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
537 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
538 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
539 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
540 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
541 (dump_case_nodes, try_switch_expansion): Likewise.
542 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
543 (vect_do_peeling): Likewise.
544 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
545 * tree-vect-stmts.c (vectorizable_load): Likewise.
546 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
547 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
548 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
549 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
550 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
551 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
552 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
553 (evrp_dom_walker::before_dom_children): Likewise.
554 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
555 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
556 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
557 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
558 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
559 (get_range_pos_neg): Likewise.
560 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
561 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
562 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
563 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
564 * config/avr/avr.c (avr_fold_builtin): Likewise.
565 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
566 * config/msp430/msp430.c (msp430_attr): Likewise.
567 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
568 * config/powerpcspe/powerpcspe-c.c
569 (altivec_resolve_overloaded_builtin): Likewise.
570 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
571 (rs6000_expand_ternop_builtin): Likewise.
572 * config/rs6000/rs6000-c.c
573 (altivec_resolve_overloaded_builtin): Likewise.
574 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
575 (rs6000_expand_ternop_builtin): Likewise.
576 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
578 2017-10-10 Bin Cheng <bin.cheng@arm.com>
580 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
581 when copying loop nest with only one inner loop.
583 2017-10-10 Richard Biener <rguenther@suse.de>
585 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
586 blocks if SCEV is active.
587 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
589 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
590 (scev_initialize): Assert we are not yet initialized.
592 2017-10-10 Bin Cheng <bin.cheng@arm.com>
594 * tree-loop-distribution.c (generate_loops_for_partition): Remove
595 inner loop's exit stmt by making it always exit the loop, otherwise
596 we would generate an infinite empty loop.
598 2017-10-10 Bin Cheng <bin.cheng@arm.com>
600 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
601 renaming variables in new preheader if it's deleted.
603 2017-10-10 Bin Cheng <bin.cheng@arm.com>
605 * tree-loop-distribution.c (struct partition): Remove unused field
606 loops of the structure.
607 (partition_alloc, partition_free): Ditto.
608 (build_rdg_partition_for_vertex): Ditto.
610 2017-10-09 Jeff Law <law@redhat.com>
612 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
613 return type to match prototype and documentation.
615 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
617 * config/rs6000/rs6000.c (processor_costs): Move to ...
618 * config/rs6000/rs6000.h: ... here.
619 (rs6000_cost): Declare.
621 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
623 * except.c (setjmp_fn): New global variable.
624 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
625 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
626 if DONT_USE_BUILTIN_SETJMP is defined.
628 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
630 * target.def (insn_cost): New hook.
631 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
632 * doc/tm.texi: Regenerate.
633 * rtlanal.c (insn_cost): Use the new hook.
635 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
637 * combine.c (combine_validate_cost): Compute the new insn_cost,
638 not just pattern_cost.
639 (try_combine): Adjust comment.
641 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
643 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
645 * combine.c (uid_insn_cost): Adjust comment.
646 (combine_validate_cost): Adjust comment. Use pattern_cost instead
648 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
649 * dse.c (find_shift_sequence): Ditto.
650 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
651 (bb_valid_for_noce_process_p): Use pattern_cost.
652 * rtl.h (insn_rtx_cost): Delete.
653 (pattern_cost): New prototype.
654 (insn_cost): New prototype.
655 * rtlanal.c (insn_rtx_cost): Rename to...
656 (pattern_cost): ... this.
659 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
661 * config/i386/i386.md (*jcc_2): Remove insn pattern.
662 (*jcc<mode>_0_r_i387): Ditto.
663 (*jccxf_r_i387): Ditto.
664 (*jcc<mode>_r_i387): Ditto.
665 (*jccu<mode>_r_i387): Ditto.
666 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
667 (*jcc): Rename from *jcc_1.
669 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
671 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
672 deferred rescans after the lvx/stvx recombination pre-pass.
674 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
676 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
677 memory operation instruction support.
678 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
679 (rs6000-ibm-aix[789]*): Likewise.
680 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
681 Document new functions.
683 2017-10-09 Richard Biener <rguenther@suse.de>
685 PR tree-optimization/82397
686 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
687 equality only for semantically equal trees.
689 2017-10-09 Richard Biener <rguenther@suse.de>
691 PR tree-optimization/82449
692 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
693 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
694 allow constant addresses.
695 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
698 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
700 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
703 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
706 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
709 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
712 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
714 2017-10-09 Jakub Jelinek <jakub@redhat.com>
717 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
718 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
720 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
722 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
723 (WI_BINARY_PREDICATE_RESULT): Likewise.
724 (wi::binary_traits::operator_result): New type.
725 (wi::binary_traits::predicate_result): Likewise.
726 (generic_wide_int::operator~, unary generic_wide_int::operator-)
727 (generic_wide_int::operator==, generic_wide_int::operator!=)
728 (generic_wide_int::operator&, generic_wide_int::and_not)
729 (generic_wide_int::operator|, generic_wide_int::or_not)
730 (generic_wide_int::operator^, generic_wide_int::operator+
731 (binary generic_wide_int::operator-, generic_wide_int::operator*):
733 (operator~, unary operator-, operator==, operator!=, operator&)
734 (operator|, operator^, operator+, binary operator-, operator*): New
736 * expr.c (get_inner_reference): Use wi::bit_and_not.
737 * fold-const.c (fold_binary_loc): Likewise.
738 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
739 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
740 (bit_value_binop): Likewise.
741 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
742 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
743 (extract_range_from_binary_expr_1): Likewise.
744 (masked_increment): Likewise.
745 (simplify_bit_ops_using_ranges): Likewise.
747 2017-10-09 Martin Jambor <mjambor@suse.cz>
750 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
751 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
752 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
754 (hsa_fixup_mov_insn_type): New function.
755 (hsa_op_with_type::get_in_type): Use it.
756 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
757 immediates in an assert.
758 (hsa_op_with_type::extend_int_to_32bit): New method.
759 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
760 types. Convert to dest type if necessary.
761 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
762 (reg_for_gimple_ssa): Pass false as min32int to
763 hsa_type_for_scalar_tree_type.
764 (gen_hsa_addr): Fixup type when creating addresable temporary.
765 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
766 (gen_hsa_unary_operation): Extend operands and convert to dest type if
767 necessary. Call hsa_fixup_mov_insn_type.
768 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
769 extend operands and convert to dest type if necessary.
770 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
771 to dest type if necessary.
772 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
773 if conversion nt necessary and size matches.
774 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
775 to dest type if necessary.
776 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
777 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
779 (gen_hsa_clrsb): Likewise.
780 (gen_hsa_ffs): Likewise.
781 (gen_hsa_divmod): Extend operands and convert to dest type if
783 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
785 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
787 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
788 Remove empty default arguments. Use a brace block as output
790 (conditional return): Ditto.
792 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
793 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
795 (group_ending_nop): Ditto.
797 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
798 (splitters for those): Ditto.
800 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
802 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
803 a conditional jump (and the compare for it) so that pc_rtx is the
805 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
806 for the deleted and renamed ctr<mode>_internal[234] patterns.
807 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
808 Delete second conditional return pattern.
809 (ctr<mode>_internal2): Delete this second bdnz pattern.
810 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
811 (ctr<mode>_internal4): Delete this second bdz pattern.
813 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
815 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
816 (always_initialized_rtx_for_ssa_name_p): New predicate.
817 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
818 (finish_out_of_ssa): Free new field of SA.
819 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
820 * tree-ssa-coalesce.c: Include tree-ssa.h.
821 (get_parm_default_def_partitions): Remove extern keyword.
822 (get_undefined_value_partitions): New function.
823 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
824 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
826 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
828 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
830 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
832 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
834 * i386.c (ix86_expand_set_or_movmem): Disable 512bit loops for targets
837 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
839 * i386.c (has_dispatch): Disable for Ryzen.
841 2017-10-08 Olivier Hainque <hainque@adacore.com>
843 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
844 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
846 (thumb_set_return_address): Likewise.
848 2017-10-08 Olivier Hainque <hainque@adacore.com>
850 * common/config/arm/arm-common.c (arm_except_unwind_info):
851 Handle DWARF2_UNWIND_INFO.
853 2017-10-07 Michael Collison <michael.collison@arm.com>
855 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
858 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
860 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
861 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
862 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
863 defined, force the creation of a new block for a dispatch label.
865 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
867 * invoke.texi (Wsuggest-attribute=cold): Document.
868 * common.opt (Wsuggest-attribute=cold): New
869 * ipa-pure-const.c (warn_function_cold): New function.
870 * predict.c (compute_function_frequency): Use it.
871 * predict.h (warn_function_cold): Declare.
873 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
875 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
878 2017-10-06 Martin Liska <mliska@suse.cz>
880 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
881 keyword for member functions.
882 (struct sanopt_tree_couple): New struct.
883 (struct sanopt_tree_couple_hash): New function.
884 (struct sanopt_ctx): Add new hash_map.
885 (has_dominating_ubsan_ptr_check): New function.
886 (record_ubsan_ptr_check_stmt): Likewise.
887 (maybe_optimize_ubsan_ptr_ifn): Likewise.
888 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
889 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
891 2017-10-06 Sudakshina Das <sudi.das@arm.com>
894 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
895 aarch64_simd_valid_immediate on CONST_VECTORs.
896 (aarch64_reg_or_bic_imm): Likewise.
898 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
900 PR rtl-optimization/82396
901 * haifa-sched.c (ready_sort_real): Disable qsort checking.
903 2017-10-06 Sebastian Pop <sebpop@gmail.com>
905 * graphite-dependences.c (scop_get_reads): Move code to...
906 (scop_get_must_writes): Move code to...
907 (scop_get_may_writes): Move code to...
908 (scop_get_reads_and_writes): ... here.
909 (scop_get_dependences): Call scop_get_reads_and_writes.
911 2017-10-06 Jakub Jelinek <jakub@redhat.com>
913 PR tree-optimization/82434
914 * fold-const.h (can_native_encode_type_p,
915 can_native_encode_string_p): Remove.
916 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
917 don't encode anything, just return what would be otherwise returned.
918 (native_encode_fixed, native_encode_complex, native_encode_vector):
920 (native_encode_string): Likewise. Inline by hand
921 can_native_encode_string_p.
922 (can_native_encode_type_p): Remove.
923 (can_native_encode_string_p): Remove.
924 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
925 STRING_CSTs using can_native_encode_string_p, test all
926 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
927 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
928 argument from native_encode_expr.
929 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
930 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
933 2017-10-06 Richard Biener <rguenther@suse.de>
935 PR tree-optimization/82397
936 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
937 operand_equal_p but rely on data_ref_compare_tree for detecting
939 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
940 to match up with dr_group_sort_cmp.
942 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
945 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
947 * config/s390/s390-builtin-types.def: Regenerate.
949 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
952 * config/s390/s390-builtin-types.def: Regenerate.
953 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
954 Change flag from B_VXE to B_VX.
955 (s390_vec_min_dbl): Remove B_VXE flag.
957 2017-10-06 Richard Biener <rguenther@suse.de>
959 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
960 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
961 translate_isl_ast_to_gimple::is_valid_rename,
962 translate_isl_ast_to_gimple::get_rename,
963 translate_isl_ast_to_gimple::get_def_bb_for_const,
964 translate_isl_ast_to_gimple::get_new_name,
965 translate_isl_ast_to_gimple::collect_all_ssa_names,
966 translate_isl_ast_to_gimple::copy_loop_phi_args,
967 translate_isl_ast_to_gimple::collect_all_ssa_names,
968 translate_isl_ast_to_gimple::copy_loop_phi_args,
969 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
970 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
971 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
972 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
973 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
974 translate_isl_ast_to_gimple::copy_cond_phi_args,
975 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
976 translate_isl_ast_to_gimple::edge_for_new_close_phis,
977 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
978 translate_isl_ast_to_gimple::rename_uses,
979 translate_isl_ast_to_gimple::rename_all_uses): Remove.
980 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
981 (set_rename_for_each_def): Likewise.
982 (graphite_copy_stmts_from_block): Handle debug stmt resetting
983 here. Handle rewriting SCEV analyzable uses here.
984 (copy_bb_and_scalar_dependences): Generate code for PHI
986 (graphite_regenerate_ast_isl): Adjust.
987 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
988 (add_write, add_read): New functions.
989 (build_cross_bb_scalars_def): Use it and simplify.
990 (build_cross_bb_scalars_use): Likewise.
991 (graphite_find_cross_bb_scalar_vars): Inline into...
992 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
993 simulating out-of-SSA. Compute liveout and add dependencies.
994 (build_scops): Force an empty entry block.
995 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
997 (sese_build_liveouts): Declare.
998 (sese_trivially_empty_bb_p): Likewise.
999 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
1000 compute liveout and debug_liveout.
1001 (sese_bad_liveouts_use): Remove.
1002 (sese_reset_debug_liveouts_bb): Likewise.
1003 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
1004 (sese_build_liveouts): Build liveout and debug_liveout and store
1006 (new_sese_info): Adjust.
1007 (free_sese_info): Likewise.
1008 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
1009 do not build liveout here.
1010 (move_sese_in_condition): Adjust region entry.
1011 (scev_analyzable_p): Match up with chrec_apply requirements.
1012 (sese_trivially_empty_bb_p): New.
1013 * tree-into-ssa.c (get_reaching_def): Properly support generating
1014 default-defs for incremental rewrite of anonymous names.
1016 2017-10-06 Richard Biener <rguenther@suse.de>
1018 * graphite-sese-to-poly.c (extract_affine): For casts increasing
1019 precision do not perform modulo reduction.
1021 2017-10-06 Richard Biener <rguenther@suse.de>
1023 PR tree-optimization/82436
1024 * tree-vect-slp.c (vect_supported_load_permutation_p): More
1025 conservatively choose the vectorization factor when checking
1026 whether we can perform the required load permutation.
1027 (vect_transform_slp_perm_load): Assert when we may not fail.
1029 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
1031 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
1032 message for incompatible -msdata=* and -mcall-* options.
1034 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
1036 * i386.c (ia32_multipass_dfa_lookahead): Default to issue rate
1037 for post-reload scheduling.
1039 2017-10-05 Tamar Christina <tamar.christina@arm.com>
1041 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
1043 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
1045 * i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
1046 to improve monte carlo in scimark.
1048 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
1050 * i386.c (ix86_size_cost, i386_cost, i486_cost, pentium_cost,
1051 lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
1052 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
1053 pentium4_cost, nocona_cost): Set reassociation width to 1.
1054 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
1055 width to 2 for fp operations and 1 otherwise.
1056 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
1058 (atom_cost): Set reassociation width to 2.
1059 (slm_cost, generic_cost): Set fp reassociation width
1060 to 2 and 1 otherwise.
1061 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
1062 (core_cost): Set fp reassociation width to 4 and vector to 2.
1063 (ix86_reassociation_width): Rewrite using cost table; special case
1064 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
1065 and TARGET_AVX128_OPTIMAL.
1066 * i386.h (processor_costs): Add
1067 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
1068 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
1069 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
1070 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
1071 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
1072 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
1074 2017-10-05 Nathan Sidwell <nathan@acm.org>
1076 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
1078 2017-10-05 Tamar Christina <tamar.christina@arm.com>
1080 * config/arm/arm.c (arm_test_fpu_data): New.
1081 (arm_run_selftests): Call arm_test_fpu_data.
1083 2017-10-04 Nathan Sidwell <nathan@acm.org>
1085 * toplev.c (toplev::main): Remove excess parens on pretty_printer
1087 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
1089 2017-10-04 Sudakshina Das <sudi.das@arm.com>
1091 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
1092 check type for aarch64_simd_valid_immediate.
1093 (aarch64_output_simd_mov_immediate): Update prototype.
1094 (aarch64_simd_valid_immediate): Update prototype.
1095 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
1096 support for ORR-immediate.
1097 (and<mode>3): modified pattern to add support for BIC-immediate.
1098 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
1099 now checks for valid immediate for BIC and ORR based on new enum
1101 (aarch64_output_simd_mov_immediate): Function now used to output
1102 BIC/ORR imm as well based on new enum argument.
1103 * config/aarch64/constraints.md (Do): New vector immediate constraint.
1105 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
1106 (aarch64_reg_or_bic_imm): Likewise.
1108 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1110 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
1111 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
1113 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
1117 PR rtl-optimization/82396
1118 * haifa-sched.c (autopref_multipass_init): Simplify
1120 (autopref_rank_data): Simplify sort order.
1121 * sched-int.h (autopref_multipass_data_): Remove
1122 multi_mem_insn_p, min_offset and max_offset.
1124 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1126 * doc/sourcebuild.texi: Document vect_peeling_profitable.
1128 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1130 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
1133 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1135 * doc/sourcebuild.texi: Document vect_long_mult.
1137 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
1139 PR tree-optimization/82413
1140 * fold-const.c (build_range_check): Use widest_int when comparing
1141 the maximum ETYPE value with HIGH.
1143 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
1145 PR rtl-optimization/82396
1146 * haifa-sched.c (autopref_multipass_init): Simplify
1148 (autopref_rank_data): Simplify sort order.
1149 * sched-int.h (autopref_multipass_data_): Remove
1150 multi_mem_insn_p, min_offset and max_offset.
1152 2017-10-04 Jakub Jelinek <jakub@redhat.com>
1154 PR tree-optimization/82381
1155 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
1156 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
1159 PR tree-optimization/82374
1160 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
1161 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
1162 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
1163 current_function_decl to the new decl.
1165 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
1167 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
1168 helper macro for IEEE float128 hardware built-in functions.
1169 (SQRTF128_ODD): Add built-in functions with the round-to-odd
1171 (TRUNCF128_ODD): Likewise.
1172 (ADDF128_ODD): Likewise.
1173 (SUBF128_ODD): Likewise.
1174 (MULF128_ODD): Likewise.
1175 (DIVF128_ODD): Likewise.
1176 (FMAF128_ODD): Likewise.
1177 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
1178 UNSPEC_TRUNC_ROUND_TO_ODD.
1179 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
1180 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
1181 floating point round to odd instructions.
1182 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
1183 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
1184 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
1185 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
1186 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
1187 (trunc<mode>sf2_hw): Change the truncate with round to odd
1188 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
1189 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
1190 to odd hardware instructions.
1191 (sub<mode>3_odd): Likewise.
1192 (mul<mode>3_odd): Likewise.
1193 (div<mode>3_odd): Likewise.
1194 (sqrt<mode>2_odd): Likewise.
1195 (fma<mode>4_odd): Likewise.
1196 (fms<mode>4_odd): Likewise.
1197 (nfma<mode>4_odd): Likewise.
1198 (nfms<mode>4_odd): Likewise.
1199 (trunc<mode>df2_odd): Change the truncate with round to odd
1200 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
1202 * doc/extend.texi (PowerPC built-in functions): Update documentation
1203 for existing IEEE float128-bit built-in functions. Add built-in
1204 functions that generate the IEEE 128-bit floating point round to
1207 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
1209 PR rtl-optimization/77729
1210 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
1211 to (X&(C1&~C2))|C2 transformations.
1213 2017-10-03 Martin Jambor <mjambor@suse.cz>
1215 PR tree-optimization/82363
1216 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
1217 mismatch, mark lacc written regardless of racc.
1219 2017-10-03 Jakub Jelinek <jakub@redhat.com>
1221 PR tree-optimization/82381
1222 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
1223 stmt_to_insert nor wheather SSA_NAMEs are default defs.
1224 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
1225 fallthrough into reassoc_stmt_dominates_stmt_p.
1228 * combine.c (combine_instructions): Don't combine in unreachable
1231 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
1234 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
1235 function to not use the have_cpu variable. Do not set cpu_index,
1236 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
1238 (rs6000_valid_attribute_p): Remove duplicate initializations of
1239 old_optimize and func_optimize.
1240 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
1241 (rs6000_activate_target_options): Make global.
1242 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
1245 2017-10-02 Jakub Jelinek <jakub@redhat.com>
1247 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
1248 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
1249 Set *poffset to 0 and *psize and *pmax_size to -1 if
1250 *poffset + *psize overflows in HOST_WIDE_INT.
1252 PR tree-optimization/82387
1253 PR tree-optimization/82388
1254 PR tree-optimization/82389
1255 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
1256 instead of live_bytes non-NULL.
1258 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
1261 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
1264 2017-10-02 Richard Biener <rguenther@suse.de>
1266 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
1267 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
1268 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
1270 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
1272 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
1273 requested precision matches the type's.
1274 * calls.c (alloc_max_size): Calculate the new candidate size as
1275 a widest_int and use wi::to_widest when comparing it with the
1276 current candidate size.
1277 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
1278 zero rather than integer_zero_node.
1279 * match.pd: Check for a no-op conversion before using wi::add
1280 rather than after. Use tree_to_uhwi when summing small shift
1281 counts into an unsigned int.
1283 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
1284 Alan Hayward <alan.hayward@arm.com>
1285 David Sherwood <david.sherwood@arm.com>
1288 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
1289 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
1290 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
1291 POINTER_AND_FP_REGS.
1293 2017-10-02 Richard Biener <rguenther@suse.de>
1295 PR tree-optimization/82355
1296 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
1297 a mapping for the enclosing loop but avoid generating one for
1299 (copy_bb_and_scalar_dependences): Remove premature codegen
1300 error on PHIs in blocks duplicated into multiple places.
1301 * graphite-scop-detection.c
1302 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
1303 in the region use it as loop and nest to analyze the DR in.
1304 (try_generate_gimple_bb): Likewise.
1305 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
1306 (add_loop_constraints): For blocks in a loop not in the region
1307 create a dimension with a single iteration.
1308 * sese.h (gbb_loop_at_index): Remove assert.
1310 2017-10-01 Kevin Buettner <kevinb@redhat.com>
1312 * omp-expand.c (adjust_context_scope): New function.
1313 (expand_parallel_call): Call adjust_context_scope.
1315 2017-10-01 Jeff Law <law@redhat.com>
1317 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
1318 dom_opt_dom_walker class with direct access to private members.
1319 Add comments. Call test_for_singularity.
1320 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
1321 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
1322 m_dummy_cond anymore.
1323 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
1325 (pass_dominator:execute): Build the dummy_cond here and pass it
1326 to the dom_opt_dom_walker ctor.
1327 (test_for_singularity): New function.
1329 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
1330 Maya Rashish <coypu@sdf.org>
1332 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
1333 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
1334 (alpha*-*-netbsd*) Use nbsd_tm_file.
1335 (arm*-*-netbsdelf*) Likewise.
1336 (i[34567]86-*-netbsdelf*) Likewise.
1337 (x86_64-*-netbsd*) Likewise.
1338 (mips*-*-netbsd*) Likewise.
1339 (powerpc-*-netbsd*) Likewise.
1340 (sh*-*-netbsd*) Likewise.
1341 (sparc-*-netbsdelf*) Likewise.
1342 (sparc64-*-netbsd*) Likewise.
1343 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
1345 (vax-*-netbsdelf*) Likewise.
1346 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
1347 (UINT_FAST8_TYPE) Likewise.
1348 (INT_FAST16_TYPE) Check CHAR_FAST16.
1349 (UINT_FAST16_TYPE) Likewise.
1351 2017-09-30 Jakub Jelinek <jakub@redhat.com>
1354 * config/i386/i386.md
1355 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
1356 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
1357 *divmodsi4_zext_2): New define_insn_and_split.
1358 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
1359 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
1360 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
1361 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
1362 New define_insn_and_split.
1363 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
1364 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
1365 operands[1] having DImode when mode is SImode.
1367 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
1368 always SImode for DIV and MOD in REG_EQUAL notes.
1370 2017-09-29 Yury Gribov <tetra2005@gmail.com>
1373 * match.pd: Fix handling of NaNs in pattern.
1375 2017-09-29 Jeff Law <law@redhat.com>
1377 * sbitmap.c (bitmap_bit_in_range_p): New function.
1378 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
1379 * tree-ssa-dse.c (live_bytes_read): New function.
1380 (dse_classify_store): Ignore reads of dead bytes.
1382 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
1383 typos and whitespace errors.
1384 * config/i386/predicates.md (address_no_seg_operand): Likewise.
1385 * config/s390/s390.c (s390_emit_prologue): Likewise.
1387 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
1390 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
1391 with a symbol for LRA.
1393 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
1395 PR rtl-optimization/82338
1396 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
1398 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
1400 * genmodes.c (calc_wider_mode): Suppress qsort macro.
1401 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
1402 (qsort_chk): Declare.
1403 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
1404 (qsort_chk): New function.
1406 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1408 PR tree-optimization/82337
1409 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
1410 phi definition if the PHI result appears in an abnormal PHI.
1411 (find_basis_for_base_expr): Don't record a basis if the LHS of the
1412 basis appears in an abnormal PHI.
1414 2017-09-29 Richard Biener <rguenther@suse.de>
1416 * graphite-isl-ast-to-gimple.c
1417 (translate_isl_ast_to_gimple::set_codegen_error): New function.
1418 (binary_op_to_tree): Use it.
1419 (get_rename_from_scev): Likewise.
1420 (copy_loop_phi_nodes): Likewise.
1421 (copy_bb_and_scalar_dependences): Likewise.
1422 (translate_pending_phi_nodes): Likewise.
1424 2017-09-29 Jakub Jelinek <jakub@redhat.com>
1427 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
1428 for movabsq $(i32 << shift), r64.
1430 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
1432 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
1433 index when encoding %esp as %rsp to avoid 0x67 prefix.
1435 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
1437 * config/i386/i386.md (*movsf_internal, *movdf_internal):
1438 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
1440 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
1442 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
1443 Extensions with more than 16 double VFP registers.
1444 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
1445 to_clear_mask and all code related to it. Replace the remaining
1446 entry by a sbitmap and adapt code accordingly.
1448 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
1450 * brig-builtins.def: Change pure attributes to const.
1452 2017-09-28 Joseph Myers <joseph@codesourcery.com>
1454 * config.gcc (default_gnu_indirect_function): Default to yes for
1455 sparc*-*-linux* with glibc.
1457 2017-09-28 Joseph Myers <joseph@codesourcery.com>
1459 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
1460 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
1461 when creating .init_array and .fini_array sections with priority
1464 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
1467 * config/aarch64/aarch64.c
1468 (aarch64_builtin_support_vector_misalignment): Always return false
1469 when misalignment is unknown.
1471 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
1473 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
1474 this function to return false if the definition used by the swap
1475 instruction is artificial, or if the memory address from which the
1476 constant value is loaded is not represented by a base address held
1477 in a register or if the base address register is a frame or stack
1478 pointer. Additionally, return false if the base address of the
1479 loaded constant is a SYMBOL_REF but is not considered to be a
1481 (replace_swapped_load_constant): New function.
1482 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
1483 loaded constant vector with a load of a swapped constant vector.
1485 2017-09-27 Carl Love <cel@us.ibm.com>
1487 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
1488 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
1489 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
1492 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
1494 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
1495 first, always call autopref_rank_data otherwise.
1497 2017-09-27 Richard Biener <rguenther@suse.de>
1499 * graphite-scop-detection.c (find_scop_parameters): Move
1500 loop bound handling ...
1501 (gather_bbs::before_dom_children): ... here, avoiding the need
1502 to build scop_info->loop_nest.
1503 (record_loop_in_sese): Remove.
1504 * sese.h (sese_info_t::loop_nest): Remove.
1505 * sese.c (new_sese_info): Do not allocate loop_nest.
1506 (free_sese_info): Do not free loop_nest.
1508 2017-09-27 Jakub Jelinek <jakub@redhat.com>
1511 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
1512 lhs from calls if the lhs has addressable type.
1514 2017-09-27 Richard Biener <rguenther@suse.de>
1516 * graphite.h (scop::max_alias_set): New member.
1517 * graphite-scop-detection.c: Remove references to non-existing
1518 --param in comments.
1519 (build_alias_sets): Record the maximum alias set used for drs.
1520 (build_scops): Support zero as unlimited for
1521 --param graphite-max-arrays-per-scop.
1522 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
1524 (build_poly_sr_1): ... here. Compute alias set based on the
1525 maximum alias set used for drs rather than
1526 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
1528 2017-09-27 Richard Biener <rguenther@suse.de>
1530 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
1531 --param loop-block-tile-size=0 to disable tiling.
1533 2017-09-27 Richard Biener <rguenther@suse.de>
1535 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
1536 (graphite-max-nb-scop-params): Document special value zero.
1537 * domwalk.h (dom_walker::STOP): New symbolical constant.
1538 (dom_walker::dom_walker): Add optional parameter for bb to
1540 (dom_walker::~dom_walker): Declare.
1541 (dom_walker::before_dom_children): Document STOP return value.
1542 (dom_walker::m_user_bb_to_rpo): New member.
1543 (dom_walker::m_bb_to_rpo): Likewise.
1544 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
1545 mapping here if not provided by the user.
1546 (dom_walker::~dom_walker): Free bb to RPO mapping if not
1547 provided by the user.
1548 (dom_walker::STOP): Define.
1549 (dom_walker::walk): Do not compute bb to RPO mapping here.
1550 Support STOP return value from before_dom_children to stop
1552 * graphite-optimize-isl.c (optimize_isl): If the schedule
1553 is the same still generate code if -fgraphite-identity
1554 or -floop-parallelize-all are given.
1555 * graphite-scop-detection.c: Include cfganal.h.
1556 (gather_bbs::gather_bbs): Get and pass through bb to RPO
1558 (gather_bbs::before_dom_children): Return STOP for BBs
1560 (build_scops): Compute bb to RPO mapping and pass it to
1561 the domwalk. Treat --param graphite-max-nb-scop-params=0
1562 as not limiting the number of params.
1563 * graphite.c (graphite_initialize): Remove limit on the
1564 number of basic-blocks in a function.
1565 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
1566 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
1567 default value of 10.
1569 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
1571 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
1572 Adjust code to eliminate needing to do the shift right 32-bits
1573 operation after XSCVDPSPN.
1575 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1577 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
1578 ((X / Y) != 0 -> X >= Y): Likewise.
1580 2017-09-26 Carl Love <cel@us.ibm.com>
1582 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
1583 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
1584 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
1585 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
1586 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
1587 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
1588 definitions and overloading.
1589 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
1590 statement for P9V_BUILTIN_XST_LEN_R.
1591 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
1592 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
1593 define_expand and define_insn for the instructions and builtins.
1594 * doc/extend.texi: Update the built-in documentation file for the new
1596 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
1597 define_insn for the instructions
1599 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
1602 * gcc/config/netbsd-protos.h: New file.
1603 * gcc/config/netbsd.c: New file.
1604 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
1605 * gcc/config/t-netbsd: New file.
1606 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
1607 (tmake_file) Add t-netbsd.
1608 (extra_objs) Add netbsd.o.
1610 2017-09-26 Janus Weil <janus@gcc.gnu.org>
1614 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
1616 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
1618 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
1619 sign extension from a vector register to a GPR by doing a 32-bit
1620 direct move and then an EXTSW.
1621 (extendsi<mode>2 splitter): Likewise.
1622 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
1623 right or vector extract after doing XSCVDPSPN. Use
1624 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
1626 (movdi_from_sf_zero_ext): Likewise.
1627 (reload_gpr_from_vsxsf): Likewise.
1628 (p8_mfvsrd_4_disf): Delete, no longer used.
1629 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
1630 then needing to move the SFmode to a GPR to use the XSCVDPSP
1631 instruction instead of FRSP and XSCVDPSPN.
1632 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
1633 it is adjacent to the other XSCVSPDP insns.
1634 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
1635 SFmode to be in traditional Altivec registers.
1636 (vsx_xscvdpspn): Eliminate useless alternative constraint.
1637 (vsx_xscvspdpn): Likewise.
1638 (vsx_xscvspdpn_scalar): Likewise.
1640 2017-09-26 Martin Jambor <mjambor@suse.cz>
1642 * tree-sra.c (compare_access_positions): Put integral types first,
1643 stabilize sorting of integral types, remove conditions putting
1644 non-full-precision integers last.
1645 (sort_and_splice_var_accesses): Disable scalarization if a
1646 non-integert would be represented by a non-full-precision integer.
1648 2017-09-26 Joseph Myers <joseph@codesourcery.com>
1650 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
1651 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
1652 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
1653 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
1654 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
1655 conditionals inside the function instead of around it. Call
1656 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
1657 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
1659 2017-09-26 Richard Biener <rguenther@suse.de>
1661 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
1663 (scop_detection::build_scop_breadth): ... this. Removed.
1664 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
1665 (scop_detection::harmful_stmt_in_bb): Likewise.
1666 (scop_detection::graphite_can_represent_stmt): Likewise.
1667 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
1668 (scop_detection::can_represent_loop): Remove recursion, fold in ...
1669 (scop_detection::can_represent_loop_1): ... this. Removed.
1670 (scop_detection::harmful_loop_in_region): Simplify after inlining
1671 the above and remove more quadraticness.
1672 (build_scops): Adjust.
1673 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
1676 2017-09-26 Jakub Jelinek <jakub@redhat.com>
1679 * config/i386/i386.c (ix86_print_operand_address_as): Only test
1680 REGNO (base) == SP_REG if base is a REG.
1683 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
1684 if it is different SSA_NAME.
1685 (optimize_range_tests_cmp_bitwise): New function.
1686 (optimize_range_tests): Call it.
1688 2017-09-26 Richard Biener <rguenther@suse.de>
1690 PR tree-optimization/82321
1691 * graphite.c (canonicalize_loop_closed_ssa): Properly check
1692 for the def being inside the loop.
1694 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1696 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
1698 * config/s390/s390-builtins.def: Fix constraint on op4.
1700 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1702 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
1703 independent expanders.
1704 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
1705 ("vec_ordered", "vec_unordered"): New expanders.
1707 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1709 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
1712 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1714 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
1715 vec_unpacks_lo_v16qi.
1716 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
1718 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1720 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
1721 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
1722 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
1724 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1726 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
1728 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
1730 ("*vec_slb<mode>"): New insn pattern.
1731 ("vec_shr_<mode>"): New expander.
1732 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
1733 and force the shift count operand to V16QImode.
1734 ("vec_srb<mode>"): Set shift count mode to V16QI.
1736 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1738 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
1739 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
1740 ("vec_widen_smult_hi_<mode>"): New expander definitions.
1742 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
1745 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
1747 2017-09-26 Richard Biener <rguenther@suse.de>
1749 PR tree-optimization/82320
1750 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
1753 2017-09-25 Jeff Law <law@redhat.com>
1755 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
1756 prototype for new argument.
1757 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
1758 mostly extracted from rs6000_emit_allocate_stack.
1759 (rs6000_emit_probe_stack_range_stack_clash): New function.
1760 (rs6000_emit_allocate_stack): Call
1761 rs6000_emit_probe_stack_range_stack_clash as needed.
1762 (rs6000_emit_probe_stack_range): Add additional argument
1763 to call to gen_probe_stack_range{si,di}.
1764 (output_probe_stack_range): New.
1765 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
1766 (output_probe_stack_range_stack_clash): New.
1767 (rs6000_emit_prologue): Emit notes into dump file as requested.
1768 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
1769 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
1770 Add additional operand and pass it to output_probe_stack_range.
1772 2017-09-25 Bin Cheng <bin.cheng@arm.com>
1774 PR tree-optimization/82163
1775 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
1776 (checking_verify_loop_closed_ssa): New parameter.
1777 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
1778 (check_loop_closed_ssa_stmt): Delete.
1779 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
1780 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
1781 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
1784 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
1786 * brig-builtins.def: Treat HSAIL barrier builtins as
1787 setjmp/longjump style functions.
1789 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
1791 * target.def (constant_alignment): New hook.
1792 * defaults.h (CONSTANT_ALIGNMENT): Delete.
1793 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
1794 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
1795 * doc/tm.texi: Regenerate.
1796 * targhooks.h (default_constant_alignment): Declare.
1797 (constant_alignment_word_strings): Likewise.
1798 * targhooks.c (default_constant_alignment): New function.
1799 (constant_alignment_word_strings): Likewise.
1800 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
1801 instead of CONSTANT_ALIGNMENT.
1802 * varasm.c (align_variable, get_variable_align, build_constant_desc)
1803 (force_const_mem): Likewise.
1804 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
1805 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
1806 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
1807 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1808 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
1810 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
1811 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1812 constant_alignment_word_strings.
1813 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
1814 (CONSTANT_ALIGNMENT): Likewise.
1815 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1816 (arm_constant_alignment): New function.
1817 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
1818 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1819 constant_alignment_word_strings.
1820 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
1821 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1822 constant_alignment_word_strings.
1823 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
1824 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1825 (cris_constant_alignment): New function.
1826 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
1827 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1828 (epiphany_constant_alignment): New function.
1829 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
1830 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1831 constant_alignment_word_strings.
1832 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
1833 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1834 constant_alignment_word_strings.
1835 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
1836 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1837 constant_alignment_word_strings.
1838 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
1839 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
1840 * config/i386/i386.c (ix86_constant_alignment): Make static.
1841 Use the same interface as the target hook.
1842 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1843 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
1844 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1845 constant_alignment_word_strings.
1846 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
1847 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
1848 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1849 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
1850 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1851 constant_alignment_word_strings.
1852 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
1853 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1854 constant_alignment_word_strings.
1855 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
1856 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1857 constant_alignment_word_strings.
1858 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
1859 * config/microblaze/microblaze.c (microblaze_constant_alignment):
1861 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1862 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
1863 * config/mips/mips.c (mips_constant_alignment): New function.
1864 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1865 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
1866 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
1867 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1868 (mmix_constant_alignment): Make static. Use the same interface
1870 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
1871 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1872 constant_alignment_word_strings.
1873 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
1874 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1875 constant_alignment_word_strings.
1876 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
1877 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1878 constant_alignment_word_strings.
1879 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
1880 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1881 (rs6000_constant_alignment): New function.
1882 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
1883 * config/riscv/riscv.c (riscv_constant_alignment): New function.
1884 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1885 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
1886 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1887 (rs6000_constant_alignment): New function.
1888 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
1889 * config/s390/s390.c (s390_constant_alignment): New function.
1890 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1891 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
1892 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1893 constant_alignment_word_strings.
1894 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
1895 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1896 (sparc_constant_alignment): New function.
1897 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
1898 * config/spu/spu.c (spu_constant_alignment): New function.
1899 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1900 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
1901 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1902 constant_alignment_word_strings.
1903 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
1904 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1905 constant_alignment_word_strings.
1906 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
1907 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1908 constant_alignment_word_strings.
1909 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
1910 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1911 (visium_constant_alignment): New function.
1912 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
1913 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1914 (xtensa_constant_alignment): New function.
1915 * system.h (CONSTANT_ALIGNMENT): Poison.
1917 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
1919 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
1920 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
1921 (rs6000_builtin_valid_without_lhs): New helper function.
1922 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1923 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
1925 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
1927 * target.h (vec_perm_indices): Use unsigned short rather than
1929 (auto_vec_perm_indices): Likewise.
1930 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
1931 Use unsigned int rather than unsigned char.
1932 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
1934 2017-09-25 Richard Biener <rguenther@suse.de>
1936 * cfgloop.h (sort_sibling_loops): Declare.
1937 * cfgloop.c (sort_sibling_loops_cmp): New helper.
1938 (sort_sibling_loops): New function sorting the sibling loop list
1940 * graphite.c (graphite_transform_loops): Sort sibling loops.
1942 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
1944 * target.def (vec_perm_const_ok): Change sel parameter to
1946 * optabs-query.c (can_vec_perm_p): Update accordingly.
1947 * doc/tm.texi: Regenerate.
1948 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
1949 auto_vec_perm_indices and remove separate nelt field.
1950 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
1951 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
1952 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
1953 (aarch64_expand_vec_perm_const): Update accordingly.
1954 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
1955 to vec_perm_indices.
1956 * config/arm/arm.c (expand_vec_perm_d): Change perm to
1957 auto_vec_perm_indices and remove separate nelt field.
1958 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
1959 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
1960 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
1962 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
1963 to vec_perm_indices.
1964 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
1965 sel to vec_perm_indices.
1966 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
1967 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
1968 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
1970 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
1973 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
1976 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
1977 on the FUNCTION_DECL function context if it has a DIE that is a
1980 2017-09-25 Richard Biener <rguenther@suse.de>
1982 PR tree-optimization/82285
1983 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
1986 2017-09-25 Tom de Vries <tom@codesourcery.com>
1990 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
1993 2017-09-25 Richard Biener <rguenther@suse.de>
1995 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
1996 ISL errors other than isl_error_quota happen. Dump if the
1997 schedule is the same.
1998 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
1999 errors instead of aborting inside ISL.
2001 2017-09-25 Iain Sandoe <iain@codesourcery.com>
2004 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
2005 of libgcc_eh for m64.
2006 * config/i386/darwin64.h: Likewise.
2008 2017-09-25 Richard Biener <rguenther@suse.de>
2011 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
2012 attribute for incomplete types nor twice for complete ones.
2014 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
2017 * config/i386/i386.c (ix86_print_operand_address_as): Encode
2018 %esp as %rsp to avoid 0x67 prefix if there is no index or base
2021 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
2024 * config/i386/i386.opt (mprefer-avx256): Use
2025 ix86_target_flags variable.
2026 * config/i386/i386.c (ix86_target_string): Move
2027 -mprefer-avx256 to flag2_opts.
2029 2017-09-22 Jakub Jelinek <jakub@redhat.com>
2032 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
2033 and x != -1 | y != -1 into (x & y) != -1.
2035 2017-09-22 Steve Ellcey <sellcey@cavium.com>
2037 * config.gcc: Add new case statement to set
2038 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
2039 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
2040 s390x-*-linux* case statements. Added aarch64 to the list of
2041 supported architectures.
2043 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2045 PR tree-optimization/82289
2046 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
2047 STMT_VINFO_RELEVANT_P.
2049 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2050 Alan Hayward <alan.hayward@arm.com>
2051 David Sherwood <david.sherwood@arm.com>
2053 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
2054 for VR_RANGE only; don't allow VR_ANTI_RANGE.
2055 (extract_range_from_binary_expr_1): Don't call
2056 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
2058 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2059 Alan Hayward <alan.hayward@arm.com>
2060 David Sherwood <david.sherwood@arm.com>
2062 * target.def (preferred_vector_alignment): New hook.
2063 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
2065 * doc/tm.texi: Regenerate.
2066 * targhooks.h (default_preferred_vector_alignment): Declare.
2067 * targhooks.c (default_preferred_vector_alignment): New function.
2068 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
2070 (DR_TARGET_ALIGNMENT): New macro.
2071 (aligned_access_p): Update commentary.
2072 (vect_known_alignment_in_bytes): New function.
2073 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
2075 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
2076 Calculate the misalignment based on the target alignment rather than
2078 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
2079 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
2080 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
2081 the target alignment, rather than masking the element misalignment
2082 with the number of elements in a vector. Also use the target
2083 alignment when calculating the maximum number of peels.
2084 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
2085 instead of TYPE_ALIGN_UNIT.
2086 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
2087 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
2088 (vect_create_addr_base_for_vector_ref): Update call accordingly.
2089 (vect_create_data_ref_ptr): Likewise.
2090 (vect_setup_realignment): Realign by ANDing with
2091 -DR_TARGET_MISALIGNMENT.
2092 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
2093 the number of peels based on DR_TARGET_ALIGNMENT.
2094 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
2095 with the guaranteed alignment boundary when deciding whether
2097 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
2098 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
2099 (ensure_base_align): Remove stmt_info parameter. Get the
2100 target base alignment from DR_TARGET_ALIGNMENT.
2101 (vectorizable_store): Update call accordingly. Interpret
2102 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
2104 (vectorizable_load): Likewise.
2106 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2107 Alan Hayward <alan.hayward@arm.com>
2108 David Sherwood <david.sherwood@arm.com>
2110 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
2111 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
2112 (vect_enhance_data_refs_alignment): Likewise.
2114 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
2116 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
2117 error. Only quit immediately if parsing is complete.
2118 (BEGIN): Initialize fatal_err and parse_done.
2119 (begin fpu, end fpu): Check number of arguments.
2120 (begin arch, end arch): Likewise.
2121 (begin cpu, end cpu): Likewise.
2122 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
2123 (optalias): Likewise.
2125 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
2127 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
2128 * config/arm/arm-isa.h: Delete. Move definitions to ...
2129 * arm-cpus.in: ... here. Use new feature and fgroup values.
2130 * config/arm/arm.c (arm_option_override): Use lower case for feature
2132 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
2133 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
2134 * config/arm/parsecpu.awk (END): Add new command 'isa'.
2136 (print_isa_bits_for): New function.
2137 (gen_isa): New function.
2138 (gen_comm_data): Use print_isa_bits_for.
2139 (define feature): New keyword.
2140 (define fgroup): New keyword.
2141 * config/arm/t-arm (TM_H): Remove.
2142 (GTM_H): Add arm-isa.h.
2143 (arm-isa.h): Add rule to generate file.
2144 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
2145 case for feature bit names.
2147 2017-09-22 Richard Biener <rguenther@suse.de>
2149 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
2151 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
2152 print of no dependency loops ...
2153 * graphite.c (graphite_transform_loops): ... here.
2154 (canonicalize_loop_closed_ssa_form): Work from inner to outer
2156 (same_close_phi_node, remove_duplicate_close_phi,
2157 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
2158 (canonicalize_loop_closed_ssa): ... here and simplify.
2159 * graphite-optimize-isl.c: Include tree-vectorizer.h.
2160 (optimize_isl): Use dump_printf_loc to tell when we stopped
2161 optimizing because of an ISL timeout.
2163 2017-09-22 Richard Biener <rguenther@suse.de>
2165 PR tree-optimization/82291
2166 * tree-if-conv.c (predicate_mem_writes): Make sure to
2167 remove writes in blocks predicated with false.
2169 2017-09-22 Richard Biener <rguenther@suse.de>
2171 * sese.c: Include cfganal.h.
2172 (if_region_set_false_region): Remove.
2173 (create_if_region_on_edge): Likewise.
2174 (move_sese_in_condition): Re-implement without destroying
2177 2017-09-22 Richard Biener <rguenther@suse.de>
2179 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
2180 Verify both BBs contain loop PHI nodes before dispatching to
2182 (graphite_regenerate_ast_isl): Do not recompute dominators,
2183 do not verify three times. Restructure for clarity.
2184 * graphite-scop-detection.c (same_close_phi_node,
2185 remove_duplicate_close_phi, make_close_phi_nodes_unique,
2186 defined_in_loop_p, canonicalize_loop_closed_ssa,
2187 canonicalize_loop_closed_ssa_form): Simplify, remove excess
2188 checking and SSA rewrite, move to ...
2189 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
2190 (graphite_initialize): Do not pass in ctx, do not reset the
2191 SCEV cache, compute only dominators.
2192 (graphite_transform_loops): Allocate ISL ctx after
2193 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
2194 Maintain post-dominators only around build_scops.
2195 * sese.c (if_region_set_false_region): Make static. Free
2196 and recompute dominators.
2197 (move_sese_in_condition): Assert we don't get called with
2198 post-dominators computed.
2199 * sese.h (if_region_set_false_region): Remove.
2201 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
2203 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
2204 mode attribute for TARGET_AVX512VL.
2206 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
2208 * config/i386/i386.opt (mprefer-avx256): New option.
2209 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
2211 (ix86_preferred_simd_mode): Return 256-bit AVX modes
2212 for TARGET_PREFER_AVX256.
2213 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
2215 2017-09-21 Jeff Law <law@redhat.com>
2217 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
2218 Fix dump output if the only stack space is for pushed registers.
2220 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2222 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
2225 2017-09-21 Martin Sebor <msebor@redhat.com>
2228 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
2229 code (in C++) or code that triggers warnings.
2231 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
2233 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
2235 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
2237 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
2238 * sched-rgn.c: Ditto.
2239 * sel-sched-ir.c: Ditto.
2241 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
2243 * toplev.h (set_random_seed): Adjust return type.
2244 * toplev.c (init_local_tick): Move eager initialization of random_seed
2245 to get_random_seed. Adjust comment.
2246 (init_random_seed): Inline to get_random_seed, delete.
2247 (get_random_seed): Initialize random_seed lazily.
2248 (set_random_seed): Do not return previous value.
2249 (print_switch_value): Do not call get_random_seed.
2251 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
2253 * cgraph.c (delete_function_version): New, broken out from...
2254 (cgraph_node::delete_function_version): ...here. Rename to
2255 cgraph_node::delete_function_version_by_decl. Update all uses.
2256 (cgraph_node::remove): Call delete_function_version.
2258 2017-09-21 Jakub Jelinek <jakub@redhat.com>
2261 * tree-inline.c (expand_call_inline): Emit clobber stmts for
2262 VAR_DECLs to which addressable non-volatile parameters are mapped
2263 and for id->retvar after the return value assignment. Clear
2264 id->retval and id->retbnd after inlining.
2266 2017-09-21 Richard Biener <rguenther@suse.de>
2268 PR tree-optimization/82276
2269 PR tree-optimization/82244
2270 * tree-vrp.c (build_assert_expr_for): Set
2271 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
2273 (remove_range_assertions): Revert earlier change.
2275 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
2278 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
2280 2017-09-21 Richard Biener <rguenther@suse.de>
2282 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
2283 Restore valid IL after code generation errors.
2284 * graphite.c (graphite_transform_loops): Diagnose code
2285 generation issues as MSG_MISSED_OPTIMIZATION and continue
2286 with processing SCOPs.
2288 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2289 Alan Hayward <alan.hayward@arm.com>
2290 David Sherwood <david.sherwood@arm.com>
2292 * calls.c (compute_argument_addresses): Use simplify_gen_binary
2293 rather than choosing between plus_constant and gen_rtx_<CODE>.
2294 * expr.c (emit_push_insn): Likewise.
2295 (expand_expr_real_2): Likewise.
2297 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2298 Alan Hayward <alan.hayward@arm.com>
2299 David Sherwood <david.sherwood@arm.com>
2301 * loop-unroll.c (split_iv): Call copy_rtx on the step.
2303 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2304 Alan Hayward <alan.hayward@arm.com>
2305 David Sherwood <david.sherwood@arm.com>
2307 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
2308 calling tree_to_uhwi.
2310 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2311 Alan Hayward <alan.hayward@arm.com>
2312 David Sherwood <david.sherwood@arm.com>
2314 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
2315 INTEGER_CST rather than a negative test for ADDR_EXPR.
2317 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2318 Alan Hayward <alan.hayward@arm.com>
2319 David Sherwood <david.sherwood@arm.com>
2321 * tree-vrp.c (extract_range_from_binary_expr_1): Check
2322 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
2324 2017-09-21 Richard Biener <rguenther@suse.de>
2326 PR tree-optimization/71351
2327 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
2328 graphite_create_new_loop_guard): Remove, fold remaining parts
2330 (translate_isl_ast_node_for): ... here and simplify.
2332 2017-09-21 Jakub Jelinek <jakub@redhat.com>
2335 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
2336 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
2337 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
2338 alternative always use QI mode, for -Os imov (=R,R) alternative
2339 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
2342 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2343 Jeff Law <law@redhat.com>
2345 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
2346 (allocate_stack_space): New function, partially extracted from
2348 (s390_emit_prologue): Track offset to most recent stack probe.
2349 Code to allocate space moved into allocate_stack_space.
2350 Dump actions when no stack is allocated.
2351 (s390_prologue_plus_offset): New function.
2352 (s390_emit_stack_probe): Likewise.
2354 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
2356 * common.opt (Wa, Wl, Wp, g, gz=): Add
2358 (gno-column-info): Remove.
2359 (gcolumn-info): Drop RejectNegative.
2361 (gno-record-gcc-switches): Remove.
2362 (grecord-gcc-switches): Drop RejectNegative.
2363 (gno-split-dwarf): Remove.
2364 (gsplit-dwarf): Drop RejectNegative.
2365 (gno-strict-dwarf): Remove.
2366 (gstrict-dwarf): Drop RejectNegative.
2367 * config/darwin.opt (gfull, gused): Add RejectNegative.
2368 * dwarf2out.c (gen_producer_string): Drop
2369 gno-record-gcc-switches handler.
2370 * optc-gen.awk: Add g to prefixes with negative forms.
2371 * opts-common.c (remapping_prefix_p): New.
2372 (find_opt): Check it.
2373 (generate_canonical_option): Test g prefix.
2374 (option_map): Add -gno- mapping.
2375 (add_misspelling_candidates): Check remapping_prefix_p.
2377 2017-09-20 Jeff Law <law@redhat.com>
2379 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
2380 thinko in stack clash protection support.
2382 * explow.c (compute_stack_clash_protection_loop_data): Use
2383 CONST_INT_P instead of explicit test. Verify object is a
2384 CONST_INT_P before looking at INTVAL.
2385 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
2386 instead of explicit test.
2388 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
2391 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
2392 address instead of to r1 and r11.
2394 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
2396 * config.gcc: Support "knm".
2397 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
2398 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2400 * config/i386/i386.c (m_KNM): Define.
2401 (processor_target_table): Add "knm".
2403 (ix86_option_override_internal): Add "knm".
2404 (ix86_issue_rate): Add PROCESSOR_KNM.
2405 (ix86_adjust_cost): Ditto.
2406 (ia32_multipass_dfa_lookahead): Ditto.
2407 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
2408 (fold_builtin_cpu): Add M_INTEL_KNM.
2409 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
2410 (processor_type): Add PROCESSOR_KNM.
2411 * config/i386/x86-tune.def: Add m_KNM.
2412 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
2414 2017-09-20 Richard Biener <rguenther@suse.de>
2416 PR tree-optimization/80213
2417 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
2418 are allowed in empty BBs as well.
2419 (canonicalize_loop_closed_ssa): Also look for other complex
2421 (scop_detection::get_sese): Include the loop-closed PHI block
2423 (scop_detection::merge_sese): Remove code adding extra blocks.
2424 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
2425 (build_scops): Assert the final returned scop is invalid.
2427 2017-09-20 Richard Biener <rguenther@suse.de>
2429 PR tree-optimization/82264
2430 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
2432 (vn_phi_lookup): Likewise.
2433 (vn_phi_insert): Likewise.
2435 2017-09-20 Jakub Jelinek <jakub@redhat.com>
2437 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
2438 that fits into uhwi or shwi, add DW_AT_const_value regardless
2439 of early_dwarf without going through RTL, using add_AT_unsigned
2442 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
2443 (DEBUG_LTO_ABBREV_SECTION): Likewise.
2444 (DEBUG_LTO_MACINFO_SECTION): Likewise.
2445 (DEBUG_MACRO_SECTION): Likewise.
2446 (DEBUG_LTO_MACRO_SECTION): Likewise.
2447 (DEBUG_STR_DWO_SECTION): Likewise.
2448 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
2449 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
2450 (DEBUG_LTO_DWO_LINE_SECTION): Define.
2451 (DEBUG_LTO_LINE_STR_SECTION): Define.
2452 (init_sections_and_labels): Initialize debug_line_str_section
2453 variable. Initialize debug_loc_section for -gdwarf-5 to
2454 DEBUG_LOCLISTS_SECTION. Formatting fixes.
2456 2017-09-20 Richard Biener <rguenther@suse.de>
2458 * graphite-sese-to-poly.c (extract_affine): Properly handle
2459 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
2461 2017-09-20 Richard Biener <rguenther@suse.de>
2463 PR tree-optimization/81373
2464 * graphite-scop-detection.c (build_cross_bb_scalars_def):
2465 Force SESE live-out defs to be handled even if they are
2468 2017-09-19 Jeff Law <law@redhat.com>
2470 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
2471 nothing for stack adjustments with REG_STACK_CHECK.
2472 * sched-deps.c (parse_add_or_inc): Reject insns with
2473 REG_STACK_CHECK from dependency breaking.
2474 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
2475 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
2476 * reg-notes.def (STACK_CHECK): New note.
2478 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
2479 (ix86_expand_prologue): Dump stack clash info as needed.
2480 Call ix86_adjust_stack_and_probe_stack_clash as needed.
2482 * function.c (dump_stack_clash_frame_info): New function.
2483 * function.h (dump_stack_clash_frame_info): Prototype.
2484 (enum stack_clash_probes): New enum.
2486 * config/alpha/alpha.c (alpha_expand_prologue): Also check
2487 flag_stack_clash_protection.
2488 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
2489 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
2490 (arm_frame_pointer_required): Likewise.
2491 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
2492 (ia64_expand_prologue): Likewise.
2493 * config/mips/mips.c (mips_expand_prologue): Likewise.
2494 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
2495 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
2496 (sparc_flat_expand_prologue): Likewise.
2497 * config/spu/spu.c (spu_expand_prologue): Likewise.
2499 * explow.c: Include "params.h".
2500 (anti_adjust_stack_and_probe_stack_clash): New function.
2501 (get_stack_check_protect): Likewise.
2502 (compute_stack_clash_protection_loop_data): Likewise.
2503 (emit_stack_clash_protection_loop_start): Likewise.
2504 (emit_stack_clash_protection_loop_end): Likewise.
2505 (allocate_dynamic_stack_space): Use get_stack_check_protect.
2506 Use anti_adjust_stack_and_probe_stack_clash.
2507 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
2508 (emit_stack_clash_protection_loop_start): Likewise.
2509 (emit_stack_clash_protection_loop_end): Likewise.
2510 * rtl.h (get_stack_check_protect): Prototype.
2511 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
2512 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
2513 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
2515 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
2517 * doc/tm.texi: Rebuilt.
2518 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
2519 get_stack_check_protect.
2520 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
2521 * config/arm/arm.c (arm_expand_prologue): Likewise.
2522 (arm_frame_pointer_required): Likewise.
2523 * config/i386/i386.c (ix86_expand_prologue): Likewise.
2524 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
2525 * config/mips/mips.c (mips_expand_prologue): Likewise.
2526 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
2527 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
2528 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
2529 (sparc_flat_expand_prologue): Likewise.
2531 * common.opt (-fstack-clash-protection): New option.
2532 * flag-types.h (enum stack_check_type): Note difference between
2533 -fstack-check= and -fstack-clash-protection.
2534 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
2535 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
2536 * toplev.c (process_options): Issue warnings/errors for cases
2537 not handled with -fstack-clash-protection.
2538 * doc/invoke.texi (-fstack-clash-protection): Document new option.
2539 (-fstack-check): Note additional problem with -fstack-check=generic.
2540 Note that -fstack-check is primarily for Ada and refer users
2541 to -fstack-clash-protection for stack-clash-protection.
2542 Document new params for stack clash protection.
2544 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
2546 * config/i386/i386.c (ix86_split_long_move): Do not handle
2547 address used for LEA in a special way.
2549 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
2551 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
2553 2017-09-19 Martin Sebor <msebor@redhat.com>
2556 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
2557 of incompatible types.
2559 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
2561 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
2562 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
2563 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2564 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
2566 2017-09-19 Richard Biener <rguenther@suse.de>
2568 PR tree-optimization/82244
2569 * tree-vrp.c (remove_range_assertions): Do not propagate
2570 a constant to abnormals but replace the assert with a copy.
2572 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
2574 PR rtl-optimization/57878
2575 PR rtl-optimization/68988
2576 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
2577 avoidance test involving non_reload_pseudos. Move frequency test
2578 below the general fragmentation avoidance test.
2580 2017-09-19 Richard Biener <rguenther@suse.de>
2582 PR tree-optimization/69728
2583 * graphite-sese-to-poly.c (schedule_error): New global.
2584 (add_loop_schedule): Handle empty domain by failing the
2586 (build_original_schedule): Handle schedule_error.
2588 2017-09-19 Richard Biener <rguenther@suse.de>
2590 * graphite-scop-detection.c (scop_detection::can_represent_loop):
2591 Do not iterate to sibling loops but only to siblings of inner
2594 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
2597 * config/m68k/m68k.md (moveq feeding equality comparison): Check
2598 that the registers are different.
2600 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
2602 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
2603 to processor_model and "amdfam17h" to arch_names_table.
2604 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
2606 2017-09-18 Jakub Jelinek <jakub@redhat.com>
2609 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
2611 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2612 Alan Hayward <alan.hayward@arm.com>
2613 David Sherwood <david.sherwood@arm.com>
2615 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
2617 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
2619 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
2620 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
2621 vect_schedule_slp_instance.
2622 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
2623 Update call to vect_slp_analyze_node_operations. Simplify return
2625 (vect_slp_analyze_bb_1): Update call accordingly.
2626 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
2627 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
2628 (vect_schedule_slp): Update call accordingly.
2630 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2631 Alan Hayward <alan.hayward@arm.com>
2632 David Sherwood <david.sherwood@arm.com>
2634 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
2635 with types that aren't in fact scalar.
2637 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2639 * tree-vect-slp.c (vect_record_max_nunits): New function,
2641 (vect_build_slp_tree_1): ...here.
2642 (vect_build_slp_tree_2): Call it for phis too.
2644 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2646 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
2647 to vect_get_vec_def_for_operand when getting the mask operand.
2649 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2650 Alan Hayward <alan.hayward@arm.com>
2651 David Sherwood <david.sherwood@arm.com>
2653 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
2656 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2657 Alan Hayward <alan.hayward@arm.com>
2658 David Sherwood <david.sherwood@arm.com>
2660 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
2661 calculation for vector booleans.
2663 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2664 Alan Hayward <alan.hayward@arm.com>
2665 David Sherwood <david.sherwood@arm.com>
2667 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
2669 (vect_transform_stmt): ...here.
2670 (vect_analyze_stmt): Use it instead of calling
2671 vectorizable_live_operation directly.
2673 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
2675 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
2676 non-SIMT targets in acc vector loops.
2678 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
2680 * configure.ac: Add arc and check if assembler supports gdwarf2.
2681 * configure: Regenerate.
2683 2017-09-18 Richard Biener <rguenther@suse.de>
2685 PR tree-optimization/82220
2686 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
2687 epilogue niters from the min_profitable_iters compute.
2689 2017-09-18 Jakub Jelinek <jakub@redhat.com>
2692 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
2693 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
2694 (ix86_init_pic_reg): Revert 2017-09-01 changes.
2696 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
2699 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
2700 switching to a new text section.
2702 2017-09-18 Richard Biener <rguenther@suse.de>
2704 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
2706 (build_alias_set): Reject aliases with no access function.
2708 2017-09-18 Richard Biener <rguenther@suse.de>
2710 PR tree-optimization/79622
2711 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
2713 (build_cross_bb_scalars_use): Likewise.
2715 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
2717 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
2719 2017-09-18 Alan Modra <amodra@gmail.com>
2722 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
2723 stack_pointer_rtx for count 0. Update comments. Break up
2724 large rtl expression.
2726 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
2728 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
2729 Increase to 20 bytes.
2730 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
2731 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
2732 or avx version of the stub.
2734 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
2737 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
2738 compute the minimum stack alignment. Also update preferred stack
2739 boundary for leaf functions.
2741 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
2743 PR tree-optimization/82228
2744 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
2747 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
2749 * common/config/nds32/nds32-common.c
2750 (nds32_option_optimization_table): Refine formatting.
2751 (nds32_option_optimization_table): Use -fsched-pressure and
2752 -fomit-frame-pointer for specific optimization level.
2754 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
2756 * config/nds32/nds32.c: Refine formatting and comments.
2757 * config/nds32/nds32.h: Likewise.
2758 * config/nds32/nds32.md: Likewise.
2759 * config/nds32/nds32-cost.c: Likewise.
2760 * config/nds32/nds32-isr.c: Likewise.
2761 * config/nds32/nds32-md-auxiliary.c: Likewise.
2762 * config/nds32/nds32-multiple.md: Likewise.
2763 * config/nds32/nds32-predicates.c: Likewise.
2765 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
2766 Jakub Jelinek <jakub@redhat.com>
2768 Add support for -std=c++2a.
2769 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
2771 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
2773 2017-09-15 Steve Ellcey <sellcey@cavium.com>
2776 * doc/extend.texi (Common Function Attributes): Add
2777 references to ARM, AArch64, and S/390 specific attributes.
2778 (Function Specific Option Pragmas): Add AArch64 and S/390
2779 to list of back ends that support the target pragma.
2781 2017-09-15 Nathan Sidwell <nathan@acm.org>
2783 * doc/standards.texi: Fix C++17 description. Update URLs for
2786 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
2788 * common.opt (Wcast-align=strict): New warning option.
2789 * doc/invoke.texi: Document -Wcast-align=strict.
2791 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
2793 * cgraph.h (cgraph_thunk_info): Add comments.
2794 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
2795 assert for VIRTUAL_* arguments stricter.
2797 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
2799 PR tree-optimization/71026
2800 * match.pd: Move RDIV patterns from fold-const.c
2801 * fold-const.c (distribute_real_division): Removed.
2802 (fold_binary_loc): Remove calls to distribute_real_divison.
2804 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2806 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
2807 c++1z and gnu++1z as deprecated. Change other references to
2808 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
2809 Change -Wc++1z-compat to -Wc++17-compat.
2810 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
2811 * dwarf2out.c (highest_c_language): Handle C++17.
2812 (gen_compile_unit_die): Likewise.
2814 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2816 PR rtl-optimization/82192
2817 * combine.c (make_extraction): Don't look through non-paradoxical
2818 SUBREGs or TRUNCATE if pos + len is or might be bigger than
2821 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2822 Alan Hayward <alan.hayward@arm.com>
2823 David Sherwood <david.sherwood@arm.com>
2825 * target.def (function_arg_offset): New hook.
2826 * targhooks.h (default_function_arg_offset): Declare.
2827 * targhooks.c (default_function_arg_offset): New function.
2828 * function.c (locate_and_pad_parm): Use
2829 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
2830 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
2831 (TARGET_FUNCTION_ARG_OFFSET): ...this.
2832 * doc/tm.texi: Regenerate.
2833 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
2834 * config/spu/spu.c (spu_function_arg_offset): New function.
2835 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
2836 * system.h (FUNCTION_ARG_OFFSET): Poison.
2838 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2839 Alan Hayard <alan.hayward@arm.com>
2840 David Sherwood <david.sherwood@arm.com>
2842 * target.def (truly_noop_truncation): New hook.
2843 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
2844 than TRULY_NOOP_TRUNCATION.
2845 * hooks.h (hook_bool_uint_uint_true): Declare.
2846 * hooks.c (hook_bool_uint_uint_true): New function.
2847 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
2848 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
2849 * doc/tm.texi: Regenerate.
2850 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
2851 rather than TRULY_NOOP_TRUNCATION in comments.
2852 (simplify_comparison): Likewise.
2853 (record_truncated_value): Likewise.
2854 * expmed.c (extract_bit_field_1): Likewise.
2855 (extract_split_bit_field): Likewise.
2856 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
2857 instead of TRULY_NOOP_TRUNCATION.
2858 * function.c (assign_parm_setup_block): Likewise.
2859 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
2860 * rtlhooks.c: Include target.h.
2861 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
2862 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
2863 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
2864 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
2865 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
2866 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
2867 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
2868 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
2869 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
2870 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
2871 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
2872 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
2873 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
2874 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
2875 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
2876 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
2877 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
2878 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
2879 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
2880 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
2881 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
2882 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
2883 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
2884 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
2885 * config/mips/mips.c (mips_truly_noop_truncation): New function.
2886 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2887 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2888 rather than TRULY_NOOP_TRUNCATION in comments.
2889 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
2890 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
2891 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
2892 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
2893 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
2894 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
2895 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
2896 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
2897 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
2898 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
2899 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
2900 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2901 rather than TRULY_NOOP_TRUNCATION in comments.
2902 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
2903 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
2904 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
2905 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
2906 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
2907 TRULY_NOOP_TRUNCATION condition.
2908 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
2909 (TRULY_NOOP_TRUNCATION): Delete.
2910 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
2911 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
2912 * config/spu/spu.c (spu_truly_noop_truncation): New function.
2913 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2914 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
2915 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
2916 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
2917 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2918 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2919 rather than TRULY_NOOP_TRUNCATION in comments.
2920 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
2921 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
2922 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
2923 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
2924 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
2925 * system.h (TRULY_NOOP_TRUNCATION): Poison.
2927 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
2930 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
2931 (*cmp_ior): Likewise.
2932 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
2933 (*ior_scc_scc_cmp): Likewise.
2934 (*and_scc_scc): Likewise.
2935 (*and_scc_scc_cmp): Likewise.
2937 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2938 Alan Hayard <alan.hayward@arm.com>
2939 David Sherwood <david.sherwood@arm.com>
2941 * target.def (can_change_mode_class): New hook.
2942 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
2943 (hard_regno_nregs): Likewise.
2944 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
2945 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
2946 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
2947 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
2948 (LOAD_EXTEND_OP): Update accordingly.
2949 * doc/tm.texi: Regenerate.
2950 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
2951 CANNOT_CHANGE_MODE_CLASS.
2952 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
2953 (REG_CAN_CHANGE_MODE_P): ...this new macro.
2954 * combine.c (simplify_set): Update accordingly.
2955 * emit-rtl.c (validate_subreg): Likewise.
2956 * recog.c (general_operand): Likewise.
2957 * regcprop.c (mode_change_ok): Likewise.
2958 * reload1.c (choose_reload_regs): Likewise.
2959 (inherit_piecemeal_p): Likewise.
2960 * rtlanal.c (simplify_subreg_regno): Likewise.
2961 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
2962 instead of CANNOT_CHANGE_MODE_CLASS.
2963 (reload_cse_simplify_operands): Likewise.
2964 * reload.c (push_reload): Use targetm.can_change_mode_class
2965 instead of CANNOT_CHANGE_MODE_CLASS.
2966 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
2967 REG_CANNOT_CHANGE_MODE_P.
2968 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2969 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
2970 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2971 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2972 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2973 (arm_can_change_mode_class): New function.
2974 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
2975 than CANNOT_CHANGE_MODE_CLASS in comments.
2976 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2977 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
2978 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
2979 (ix86_can_change_mode_class): ...this new function, inverting the
2980 sense of the return value.
2981 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2982 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2983 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2984 (ia64_can_change_mode_class): New function.
2985 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2986 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
2987 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
2988 (m32c_can_change_mode_class): ...this new function, inverting the
2989 sense of the return value.
2990 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2991 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2992 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
2993 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
2994 (mips_can_change_mode_class): ...this new function, inverting the
2995 sense of the return value.
2996 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2997 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2998 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2999 (msp430_can_change_mode_class): New function.
3000 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3001 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
3002 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3003 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3004 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3005 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
3006 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3007 (pa_cannot_change_mode_class): Replace with...
3008 (pa_can_change_mode_class): ...this new function, inverting the
3009 sense of the return value.
3010 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
3011 than CANNOT_CHANGE_MODE_CLASS in comments.
3012 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3013 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
3014 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3015 (pdp11_cannot_change_mode_class): Replace with...
3016 (pdp11_can_change_mode_class): ...this new function, inverting the
3017 sense of the return value.
3018 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3019 * config/powerpcspe/powerpcspe-protos.h
3020 (rs6000_cannot_change_mode_class_ptr): Delete.
3021 * config/powerpcspe/powerpcspe.c
3022 (rs6000_cannot_change_mode_class_ptr): Delete.
3023 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3024 (rs6000_option_override_internal): Assign to
3025 targetm.can_change_mode_class instead of
3026 rs6000_cannot_change_mode_class_ptr.
3027 (rs6000_cannot_change_mode_class): Replace with...
3028 (rs6000_can_change_mode_class): ...this new function, inverting the
3029 sense of the return value.
3030 (rs6000_debug_cannot_change_mode_class): Replace with...
3031 (rs6000_debug_can_change_mode_class): ...this new function.
3032 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3033 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
3034 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3035 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3036 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
3038 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
3039 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3040 (rs6000_option_override_internal): Assign to
3041 targetm.can_change_mode_class instead of
3042 rs6000_cannot_change_mode_class_ptr.
3043 (rs6000_cannot_change_mode_class): Replace with...
3044 (rs6000_can_change_mode_class): ...this new function, inverting the
3045 sense of the return value.
3046 (rs6000_debug_cannot_change_mode_class): Replace with...
3047 (rs6000_debug_can_change_mode_class): ...this new function.
3048 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3049 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
3050 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
3051 (s390_can_change_mode_class): ...this new function, inverting the
3052 sense of the return value.
3053 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3054 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3055 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
3056 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3057 (sh_cannot_change_mode_class): Replace with...
3058 (sh_can_change_mode_class): ...this new function, inverting the
3059 sense of the return value.
3060 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3061 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3062 (sparc_can_change_mode_class): New function.
3063 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3064 * config/spu/spu.c (spu_can_change_mode_class): New function.
3065 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3066 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3067 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3068 (visium_can_change_mode_class): New function.
3069 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
3071 2017-09-15 Richard Biener <rguenther@suse.de>
3073 PR tree-optimization/82217
3074 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
3075 but not undefined case.
3077 2017-09-15 Jakub Jelinek <jakub@redhat.com>
3080 * postreload.c (reload_cse_simplify_operands): Skip
3081 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
3083 2017-09-15 Richard Biener <rguenther@suse.de>
3085 PR tree-optimization/68823
3086 * graphite-scop-detection.c (build_alias_set): If we have a
3087 possible dependence check whether we can handle them by just
3088 looking at the DRs DR_ACCESS_FNs.
3089 (build_scops): If build_alias_set fails, fail the SCOP.
3091 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
3093 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
3094 to support float128 built-in functions that require the ISA 3.0
3096 (BU_FLOAT128_3_HW): Likewise.
3097 (SQRTF128): Add support for the IEEE 128-bit square root and fma
3099 (FMAF128): Likewise.
3101 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
3102 support for built-in functions that need the ISA 3.0 IEEE 128-bit
3103 floating point instructions.
3104 (rs6000_invalid_builtin): Likewise.
3105 (rs6000_builtin_mask_names): Likewise.
3106 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
3107 (RS6000_BTM_FLOAT128_HW): Likewise.
3108 (RS6000_BTM_COMMON): Likewise.
3109 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
3111 * doc/extend.texi (RS/6000 built-in functions): Document the
3112 IEEE 128-bit floating point square root and fused multiply-add
3115 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
3117 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
3118 reg (r2) isn't in the set of registers defined in the prologue.
3120 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3121 Alan Hayward <alan.hayward@arm.com>
3122 David Sherwood <david.sherwood@arm.com>
3124 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
3125 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
3126 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
3127 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
3128 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
3130 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
3131 max_vectorization_factor.
3132 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
3134 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3135 Alan Hayward <alan.hayward@arm.com>
3136 David Sherwood <david.sherwood@arm.com>
3138 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
3139 (vect_worthwhile_without_simd_p): Declare.
3140 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
3141 (vectorizable_reduction): Use it.
3142 * tree-vect-stmts.c (vectorizable_shift): Likewise.
3143 (vectorizable_operation): Likewise.
3145 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3146 Alan Hayward <alan.hayward@arm.com>
3147 David Sherwood <david.sherwood@arm.com>
3149 * tree-vectorizer.h (vect_get_num_copies): New function.
3150 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
3151 * tree-vect-loop.c (vectorizable_reduction): Likewise.
3152 (vectorizable_induction): Likewise.
3153 (vectorizable_live_operation): Likewise.
3154 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
3155 (vectorizable_bswap): Likewise.
3156 (vectorizable_call): Likewise.
3157 (vectorizable_conversion): Likewise.
3158 (vectorizable_assignment): Likewise.
3159 (vectorizable_shift): Likewise.
3160 (vectorizable_operation): Likewise.
3161 (vectorizable_store): Likewise.
3162 (vectorizable_load): Likewise.
3163 (vectorizable_condition): Likewise.
3164 (vectorizable_comparison): Likewise.
3165 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
3167 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3168 Alan Hayward <alan.hayward@arm.com>
3169 David Sherwood <david.sherwood@arm.com>
3171 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
3172 of vect_init_vector.
3174 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3175 Alan Hayward <alan.hayward@arm.com>
3176 David Sherwood <david.sherwood@arm.com>
3178 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
3179 an inline wrapper that provides a location.
3180 (gimple_build_vector): Likewise.
3181 * gimple-fold.c (gimple_build_vector_from_val): New function.
3182 (gimple_build_vector): Likewise.
3183 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
3184 functions to build the initial value. Always return a gimple value.
3185 (get_initial_defs_for_reduction): Likewise. Only compute
3187 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
3188 vect_init_vector on the results from get_initial_def(s)_for_reduction.
3189 (vectorizable_induction): Use gimple_build_vector rather than
3192 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3193 Alan Hayward <alan.hayward@arm.com>
3194 David Sherwood <david.sherwood@arm.com>
3196 * target.h (vec_perm_indices): New typedef.
3197 (auto_vec_perm_indices): Likewise.
3198 * optabs-query.h: Include target.h
3199 (can_vec_perm_p): Take a vec_perm_indices *.
3200 * optabs-query.c (can_vec_perm_p): Likewise.
3201 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
3202 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
3203 * tree-vect-generic.c (lower_vec_perm): Likewise.
3204 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
3205 (vect_grouped_load_supported): Likewise.
3206 (vect_shift_permute_load_chain): Likewise.
3207 (vect_permute_store_chain): Use auto_vec_perm_indices.
3208 (vect_permute_load_chain): Likewise.
3209 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
3210 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
3211 Update uses of can_vec_perm_p.
3212 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
3213 mode with a number of elements. Take a vec_perm_indices *.
3214 (vect_create_epilog_for_reduction): Update accordingly.
3215 Use auto_vec_perm_indices.
3216 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
3217 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
3218 (vect_transform_slp_perm_load): Likewise.
3219 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
3220 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
3221 (vect_gen_perm_mask_checked): Likewise.
3222 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
3223 (vect_gen_perm_mask_checked): Likewise.
3224 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
3225 (vectorizable_store): Likewise.
3226 (vectorizable_load): Likewise.
3227 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
3228 (vectorizable_bswap): Likewise.
3230 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3231 Alan Hayward <alan.hayward@arm.com>
3232 David Sherwood <david.sherwood@arm.com>
3234 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
3235 * tree.c (build_vector): Likewise.
3236 (build_vector_from_ctor): Update accordingly.
3237 (build_vector_from_val): Likewise.
3238 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
3239 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
3240 * tree-vect-generic.c (add_rshift): Likewise.
3241 (expand_vector_divmod): Likewise.
3242 (optimize_vector_constructor): Likewise.
3243 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
3244 (vect_transform_slp_perm_load): Likewise.
3245 (vect_schedule_slp_instance): Likewise.
3246 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
3247 (vectorizable_call): Likewise.
3248 (vect_gen_perm_mask_any): Likewise. Add elements in order.
3249 * expmed.c (make_tree): Likewise.
3250 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
3251 a vector passed to build_vector.
3252 (fold_convert_const): Likewise.
3253 (exact_inverse): Likewise.
3254 (fold_ternary_loc): Likewise.
3255 (fold_relational_const): Likewise.
3256 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
3257 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
3258 (const_unop): Likewise. Store the reduction accumulator in a
3259 variable rather than an array.
3260 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
3261 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
3262 the new vector, rather than constructing it after the input arrays.
3263 (native_interpret_vector): Use auto_vec<tree> when building
3264 a vector passed to build_vector. Add elements in order.
3265 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
3266 auto_vec<tree> when building a vector passed to build_vector.
3267 (vect_create_epilog_for_reduction): Likewise.
3268 (vectorizable_induction): Likewise.
3269 (get_initial_def_for_reduction): Likewise. Fix indentation of
3271 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
3273 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
3274 passed to build_vector.
3276 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3277 Alan Hayward <alan.hayward@arm.com>
3278 David Sherwood <david.sherwood@arm.com>
3280 * tree-core.h (tree_base::u): Add an "nelts" field.
3281 (tree_vector): Use VECTOR_CST_NELTS as the length.
3282 * tree.c (tree_size): Likewise.
3283 (make_vector): Initialize VECTOR_CST_NELTS.
3284 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
3285 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
3286 TYPE_VECTOR_SUBPARTS.
3287 * expr.c (const_vector_mask_from_tree): Consistently use "units"
3288 as the number of units, setting it from VECTOR_CST_NELTS.
3289 (const_vector_from_tree): Likewise.
3290 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
3291 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
3292 (fold_negate_expr_1): Likewise.
3293 (fold_convert_const): Likewise.
3294 (const_binop): Likewise. Differentiate the number of output and
3296 (const_unop): Likewise.
3297 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
3298 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
3299 in cases that did the opposite.
3301 2017-09-14 Richard Biener <rguenther@suse.de>
3303 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
3306 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
3308 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
3310 2017-09-14 Jakub Jelinek <jakub@redhat.com>
3313 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
3314 if and where to split a bb, except for splitting before debug insn
3315 sequences followed by non-label real insn. Delete debug insns
3316 in between basic blocks.
3318 * combine.c (make_compound_operation_int): Formatting fixes.
3320 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
3321 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3322 * config/netbsd.h (LINK_EH_SPEC): Likewise.
3323 * config/sol2.h (LINK_EH_SPEC): Likewise.
3324 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3325 * config/s390/linux.h (LINK_SPEC): Likewise.
3326 * config/freebsd.h (LINK_EH_SPEC): Likewise.
3327 * config/openbsd.h (LINK_EH_SPEC): Likewise.
3328 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3329 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
3330 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
3331 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3332 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
3333 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
3335 2017-09-13 Jakub Jelinek <jakub@redhat.com>
3337 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
3339 (ENDFILE_LINUX_SPEC): Likewise.
3340 (LINK_EH_SPEC): Likewise.
3341 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
3342 (LINK_OS_LINUX_SPEC32): Likewise.
3343 (LINK_OS_LINUX_SPEC64): Likewise.
3344 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
3345 (LINK_OS_LINUX_SPEC): Likewise.
3347 2017-09-13 Martin Liska <mliska@suse.cz>
3350 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
3351 CASE_HIGH is NULL_TREE.
3353 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
3354 Alan Hayward <alan.hayward@arm.com>
3355 David Sherwood <david.sherwood@arm.com>
3357 * target.def (secondary_memory_needed): New hook.
3358 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
3359 instead of SECONDARY_MEMORY_NEEDED.
3360 (secondary_memory_needed_mode): Likewise.
3361 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
3362 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
3363 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
3364 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
3365 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
3366 * doc/tm.texi: Regenerate.
3367 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
3368 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
3369 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3370 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
3371 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
3372 * config/i386/i386.c (inline_secondary_memory_needed): Put the
3373 mode argument first and change the reg_class arguments to reg_class_t.
3374 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
3375 Make static. Update the call to inline_secondary_memory_needed.
3376 (ix86_register_move_cost): Update the call to
3377 inline_secondary_memory_needed.
3378 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3379 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
3381 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
3382 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
3384 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
3385 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
3386 * config/mips/mips.c (mips_secondary_memory_needed): Make static
3387 and match hook interface. Add comment from mips.h.
3388 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3389 * config/mmix/mmix.md (truncdfsf2): Refer to
3390 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
3392 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
3393 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
3394 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3395 (pa_secondary_memory_needed): New function.
3396 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
3397 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
3398 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3399 (pdp11_secondary_memory_needed): Make static and match hook interface.
3400 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
3401 * config/powerpcspe/powerpcspe-protos.h
3402 (rs6000_secondary_memory_needed_ptr): Delete.
3403 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
3405 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3406 (rs6000_option_override_internal): Assign to
3407 targetm.secondary_memory_needed rather than
3408 rs6000_secondary_memory_needed_ptr.
3409 (rs6000_secondary_memory_needed): Match hook interface.
3410 (rs6000_debug_secondary_memory_needed): Likewise.
3411 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
3412 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
3413 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
3414 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3415 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
3416 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
3418 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
3419 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3420 (rs6000_option_override_internal): Assign to
3421 targetm.secondary_memory_needed rather than
3422 rs6000_secondary_memory_needed_ptr.
3423 (rs6000_secondary_memory_needed): Match hook interface.
3424 (rs6000_debug_secondary_memory_needed): Likewise.
3425 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
3426 * config/s390/s390.c (s390_secondary_memory_needed): New function.
3427 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3428 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
3429 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3430 (sparc_secondary_memory_needed): New function.
3431 * lra-constraints.c (check_and_process_move): Refer to
3432 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
3434 (curr_insn_transform): Likewise.
3435 (process_alt_operands): Use targetm.secondary_memory_needed
3436 instead of TARGET_SECONDARY_MEMORY_NEEDED.
3437 (check_secondary_memory_needed_p): Likewise.
3438 (choose_split_class): Likewise.
3439 * reload.c: Unconditionally include code that was previously
3440 conditional on SECONDARY_MEMORY_NEEDED.
3441 (push_secondary_reload): Use targetm.secondary_memory_needed
3442 instead of TARGET_SECONDARY_MEMORY_NEEDED.
3443 (push_reload): Likewise.
3444 * reload1.c: Unconditionally include code that was previously
3445 conditional on SECONDARY_MEMORY_NEEDED.
3446 (choose_reload_regs): Use targetm.secondary_memory_needed
3447 instead of TARGET_SECONDARY_MEMORY_NEEDED.
3448 (gen_reload): Likewise.
3449 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
3451 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
3452 Alan Hayward <alan.hayward@arm.com>
3453 David Sherwood <david.sherwood@arm.com>
3455 * target.def (secondary_memory_needed_mode): New hook:
3456 * targhooks.c (default_secondary_memory_needed_mode): Declare.
3457 * targhooks.h (default_secondary_memory_needed_mode): New function.
3458 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
3459 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
3460 * doc/tm.texi: Regenerate.
3461 * lra-constraints.c (check_and_process_move): Use
3462 targetm.secondary_memory_needed_mode instead of
3463 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
3464 (curr_insn_transform): Likewise.
3465 * reload.c (get_secondary_mem): Likewise.
3466 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3467 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
3469 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3470 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3471 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
3472 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3473 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
3475 * config/powerpcspe/powerpcspe-protos.h
3476 (rs6000_secondary_memory_needed_mode): Delete.
3477 * config/powerpcspe/powerpcspe.c
3478 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3479 (rs6000_secondary_memory_needed_mode): Make static.
3480 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3481 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
3483 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
3485 (rs6000_secondary_memory_needed_mode): Make static.
3486 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3487 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
3488 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3489 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3490 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
3492 (sparc_secondary_memory_needed_mode): New function.
3493 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
3495 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
3497 * config/aarch64/constraints.md (Umq): New constraint.
3498 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
3500 (mov<mode>): Update condition.
3502 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3504 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
3505 when bitposition is the same.
3507 2017-09-13 Richard Biener <rguenther@suse.de>
3509 * dwarf2out.c (output_die_symbol): Remove.
3510 (output_die): Do not output a DIEs symbol.
3512 2017-09-13 Richard Biener <rguenther@suse.de>
3515 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
3516 default-def to avoid breaking iterator update with the weird
3517 interaction with cgraph_update_edges_for_call_stmt_node.
3519 2017-09-13 Richard Biener <rguenther@suse.de>
3521 * tree-cfg.c (verify_gimple_assign_binary): Add verification
3522 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
3523 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
3524 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
3526 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
3528 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
3529 Disable pc relative literal load irrespective of
3530 TARGET_FIX_ERR_A53_84341 for default.
3532 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
3534 * config/sparc/sparc.c (output_return): Output the source location of
3535 the insn in the delay slot, if any.
3536 (output_sibcall): Likewise.
3538 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
3541 * common.opt (-static-pie): New alias.
3542 (shared): Negate static-pie.
3543 (-no-pie): Update help text.
3545 (static-pie): New option.
3546 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
3547 -static-pie support.
3548 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
3549 (LINK_EH_SPEC): Likewise.
3550 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3551 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
3552 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
3553 * gcc.c (LINK_COMMAND_SPEC): Likewise.
3554 (init_gcc_specs): Likewise.
3555 (init_spec): Likewise.
3556 (display_help): Update help message for -pie.
3557 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
3560 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
3562 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
3563 (movdi_aarch64): Likewise.
3564 (movti_aarch64): Likewise.
3566 2017-09-12 Simon Wright <simon@pushface.org>
3569 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
3570 calculation of the minor version, always output as 0.
3572 2017-09-12 Jakub Jelinek <jakub@redhat.com>
3575 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
3576 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
3577 on it early, rather than manual conversion late. For
3578 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
3579 instead of performing manual conversion.
3581 2017-09-12 Carl Love <cel@us.ibm.com>
3583 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
3584 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
3586 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
3587 VMULOSW): Add definitions.
3588 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
3589 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
3590 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
3591 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
3592 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
3594 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
3596 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
3598 (movti_aarch64): Likewise.
3599 (movdf_aarch64): Likewise.
3600 (movtf_aarch64): Likewise.
3601 (load_pairdi): Likewise.
3602 (store_pairdi): Likewise.
3603 (load_pairdf): Likewise.
3604 (store_pairdf): Likewise.
3605 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
3606 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
3607 (ldr_got_small_<mode>): Likewise.
3608 (ldr_got_small_28k_<mode>): Likewise.
3609 (ldr_got_tiny): Likewise.
3610 * config/aarch64/iterators.md (ldst_sz): New.
3611 (ldpstp_sz): Likewise.
3612 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
3614 (thunderx_load): Split load_8 to load_16.
3615 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
3617 (thunderx2t99_storepair_basic): Split store_8 to store_16.
3618 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
3619 (xgene1_store_pair): Split store_8 to store_16.
3620 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
3621 (falkor_st_0_st_sd): Split store_8 to store_16.
3623 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
3625 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
3626 and store1/2/3/4 to store_4/8/12/16.
3627 * config/aarch64/aarch64.md: Update for rename.
3628 * config/arm/arm.md: Likewise.: Likewise.
3629 * config/arm/arm.c: Likewise.
3630 * config/arm/thumb1.md: Likewise.
3631 * config/arm/thumb2.md: Likewise.
3632 * config/arm/vfp.md: Likewise.
3633 * config/arm/arm-generic.md: Likewise.
3634 * config/arm/arm1020e.md: Likewise.
3635 * config/arm/arm1026ejs.md: Likewise.
3636 * config/arm/arm1136jfs.md: Likewise.
3637 * config/arm/arm926ejs.md: Likewise.
3638 * config/arm/cortex-a15.md: Likewise.
3639 * config/arm/cortex-a17.md: Likewise.
3640 * config/arm/cortex-a5.md: Likewise.
3641 * config/arm/cortex-a53.md: Likewise.
3642 * config/arm/cortex-a57.md: Likewise.
3643 * config/arm/cortex-a7.md: Likewise.
3644 * config/arm/cortex-a8.md: Likewise.
3645 * config/arm/cortex-a9.md: Likewise.
3646 * config/arm/cortex-m4.md: Likewise.
3647 * config/arm/cortex-m7.md: Likewise.
3648 * config/arm/cortex-r4.md: Likewise.
3649 * config/arm/exynos-m1.md: Likewise.
3650 * config/arm/fa526.md: Likewise.
3651 * config/arm/fa606te.md: Likewise.
3652 * config/arm/fa626te.md: Likewise.
3653 * config/arm/fa726te.md: Likewise.
3654 * config/arm/fmp626.md: Likewise.
3655 * config/arm/iwmmxt.md: Likewise.
3656 * config/arm/ldmstm.md: Likewise.
3657 * config/arm/marvell-pj4.md: Likewise.
3658 * config/arm/xgene1.md: Likewise.
3659 * config/aarch64/thunderx.md: Likewise.
3660 * config/aarch64/thunderx2t99.md: Likewise.
3661 * config/aarch64/falkor.md: Likewise.
3663 2017-09-12 Martin Liska <mliska@suse.cz>
3665 * attribs.c (private_lookup_attribute): New function.
3666 * attribs.h (private_lookup_attribute): Declared here.
3667 (lookup_attribute): Called from this place.
3669 2017-09-12 Richard Biener <rguenther@suse.de>
3671 PR tree-optimization/82157
3672 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
3673 stmts with side-effects.
3675 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3676 Alan Hayward <alan.hayward@arm.com>
3677 David Sherwood <david.sherwood@arm.com>
3679 * target.def (hard_regno_nregs): New hook.
3680 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
3681 * targhooks.h (default_hard_regno_nregs): Declare.
3682 * targhooks.c (default_hard_regno_nregs): New function.
3683 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
3684 (TARGET_HARD_REGNO_NREGS): ...this hook.
3685 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
3686 (CLASS_MAX_NREGS): Likewise.
3687 * doc/tm.texi: Regenerate.
3688 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
3689 instead of HARD_REGNO_NREGS.
3690 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
3691 HARD_REGNO_NREGS in the comment.
3692 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
3693 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
3694 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
3695 Return an unsigned int.
3696 (TARGET_HARD_REGNO_NREGS): Redefine.
3697 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
3698 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
3699 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
3700 (arc_hard_regno_nregs): New function.
3701 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
3702 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
3703 (arm_hard_regno_nregs): New function.
3704 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
3705 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
3706 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
3707 (TARGET_HARD_REGNO_NREGS): Redefine.
3708 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
3709 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
3710 (HARD_REGNO_NREGS): Delete.
3711 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
3712 (cr16_hard_regno_nregs): New function.
3713 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
3714 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
3715 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
3716 (cris_hard_regno_nregs): New function.
3717 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
3718 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
3719 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
3720 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
3721 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
3722 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
3723 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
3724 (frv_hard_regno_nregs): Make static. Take and return an
3726 (frv_class_max_nregs): Remove outdated copy of documentation.
3727 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
3728 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
3729 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
3730 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
3731 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
3732 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
3733 (TARGET_HARD_REGNO_NREGS): Redefine.
3734 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
3735 (CLASS_MAX_NREGS): Update comment.
3736 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
3737 (ia64_hard_regno_nregs): New function.
3738 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
3739 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
3740 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
3741 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
3742 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
3744 (m32c_hard_regno_nregs): Likewise. Make static.
3745 (TARGET_HARD_REGNO_NREGS): Redefine.
3746 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
3747 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
3748 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
3749 (m68k_hard_regno_nregs): New function.
3750 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
3751 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
3752 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
3753 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
3754 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
3755 Take and return an unsigned int.
3756 (TARGET_HARD_REGNO_NREGS): Redefine.
3757 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
3758 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
3759 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
3760 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
3761 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
3762 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
3763 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
3764 (msp430_hard_regno_nregs): Make static. Take and return an
3766 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
3767 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
3768 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
3769 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
3770 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
3771 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
3772 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
3773 (TARGET_HARD_REGNO_NREGS): Redefine.
3774 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
3775 (PA_HARD_REGNO_NREGS): ...this.
3776 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
3777 (PA_HARD_REGNO_NREGS): ...this.
3778 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
3779 (pa_hard_regno_nregs): New function.
3780 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
3781 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
3782 (pdp11_hard_regno_nregs): New function.
3783 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
3784 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
3785 (rs6000_hard_regno_nregs_hook): New function.
3786 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
3787 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
3788 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
3789 Take and return an unsigned int. Move earlier in file.
3790 (TARGET_HARD_REGNO_NREGS): Redefine.
3791 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
3792 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
3793 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
3794 (rl78_hard_regno_nregs): Make static. Take and return an
3796 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
3797 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
3798 (rs6000_hard_regno_nregs_hook): New function.
3799 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
3800 * config/rx/rx.c (rx_hard_regno_nregs): New function.
3801 (TARGET_HARD_REGNO_NREGS): Redefine.
3802 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
3803 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
3804 instead of HARD_REGNO_NREGS.
3805 (s390_hard_regno_nregs): New function.
3806 (s390_hard_regno_mode_ok): Add comment from s390.h.
3807 (TARGET_HARD_REGNO_NREGS): Redefine.
3808 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
3809 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
3810 (sh_hard_regno_nregs): New function.
3811 (sh_pass_in_reg_p): Use it.
3812 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
3813 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
3814 (sparc_hard_regno_nregs): New function.
3815 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
3816 * config/spu/spu.c (spu_hard_regno_nregs): New function.
3817 (spu_function_arg_advance): Use it, supplying a valid register number.
3818 (TARGET_HARD_REGNO_NREGS): Redefine.
3819 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
3820 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
3821 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
3822 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
3823 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
3824 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
3825 (CLASS_MAX_NREGS): Remove copy of old documentation.
3826 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
3827 (visium_hard_regno_nregs): New function.
3828 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
3829 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
3830 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
3831 xtensa_hard_regno_nregs): New function.
3832 * system.h (HARD_REGNO_NREGS): Poison.
3834 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3836 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
3837 hard_regno_nregs instead of HARD_REGNO_NREGS.
3838 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
3839 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
3840 (c6x_expand_epilogue): Likewise.
3841 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
3842 (frv_read_iacc_argument): Likewise.
3843 * config/sh/sh.c: Include regs.h.
3844 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
3845 (regs_used): Likewise.
3846 (output_stack_adjust): Likewise.
3847 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
3848 * expmed.c: Include regs.h.
3849 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
3850 * ree.c: Include regs.h.
3851 (combine_reaching_defs): Use hard_regno_nregs instead of
3853 (add_removable_extension): Likewise.
3855 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3857 * regs.h (hard_regno_nregs): Turn into a function.
3858 (end_hard_regno): Update accordingly.
3859 * caller-save.c (setup_save_areas): Likewise.
3860 (save_call_clobbered_regs): Likewise.
3861 (replace_reg_with_saved_mem): Likewise.
3862 (insert_restore): Likewise.
3863 (insert_save): Likewise.
3864 * combine.c (can_change_dest_mode): Likewise.
3865 (move_deaths): Likewise.
3866 (distribute_notes): Likewise.
3867 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
3868 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
3869 (rs6000_split_multireg_move): Likewise.
3870 (rs6000_register_move_cost): Likewise.
3871 (rs6000_memory_move_cost): Likewise.
3872 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
3873 (rs6000_split_multireg_move): Likewise.
3874 (rs6000_register_move_cost): Likewise.
3875 (rs6000_memory_move_cost): Likewise.
3876 * cselib.c (cselib_reset_table): Likewise.
3877 (cselib_lookup_1): Likewise.
3878 * emit-rtl.c (set_mode_and_regno): Likewise.
3879 * function.c (aggregate_value_p): Likewise.
3880 * ira-color.c (setup_profitable_hard_regs): Likewise.
3881 (check_hard_reg_p): Likewise.
3882 (calculate_saved_nregs): Likewise.
3883 (assign_hard_reg): Likewise.
3884 (improve_allocation): Likewise.
3885 (calculate_spill_cost): Likewise.
3886 * ira-emit.c (modify_move_list): Likewise.
3887 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
3888 (ira_hard_reg_in_set_p): Likewise.
3889 * ira.c (setup_reg_mode_hard_regset): Likewise.
3890 (clarify_prohibited_class_mode_regs): Likewise.
3891 (check_allocation): Likewise.
3892 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3893 (lra_setup_reg_renumber): Likewise.
3894 (setup_try_hard_regno_pseudos): Likewise.
3895 (spill_for): Likewise.
3896 (assign_hard_regno): Likewise.
3897 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
3898 * lra-constraints.c (in_class_p): Likewise.
3899 (lra_constraint_offset): Likewise.
3900 (simplify_operand_subreg): Likewise.
3901 (lra_constraints): Likewise.
3902 (split_reg): Likewise.
3903 (split_if_necessary): Likewise.
3904 (invariant_p): Likewise.
3905 (inherit_in_ebb): Likewise.
3906 * lra-lives.c (process_bb_lives): Likewise.
3907 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
3908 (get_hard_regs): Likewise.
3909 (do_remat): Likewise.
3910 * lra-spills.c (assign_spill_hard_regs): Likewise.
3911 * mode-switching.c (create_pre_exit): Likewise.
3912 * postreload.c (reload_combine_recognize_pattern): Likewise.
3913 * recog.c (peep2_find_free_register): Likewise.
3914 * regcprop.c (kill_value_regno): Likewise.
3915 (set_value_regno): Likewise.
3916 (copy_value): Likewise.
3917 (maybe_mode_change): Likewise.
3918 (find_oldest_value_reg): Likewise.
3919 (copyprop_hardreg_forward_1): Likewise.
3920 * regrename.c (check_new_reg_p): Likewise.
3921 (regrename_do_replace): Likewise.
3922 * reload.c (push_reload): Likewise.
3923 (combine_reloads): Likewise.
3924 (find_dummy_reload): Likewise.
3925 (operands_match_p): Likewise.
3926 (find_reloads): Likewise.
3927 (find_equiv_reg): Likewise.
3928 (reload_adjust_reg_for_mode): Likewise.
3929 * reload1.c (count_pseudo): Likewise.
3930 (count_spilled_pseudo): Likewise.
3931 (find_reg): Likewise.
3932 (clear_reload_reg_in_use): Likewise.
3933 (free_for_value_p): Likewise.
3934 (allocate_reload_reg): Likewise.
3935 (choose_reload_regs): Likewise.
3936 (reload_adjust_reg_for_temp): Likewise.
3937 (emit_reload_insns): Likewise.
3938 (delete_output_reload): Likewise.
3939 * rtlanal.c (subreg_get_info): Likewise.
3940 * sched-deps.c (sched_analyze_reg): Likewise.
3941 * sel-sched.c (init_regs_for_mode): Likewise.
3942 (mark_unavailable_hard_regs): Likewise.
3943 (choose_best_reg_1): Likewise.
3944 (verify_target_availability): Likewise.
3945 * valtrack.c (dead_debug_insert_temp): Likewise.
3946 * var-tracking.c (track_loc_p): Likewise.
3947 (emit_note_insn_var_location): Likewise.
3948 * varasm.c (make_decl_rtl): Likewise.
3949 * reginfo.c (choose_hard_reg_mode): Likewise.
3950 (init_reg_modes_target): Refer directly to
3951 this_target_regs->x_hard_regno_nregs.
3953 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3955 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
3956 instead of hard_regno_nregs.
3958 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3960 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
3961 end_hard_regno instead of hard_regno_nregs.
3962 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
3963 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
3964 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
3965 * ira-color.c (improve_allocation): Likewise.
3966 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3967 * lra-lives.c (mark_regno_live): Likewise.
3968 (mark_regno_dead): Likewise.
3969 * lra-remat.c (operand_to_remat): Likewise.
3970 * lra.c (collect_non_operand_hard_regs): Likewise.
3971 * postreload.c (reload_combine_note_store): Likewise.
3972 (move2add_valid_value_p): Likewise.
3973 * reload.c (regno_clobbered_p): Likewise.
3975 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3977 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
3979 * config/v850/v850.c (v850_reorg): Likewise.
3980 * reload.c (refers_to_regno_for_reload_p): Likewise.
3981 (find_equiv_reg): Likewise.
3982 * reload1.c (reload_reg_reaches_end_p): Likewise.
3984 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3986 * caller-save.c (add_used_regs): Use REG_NREGS instead of
3988 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
3989 * config/arm/arm.c (output_move_neon): Likewise.
3990 (arm_attr_length_move_neon): Likewise.
3991 (neon_split_vcombine): Likewise.
3992 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
3993 (c6x_mark_reg_written): Likewise.
3994 (c6x_dwarf_register_span): Likewise.
3995 * config/i386/i386.c (ix86_save_reg): Likewise.
3996 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
3997 (rws_access_reg): Likewise.
3998 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
3999 * mode-switching.c (create_pre_exit): Likewise.
4000 * ree.c (combine_reaching_defs): Likewise.
4001 (add_removable_extension): Likewise.
4002 * regcprop.c (find_oldest_value_reg): Likewise.
4003 (copyprop_hardreg_forward_1): Likewise.
4004 * reload.c (reload_inner_reg_of_subreg): Likewise.
4005 (push_reload): Likewise.
4006 (combine_reloads): Likewise.
4007 (find_dummy_reload): Likewise.
4008 (reload_adjust_reg_for_mode): Likewise.
4009 * reload1.c (find_reload_regs): Likewise.
4010 (forget_old_reloads_1): Likewise.
4011 (reload_reg_free_for_value_p): Likewise.
4012 (reload_adjust_reg_for_temp): Likewise.
4013 (emit_reload_insns): Likewise.
4014 (delete_output_reload): Likewise.
4015 * sel-sched.c (choose_best_reg_1): Likewise.
4016 (choose_best_pseudo_reg): Likewise.
4018 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4019 Alan Hayward <alan.hayward@arm.com>
4020 David Sherwood <david.sherwood@arm.com>
4022 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
4023 * target.def (slow_unaligned_access): New hook.
4024 * targhooks.h (default_slow_unaligned_access): Declare.
4025 * targhooks.c (default_slow_unaligned_access): New function.
4026 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
4027 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
4028 * doc/tm.texi: Regenerate.
4029 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
4030 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
4031 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
4033 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
4034 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
4036 (rs6000_slow_unaligned_access): New function.
4037 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
4038 (expand_block_compare): Likewise.
4039 (expand_strn_compare): Likewise.
4040 (rs6000_rtx_costs): Likewise.
4041 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
4042 (riscv_slow_unaligned_access): Likewise.
4043 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
4044 (riscv_slow_unaligned_access_p): ...this and make static.
4045 (riscv_option_override): Update accordingly.
4046 (riscv_slow_unaligned_access): New function.
4047 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
4048 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
4049 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
4050 (rs6000_slow_unaligned_access): New function.
4051 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
4052 (rs6000_rtx_costs): Likewise.
4053 * config/rs6000/rs6000-string.c (expand_block_compare)
4054 (expand_strn_compare): Use targetm.slow_unaligned_access instead
4055 of SLOW_UNALIGNED_ACCESS.
4056 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
4057 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
4058 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
4059 of SLOW_UNALIGNED_ACCESS.
4060 * expmed.c (simple_mem_bitfield_p): Likewise.
4061 * expr.c (alignment_for_piecewise_move): Likewise.
4062 (emit_group_load_1): Likewise.
4063 (emit_group_store): Likewise.
4064 (copy_blkmode_from_reg): Likewise.
4065 (emit_push_insn): Likewise.
4066 (expand_assignment): Likewise.
4067 (store_field): Likewise.
4068 (expand_expr_real_1): Likewise.
4069 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
4070 * lra-constraints.c (simplify_operand_subreg): Likewise.
4071 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
4072 * gimple-ssa-store-merging.c: Likewise in block comment at start
4074 * tree-ssa-strlen.c: Include target.h.
4075 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
4076 of SLOW_UNALIGNED_ACCESS.
4077 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
4079 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4081 PR rtl-optimization/82185
4082 * expmed.c (emit_store_flag_int): Only test tem if it has been
4085 2017-09-12 Richard Biener <rguenther@suse.de>
4088 * match.pd ((FTYPE) N CMP CST): Fix typo.
4090 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
4092 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
4094 (mips_near_type_p): Add 'short_call' attribute as a synonym
4096 * doc/extend.texi (short_call): Document new function attribute.
4098 2017-09-12 Jakub Jelinek <jakub@redhat.com>
4101 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
4102 assertion check that in the condition.
4103 (get_atomic_generic_size): Likewise. Before testing if parameter
4104 has pointer type, if it has array type, call for C++
4105 default_conversion to perform array-to-pointer conversion.
4107 2017-09-12 Richard Biener <rguenther@suse.de>
4109 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
4110 for operations we cannot scalarize.
4112 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
4114 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
4115 vectors heap vectors. Clean up comments.
4116 Make visited_bbs a reference.
4117 (profitable_jump_thread_path): Make GC
4118 vectors heap vectors. Clean up comments.
4120 (convert_and_register_jump_thread_path): Make GC vectors heap
4122 (check_subpath_and_update_thread_path): Same. Clean up comments.
4123 Make visited_bbs a reference.
4124 (handle_phi): Abstract common code to to
4125 register_jump_thread_path_if_profitable.
4126 Rename VAR_BB to DEF_BB.
4128 Make GC vectors heap vectors.
4129 Make visited_bbs a reference.
4130 (handle_assignment): Same.
4131 (register_jump_thread_path_if_profitable): New.
4132 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
4134 Make GC vectors heap vectors. Clean up comments.
4135 Make visited_bbs a reference.
4136 (find_jump_threads_backwards): Make visited_bbs live in the stack.
4137 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
4140 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
4143 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
4144 words of E_DImode object are reachable by xtensa_uimm8x4 access.
4146 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
4148 Revert r251800 and r251799.
4150 2017-09-11 Martin Jambor <mjambor@suse.cz>
4153 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
4154 arguments in advance.
4155 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
4156 use it to find predecessor edges.
4157 (naive_outof_ssa): Collect vector of predecessors.
4159 2017-09-08 Jason Merrill <jason@redhat.com>
4161 PR c++/70029 - ICE with ref-qualifier and -flto
4162 * langhooks.h (struct lang_hooks_for_types): Add
4163 copy_lang_qualifiers.
4164 * attribs.c (build_type_attribute_qual_variant): Use it.
4165 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
4167 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
4168 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
4170 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
4173 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
4174 (*mulsi3_sp64): New instruction.
4175 (mulsi3): New expander.
4177 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
4179 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
4181 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4183 * sancov.c: Include memmodel.h.
4185 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
4188 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
4191 2017-09-07 Carl Love <cel@us.ibm.com>
4193 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
4194 the sldi instruction.
4196 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
4198 * sancov.c: Include tm_p.h.
4200 2017-09-07 Jakub Jelinek <jakub@redhat.com>
4203 * output.h (switch_to_other_text_partition): New declaration.
4204 * varasm.c (switch_to_other_text_partition): New function.
4205 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
4206 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
4207 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
4208 to the other text partition before emitting LCL label and switch back
4209 after emitting the word after it.
4211 2017-09-07 Richard Biener <rguenther@suse.de>
4213 * passes.def (pass_split_crit_edges): Remove instance before PRE.
4214 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
4215 critical edges here, after loop init.
4216 (pass_data_pre): Remove PROP_no_crit_edges flags.
4217 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
4218 for valueization of call args to avoid leaking VN_TOP.
4219 (visit_use): Assert we do not visit default defs.
4220 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
4221 Use error_mark_node to more easily detect leaking VN_TOP.
4222 All default-defs are varying, not VN_TOP. Mark them visited.
4223 (run_scc_vn): Make code match comment.
4225 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
4227 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
4228 OPTION_MASK_FLOAT128_KEYWORD.
4229 (POWERPC_MASKS): Likewise.
4230 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
4231 support for the -mfloat128-type option, and make -mfloat128
4232 default on PowerPC Linux systems. Define or undefine
4233 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
4234 Define __float128 to be __ieee128 if IEEE 128-bit support is
4235 enabled, or undefine it.
4236 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
4237 Delete defining __FLOAT128_TYPE__.
4238 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
4239 -mfloat128-type option and make -mfloat128 default on PowerPC
4241 (TARGET_FLOAT128_TYPE): Likewise.
4242 (-mfloat128-type): Likewise.
4243 * config/rs6000/rs6000.c (rs6000_option_override_internal):
4244 Delete the -mfloat128-type option and make -mfloat128 default on
4245 PowerPC Linux systems. Always use __ieee128 to be the keyword for
4246 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
4247 128-bit floating point is enabled. Change tests from using
4248 -mfloat128-type to -mfloat128.
4249 (rs6000_mangle_type): Use the correct mangling for the __float128
4250 type even if normal long double is restricted to 64-bits.
4251 (floatn_mode): Enable the _Float128 type by default on VSX Linux
4253 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
4254 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
4255 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
4257 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
4258 documentation for -mfloat128.
4260 2017-09-06 Olivier Hainque <hainque@adacore.com>
4262 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
4264 2017-09-06 Wish Wu <wishwu007@gmail.com>
4265 Jakub Jelinek <jakub@redhat.com>
4267 * asan.c (initialize_sanitizer_builtins): Add
4268 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
4269 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
4270 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
4271 BT_FN_VOID_UINT64_PTR variables.
4272 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
4273 (BT_FN_VOID_UINT16_UINT16): Likewise.
4274 (BT_FN_VOID_UINT32_UINT32): Likewise.
4275 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
4276 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
4277 (BT_FN_VOID_UINT64_PTR): Likewise.
4278 * common.opt (flag_sanitize_coverage): New variable.
4279 (fsanitize-coverage=trace-pc): Remove.
4280 (fsanitize-coverage=): Add.
4281 * flag-types.h (enum sanitize_coverage_code): New enum.
4282 * fold-const.c (fold_range_test): Disable non-short-circuit
4283 optimization if flag_sanitize_coverage.
4284 (fold_truth_andor): Likewise.
4285 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
4286 * opts.c (COVERAGE_SANITIZER_OPT): Define.
4287 (coverage_sanitizer_opts): New array.
4288 (get_closest_sanitizer_option): Add OPTS argument, handle also
4289 OPT_fsanitize_coverage_.
4290 (parse_sanitizer_options): Adjusted to also handle
4291 OPT_fsanitize_coverage_.
4292 (common_handle_option): Add OPT_fsanitize_coverage_.
4293 * sancov.c (instrument_comparison, instrument_switch): New function.
4294 (sancov_pass): Add trace-cmp support.
4295 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
4296 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
4297 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
4298 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
4299 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
4300 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
4301 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
4302 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
4303 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
4304 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
4306 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
4308 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
4309 error. Only quit immediately if parsing is complete.
4310 (BEGIN): Initialize fatal_err and parse_done.
4311 (begin fpu, end fpu): Check number of arguments.
4312 (begin arch, end arch): Likewise.
4313 (begin cpu, end cpu): Likewise.
4314 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
4315 (optalias): Likewise.
4317 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
4319 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
4320 * config/arm/arm-isa.h: Delete. Move definitions to ...
4321 * arm-cpus.in: ... here. Use new feature and fgroup values.
4322 * config/arm/arm.c (arm_option_override): Use lower case for feature
4324 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
4325 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
4326 * config/arm/parsecpu.awk (END): Add new command 'isa'.
4328 (print_isa_bits_for): New function.
4329 (gen_isa): New function.
4330 (gen_comm_data): Use print_isa_bits_for.
4331 (define feature): New keyword.
4332 (define fgroup): New keyword.
4333 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
4334 (arm-isa.h): Add rule to generate file.
4335 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
4336 case for feature bit names.
4338 2017-09-06 Richard Biener <rguenther@suse.de>
4340 * tree-ssa-pre.c (NECESSARY): Remove.
4341 (create_expression_by_pieces): Do not touch pass-local flags.
4342 (insert_into_preds_of_block): Likewise.
4343 (do_pre_regular_insertion): Likewise.
4344 (eliminate_insert): Likewise.
4345 (eliminate_dom_walker::before_dom_children): Likewise.
4346 (fini_eliminate): Do not look at inserted_exprs.
4347 (mark_operand_necessary): Remove.
4348 (remove_dead_inserted_code): Replace with simple work-list
4349 algorithm based on inserted_exprs and SSA uses.
4350 (pass_pre::execute): Re-order fini_eliminate and
4351 remove_dead_inserted_code.
4353 2017-09-06 Olivier Hainque <hainque@adacore.com>
4355 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
4356 for VxWorks 7. Adjust surrounding comments.
4358 2017-09-06 Richard Biener <rguenther@suse.de>
4360 * gimple-ssa-strength-reduction.c
4361 (find_candidates_dom_walker::before_dom_children): Also allow
4364 2017-09-06 Richard Biener <rguenther@suse.de>
4366 PR tree-optimization/82108
4367 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
4368 for gap in the non-permutation SLP case.
4370 2017-09-06 Martin Jambor <mjambor@suse.cz>
4372 PR tree-optimization/82078
4373 * tree-sra.c (sort_and_splice_var_accesses): Move call to
4374 add_access_to_work_queue...
4375 (build_accesses_from_assign): ...here.
4376 (propagate_all_subaccesses): Make sure racc is the group
4377 representative, if there is one.
4379 2017-09-06 Jakub Jelinek <jakub@redhat.com>
4382 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
4385 2017-09-06 Richard Biener <rguenther@suse.de>
4387 * gimple-ssa-strength-reduction.c
4388 (find_candidates_dom_walker::before_doom_children): Use a
4389 type and not a mode check.
4391 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
4394 * config/arm/predicates.md (arm_general_adddi_operand): Create new
4396 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
4398 2017-09-05 Jeff Law <law@redhat.com>
4400 PR tree-optimization/64910
4401 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
4402 cases where we have 3 or more operands.
4404 2017-09-05 Jakub Jelinek <jakub@redhat.com>
4407 * omp-low.c (lower_omp_for): Recompute tree invariant if
4408 gimple_omp_for_initial/final is ADDR_EXPR.
4411 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
4412 into gimple val before gimplification fo the COND_EXPR.
4414 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
4416 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
4417 REGION_COPY argument.
4418 (thread_through_all_blocks): Remove unused argument to
4419 duplicate_thread_path.
4421 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4422 Alan Hayward <alan.hayward@arm.com>
4423 David Sherwood <david.sherwood@arm.com>
4425 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
4426 Take a scalar_mode rather than a machine_mode.
4427 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
4428 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
4429 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
4430 (aarch64_gen_adjusted_ldpstp): Likewise.
4431 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
4433 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4434 Alan Hayward <alan.hayward@arm.com>
4435 David Sherwood <david.sherwood@arm.com>
4437 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
4438 Take a scalar_int_mode instead of a machine_mode.
4439 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
4440 (aarch64_output_scalar_simd_mov_immediate): Likewise.
4441 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
4442 (aarch64_simd_attr_length_rglist): Delete.
4443 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
4444 a scalar_int_mode instead of a machine_mode.
4445 (aarch64_add_offset): Likewise.
4446 (aarch64_internal_mov_immediate): Likewise
4447 (aarch64_add_constant_internal): Likewise.
4448 (aarch64_add_constant): Likewise.
4449 (aarch64_movw_imm): Likewise.
4450 (aarch64_rtx_arith_op_extract_p): Likewise.
4451 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
4452 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
4453 Remove assert that the mode isn't a vector.
4454 (aarch64_output_scalar_simd_mov_immediate): Likewise.
4455 (aarch64_expand_mov_immediate): Update calls after above changes.
4456 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
4457 (aarch64_and_bitmask_imm): Check for scalar integer modes.
4458 (aarch64_move_imm): Likewise.
4459 (aarch64_can_const_movi_rtx_p): Likewise.
4460 (aarch64_strip_extend): Likewise.
4461 (aarch64_extr_rtx_p): Likewise.
4462 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
4463 a CONST_INT when the mode parameter is VOIDmode.
4464 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
4466 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4468 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
4469 * stor-layout.c (bitwise_mode_for_mode): Likewise.
4470 (bitwise_type_for_mode): Update accordingly.
4472 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4474 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
4475 * stor-layout.c (mode_for_size_tree): Likewise.
4476 (mode_for_array): Update accordingly.
4477 (layout_decl): Likewise.
4478 (compute_record_mode): Likewise. Only set the mode once.
4480 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4482 * target.def (get_mask_mode): Change return type to opt_mode.
4484 * doc/tm.texi: Regenerate.
4485 * targhooks.h (default_get_mask_mode): Return an opt_mode.
4486 * targhooks.c (default_get_mask_mode): Likewise.
4487 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
4488 * optabs-query.c (can_vec_mask_load_store_p): Update use of
4489 targetm.get_mask_mode.
4490 * tree.c (build_truth_vector_type): Likewise.
4492 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4494 * machmode.h (mode_for_vector): Return an opt_mode.
4495 * stor-layout.c (mode_for_vector): Likewise.
4496 (mode_for_int_vector): Update accordingly.
4497 (layout_type): Likewise.
4498 * config/i386/i386.c (emit_memmov): Likewise.
4499 (ix86_expand_set_or_movmem): Likewise.
4500 (ix86_expand_vector_init): Likewise.
4501 (ix86_get_mask_mode): Likewise.
4502 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
4504 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
4505 * expmed.c (extract_bit_field_1): Likewise.
4506 * expr.c (expand_expr_real_2): Likewise.
4507 * optabs-query.c (can_vec_perm_p): Likewise.
4508 (can_vec_mask_load_store_p): Likewise.
4509 * optabs.c (expand_vec_perm): Likewise.
4510 * targhooks.c (default_get_mask_mode): Likewise.
4511 * tree-vect-stmts.c (vectorizable_store): Likewise.
4512 (vectorizable_load): Likewise.
4513 (get_vectype_for_scalar_type_and_size): Likewise.
4515 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4517 * machmode.h (mode_for_int_vector): New function.
4518 * stor-layout.c (mode_for_int_vector): Likewise.
4519 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
4520 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
4521 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
4522 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
4523 (s390_expand_vcond): Likewise.
4525 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4527 * machmode.h (opt_machine_mode): New type.
4528 (opt_mode<T>): Allow construction from anything that can be
4530 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
4531 (mode_for_size): Return an opt_machine_mode.
4532 * stor-layout.c (mode_for_size): Likewise.
4533 (mode_for_size_tree): Update call accordingly.
4534 (bitwise_mode_for_mode): Likewise.
4535 (make_fract_type): Likewise.
4536 (make_accum_type): Likewise.
4537 * caller-save.c (replace_reg_with_saved_mem): Update call
4539 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4540 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4541 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4542 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4543 * expmed.c (extract_bit_field_1): Likewise.
4544 * reload.c (get_secondary_mem): Likewise.
4545 * varasm.c (assemble_integer): Likewise.
4546 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
4549 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4551 * machmode.h (decimal_float_mode_for_size): New function.
4552 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
4553 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
4554 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
4555 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
4556 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
4557 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
4559 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4561 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
4562 (get_builtin_sync_mode): Likewise.
4563 (expand_ifn_atomic_compare_exchange): Likewise.
4564 (expand_builtin_atomic_clear): Likewise.
4565 (expand_builtin_atomic_test_and_set): Likewise.
4566 (fold_builtin_atomic_always_lock_free): Likewise.
4567 * calls.c (compute_argument_addresses): Likewise.
4568 (emit_library_call_value_1): Likewise.
4569 (store_one_arg): Likewise.
4570 * combine.c (combine_instructions): Likewise.
4571 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
4572 * config/arm/arm.c (arm_function_value): Likewise.
4573 (aapcs_allocate_return_reg): Likewise.
4574 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
4575 * config/i386/i386.c (construct_container): Likewise.
4576 (ix86_gimplify_va_arg): Likewise.
4577 (ix86_expand_sse_cmp): Likewise.
4578 (emit_memmov): Likewise.
4579 (emit_memset): Likewise.
4580 (expand_small_movmem_or_setmem): Likewise.
4581 (ix86_expand_pextr): Likewise.
4582 (ix86_expand_pinsr): Likewise.
4583 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
4584 * config/microblaze/microblaze.c (microblaze_block_move_straight):
4586 * config/mips/mips.c (mips_function_value_1) Likewise.
4587 (mips_block_move_straight): Likewise.
4588 (mips_expand_ins_as_unaligned_store): Likewise.
4589 * config/powerpcspe/powerpcspe.c
4590 (rs6000_darwin64_record_arg_advance_flush): Likewise.
4591 (rs6000_darwin64_record_arg_flush): Likewise.
4592 * config/rs6000/rs6000.c
4593 (rs6000_darwin64_record_arg_advance_flush): Likewise.
4594 (rs6000_darwin64_record_arg_flush): Likewise.
4595 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
4596 (sparc_function_value_1): Likewise.
4597 * config/spu/spu.c (adjust_operand): Likewise.
4598 (spu_emit_branch_or_set): Likewise.
4599 (arith_immediate_p): Likewise.
4600 * emit-rtl.c (gen_lowpart_common): Likewise.
4601 * expr.c (expand_expr_real_1): Likewise.
4602 * function.c (assign_parm_setup_block): Likewise.
4603 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
4604 * reload1.c (alter_reg): Likewise.
4605 * stor-layout.c (mode_for_vector): Likewise.
4606 (layout_type): Likewise.
4608 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4610 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
4611 (spu_convert_move): Likewise.
4612 * lower-subreg.c (resolve_simple_move): Likewise.
4614 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4617 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
4618 define_insn to a define_expand.
4619 (altivec_vsum2sws_direct): New define_insn.
4620 (altivec_vsumsws): Convert from a define_insn to a define_expand.
4622 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
4624 * config/arm/arm.c (arm_option_params_internal): Improve setting of
4627 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
4632 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
4633 realign stack if stack alignment needed is less than incoming
4636 2017-09-05 Marek Polacek <polacek@redhat.com>
4639 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
4642 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
4644 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
4646 2017-09-05 Richard Biener <rguenther@suse.de>
4648 PR tree-optimization/82084
4649 * fold-const.c (can_native_encode_string_p): Handle wide characters.
4651 2017-09-05 Richard Biener <rguenther@suse.de>
4653 PR tree-optimization/82102
4654 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
4656 2017-09-05 Martin Liska <mliska@suse.cz>
4658 PR tree-optimization/82032
4659 * tree-cfg.c (generate_range_test): New function.
4660 * tree-cfg.h (generate_range_test): Declared here.
4661 * tree-cfgcleanup.c (convert_single_case_switch): New function.
4662 (cleanup_control_expr_graph): Use it.
4663 * tree-switch-conversion.c (try_switch_expansion): Remove
4665 (emit_case_nodes): Use generate_range_test.
4667 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
4670 * config/i386/i386.md (*<btsc><mode>_mask): Add
4671 TARGET_USE_BT to insn constraint.
4672 (*btr<mode>_mask): Ditto.
4674 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
4676 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
4677 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
4679 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
4682 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
4683 TARGET_NEON and TARGET_IWMMXT.
4684 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
4685 TARGET_NEON and TARGET_IWMMXT.
4686 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
4688 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
4690 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
4691 (ix86_rewrite_tls_address): Ditto.
4692 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
4693 (ix86_rewrite_tls_address_1): Ditto.
4694 (ix86_rewrite_tls_address): Ditto.
4695 * config/i386/predicates.md (tls_address_pattern): New predicate.
4696 * config/i386/i386.md (TLS address splitter): New splitter.
4698 2017-09-04 Richard Biener <rguenther@suse.de>
4700 PR tree-optimization/82084
4701 * fold-const.h (can_native_encode_string_p): Declare.
4702 * fold-const.c (can_native_encode_string_p): Factor out from ...
4703 (native_encode_string): ... here.
4704 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
4705 vectorizing stores from constants we later cannot handle.
4707 2017-09-04 Marek Polacek <polacek@redhat.com>
4710 * doc/invoke.texi: Update -Wtautological-compare documentation.
4712 2017-09-04 Jeff Law <law@redhat.com>
4714 PR tree-optimization/64910
4715 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
4716 swap the first and last operand if the last is a constant.
4718 2017-09-04 Marek Polacek <polacek@redhat.com>
4721 * convert.c (do_narrow): When sanitizing signed integer overflows,
4722 bail out for signed types.
4723 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
4725 2017-09-04 Richard Biener <rguenther@suse.de>
4727 PR tree-optimization/82060
4728 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
4729 Move devirtualization after stmt folding and before EH/AB/noreturn
4730 cleanup to get the stmt refs canonicalized. Use a bool instead
4731 of gimple_modified_p since that doesn't work for NOPs. Schedule
4732 NOPs generated by folding for removal.
4734 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4735 Alan Hayward <alan.hayward@arm.com>
4736 David Sherwood <david.sherwood@arm.com>
4738 * coretypes.h (pad_direction): New enum.
4739 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
4740 (FUNCTION_ARG_PADDING): Likewise.
4741 * target.def (function_arg_padding): New hook.
4742 * targhooks.h (default_function_arg_padding): Declare.
4743 * targhooks.c (default_function_arg_padding): New function.
4744 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
4745 (TARGET_FUNCTION_ARG_PADDING): ...this.
4746 * doc/tm.texi: Regenerate.
4747 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
4748 instead of direction.
4749 (compute_argument_addresses): Likewise.
4750 (load_register_parameters): Likewise.
4751 (emit_library_call_value_1): Likewise.
4752 (store_one_arg): Use targetm.calls.function_arg_padding instead
4753 of FUNCTION_ARG_PADDING.
4754 (must_pass_in_stack_var_size_or_pad): Likewise.
4755 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
4756 (emit_group_store): Likewise.
4757 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
4758 instead of FUNCTION_ARG_PADDING.
4759 (emit_push_insn): Likewise, and propagate enum change throughout
4761 * function.h (direction): Delete.
4762 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
4764 * function.c (assign_parm_find_stack_rtl): Likewise.
4765 (assign_parm_setup_block_p): Likewise.
4766 (assign_parm_setup_block): Likewise.
4767 (gimplify_parameters): Likewise.
4768 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
4769 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
4771 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
4772 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
4773 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
4774 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
4775 (aarch64_function_arg_padding): ...this new function.
4776 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
4777 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4778 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
4779 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
4780 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
4781 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4782 (arm_pad_arg_upward): Replace with...
4783 (arm_function_arg_padding): ...this new function.
4784 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
4786 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
4787 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
4788 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4789 (ia64_hpux_function_arg_padding): Replace with...
4790 (ia64_function_arg_padding): ...this new function. Use pad_direction
4791 instead of direction. Check for TARGET_HPUX.
4792 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
4793 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4794 (iq2000_function_arg_padding): New function.
4795 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
4796 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
4797 (mips_function_arg_padding): ...this new function.
4798 (mips_pad_reg_upward): Update accordingly.
4799 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4800 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
4801 targetm.calls.function_arg_padding.
4802 (FUNCTION_ARG_PADDING): Delete.
4803 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
4804 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
4805 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
4806 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
4807 (nios2_block_reg_padding): Return pad_direction instead of direction.
4808 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
4809 instead of direction.
4810 (nios2_function_arg_padding): Likewise. Make static.
4811 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4812 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
4813 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
4814 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
4815 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4816 (pa_function_arg_padding): Make static. Return pad_direction instead
4818 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
4819 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
4820 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
4821 instead of direction. Use targetm.calls.function_arg_padding.
4822 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
4823 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
4824 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
4825 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
4826 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
4828 (function_arg_padding): Rename to...
4829 (rs6000_function_arg_padding): ...this. Make static. Return
4830 pad_direction instead of direction.
4831 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
4832 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
4833 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
4834 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
4835 instead of direction. Use targetm.calls.function_arg_padding.
4836 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
4837 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
4838 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
4839 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
4840 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4841 (function_arg_padding): Rename to...
4842 (rs6000_function_arg_padding): ...this. Make static. Return
4843 pad_direction instead of direction.
4844 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
4845 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
4846 * config/s390/s390.c (s390_function_arg_padding): New function.
4847 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4848 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
4849 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
4850 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4851 (function_arg_padding): Rename to...
4852 (sparc_function_arg_padding): ...this. Make static. Return
4853 pad_direction instead of direction.
4854 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
4855 * config/spu/spu.c (spu_function_arg_padding): New function.
4856 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4857 * system.h (FUNCTION_ARG_PADDING): Poison.
4859 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4860 Alan Hayward <alan.hayward@arm.com>
4861 David Sherwood <david.sherwood@arm.com>
4863 * target.def (modes_tieable_p): New hook.
4864 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
4865 (TARGET_MODES_TIEABLE_P): ...this.
4866 * doc/tm.texi.in: Regenerate.
4867 * hooks.h (hook_bool_mode_mode_true): Declare.
4868 * hooks.c (hook_bool_mode_mode_true): New function.
4869 * combine.c (subst): Use targetm.modes_tieable_p instead of
4871 * dse.c (find_shift_sequence): Likewise.
4872 * expmed.c (extract_low_bits): Likewise.
4873 * lower-subreg.c: Include target.h.
4874 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
4876 * rtlanal.c (rtx_cost): Likewise.
4877 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
4878 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
4879 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
4880 (TARGET_MODES_TIEABLE_P): Redefine.
4881 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
4882 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
4883 (TARGET_MODES_TIEABLE_P): Redefine.
4884 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
4885 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
4886 (arc_modes_tieable_p): New function.
4887 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
4888 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
4889 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
4890 (arm_modes_tieable_p): Make static.
4891 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
4892 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
4893 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
4894 (TARGET_MODES_TIEABLE_P): Redefine.
4895 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
4896 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
4897 (TARGET_MODES_TIEABLE_P): Redefine.
4898 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
4899 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
4900 (cr16_modes_tieable_p): New function.
4901 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
4902 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
4903 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
4904 (TRULY_NOOP_TRUNCATION): Update comment.
4905 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
4906 (TRULY_NOOP_TRUNCATION): Update comment.
4907 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
4908 (frv_modes_tieable_p): New function.
4909 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
4910 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
4911 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
4912 (TARGET_MODES_TIEABLE_P): Redefine.
4913 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
4914 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
4915 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
4916 (TARGET_MODES_TIEABLE_P): Redefine.
4917 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
4918 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
4919 (ia64_modes_tieable_p): New function.
4920 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
4921 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
4922 (iq2000_modes_tieable_p): New function.
4923 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
4924 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
4925 (lm32_modes_tieable_p): New function.
4926 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
4927 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
4928 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
4929 (TARGET_MODES_TIEABLE_P): Redefine.
4930 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
4931 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
4932 (m32r_modes_tieable_p): New function.
4933 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
4934 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
4935 (m68k_modes_tieable_p): New function.
4936 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
4937 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
4938 (mcore_modes_tieable_p): New function.
4939 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
4940 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
4942 (TARGET_MODES_TIEABLE_P): Redefine.
4943 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
4944 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
4945 * config/mips/mips.c (mips_modes_tieable_p): Make static.
4946 (TARGET_MODES_TIEABLE_P): Redefine.
4947 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
4948 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
4949 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
4950 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
4951 (mn10300_modes_tieable_p): ...this and make static.
4952 (TARGET_MODES_TIEABLE_P): Redefine.
4953 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
4954 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
4955 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
4956 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
4957 (msp430_modes_tieable_p): Make static.
4958 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
4959 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
4960 (TARGET_MODES_TIEABLE_P): Redefine.
4961 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
4962 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
4963 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
4964 (TARGET_MODES_TIEABLE_P): Redefine.
4965 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
4966 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
4967 * config/pa/pa.c (pa_modes_tieable_p): Make static.
4968 (TARGET_MODES_TIEABLE_P): Redefine.
4969 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
4970 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
4971 (pdp11_modes_tieable_p): New function.
4972 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
4973 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
4974 (rs6000_modes_tieable_p): New function.
4975 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
4976 * config/powerpcspe/powerpcspe.md: Update comment.
4977 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
4978 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
4979 (TARGET_MODES_TIEABLE_P): Redefine.
4980 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
4981 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
4982 (rl78_modes_tieable_p): New function.
4983 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
4984 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
4985 (rs6000_modes_tieable_p): New function.
4986 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
4987 * config/rs6000/rs6000.md: Update comment.
4988 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
4989 * config/rx/rx.c (rx_modes_tieable_p): New function.
4990 (TARGET_MODES_TIEABLE_P): Redefine.
4991 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
4992 * config/s390/s390.c (s390_modes_tieable_p): New function.
4993 (TARGET_MODES_TIEABLE_P): Redefine.
4994 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
4995 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
4996 (sh_modes_tieable_p): New function.
4997 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
4998 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
4999 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
5000 (sparc_modes_tieable_p): Make static.
5001 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
5002 * config/spu/spu.c (spu_modes_tieable_p): New function.
5003 (TARGET_MODES_TIEABLE_P): Redefine.
5004 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
5005 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
5006 (TARGET_MODES_TIEABLE_P): Redefine.
5007 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
5008 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
5009 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
5010 * config/v850/v850.c (v850_modes_tieable_p): New function.
5011 (TARGET_MODES_TIEABLE_P): Redefine.
5012 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
5013 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
5014 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
5015 (visium_modes_tieable_p): New function.
5016 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
5017 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
5018 (xtensa_modes_tieable_p): New function.
5019 * system.h (MODES_TIEABLE_P): Poison.
5021 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5022 Alan Hayward <alan.hayward@arm.com>
5023 David Sherwood <david.sherwood@arm.com>
5025 * target.def (hard_regno_mode_ok): New hook.
5026 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
5027 (TARGET_HARD_REGNO_MODE_OK): ...this.
5028 * doc/tm.texi.in: Regenerate.
5029 * hooks.h (hook_bool_uint_mode_true): Declare.
5030 * hooks.c (hook_bool_uint_mode_true): New function.
5031 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
5033 * genpreds.c (write_insn_preds_c): Add an include of target.h.
5034 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
5035 instead of HARD_REGNO_MODE_OK.
5036 * caller-save.c: Include target.h.
5037 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
5039 * combine.c (can_combine_p): Likewise.
5040 (combinable_i3pat): Likewise.
5041 (can_change_dest_mode): Likewise.
5042 * expr.c (init_expr_target): Likewise.
5043 (convert_move): Likewise.
5044 (convert_modes): Likewise.
5045 * ira.c (setup_prohibited_class_mode_regs): Likewise.
5046 (setup_prohibited_mode_move_regs): Likewise.
5047 * ira.h (target_ira): Likewise.
5048 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5049 * lra-constraints.c (process_alt_operands): Likewise.
5050 (split_reg): Likewise.
5051 * recog.c (peep2_find_free_register): Likewise.
5052 * ree.c (combine_reaching_defs): Likewise.
5053 * regcprop.c (maybe_mode_change): Likewise.
5054 * reginfo.c (init_reg_sets_1): Likewise.
5055 (choose_hard_reg_mode): Likewise.
5056 (simplifiable_subregs): Likewise.
5057 * regrename.c (check_new_reg_p): Likewise.
5058 * reload.c (find_valid_class): Likewise.
5059 (find_valid_class_1): Likewise.
5060 (reload_inner_reg_of_subreg): Likewise.
5061 (push_reload): Likewise.
5062 (combine_reloads): Likewise.
5063 (find_dummy_reload): Likewise.
5064 (find_reloads): Likewise.
5065 * reload1.c (find_reg): Likewise.
5066 (set_reload_reg): Likewise.
5067 (allocate_reload_reg): Likewise.
5068 (choose_reload_regs): Likewise.
5069 (reload_adjust_reg_for_temp): Likewise.
5070 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
5071 (simplify_subreg_regno): Likewise.
5072 * sel-sched.c (init_regs_for_mode): Likewise.
5073 * varasm.c (make_decl_rtl): Likewise.
5074 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
5075 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
5077 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
5078 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
5079 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5080 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
5081 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
5082 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5083 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
5084 (arc_mode_class): Delete.
5085 (HARD_REGNO_MODE_OK): Delete.
5086 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5087 (arc_hard_regno_mode_ok): Rename old array to...
5088 (arc_hard_regno_mode_ok_modes): ...this.
5089 (arc_conditional_register_usage): Update accordingly.
5090 (arc_mode_class): Make static.
5091 (arc_hard_regno_mode_ok): New function.
5092 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
5093 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
5094 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5095 (arm_hard_regno_mode_ok): Make static.
5096 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
5098 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
5099 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
5100 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
5102 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5103 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
5104 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
5105 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
5106 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
5107 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5108 * config/bfin/predicates.md (valid_reg_operand): Use
5109 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
5110 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
5111 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
5112 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5113 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
5114 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
5115 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5116 (cr16_hard_regno_mode_ok): Make static and return a bool.
5117 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
5118 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5119 (cris_hard_regno_mode_ok): New function.
5120 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
5121 (epiphany_mode_class): Delete.
5122 (HARD_REGNO_MODE_OK): Delete.
5123 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
5124 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5125 (hard_regno_mode_ok): Rename to...
5126 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
5127 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
5128 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
5130 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
5131 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
5132 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5133 (frv_hard_regno_mode_ok): Make static and return a bool.
5134 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
5136 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
5137 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
5138 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
5139 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
5141 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5142 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
5143 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
5144 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
5146 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5147 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
5148 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5149 (ia64_hard_regno_mode_ok): New function.
5150 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
5151 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5152 (iq2000_hard_regno_mode_ok): New function.
5153 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
5154 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5155 (lm32_hard_regno_mode_ok): New function.
5156 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
5157 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
5158 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
5159 instead of HARD_REGNO_MODE_OK.
5160 (m32c_hard_regno_ok): Rename to...
5161 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
5162 (m32c_cannot_change_mode_class): Update accordingly.
5163 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5164 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
5165 (m32r_mode_class): Delete.
5166 (HARD_REGNO_MODE_OK): Delete.
5167 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5168 (m32r_hard_regno_mode_ok): Rename to...
5169 (m32r_hard_regno_modes): ...this.
5170 (m32r_mode_class): Make static.
5171 (m32r_hard_regno_mode_ok): New function.
5172 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
5173 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
5174 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5175 (m68k_hard_regno_mode_ok): Make static.
5176 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
5177 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5178 (mcore_hard_regno_mode_ok): New function.
5179 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
5180 (HARD_REGNO_MODE_OK): Delete.
5181 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
5183 (microblaze_hard_regno_mode_ok_p): ...this and make static.
5184 (microblaze_hard_regno_mode_ok): New function.
5185 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5186 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
5187 (mips_hard_regno_mode_ok): Delete.
5188 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
5189 (mips_hard_regno_mode_ok_p): ...this and make static.
5190 (mips_hard_regno_mode_ok_p): Rename to...
5191 (mips_hard_regno_mode_ok_uncached): ...this.
5192 (mips_hard_regno_mode_ok): New function.
5193 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
5194 of HARD_REGNO_MODE_OK.
5195 (mips_option_override): Update after above name changes.
5196 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5197 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
5198 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
5199 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
5200 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
5201 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5202 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
5203 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
5204 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
5205 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5206 (msp430_hard_regno_mode_ok): Make static and return a bool.
5207 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
5208 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
5209 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
5211 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5212 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
5213 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
5214 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
5215 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
5216 (PA_HARD_REGNO_MODE_OK): ...this
5217 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
5218 (PA_HARD_REGNO_MODE_OK): ...this.
5219 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5220 (pa_hard_regno_mode_ok): New function.
5221 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
5222 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5223 (pdp11_hard_regno_mode_ok): New function.
5224 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
5225 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
5227 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
5229 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5230 (rs6000_hard_regno_mode_ok): Rename to...
5231 (rs6000_hard_regno_mode_ok_uncached): ...this.
5232 (rs6000_init_hard_regno_mode_ok): Update accordingly.
5233 (rs6000_hard_regno_mode_ok): New function.
5234 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
5235 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
5236 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
5237 (riscv_hard_regno_mode_ok): ...this and make static.
5238 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5239 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
5240 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
5241 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5242 (rl78_hard_regno_mode_ok): Make static and return bool.
5243 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
5244 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
5246 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
5247 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5248 (rs6000_hard_regno_mode_ok): Rename to...
5249 (rs6000_hard_regno_mode_ok_uncached): ...this.
5250 (rs6000_init_hard_regno_mode_ok): Update accordingly.
5251 (rs6000_hard_regno_mode_ok): New function.
5252 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
5253 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
5254 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5255 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
5256 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
5257 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
5258 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5259 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
5260 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
5261 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5262 (sh_hard_regno_mode_ok): Make static.
5263 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
5264 instead of HARD_REGNO_MODE_OK.
5265 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
5266 (sparc_mode_class): Delete.
5267 (HARD_REGNO_MODE_OK): Delete.
5268 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5269 (hard_regno_mode_classes): Make static.
5270 (sparc_mode_class): Likewise.
5271 (sparc_hard_regno_mode_ok): New function.
5272 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
5273 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
5274 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
5276 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5277 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
5278 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
5279 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
5280 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
5281 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5282 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
5283 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
5284 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5285 (visium_hard_regno_mode_ok): New function.
5286 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
5287 instead of HARD_REGNO_MODE_OK.
5288 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
5289 (HARD_REGNO_MODE_OK): Delete.
5290 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
5291 (xtensa_hard_regno_mode_ok_p): ...this and make static.
5292 (xtensa_option_override): Update accordingly.
5293 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5294 (xtensa_hard_regno_mode_ok): New function.
5295 * system.h (HARD_REGNO_MODE_OK): Poison.
5297 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5298 Alan Hayward <alan.hayward@arm.com>
5299 David Sherwood <david.sherwood@arm.com>
5301 * target.def (hard_regno_call_part_clobbered): New hook.
5302 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
5303 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
5304 * doc/tm.texi: Regenerate.
5305 * hooks.h (hook_bool_uint_mode_false): Declare.
5306 * hooks.c (hook_bool_uint_mode_false): New function.
5307 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5308 * cselib.c (cselib_process_insn): Use
5309 targetm.hard_regno_call_part_clobbered instead of
5310 HARD_REGNO_CALL_PART_CLOBBERED.
5311 * ira-conflicts.c (ira_build_conflicts): Likewise.
5312 * ira-costs.c (ira_tune_allocno_costs): Likewise.
5313 * lra-constraints.c (need_for_call_save_p): Likewise.
5314 * lra-lives.c: Include target.h.
5315 (check_pseudos_live_through_calls): Use
5316 targetm.hard_regno_call_part_clobbered instead of
5317 HARD_REGNO_CALL_PART_CLOBBERED.
5318 * regcprop.c: Include target.h.
5319 (copyprop_hardreg_forward_1): Use
5320 targetm.hard_regno_call_part_clobbered instead of
5321 HARD_REGNO_CALL_PART_CLOBBERED.
5322 * reginfo.c (choose_hard_reg_mode): Likewise.
5323 * regrename.c (check_new_reg_p): Likewise.
5324 * reload.c (find_equiv_reg): Likewise.
5325 * reload1.c (emit_reload_insns): Likewise.
5326 * sched-deps.c (deps_analyze_insn): Likewise.
5327 * sel-sched.c (init_regs_for_mode): Likewise.
5328 (mark_unavailable_hard_regs): Likewise.
5329 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
5330 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5331 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
5333 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5334 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5335 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
5337 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
5339 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5340 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5341 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
5343 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5344 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5345 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
5347 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5348 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
5350 * config/powerpcspe/powerpcspe.c
5351 (rs6000_hard_regno_call_part_clobbered): New function.
5352 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5353 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5354 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
5356 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5357 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5358 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
5360 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5361 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5362 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
5364 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5365 Alan Hayward <alan.hayward@arm.com>
5366 David Sherwood <david.sherwood@arm.com>
5368 * rtl.h (subreg_memory_offset): Declare.
5369 * emit-rtl.c (subreg_memory_offset): New function.
5370 * expmed.c (store_bit_field_1): Use it.
5371 * expr.c (undefined_operand_subword_p): Likewise.
5372 * simplify-rtx.c (simplify_subreg): Likewise.
5374 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
5376 PR rtl-optimization/57448
5379 * optabs.c (expand_atomic_load): Place compiler memory barriers if
5380 using atomic_load pattern.
5381 (expand_atomic_store): Likewise.
5383 2017-09-04 Jakub Jelinek <jakub@redhat.com>
5386 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
5387 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
5388 handling. Use replace_call_with_value with NULL instead of
5389 gsi_replace, unlink_stmt_vdef and release_defs.
5391 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
5394 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
5396 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5399 * rtl.h (emit_library_call_value_1): Declare.
5400 (emit_library_call): Replace declaration with a series of overloads.
5401 Remove the parameter count argument.
5402 (emit_library_call_value): Likewise.
5403 * calls.c (emit_library_call_value_1): Make global. Replace varargs
5404 with an "rtx_mode_t *".
5405 (emit_library_call_value): Delete.
5406 (emit_library_call): Likewise.
5407 * asan.c (asan_emit_stack_protection): Update calls accordingly.
5408 (asan_emit_allocas_unpoison): Likewise.
5409 * builtins.c (expand_builtin_powi): Likewise.
5410 (expand_asan_emit_allocas_unpoison): Likewise.
5411 * cfgexpand.c (expand_main_function): Likewise.
5412 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
5413 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
5414 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
5415 * config/arm/arm.c (arm_trampoline_init): Likewise.
5416 (arm_call_tls_get_addr): Likewise.
5417 (arm_expand_divmod_libfunc): Likewise.
5418 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
5419 (smulsi3_highpart): Likewise.
5420 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
5421 (c6x_expand_compare): Likewise.
5422 (c6x_expand_movmem): Likewise.
5423 * config/frv/frv.c (frv_trampoline_init): Likewise.
5424 * config/i386/i386.c (ix86_trampoline_init): Likewise.
5425 (ix86_expand_divmod_libfunc): Likewise.
5426 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
5427 (ia64_expand_compare): Likewise.
5428 (ia64_profile_hook): Likewise.
5429 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
5430 (nonlocal_goto): Likewise.
5431 (restore_stack_nonlocal): Likewise.
5432 * config/m32r/m32r.c (block_move_call): Likewise.
5433 (m32r_trampoline_init): Likewise.
5434 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
5435 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
5436 (m68k_call_m68k_read_tp): Likewise.
5437 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
5438 (microblaze_expand_divide): Likewise.
5439 * config/mips/mips.h (mips_args): Likewise.
5440 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
5441 (MIPS_ICACHE_SYNC): Likewise.
5442 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
5443 (nios2_trampoline_init): Likewise.
5444 * config/pa/pa.c (hppa_tls_call): Likewise.
5445 (pa_trampoline_init): Likewise.
5446 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
5447 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
5448 (expand_strn_compare): Likewise.
5449 (rs6000_generate_compare): Likewise.
5450 (rs6000_expand_float128_convert): Likewise.
5451 (output_profile_hook): Likewise.
5452 (rs6000_trampoline_init): Likewise.
5453 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
5454 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
5455 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
5456 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
5457 (rs6000_generate_compare): Likewise.
5458 (rs6000_expand_float128_convert): Likewise.
5459 (output_profile_hook): Likewise.
5460 (rs6000_trampoline_init): Likewise.
5461 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
5462 * config/sh/sh.c (sh_trampoline_init): Likewise.
5463 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
5464 (sparc_emit_float_lib_cmp): Likewise.
5465 (sparc32_initialize_trampoline): Likewise.
5466 (sparc64_initialize_trampoline): Likewise.
5467 (sparc_profile_hook): Likewise.
5468 * config/spu/spu.c (ea_load_store): Likewise.
5469 * config/spu/spu.md (floatunssidf2): Likewise.
5470 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
5471 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
5472 * config/visium/visium.c (expand_block_move_4): Likewise.
5473 (expand_block_move_2): Likewise.
5474 (expand_block_move_1): Likewise.
5475 (expand_block_set_4): Likewise.
5476 (expand_block_set_2): Likewise.
5477 (expand_block_set_1): Likewise.
5478 (visium_trampoline_init): Likewise.
5479 (visium_profile_hook): Likewise.
5480 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
5481 (xtensa_setup_frame_addresses): Likewise.
5482 (xtensa_trampoline_init): Likewise.
5483 * except.c (sjlj_emit_function_enter): Likewise.
5484 (sjlj_emit_function_exit): Likewise.
5485 * explow.c (allocate_dynamic_stack_space): Likewise.
5486 (probe_stack_range): Likewise.
5487 * expr.c (convert_mode_scalar): Likewise.
5488 * optabs.c (expand_binop): Likewise.
5489 (expand_twoval_binop_libfunc): Likewise.
5490 (expand_unop): Likewise.
5491 (prepare_cmp_insn): Likewise.
5492 (prepare_float_lib_cmp): Likewise.
5493 (expand_float): Likewise.
5494 (expand_fix): Likewise.
5495 (expand_fixed_convert): Likewise.
5496 (maybe_emit_sync_lock_test_and_set): Likewise.
5497 (expand_atomic_compare_and_swap): Likewise.
5498 (expand_mem_thread_fence): Likewise.
5499 (expand_atomic_fetch_op): Likewise.
5501 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
5503 * doc/generic.texi (OpenACC): Adjust URL.
5504 * doc/invoke.texi (C Dialect Options): Ditto.
5506 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
5508 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
5509 predicate for operand 1. Add (m,<S>) constraint.
5510 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
5511 Prevent memory operand 1 with register operand 2.
5513 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
5515 PR rtl-optimization/82024
5516 * combine.c (try_combine): If the combination result is a PARALLEL,
5517 and we only need to retain the SET in there that would be placed
5518 at I2, check that we can place that at I3 instead, before doing so.
5520 2017-09-01 Jakub Jelinek <jakub@redhat.com>
5523 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
5525 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
5526 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
5529 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
5530 Jeff Law <law@redhat.com>
5532 * varasm.c (bss_initializer_p): Do not put constants into .bss
5533 (categorize_decl_for_section): Handle bss_initializer_p returning
5534 false when DECL_INITIAL is NULL.
5536 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5539 * config/s390/s390.c (s390_can_inline_p): New function.
5541 2017-09-01 Jeff Law <law@redhat.com>
5543 PR tree-optimization/82052
5544 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
5545 Always initialize the returned slot after a hash table miss
5546 when INSERT is true.
5548 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
5550 * config/s390/s390.md (mem_signal_fence): Remove.
5551 * doc/md.texi (mem_signal_fence): Remove.
5552 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
5554 * target-insns.def (mem_signal_fence): Remove.
5556 2017-09-01 Jakub Jelinek <jakub@redhat.com>
5559 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
5562 * asan.c (create_odr_indicator): Strip name encoding from assembler
5563 name before appending it after __odr_asan_.
5565 2017-09-01 Martin Liska <mliska@suse.cz>
5567 PR tree-optimization/82059
5568 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
5569 frequency only when an edge is redirected.
5571 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5573 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
5574 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
5575 (arc_conditional_register_usage): Remove ARC600 lp_count
5577 (arc_file_start): Emit Tag_ARC_CPU_variation.
5578 (arc_can_use_doloop_p): New conditions to use ZOLs.
5579 (hwloop_fail): New function.
5580 (hwloop_optimize): Likewise.
5581 (hwloop_pattern_reg): Likewise.
5582 (arc_doloop_hooks): New struct, to be used with reorg_loops.
5583 (arc_reorg_loops): New function, calls reorg_loops.
5584 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
5585 (arc600_corereg_hazard): Remove ZOL checking, case handled by
5587 (arc_loop_hazard): Remove function, functionality moved into
5589 (arc_hazard): Remove arc_loop_hazard call.
5590 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
5591 into hwloop_optimize.
5592 (arc_label_align): Remove ZOL handling.
5593 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
5594 * config/arc/arc.md (doloop_begin): Remove pattern.
5595 (doloop_begin_i): Likewise.
5596 (doloop_end_i): Likewise.
5597 (doloop_fallback): Likewise.
5598 (doloop_fallback_m): Likewise.
5599 (doloop_end): Reimplement expand.
5600 (arc_lp): New pattern for LP instruction.
5601 (loop_end): New pattern.
5602 (loop_fail): Likewise.
5603 (decrement_and_branch_until_zero): Likewise.
5604 * config/arc/arc.opt (mlpc-width): New option.
5605 * doc/invoke.texi (mlpc-width): Document option.
5607 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5609 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
5610 (arc_ccfsm_advance): Fix checking for delay slots.
5611 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
5613 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5615 * config/arc/arc.md (movqi_insn): Add stores to save constant long
5617 (movhi_insn): Update store instruction constraint which are saving
5618 6-bit short immediates.
5619 (movsi_insn): Consider also short scaled load operations.
5620 (zero_extendhisi2_i): Use Usd constraint instead of T.
5621 (extendhisi2_i): Add q constraint.
5622 (arc_clzsi2): Add type and length attributes.
5623 (arc_ctzsi2): Likewise.
5624 * config/arc/constraints.md (Usc): Update constraint, the
5625 assembler can parse two relocations for a single instruction.
5627 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5629 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
5630 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
5632 2017-08-31 Olivier Hainque <hainque@adacore.com>
5634 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
5635 match as powerpc-wrs-vxworks*.
5637 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
5639 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
5640 register constraint for by-element operand.
5641 (aarch64_mls_elt_merge<mode>): Likewise.
5643 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
5645 * config/arc/arc.c (arc_can_follow_jump): Check for short
5648 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
5650 * config.gcc: Use g.opt for arc.
5651 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
5652 functionality moved to ...
5653 (legitimate_scaled_address_p): New function, ...here.
5654 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
5655 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
5656 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
5658 (arc_override_options): Handle G option.
5659 (arc_output_pic_addr_const): Correct function definition.
5660 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
5661 (arc_decl_anon_ns_mem_p): Delete.
5662 (arc_in_small_data_p): Overhaul this function to take into
5663 consideration the value given via G option.
5664 (arc_rewrite_small_data_1): Renamed and corrected old
5665 arc_rewrite_small_data function.
5666 (arc_rewrite_small_data): New function.
5667 (small_data_pattern): Don't use pic_offset_table_rtx.
5668 * config/arc/arc.h (CC1_SPEC): Recognize G option.
5669 * config/arc/simdext.md (movmisalignv2hi): Use
5670 prepare_move_operands function.
5672 (movmisalign*): Likewise.
5673 * doc/invoke.texi (ARC options): Document -G option.
5675 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
5677 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
5679 * config/arc/arc.c (arc_print_operand): Output scalled address for
5680 sdata whenever is possible.
5681 (arc_in_small_data_p): Allow sdata for 64bit datum when double
5682 load/stores are available.
5683 (compact_sda_memory_operand): Check for the alignment required by
5684 code density instructions.
5685 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
5687 * config/arc/constraints.md (Usd): Update constraint.
5688 (Us0): New constraint.
5689 (Usc): Update constraint.
5691 2017-08-31 Richard Biener <rguenther@suse.de>
5694 * dwarf2out.c (dwarf2out_early_global_decl): Process each
5697 2017-08-31 Tamar Christina <tamar.christina@arm.com>
5699 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
5700 Resize type_signature.
5702 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
5703 Alan Hayward <alan.hayward@arm.com>
5704 David Sherwood <david.sherwood@arm.com>
5706 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
5707 subregs whose inner modes can be stored in GPRs.
5708 (aarch64_classify_index): Likewise.
5710 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
5711 Alan Hayward <alan.hayward@arm.com>
5712 David Sherwood <david.sherwood@arm.com>
5714 * config/aarch64/iterators.md (V_cmp_result): Rename to...
5715 (V_INT_EQUIV): ...this.
5716 (v_cmp_result): Rename to...
5717 (v_int_equiv): ...this.
5718 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
5719 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
5720 (copysign<mode>3): Likewise.
5721 (aarch64_simd_bsl<mode>_internal): Likewise.
5722 (aarch64_simd_bsl<mode>): Likewise.
5723 (vec_cmp<mode><mode>): Likewise.
5724 (vcond<mode><mode>): Likewise.
5725 (vcond<v_cmp_mixed><mode>): Likewise.
5726 (vcondu<mode><v_cmp_mixed>): Likewise.
5727 (aarch64_cm<optab><mode>): Likewise.
5728 (aarch64_cmtst<mode>): Likewise.
5729 (aarch64_fac<optab><mode>): Likewise.
5730 (vec_perm_const<mode>): Likewise.
5731 (vcond_mask_<mode><v_cmp_result>): Rename to...
5732 (vcond_mask_<mode><v_int_equiv>): ...this.
5733 (vec_cmp<mode><v_cmp_result>): Rename to...
5734 (vec_cmp<mode><v_int_equiv>): ...this.
5736 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
5737 Alan Hayward <alan.hayward@arm.com>
5738 David Sherwood <david.sherwood@arm.com>
5740 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
5742 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
5743 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
5744 (UNSPEC_LD4_DREG): New unspecs.
5745 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
5746 (aarch64_ld2<mode>_dreg_be): Replace with...
5747 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
5749 (aarch64_ld3<mode>_dreg_le)
5750 (aarch64_ld3<mode>_dreg_be): Replace with...
5751 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
5753 (aarch64_ld4<mode>_dreg_le)
5754 (aarch64_ld4<mode>_dreg_be): Replace with...
5755 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
5758 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5760 PR tree-optimization/81987
5761 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
5762 insert an initializer in a location not dominated by the stride
5765 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
5767 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
5768 on the entire header of the finally block in the fallthru case.
5770 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
5772 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
5774 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
5776 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
5777 rs6000_emit_move_from_cr and call renamed function.
5778 (rs6000_emit_prologue): Call renamed functions.
5779 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
5780 movesi_from_cr, remove volatile CRs.
5782 2017-08-30 Jon Beniston <jon@beniston.com>
5783 Richard Biener <rguenther@suse.de>
5785 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
5786 of VECTOR_MODE_P check.
5787 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
5788 element vector types.
5790 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5792 * df.h (df_read_modify_subreg_p): Remove in favor of...
5793 * rtl.h (read_modify_subreg_p): ...this new function. Take a
5794 const_rtx instead of an rtx.
5795 * cprop.c (local_cprop_find_used_regs): Update accordingly.
5796 * df-problems.c (df_word_lr_mark_ref): Likewise.
5797 * ira-lives.c (mark_pseudo_reg_live): Likewise.
5798 (mark_pseudo_reg_dead): Likewise.
5799 (mark_ref_dead): Likewise.
5800 * reginfo.c (init_subregs_of_mode): Likewise.
5801 * sched-deps.c (sched_analyze_1): Likewise.
5802 * df-scan.c (df_def_record_1): Likewise.
5803 (df_uses_record): Likewise.
5804 (df_read_modify_subreg_p): Remove in favor of...
5805 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
5806 const_rtx instead of an rtx.
5808 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5809 Alan Hayward <alan.hayward@arm.com>
5810 David Sherwood <david.sherwood@arm.com>
5812 * rtl.h (partial_subreg_p): New function.
5813 * caller-save.c (save_call_clobbered_regs): Use it.
5814 * calls.c (expand_call): Likewise.
5815 * combine.c (combinable_i3pat): Likewise.
5816 (simplify_set): Likewise.
5817 (make_extraction): Likewise.
5818 (make_compound_operation_int): Likewise.
5819 (gen_lowpart_or_truncate): Likewise.
5820 (force_to_mode): Likewise.
5821 (make_field_assignment): Likewise.
5822 (reg_truncated_to_mode): Likewise.
5823 (record_truncated_value): Likewise.
5824 (move_deaths): Likewise.
5825 * cse.c (record_jump_cond): Likewise.
5826 (cse_insn): Likewise.
5827 * cselib.c (cselib_lookup_1): Likewise.
5828 * expmed.c (extract_bit_field_using_extv): Likewise.
5829 * function.c (assign_parm_setup_reg): Likewise.
5830 * ifcvt.c (noce_convert_multiple_sets): Likewise.
5831 * ira-build.c (create_insn_allocnos): Likewise.
5832 * lra-coalesce.c (merge_pseudos): Likewise.
5833 * lra-constraints.c (match_reload): Likewise.
5834 (simplify_operand_subreg): Likewise.
5835 (curr_insn_transform): Likewise.
5836 * lra-lives.c (process_bb_lives): Likewise.
5837 * lra.c (new_insn_reg): Likewise.
5838 (lra_substitute_pseudo): Likewise.
5839 * regcprop.c (mode_change_ok): Likewise.
5840 (maybe_mode_change): Likewise.
5841 (copyprop_hardreg_forward_1): Likewise.
5842 * reload.c (push_reload): Likewise.
5843 (find_reloads): Likewise.
5844 (find_reloads_subreg_address): Likewise.
5845 * reload1.c (alter_reg): Likewise.
5846 (eliminate_regs_1): Likewise.
5847 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5849 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
5851 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
5852 back to if statements, including unpack.
5854 2017-08-30 Martin Liska <mliska@suse.cz>
5857 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
5859 (func_checker::compare_asm_inputs_outputs): ... this function.
5860 (func_checker::compare_gimple_asm): Use the function to compare
5861 also ASM constrains.
5862 * ipa-icf-gimple.h: Rename the function.
5864 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5865 Alan Hayward <alan.hayward@arm.com>
5866 David Sherwood <david.sherwood@arm.com>
5868 * coretypes.h (complex_mode): New type.
5869 * gdbhooks.py (build_pretty_printer): Handle it.
5870 * machmode.h (complex_mode): New class.
5871 (complex_mode::includes_p): New function.
5872 (is_complex_int_mode): Likewise.
5873 (is_complex_float_mode): Likewise.
5874 * genmodes.c (get_mode_class): Handle complex mode classes.
5875 * function.c (expand_function_end): Use is_complex_int_mode.
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 * coretypes.h (scalar_mode_pod): New typedef.
5882 * gdbhooks.py (build_pretty_printer): Handle it.
5883 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
5884 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
5885 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
5886 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
5889 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5890 Alan Hayward <alan.hayward@arm.com>
5891 David Sherwood <david.sherwood@arm.com>
5893 * machmode.h (mode_for_vector): Take a scalar_mode instead
5895 * stor-layout.c (mode_for_vector): Likewise.
5896 * explow.c (promote_mode): Use as_a <scalar_mode>.
5897 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
5899 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5900 Alan Hayward <alan.hayward@arm.com>
5901 David Sherwood <david.sherwood@arm.com>
5903 * target.def (preferred_simd_mode): Take a scalar_mode
5904 instead of a machine_mode.
5905 * targhooks.h (default_preferred_simd_mode): Likewise.
5906 * targhooks.c (default_preferred_simd_mode): Likewise.
5907 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
5908 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
5909 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
5910 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
5911 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
5912 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
5913 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
5914 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
5916 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
5917 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
5918 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
5919 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
5920 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
5921 * doc/tm.texi: Regenerate.
5922 * optabs-query.c (can_vec_mask_load_store_p): Return false for
5925 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5926 Alan Hayward <alan.hayward@arm.com>
5927 David Sherwood <david.sherwood@arm.com>
5929 * target.def (scalar_mode_supported_p): Take a scalar_mode
5930 instead of a machine_mode.
5931 * targhooks.h (default_scalar_mode_supported_p): Likewise.
5932 * targhooks.c (default_scalar_mode_supported_p): Likewise.
5933 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
5934 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
5935 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
5936 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
5937 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
5938 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
5939 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
5940 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
5941 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
5942 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
5943 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
5944 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
5946 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
5947 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
5948 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
5949 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
5950 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
5952 * doc/tm.texi: Regenerate.
5954 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5955 Alan Hayward <alan.hayward@arm.com>
5956 David Sherwood <david.sherwood@arm.com>
5958 * coretypes.h (opt_scalar_mode): New typedef.
5959 * gdbhooks.py (build_pretty_printers): Handle it.
5960 * machmode.h (mode_iterator::get_2xwider): Add overload for
5962 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
5964 * expr.c (convert_mode_scalar): Likewise.
5965 * omp-low.c (omp_clause_aligned_alignment): Likewise.
5966 * optabs.c (expand_float): Likewise.
5967 (expand_fix): Likewise.
5968 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5970 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5971 Alan Hayward <alan.hayward@arm.com>
5972 David Sherwood <david.sherwood@arm.com>
5974 * optabs.c (expand_float): Explicitly check for scalars before
5975 using a branching expansion.
5976 (expand_fix): Likewise.
5978 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5979 Alan Hayward <alan.hayward@arm.com>
5980 David Sherwood <david.sherwood@arm.com>
5982 * expr.c (convert_mode): Split scalar handling out into...
5983 (convert_mode_scalar): ...this new function. Treat the modes
5986 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5987 Alan Hayward <alan.hayward@arm.com>
5988 David Sherwood <david.sherwood@arm.com>
5990 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
5992 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
5994 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5995 Alan Hayward <alan.hayward@arm.com>
5996 David Sherwood <david.sherwood@arm.com>
5998 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
5999 rather than a machine_mode.
6000 (fixed_from_string): Likewise.
6001 (fixed_convert): Likewise.
6002 (fixed_convert_from_int): Likewise.
6003 (fixed_convert_from_real): Likewise.
6004 (real_convert_from_fixed): Likewise.
6005 * fixed-value.c (fixed_from_double_int): Likewise.
6006 (fixed_from_string): Likewise.
6007 (fixed_convert): Likewise.
6008 (fixed_convert_from_int): Likewise.
6009 (fixed_convert_from_real): Likewise.
6010 (real_convert_from_fixed): Likewise.
6011 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
6013 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6014 Alan Hayward <alan.hayward@arm.com>
6015 David Sherwood <david.sherwood@arm.com>
6017 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
6018 of separate mode class checks. Do not allow vector modes here.
6019 (immed_wide_int_const): Use as_a <scalar_mode>.
6020 * explow.c (trunc_int_for_mode): Likewise.
6021 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
6022 (wi::shwi): Likewise.
6023 (wi::min_value): Likewise.
6024 (wi::max_value): Likewise.
6025 * dwarf2out.c (loc_descriptor): Likewise.
6026 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
6029 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6030 Alan Hayward <alan.hayward@arm.com>
6031 David Sherwood <david.sherwood@arm.com>
6033 * tree.h (SCALAR_TYPE_MODE): New macro.
6034 * expr.c (expand_expr_addr_expr_1): Use it.
6035 (expand_expr_real_2): Likewise.
6036 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
6037 (fold_convert_const_fixed_from_int): Likewise.
6038 (fold_convert_const_fixed_from_real): Likewise.
6039 (native_encode_fixed): Likewise
6040 (native_encode_complex): Likewise
6041 (native_encode_vector): Likewise.
6042 (native_interpret_fixed): Likewise.
6043 (native_interpret_real): Likewise.
6044 (native_interpret_complex): Likewise.
6045 (native_interpret_vector): Likewise.
6046 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
6047 (simd_clone_adjust_argument_types): Likewise.
6048 (simd_clone_init_simd_arrays): Likewise.
6049 (simd_clone_adjust): Likewise.
6050 * stor-layout.c (layout_type): Likewise.
6051 * tree.c (build_minus_one_cst): Likewise.
6052 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
6053 * tree-inline.c (estimate_move_cost): Likewise.
6054 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
6055 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
6056 (vectorizable_reduction): Likewise.
6057 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
6058 (vect_recog_mixed_size_cond_pattern): Likewise.
6059 (check_bool_pattern): Likewise.
6060 (adjust_bool_pattern): Likewise.
6061 (search_type_for_mask_1): Likewise.
6062 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
6063 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6064 (vectorizable_load): Likewise.
6065 (vectorizable_store): Likewise.
6066 * ubsan.c (ubsan_encode_value): Likewise.
6067 * varasm.c (output_constant): Likewise.
6069 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6070 Alan Hayward <alan.hayward@arm.com>
6071 David Sherwood <david.sherwood@arm.com>
6073 * coretypes.h (scalar_mode): New class.
6074 * machmode.h (scalar_mode): Likewise.
6075 (scalar_mode::includes_p): New function.
6076 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
6077 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
6078 * genmodes.c (get_mode_class): Handle remaining scalar modes.
6079 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
6080 * expmed.c (store_bit_field_1): Likewise.
6081 (extract_bit_field_1): Likewise.
6082 * expr.c (write_complex_part): Likewise.
6083 (read_complex_part): Likewise.
6084 (emit_move_complex_push): Likewise.
6085 (expand_expr_real_2): Likewise.
6086 * function.c (assign_parm_setup_reg): Likewise.
6087 (assign_parms_unsplit_complex): Likewise.
6088 * optabs.c (expand_binop): Likewise.
6089 * rtlanal.c (subreg_get_info): Likewise.
6090 * simplify-rtx.c (simplify_immed_subreg): Likewise.
6091 * varasm.c (output_constant_pool_2): Likewise.
6093 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6094 Alan Hayward <alan.hayward@arm.com>
6095 David Sherwood <david.sherwood@arm.com>
6097 * expmed.c (extract_high_half): Use scalar_int_mode and remove
6099 (expmed_mult_highpart_optab): Likewise.
6100 (expmed_mult_highpart): Likewise.
6102 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6103 Alan Hayward <alan.hayward@arm.com>
6104 David Sherwood <david.sherwood@arm.com>
6106 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
6107 instead of a machine_mode.
6108 (builtin_memset_read_str): Likewise.
6109 * builtins.c (c_readstr): Likewise.
6110 (builtin_memcpy_read_str): Likewise.
6111 (builtin_strncpy_read_str): Likewise.
6112 (builtin_memset_read_str): Likewise.
6113 (builtin_memset_gen_str): Likewise.
6114 (expand_builtin_signbit): Use scalar_int_mode for local variables.
6115 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
6116 instead of a machine_mode.
6117 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
6119 (make_extraction): Likewise.
6120 (try_widen_shift_mode): Take and return scalar_int_modes instead
6122 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
6123 a scalar_int_mode instead of a machine_mode.
6124 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
6125 (avr_addr_space_pointer_mode): Likewise.
6126 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
6127 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
6128 (msp430_unwind_word_mode): Likewise.
6129 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
6130 (spu_addr_space_pointer_mode): Likewise.
6131 (spu_addr_space_address_mode): Likewise.
6132 (spu_libgcc_cmp_return_mode): Likewise.
6133 (spu_libgcc_shift_count_mode): Likewise.
6134 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
6135 (rl78_addr_space_pointer_mode): Likewise.
6136 (fl78_unwind_word_mode): Likewise.
6137 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
6139 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
6140 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
6141 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
6142 (mips_valid_pointer_mode): Likewise.
6143 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
6144 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
6145 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
6147 (ft32_addr_space_address_mode): Likewise.
6148 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
6149 scalar_int_mode instead of a machine_mode.
6150 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
6152 (m32c_addr_space_address_mode): Likewise.
6153 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
6154 (rs6000_eh_return_filter_mode): Likewise.
6155 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
6156 (rs6000_eh_return_filter_mode): Likewise.
6157 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
6158 (s390_libgcc_shift_count_mode): Likewise.
6159 (s390_unwind_word_mode): Likewise.
6160 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
6162 * target.def (mode_rep_extended): Likewise.
6163 (valid_pointer_mode): Likewise.
6164 (addr_space.valid_pointer_mode): Likewise.
6165 (eh_return_filter_mode): Return a scalar_int_mode rather than
6167 (libgcc_cmp_return_mode): Likewise.
6168 (libgcc_shift_count_mode): Likewise.
6169 (unwind_word_mode): Likewise.
6170 (addr_space.pointer_mode): Likewise.
6171 (addr_space.address_mode): Likewise.
6172 * doc/tm.texi: Regenerate.
6173 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
6175 (do_jump): Use scalar_int_mode for local variables.
6176 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
6177 rather than a machine_mode.
6178 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
6179 (scompare_loc_descriptor_wide): Likewise.
6180 (scompare_loc_descriptor_narrow): Likewise.
6181 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
6183 * except.c (sjlj_emit_dispatch_table): Likewise.
6184 (expand_builtin_eh_copy_values): Likewise.
6185 * explow.c (convert_memory_address_addr_space_1): Likewise.
6186 Take a scalar_int_mode rather than a machine_mode.
6187 (convert_memory_address_addr_space): Take a scalar_int_mode rather
6188 than a machine_mode.
6189 (memory_address_addr_space): Use scalar_int_mode for local variables.
6190 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
6191 rather than a machine_mode.
6192 * expmed.c (mask_rtx): Likewise.
6193 (init_expmed_one_conv): Likewise.
6194 (expand_mult_highpart_adjust): Likewise.
6195 (extract_high_half): Likewise.
6196 (expmed_mult_highpart_optab): Likewise.
6197 (expmed_mult_highpart): Likewise.
6198 (expand_smod_pow2): Likewise.
6199 (expand_sdiv_pow2): Likewise.
6200 (emit_store_flag_int): Likewise.
6201 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
6203 (extract_low_bits): Likewise.
6204 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
6206 * expr.c (pieces_addr::adjust): Likewise.
6207 (can_store_by_pieces): Likewise.
6208 (store_by_pieces): Likewise.
6209 (clear_by_pieces_1): Likewise.
6210 (expand_expr_addr_expr_1): Likewise.
6211 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
6212 (expand_expr_real_1): Likewise.
6213 (try_casesi): Likewise.
6214 * final.c (shorten_branches): Likewise.
6215 * fold-const.c (fold_convert_const_int_from_fixed): Change the
6216 type of "mode" to machine_mode.
6217 * internal-fn.c (expand_arith_overflow_result_store): Take a
6218 scalar_int_mode rather than a machine_mode.
6219 (expand_mul_overflow): Use scalar_int_mode for local variables.
6220 * loop-doloop.c (doloop_modify): Likewise.
6221 (doloop_optimize): Likewise.
6222 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
6223 than a machine_mode.
6224 (expand_doubleword_shift_condmove): Likewise.
6225 (expand_doubleword_shift): Likewise.
6226 (expand_doubleword_clz): Likewise.
6227 (expand_doubleword_popcount): Likewise.
6228 (expand_doubleword_parity): Likewise.
6229 (expand_absneg_bit): Use scalar_int_mode for local variables.
6230 (prepare_float_lib_cmp): Likewise.
6231 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
6232 rather than a machine_mode.
6233 (convert_memory_address_addr_space): Likewise.
6234 (get_mode_bounds): Likewise.
6235 (get_address_mode): Return a scalar_int_mode rather than a
6237 * rtlanal.c (get_address_mode): Likewise.
6238 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
6239 than a machine_mode.
6240 * targhooks.c (default_mode_rep_extended): Likewise.
6241 (default_valid_pointer_mode): Likewise.
6242 (default_addr_space_valid_pointer_mode): Likewise.
6243 (default_eh_return_filter_mode): Return a scalar_int_mode rather
6244 than a machine_mode.
6245 (default_libgcc_cmp_return_mode): Likewise.
6246 (default_libgcc_shift_count_mode): Likewise.
6247 (default_unwind_word_mode): Likewise.
6248 (default_addr_space_pointer_mode): Likewise.
6249 (default_addr_space_address_mode): Likewise.
6250 * targhooks.h (default_eh_return_filter_mode): Likewise.
6251 (default_libgcc_cmp_return_mode): Likewise.
6252 (default_libgcc_shift_count_mode): Likewise.
6253 (default_unwind_word_mode): Likewise.
6254 (default_addr_space_pointer_mode): Likewise.
6255 (default_addr_space_address_mode): Likewise.
6256 (default_mode_rep_extended): Take a scalar_int_mode rather than
6258 (default_valid_pointer_mode): Likewise.
6259 (default_addr_space_valid_pointer_mode): Likewise.
6260 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
6262 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
6263 rather than a machine_mode.
6264 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
6265 for local variables.
6266 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6267 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
6268 than a machine_mode.
6270 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6271 Alan Hayward <alan.hayward@arm.com>
6272 David Sherwood <david.sherwood@arm.com>
6274 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
6275 the mode argument to scalar_int_mode.
6276 (do_jump_by_parts_zero_rtx): Likewise.
6277 (do_jump_by_parts_equality_rtx): Likewise.
6278 (do_jump_by_parts_greater): Take a mode argument.
6279 (do_jump_by_parts_equality): Likewise.
6280 (do_jump_1): Update calls accordingly.
6282 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6283 Alan Hayward <alan.hayward@arm.com>
6284 David Sherwood <david.sherwood@arm.com>
6286 * is-a.h (safe_dyn_cast): New function.
6287 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
6288 (jump_table_for_label): Likewise.
6289 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
6290 instead of an rtx_insn *.
6291 (shorten_branches): Use dyn_cast instead of LABEL_P and
6292 JUMP_TABLE_DATA_P. Use jump_table_for_label and
6293 rtx_jump_table_data::get_data_mode.
6294 (final_scan_insn): Likewise.
6296 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6297 Alan Hayward <alan.hayward@arm.com>
6298 David Sherwood <david.sherwood@arm.com>
6300 * combine.c (try_combine): Use is_a <scalar_int_mode> when
6301 trying to combine a full-register integer set with a subreg
6304 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6305 Alan Hayward <alan.hayward@arm.com>
6306 David Sherwood <david.sherwood@arm.com>
6308 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
6309 that is always either address_mode or pointer_mode.
6311 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6312 Alan Hayward <alan.hayward@arm.com>
6313 David Sherwood <david.sherwood@arm.com>
6315 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
6316 when the two are known to be equal.
6318 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6319 Alan Hayward <alan.hayward@arm.com>
6320 David Sherwood <david.sherwood@arm.com>
6322 * simplify-rtx.c (simplify_const_unary_operation): Use
6323 is_a <scalar_int_mode> instead of checking for a nonzero
6324 precision. Forcibly convert op_mode to a scalar_int_mode
6325 in that case. More clearly differentiate the operand and
6326 result modes and use the former when deciding what the value
6327 of a count-bits operation should be. Use is_int_mode instead
6328 of checking for a MODE_INT. Remove redundant check for whether
6329 this mode has a zero precision.
6331 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6332 Alan Hayward <alan.hayward@arm.com>
6333 David Sherwood <david.sherwood@arm.com>
6335 * optabs.c (widen_leading): Change the type of the mode argument
6336 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
6337 (widen_bswap): Likewise.
6338 (expand_parity): Likewise.
6339 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
6340 (expand_ffs): Likewise.
6341 (epand_unop): Check for scalar integer modes before calling the
6344 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6345 Alan Hayward <alan.hayward@arm.com>
6346 David Sherwood <david.sherwood@arm.com>
6348 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
6350 (expand_expr_real_1): Update call accordingly.
6352 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6353 Alan Hayward <alan.hayward@arm.com>
6354 David Sherwood <david.sherwood@arm.com>
6356 * expmed.c (store_bit_field_using_insv): Add op0_mode and
6357 value_mode arguments. Use scalar_int_mode internally.
6358 (store_bit_field_1): Rename the new integer mode from imode
6359 to op0_mode and use it instead of GET_MODE (op0). Update calls
6360 to store_split_bit_field, store_bit_field_using_insv and
6361 store_fixed_bit_field.
6362 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
6363 Use scalar_int_mode internally. Use a bit count rather than a mode
6364 when calculating the largest bit size for get_best_mode.
6365 Update calls to store_split_bit_field and store_fixed_bit_field_1.
6366 (store_fixed_bit_field_1): Add mode and value_mode arguments.
6367 Remove assertion that OP0 has a scalar integer mode.
6368 (store_split_bit_field): Add op0_mode and value_mode arguments.
6369 Update calls to extract_fixed_bit_field.
6370 (extract_bit_field_using_extv): Add an op0_mode argument.
6371 Use scalar_int_mode internally.
6372 (extract_bit_field_1): Rename the new integer mode from imode to
6373 op0_mode and use it instead of GET_MODE (op0). Update calls to
6374 extract_split_bit_field, extract_bit_field_using_extv and
6375 extract_fixed_bit_field.
6376 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
6377 to extract_split_bit_field and extract_fixed_bit_field_1.
6378 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
6379 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
6381 (extract_split_bit_field): Add an op0_mode argument. Update call
6382 to extract_fixed_bit_field.
6384 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6385 Alan Hayward <alan.hayward@arm.com>
6386 David Sherwood <david.sherwood@arm.com>
6388 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
6389 * explow.c (hard_function_value): Likewise.
6390 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
6391 convert_to_mode call outside the loop.
6392 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
6393 for the mode iterator. Require the mode specified by max_pieces
6395 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
6397 (copy_blkmode_to_reg): Likewise.
6398 (set_storage_via_setmem): Likewise.
6399 * optabs.c (prepare_cmp_insn): Likewise.
6400 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6401 * stor-layout.c (finish_bitfield_representative): Likewise.
6403 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6404 Alan Hayward <alan.hayward@arm.com>
6405 David Sherwood <david.sherwood@arm.com>
6407 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
6408 * expr.c (convert_move): Use them.
6409 (convert_modes): Likewise.
6410 (store_expr_with_bounds): Likewise.
6412 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6413 Alan Hayward <alan.hayward@arm.com>
6414 David Sherwood <david.sherwood@arm.com>
6416 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
6417 parameter for the mode of "x". Remove the "known_x", "known_mode"
6418 and "known_ret" arguments. Change the type of the mode argument
6420 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
6421 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
6422 (reg_num_sign_bit_copies_for_combine): Likewise.
6423 * rtlanal.c (nonzero_bits1): Likewise.
6424 (num_sign_bit_copies1): Likewise.
6425 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
6426 (reg_num_sign_bit_copies_general): Likewise.
6427 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
6428 (reg_nonzero_bits_general): Likewise.
6430 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6431 Alan Hayward <alan.hayward@arm.com>
6432 David Sherwood <david.sherwood@arm.com>
6434 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
6435 than in subroutines. Return 1 for non-integer modes.
6436 (cached_num_sign_bit_copies): Change the type of the mode parameter
6438 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
6439 classes. Handle CONST_INT_P first and then check whether X also
6440 has a scalar integer mode. Check the same thing for inner registers
6441 of a SUBREG and for values that are being extended or truncated.
6443 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6444 Alan Hayward <alan.hayward@arm.com>
6445 David Sherwood <david.sherwood@arm.com>
6447 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
6448 in subroutines. Return the mode mask for non-integer modes.
6449 (cached_nonzero_bits): Change the type of the mode parameter
6451 (nonzero_bits1): Likewise. Remove early exit for other mode
6452 classes. Handle CONST_INT_P first and then check whether X
6453 also has a scalar integer mode.
6455 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6456 Alan Hayward <alan.hayward@arm.com>
6457 David Sherwood <david.sherwood@arm.com>
6459 * expr.c (widest_int_mode_for_size): Make the comment match the code.
6460 Return a scalar_int_mode and assert that the size is greater than
6462 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
6463 (op_by_pieces_d::op_by_pieces_d): Likewise.
6464 (op_by_pieces_d::run): Likewise.
6465 (can_store_by_pieces): Likewise.
6467 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6468 Alan Hayward <alan.hayward@arm.com>
6469 David Sherwood <david.sherwood@arm.com>
6471 * combine.c (extract_left_shift): Add a mode argument and update
6473 (make_compound_operation_int): Change the type of the mode parameter
6474 to scalar_int_mode and update the call to extract_left_shift.
6476 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6477 Alan Hayward <alan.hayward@arm.com>
6478 David Sherwood <david.sherwood@arm.com>
6480 * combine.c (simplify_and_const_int): Change the type of the mode
6481 parameter to scalar_int_mode.
6482 (simplify_and_const_int_1): Likewise. Update recursive call.
6484 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6485 Alan Hayward <alan.hayward@arm.com>
6486 David Sherwood <david.sherwood@arm.com>
6488 * combine.c (simplify_compare_const): Check that the mode is a
6489 scalar_int_mode (rather than VOIDmode) before testing its
6491 (simplify_comparison): Move COMPARISON_P handling out of the
6492 loop and restrict the latter part of the loop to scalar_int_modes.
6493 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
6494 and when considering SUBREG_REGs. Use is_int_mode instead of
6495 checking GET_MODE_CLASS against MODE_INT.
6497 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6498 Alan Hayward <alan.hayward@arm.com>
6499 David Sherwood <david.sherwood@arm.com>
6501 * combine.c (try_widen_shift_mode): Move check for equal modes to...
6502 (simplify_shift_const_1): ...here. Use scalar_int_mode for
6503 shift_unit_mode and for modes involved in scalar shifts.
6505 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6506 Alan Hayward <alan.hayward@arm.com>
6507 David Sherwood <david.sherwood@arm.com>
6509 * combine.c (force_int_to_mode): New function, split out from...
6510 (force_to_mode): ...here. Keep xmode up-to-date and use it
6511 instead of GET_MODE (x).
6513 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6514 Alan Hayward <alan.hayward@arm.com>
6515 David Sherwood <david.sherwood@arm.com>
6517 * optabs-query.h (extraction_insn::struct_mode): Change type to
6518 opt_scalar_int_mode and update comment.
6519 (extraction_insn::field_mode): Change type to scalar_int_mode.
6520 (extraction_insn::pos_mode): Likewise.
6521 * combine.c (make_extraction): Update accordingly.
6522 * optabs-query.c (get_traditional_extraction_insn): Likewise.
6523 (get_optab_extraction_insn): Likewise.
6524 * recog.c (simplify_while_replacing): Likewise.
6525 * expmed.c (narrow_bit_field_mem): Change the type of the mode
6526 parameter to opt_scalar_int_mode.
6528 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6529 Alan Hayward <alan.hayward@arm.com>
6530 David Sherwood <david.sherwood@arm.com>
6532 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
6533 to a scalar_int_mode instead of a machine_mode.
6534 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
6535 (get_best_mode): Return a boolean and use a pointer argument to store
6536 the selected mode. Replace the limit mode parameter with a bit limit.
6537 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
6538 for the values returned by bit_field_mode_iterator::next_mode.
6539 (store_bit_field): Update call to get_best_mode.
6540 (store_fixed_bit_field): Likewise.
6541 (extract_fixed_bit_field): Likewise.
6542 * expr.c (optimize_bitfield_assignment_op): Likewise.
6543 * fold-const.c (optimize_bit_field_compare): Likewise.
6544 (fold_truth_andor_1): Likewise.
6545 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
6546 Update for new type of m_mode.
6547 (get_best_mode): As above.
6549 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6550 Alan Hayward <alan.hayward@arm.com>
6551 David Sherwood <david.sherwood@arm.com>
6553 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
6554 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
6555 (store_bit_field): Check is_a <scalar_int_mode> before calling
6556 strict_volatile_bitfield_p.
6557 (extract_bit_field): Likewise.
6559 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6560 Alan Hayward <alan.hayward@arm.com>
6561 David Sherwood <david.sherwood@arm.com>
6563 * target.def (cstore_mode): Return a scalar_int_mode.
6564 * doc/tm.texi: Regenerate.
6565 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
6566 * targhooks.h (default_cstore_mode): Likewise.
6567 * targhooks.c (default_cstore_mode): Likewise, using a forced
6569 * expmed.c (emit_cstore): Expect the target of the cstore to be
6572 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6573 Alan Hayward <alan.hayward@arm.com>
6574 David Sherwood <david.sherwood@arm.com>
6576 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
6578 (niter_desc): Likewise mode.
6579 (iv_analyze): Add a mode parameter.
6581 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
6582 and change its type to scalar_int_mode.
6583 * loop-iv.c: Update commentary at head of file.
6584 (iv_constant): Pass the mode paraeter before the rtx it describes
6585 and change its type to scalar_int_mode. Remove VOIDmode handling.
6586 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
6587 (iv_extend): Likewise.
6588 (shorten_into_mode): Likewise.
6589 (iv_add): Use scalar_int_mode.
6590 (iv_mult): Likewise.
6591 (iv_shift): Likewise.
6592 (canonicalize_iv_subregs): Likewise.
6593 (get_biv_step_1): Pass the outer_mode parameter before the rtx
6594 it describes and change its mode to scalar_int_mode. Also change
6595 the type of the returned inner_mode to scalar_int_mode.
6596 (get_biv_step): Likewise, turning outer_mode from a pointer
6597 into a direct parameter. Update call to get_biv_step_1.
6598 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
6599 iv_constant and get_biv_step.
6600 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
6601 and change its type to scalar_int_mode. Don't initialise iv->mode
6602 to VOIDmode and remove later checks for its still being VOIDmode.
6603 Update calls to iv_analyze_op and iv_analyze_expr. Check
6604 is_a <scalar_int_mode> when changing the mode under consideration.
6605 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
6606 Update call to iv_analyze_expr.
6607 (iv_analyze_op): Add a mode parameter. Reject subregs whose
6608 inner register is not also a scalar_int_mode. Update call to
6610 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
6611 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
6612 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
6613 separate mode class checks. Update calls to iv_analyze. Remove
6614 fix-up of VOIDmodes after iv_analyze_biv.
6615 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
6616 don't have a scalar_int_mode. Update call to biv_p.
6618 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6619 Alan Hayward <alan.hayward@arm.com>
6620 David Sherwood <david.sherwood@arm.com>
6622 * cfgexpand.c (convert_debug_memory_address): Use
6623 as_a <scalar_int_mode>.
6624 * combine.c (expand_compound_operation): Likewise.
6625 (make_extraction): Likewise.
6626 (change_zero_ext): Likewise.
6627 (simplify_comparison): Likewise.
6628 * cse.c (cse_insn): Likewise.
6629 * dwarf2out.c (minmax_loc_descriptor): Likewise.
6630 (mem_loc_descriptor): Likewise.
6631 (loc_descriptor): Likewise.
6632 * expmed.c (init_expmed_one_mode): Likewise.
6633 (synth_mult): Likewise.
6634 (emit_store_flag_1): Likewise.
6635 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
6636 of a comparison with size.
6637 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
6638 (reduce_to_bit_field_precision): Likewise.
6639 * function.c (expand_function_end): Likewise.
6640 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
6641 * loop-doloop.c (doloop_modify): Likewise.
6642 * optabs.c (expand_binop): Likewise.
6643 (expand_unop): Likewise.
6644 (expand_copysign_absneg): Likewise.
6645 (prepare_cmp_insn): Likewise.
6646 (maybe_legitimize_operand): Likewise.
6647 * recog.c (const_scalar_int_operand): Likewise.
6648 * rtlanal.c (get_address_mode): Likewise.
6649 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6650 (simplify_cond_clz_ctz): Likewise.
6651 * tree-nested.c (get_nl_goto_field): Likewise.
6652 * tree.c (build_vector_type_for_mode): Likewise.
6653 * var-tracking.c (use_narrower_mode): Likewise.
6655 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6656 Alan Hayward <alan.hayward@arm.com>
6657 David Sherwood <david.sherwood@arm.com>
6659 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
6660 * builtins.c (expand_builtin_signbit): Use it.
6661 * cfgexpand.c (expand_debug_expr): Likewise.
6662 * dojump.c (do_jump): Likewise.
6663 (do_compare_and_jump): Likewise.
6664 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
6665 * expmed.c (make_tree): Likewise.
6666 * expr.c (expand_expr_real_2): Likewise.
6667 (expand_expr_real_1): Likewise.
6668 (try_casesi): Likewise.
6669 * fold-const-call.c (fold_const_call_ss): Likewise.
6670 * fold-const.c (unextend): Likewise.
6671 (extract_muldiv_1): Likewise.
6672 (fold_single_bit_test): Likewise.
6673 (native_encode_int): Likewise.
6674 (native_encode_string): Likewise.
6675 (native_interpret_int): Likewise.
6676 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
6677 * internal-fn.c (expand_addsub_overflow): Likewise.
6678 (expand_neg_overflow): Likewise.
6679 (expand_mul_overflow): Likewise.
6680 (expand_arith_overflow): Likewise.
6681 * match.pd: Likewise.
6682 * stor-layout.c (layout_type): Likewise.
6683 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
6684 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
6685 * tree-ssanames.c (get_range_info): Likewise.
6686 * tree-switch-conversion.c (array_value_type) Likewise.
6687 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
6688 (vect_recog_divmod_pattern): Likewise.
6689 (vect_recog_mixed_size_cond_pattern): Likewise.
6690 * tree-vrp.c (extract_range_basic): Likewise.
6691 (simplify_float_conversion_using_ranges): Likewise.
6692 * tree.c (int_fits_type_p): Likewise.
6693 * ubsan.c (instrument_bool_enum_load): Likewise.
6694 * varasm.c (mergeable_string_section): Likewise.
6695 (narrowing_initializer_constant_valid_p): Likewise.
6696 (output_constant): Likewise.
6698 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6699 Alan Hayward <alan.hayward@arm.com>
6700 David Sherwood <david.sherwood@arm.com>
6702 * machmode.h (NARROWEST_INT_MODE): New macro.
6703 * expr.c (alignment_for_piecewise_move): Use it instead of
6704 GET_CLASS_NARROWEST_MODE (MODE_INT).
6705 (push_block): Likewise.
6706 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
6708 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6710 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6711 Alan Hayward <alan.hayward@arm.com>
6712 David Sherwood <david.sherwood@arm.com>
6714 * postreload.c (move2add_valid_value_p): Change the type of the
6715 mode parameter to scalar_int_mode.
6716 (move2add_use_add2_insn): Add a mode parameter and use it instead
6718 (move2add_use_add3_insn): Likewise.
6719 (reload_cse_move2add): Update accordingly.
6721 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6722 Alan Hayward <alan.hayward@arm.com>
6723 David Sherwood <david.sherwood@arm.com>
6725 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
6727 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
6728 * optabs.c (expand_unop): Likewise.
6730 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6731 Alan Hayward <alan.hayward@arm.com>
6732 David Sherwood <david.sherwood@arm.com>
6734 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
6735 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
6736 (popcount_loc_descriptor): Likewise.
6737 (bswap_loc_descriptor): Likewise.
6738 (rotate_loc_descriptor): Likewise.
6739 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
6740 calling the functions above.
6742 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6743 Alan Hayward <alan.hayward@arm.com>
6744 David Sherwood <david.sherwood@arm.com>
6746 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
6748 (try_combine): Likewise.
6749 (simplify_if_then_else): Likewise.
6750 * cse.c (cse_insn): Likewise.
6751 * dwarf2out.c (mem_loc_descriptor): Likewise.
6752 * emit-rtl.c (gen_lowpart_common): Likewise.
6753 * simplify-rtx.c (simplify_truncation): Likewise.
6754 (simplify_binary_operation_1): Likewise.
6755 (simplify_const_relational_operation): Likewise.
6756 (simplify_ternary_operation): Likewise.
6757 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
6759 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6760 Alan Hayward <alan.hayward@arm.com>
6761 David Sherwood <david.sherwood@arm.com>
6763 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
6764 * reload.c (push_reload): Likewise.
6765 (find_reloads): Likewise.
6767 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6768 Alan Hayward <alan.hayward@arm.com>
6769 David Sherwood <david.sherwood@arm.com>
6771 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
6772 (make_compound_operation_int): Likewise.
6773 (change_zero_ext): Likewise.
6774 * expr.c (convert_move): Likewise.
6775 (convert_modes): Likewise.
6776 * fwprop.c (forward_propagate_subreg): Likewise.
6777 * loop-iv.c (get_biv_step_1): Likewise.
6778 * optabs.c (widen_operand): Likewise.
6779 * postreload.c (move2add_valid_value_p): Likewise.
6780 * recog.c (simplify_while_replacing): Likewise.
6781 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6782 (simplify_binary_operation_1): Likewise. Remove redundant
6783 mode equality check.
6785 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6786 Alan Hayward <alan.hayward@arm.com>
6787 David Sherwood <david.sherwood@arm.com>
6789 * combine.c (combine_simplify_rtx): Add checks for
6790 is_a <scalar_int_mode>.
6791 (simplify_if_then_else): Likewise.
6792 (make_field_assignment): Likewise.
6793 (simplify_comparison): Likewise.
6794 * ifcvt.c (noce_try_bitop): Likewise.
6795 * loop-invariant.c (canonicalize_address_mult): Likewise.
6796 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6798 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6799 Alan Hayward <alan.hayward@arm.com>
6800 David Sherwood <david.sherwood@arm.com>
6802 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
6803 is_a <scalar_int_mode> instead of != BLKmode.
6805 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6806 Alan Hayward <alan.hayward@arm.com>
6807 David Sherwood <david.sherwood@arm.com>
6809 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
6810 instead of != VOIDmode.
6811 * combine.c (if_then_else_cond): Likewise.
6812 (change_zero_ext): Likewise.
6813 * dwarf2out.c (mem_loc_descriptor): Likewise.
6814 (loc_descriptor): Likewise.
6815 * rtlanal.c (canonicalize_condition): Likewise.
6816 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
6818 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6819 Alan Hayward <alan.hayward@arm.com>
6820 David Sherwood <david.sherwood@arm.com>
6822 * simplify-rtx.c (simplify_binary_operation_1): Use
6823 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
6825 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6826 Alan Hayward <alan.hayward@arm.com>
6827 David Sherwood <david.sherwood@arm.com>
6829 * wide-int.h (int_traits<unsigned char>) New class.
6830 (int_traits<unsigned short>) Likewise.
6831 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
6832 Use GET_MODE_UNIT_PRECISION and remove redundant test for
6834 * combine.c (set_nonzero_bits_and_sign_copies): Use
6835 is_a <scalar_int_mode>.
6836 (find_split_point): Likewise.
6837 (combine_simplify_rtx): Likewise.
6838 (simplify_logical): Likewise.
6839 (expand_compound_operation): Likewise.
6840 (expand_field_assignment): Likewise.
6841 (make_compound_operation): Likewise.
6842 (extended_count): Likewise.
6843 (change_zero_ext): Likewise.
6844 (simplify_comparison): Likewise.
6845 * dwarf2out.c (scompare_loc_descriptor): Likewise.
6846 (ucompare_loc_descriptor): Likewise.
6847 (minmax_loc_descriptor): Likewise.
6848 (mem_loc_descriptor): Likewise.
6849 (loc_descriptor): Likewise.
6850 * expmed.c (init_expmed_one_mode): Likewise.
6851 * lra-constraints.c (lra_constraint_offset): Likewise.
6852 * optabs.c (prepare_libcall_arg): Likewise.
6853 * postreload.c (move2add_note_store): Likewise.
6854 * reload.c (operands_match_p): Likewise.
6855 * rtl.h (load_extend_op): Likewise.
6856 * rtlhooks.c (gen_lowpart_general): Likewise.
6857 * simplify-rtx.c (simplify_truncation): Likewise.
6858 (simplify_unary_operation_1): Likewise.
6859 (simplify_binary_operation_1): Likewise.
6860 (simplify_const_binary_operation): Likewise.
6861 (simplify_const_relational_operation): Likewise.
6862 (simplify_subreg): Likewise.
6863 * stor-layout.c (bitwise_mode_for_mode): Likewise.
6864 * var-tracking.c (adjust_mems): Likewise.
6865 (prepare_call_arguments): Likewise.
6867 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6868 Alan Hayward <alan.hayward@arm.com>
6869 David Sherwood <david.sherwood@arm.com>
6871 * machmode.h (is_int_mode): New fuction.
6872 * combine.c (find_split_point): Use it.
6873 (combine_simplify_rtx): Likewise.
6874 (simplify_if_then_else): Likewise.
6875 (simplify_set): Likewise.
6876 (simplify_shift_const_1): Likewise.
6877 (simplify_comparison): Likewise.
6878 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
6879 * cse.c (notreg_cost): Likewise.
6880 (cse_insn): Likewise.
6881 * cselib.c (cselib_lookup_1): Likewise.
6882 * dojump.c (do_jump_1): Likewise.
6883 (do_compare_rtx_and_jump): Likewise.
6884 * dse.c (get_call_args): Likewise.
6885 * dwarf2out.c (rtl_for_decl_init): Likewise.
6886 (native_encode_initializer): Likewise.
6887 * expmed.c (emit_store_flag_1): Likewise.
6888 (emit_store_flag): Likewise.
6889 * expr.c (convert_modes): Likewise.
6890 (store_field): Likewise.
6891 (expand_expr_real_1): Likewise.
6892 * fold-const.c (fold_read_from_constant_string): Likewise.
6893 * gimple-ssa-sprintf.c (get_format_string): Likewise.
6894 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
6895 * optabs.c (expand_binop): Likewise.
6896 (expand_unop): Likewise.
6897 (expand_abs_nojump): Likewise.
6898 (expand_one_cmpl_abs_nojump): Likewise.
6899 * simplify-rtx.c (mode_signbit_p): Likewise.
6900 (val_signbit_p): Likewise.
6901 (val_signbit_known_set_p): Likewise.
6902 (val_signbit_known_clear_p): Likewise.
6903 (simplify_relational_operation_1): Likewise.
6904 * tree.c (vector_type_mode): Likewise.
6906 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6907 Alan Hayward <alan.hayward@arm.com>
6908 David Sherwood <david.sherwood@arm.com>
6910 * machmode.h (smallest_mode_for_size): Fix formatting.
6911 (smallest_int_mode_for_size): New function.
6912 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
6913 instead of smallest_mode_for_size.
6914 * combine.c (make_extraction): Likewise.
6915 * config/arc/arc.c (arc_expand_movmem): Likewise.
6916 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
6917 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
6918 * config/s390/s390.c (s390_expand_insv): Likewise.
6919 * config/sparc/sparc.c (assign_int_registers): Likewise.
6920 * config/spu/spu.c (spu_function_value): Likewise.
6921 (spu_function_arg): Likewise.
6922 * coverage.c (get_gcov_type): Likewise.
6923 (get_gcov_unsigned_t): Likewise.
6924 * dse.c (find_shift_sequence): Likewise.
6925 * expmed.c (store_bit_field_1): Likewise.
6926 * expr.c (convert_move): Likewise.
6927 (store_field): Likewise.
6928 * internal-fn.c (expand_arith_overflow): Likewise.
6929 * optabs-query.c (get_best_extraction_insn): Likewise.
6930 * optabs.c (expand_twoval_binop_libfunc): Likewise.
6931 * stor-layout.c (layout_type): Likewise.
6932 (initialize_sizetypes): Likewise.
6933 * targhooks.c (default_get_mask_mode): Likewise.
6934 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
6936 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6937 Alan Hayward <alan.hayward@arm.com>
6938 David Sherwood <david.sherwood@arm.com>
6940 * machmode.h (opt_mode::else_blk): New function.
6941 (int_mode_for_mode): Declare.
6942 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
6943 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
6945 * cfgexpand.c (expand_debug_expr): Likewise.
6946 * combine.c (gen_lowpart_or_truncate): Likewise.
6947 (gen_lowpart_for_combine): Likewise.
6948 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
6949 * config/avr/avr.c (avr_to_int_mode): Likewise.
6950 (avr_out_plus_1): Likewise.
6951 (avr_out_plus): Likewise.
6952 (avr_out_round): Likewise.
6953 * config/i386/i386.c (ix86_split_to_parts): Likewise.
6954 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
6955 (s390_expand_vcond): Likewise.
6956 * config/spu/spu.c (spu_split_immediate): Likewise.
6957 (spu_expand_mov): Likewise.
6958 * dse.c (get_stored_val): Likewise.
6959 * expmed.c (store_bit_field_1): Likewise.
6960 (convert_extracted_bit_field): Use int_mode_for_mode instead of
6962 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
6963 (extract_low_bits): Likewise.
6964 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
6965 handling rather than repeating the check.
6966 (emit_group_store): Likewise.
6967 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
6968 * optabs.c (expand_absneg_bit): Likewise.
6969 (expand_copysign_absneg): Likewise.
6970 (expand_copysign_bit): Likewise.
6971 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
6972 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
6973 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
6974 * var-tracking.c (prepare_call_arguments): Likewise.
6975 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
6976 int_mode_for_mode instead of mode_for_size.
6977 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
6979 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6980 Alan Hayward <alan.hayward@arm.com>
6981 David Sherwood <david.sherwood@arm.com>
6983 * machmode.h (int_mode_for_size): New function.
6984 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
6985 instead of mode_for_size.
6986 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
6988 * combine.c (expand_field_assignment): Use int_mode_for_size
6989 instead of mode_for_size.
6990 (make_extraction): Likewise.
6991 (simplify_shift_const_1): Likewise.
6992 (simplify_comparison): Likewise.
6993 * dojump.c (do_jump): Likewise.
6994 * dwarf2out.c (mem_loc_descriptor): Likewise.
6995 * emit-rtl.c (init_derived_machine_modes): Likewise.
6996 * expmed.c (flip_storage_order): Likewise.
6997 (convert_extracted_bit_field): Likewise.
6998 * expr.c (copy_blkmode_from_reg): Likewise.
6999 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
7000 * internal-fn.c (expand_mul_overflow): Likewise.
7001 * lower-subreg.c (simple_move): Likewise.
7002 * optabs-libfuncs.c (init_optabs): Likewise.
7003 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7004 * tree.c (vector_type_mode): Likewise.
7005 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
7006 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
7007 * tree-vect-generic.c (expand_vector_parallel): Likewise.
7008 * tree-vect-stmts.c (vectorizable_load): Likewise.
7009 (vectorizable_store): Likewise.
7011 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7012 Alan Hayward <alan.hayward@arm.com>
7013 David Sherwood <david.sherwood@arm.com>
7015 * coretypes.h (pod_mode): New type.
7016 (scalar_int_mode_pod): New typedef.
7017 * machmode.h (pod_mode): New class.
7018 (int_n_data_t::m): Change type to scalar_int_mode_pod.
7019 * genmodes.c (emit_mode_int_n): Update accordingly.
7020 * lower-subreg.h (target_lower_subreg): Change type to
7021 scalar_int_mode_pod.
7022 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
7023 scalar_int_mode_pod.
7025 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7026 Alan Hayward <alan.hayward@arm.com>
7027 David Sherwood <david.sherwood@arm.com>
7029 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
7030 machine_mode to scalar_int_mode.
7031 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
7032 (rs6000_option_override_internal): Remove cast to int.
7033 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
7034 machine_mode to scalar_int_mode.
7035 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
7036 (rs6000_option_override_internal): Remove cast to int.
7037 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
7038 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
7040 * config/s390/s390.c (s390_expand_builtin): Likewise.
7041 * coretypes.h (scalar_int_mode): New type.
7042 (opt_scalar_int_mode): New typedef.
7043 * machmode.h (scalar_int_mode): New class.
7044 (scalar_int_mode::includes_p): New function.
7045 (byte_mode): Change type to scalar_int_mode.
7046 (word_mode): Likewise.
7047 (ptr_mode): Likewise.
7048 * emit-rtl.c (byte_mode): Likewise.
7049 (word_mode): Likewise.
7050 (ptr_mode): Likewise.
7051 (init_derived_machine_modes): Update accordingly.
7052 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
7053 and MODE_PARTIAL_INT.
7054 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
7055 opt_scalar_int_mode.
7057 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7058 Alan Hayward <alan.hayward@arm.com>
7059 David Sherwood <david.sherwood@arm.com>
7061 * target.def (libgcc_floating_mode_supported_p): Take a
7063 * doc/tm.texi: Regenerate.
7064 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
7066 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
7067 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
7070 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7071 Alan Hayward <alan.hayward@arm.com>
7072 David Sherwood <david.sherwood@arm.com>
7074 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
7075 * doc/tm.texi: Regenerate.
7076 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
7077 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
7078 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
7079 * targhooks.h (default_floatn_mode): Likewise.
7080 * targhooks.c (default_floatn_mode): Likewise.
7081 * tree.c (build_common_tree_nodes): Update accordingly.
7083 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7084 Alan Hayward <alan.hayward@arm.com>
7085 David Sherwood <david.sherwood@arm.com>
7087 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
7088 (mode_iterator::iterate_p): Likewise.
7089 (mode_iterator::get_wider): Likewise.
7090 * expr.c (init_expr_target): Use opt_scalar_float_mode.
7092 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7093 Alan Hayward <alan.hayward@arm.com>
7094 David Sherwood <david.sherwood@arm.com>
7096 * coretypes.h (opt_scalar_float_mode): New typedef.
7097 * machmode.h (float_mode_for_size): New function.
7098 * emit-rtl.c (double_mode): Delete.
7099 (init_emit_once): Use float_mode_for_size.
7100 * stor-layout.c (layout_type): Likewise.
7101 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
7103 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7104 Alan Hayward <alan.hayward@arm.com>
7105 David Sherwood <david.sherwood@arm.com>
7107 * output.h (assemble_real): Take a scalar_float_mode.
7108 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
7109 * config/arm/arm.md (consttable_4): Likewise.
7110 (consttable_8): Likewise.
7111 (consttable_16): Likewise.
7112 * config/mips/mips.md (consttable_float): Likewise.
7113 * config/s390/s390.c (s390_output_pool_entry): Likewise.
7114 * varasm.c (assemble_real): Take a scalar_float_mode.
7115 (output_constant_pool_2): Update accordingly.
7116 (output_constant): Likewise.
7118 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7119 Alan Hayward <alan.hayward@arm.com>
7120 David Sherwood <david.sherwood@arm.com>
7122 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
7123 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
7124 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
7125 (native_encode_real): Likewise.
7126 (native_interpret_real): Likewise.
7127 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
7128 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7130 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7131 Alan Hayward <alan.hayward@arm.com>
7132 David Sherwood <david.sherwood@arm.com>
7134 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
7135 <scalar_float_mode>. Simplify.
7136 (gen_extend_conv_libfunc): Likewise.
7138 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7139 Alan Hayward <alan.hayward@arm.com>
7140 David Sherwood <david.sherwood@arm.com>
7142 * coretypes.h (scalar_float_mode): New type.
7143 * machmode.h (mode_traits::from_int): Use machine_mode if
7144 USE_ENUM_MODES is defined.
7145 (is_a): New function.
7147 (dyn_cast): Likewise.
7148 (scalar_float_mode): New class.
7149 (scalar_float_mode::includes_p): New function.
7150 (is_float_mode): Likewise.
7151 * gdbhooks.py (MachineModePrinter): New class.
7152 (build_pretty_printer): Use it for scalar_float_mode.
7153 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
7154 (format_helper::format_helper): Turn into a template.
7155 * genmodes.c (get_mode_class): New function.
7156 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
7157 or machine_mode if none.
7158 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
7159 as_a <scalar_float_mode>.
7160 * dwarf2out.c (mem_loc_descriptor): Likewise.
7161 (insert_float): Likewise.
7162 (add_const_value_attribute): Likewise.
7163 * simplify-rtx.c (simplify_immed_subreg): Likewise.
7164 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
7165 (expand_unop): Update accordingly.
7166 (expand_abs_nojump): Likewise.
7167 (expand_copysign_absneg): Take a scalar_float_mode.
7168 (expand_copysign_bit): Likewise.
7169 (expand_copysign): Update accordingly.
7171 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7172 Alan Hayward <alan.hayward@arm.com>
7173 David Sherwood <david.sherwood@arm.com>
7175 * coretypes.h (opt_mode): New class.
7176 * machmode.h (opt_mode): Likewise.
7177 (opt_mode::else_void): New function.
7178 (opt_mode::require): Likewise.
7179 (opt_mode::exists): Likewise.
7180 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
7181 (GET_MODE_2XWIDER_MODE): Likewise.
7182 (mode_iterator::get_wider): Update accordingly.
7183 (mode_iterator::get_2xwider): Likewise.
7184 (mode_iterator::get_known_wider): Likewise, turning into a template.
7185 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
7186 forcing a wider mode to exist.
7187 * config/cr16/cr16.h (LONG_REG_P): Likewise.
7188 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
7189 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
7190 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
7191 * lower-subreg.c (init_lower_subreg): Likewise.
7192 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
7193 on the final iteration.
7194 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
7195 a wider mode exists before asking for a move pattern.
7196 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
7197 forcing a wider mode to exist.
7198 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
7199 returning false if no such mode exists.
7200 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
7201 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
7202 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
7203 Avoid checking for a MODE_INT if we already know the mode is not a
7205 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
7206 forcing a wider mode to exist.
7207 (expmed_mult_highpart_optab): Likewise.
7208 (expmed_mult_highpart): Likewise.
7209 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
7211 * lto-streamer-in.c (lto_input_mode_table): Likewise.
7212 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
7213 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
7214 * internal-fn.c (expand_mul_overflow): Update use of
7215 GET_MODE_2XWIDER_MODE.
7216 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7217 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
7218 GET_MODE_WIDER_MODE.
7219 (convert_plusminus_to_widen): Likewise.
7220 * tree-switch-conversion.c (array_value_type): Likewise.
7221 * var-tracking.c (emit_note_insn_var_location): Likewise.
7222 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7223 Return false inside rather than outside the loop if no wider mode
7225 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
7226 and GET_MODE_2XWIDER_MODE
7227 (can_compare_p): Use else_void.
7228 * gdbhooks.py (OptMachineModePrinter): New class.
7229 (build_pretty_printer): Use it for opt_mode.
7231 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7232 Alan Hayward <alan.hayward@arm.com>
7233 David Sherwood <david.sherwood@arm.com>
7235 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
7236 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
7238 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7239 Alan Hayward <alan.hayward@arm.com>
7240 David Sherwood <david.sherwood@arm.com>
7242 * machmode.h (mode_traits): New structure.
7243 (get_narrowest_mode): New function.
7244 (mode_iterator::start): Likewise.
7245 (mode_iterator::iterate_p): Likewise.
7246 (mode_iterator::get_wider): Likewise.
7247 (mode_iterator::get_known_wider): Likewise.
7248 (mode_iterator::get_2xwider): Likewise.
7249 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
7250 (FOR_EACH_MODE): Likewise.
7251 (FOR_EACH_MODE_FROM): Likewise.
7252 (FOR_EACH_MODE_UNTIL): Likewise.
7253 (FOR_EACH_WIDER_MODE): Likewise.
7254 (FOR_EACH_2XWIDER_MODE): Likewise.
7255 * builtins.c (expand_builtin_strlen): Use new mode iterators.
7256 * combine.c (simplify_comparison): Likewise
7257 * config/i386/i386.c (type_natural_mode): Likewise.
7258 * cse.c (cse_insn): Likewise.
7259 * dse.c (find_shift_sequence): Likewise.
7260 * emit-rtl.c (init_derived_machine_modes): Likewise.
7261 (init_emit_once): Likewise.
7262 * explow.c (hard_function_value): Likewise.
7263 * expmed.c (extract_fixed_bit_field_1): Likewise.
7264 (extract_bit_field_1): Likewise.
7265 (expand_divmod): Likewise.
7266 (emit_store_flag_1): Likewise.
7267 * expr.c (init_expr_target): Likewise.
7268 (convert_move): Likewise.
7269 (alignment_for_piecewise_move): Likewise.
7270 (widest_int_mode_for_size): Likewise.
7271 (emit_block_move_via_movmem): Likewise.
7272 (copy_blkmode_to_reg): Likewise.
7273 (set_storage_via_setmem): Likewise.
7274 (compress_float_constant): Likewise.
7275 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7276 * optabs-query.c (get_best_extraction_insn): Likewise.
7277 * optabs.c (expand_binop): Likewise.
7278 (expand_twoval_unop): Likewise.
7279 (expand_twoval_binop): Likewise.
7280 (widen_leading): Likewise.
7281 (widen_bswap): Likewise.
7282 (expand_parity): Likewise.
7283 (expand_unop): Likewise.
7284 (prepare_cmp_insn): Likewise.
7285 (prepare_float_lib_cmp): Likewise.
7286 (expand_float): Likewise.
7287 (expand_fix): Likewise.
7288 (expand_sfix_optab): Likewise.
7289 * postreload.c (move2add_use_add2_insn): Likewise.
7290 * reg-stack.c (reg_to_stack): Likewise.
7291 * reginfo.c (choose_hard_reg_mode): Likewise.
7292 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
7293 * stor-layout.c (mode_for_size): Likewise.
7294 (smallest_mode_for_size): Likewise.
7295 (mode_for_vector): Likewise.
7296 (finish_bitfield_representative): Likewise.
7297 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
7298 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
7299 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7300 * var-tracking.c (prepare_call_arguments): Likewise.
7302 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7303 Alan Hayward <alan.hayward@arm.com>
7304 David Sherwood <david.sherwood@arm.com>
7306 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
7307 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
7308 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
7309 * machmode.h (mode_size): Move earlier in file.
7310 (mode_precision): Likewise.
7311 (mode_inner): Likewise.
7312 (mode_nunits): Likewise.
7313 (mode_unit_size): Likewise.
7314 (unit_unit_precision): Likewise.
7315 (mode_wider): Likewise.
7316 (mode_2xwider): Likewise.
7317 (machine_mode): New class.
7318 (mode_to_bytes): New function.
7319 (mode_to_bits): Likewise.
7320 (mode_to_precision): Likewise.
7321 (mode_to_inner): Likewise.
7322 (mode_to_unit_size): Likewise.
7323 (mode_to_unit_precision): Likewise.
7324 (mode_to_nunits): Likewise.
7325 (GET_MODE_SIZE): Use mode_to_bytes.
7326 (GET_MODE_BITSIZE): Use mode_to_bits.
7327 (GET_MODE_PRECISION): Use mode_to_precision.
7328 (GET_MODE_INNER): Use mode_to_inner.
7329 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
7330 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
7331 (GET_MODE_NUNITS): Use mode_to_nunits.
7332 * system.h (ALWAYS_INLINE): New macro.
7333 * config/powerpcspe/powerpcspe-c.c
7334 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
7335 int for arg1_mode and arg2_mode.
7337 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7338 Alan Hayward <alan.hayward@arm.com>
7339 David Sherwood <david.sherwood@arm.com>
7341 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
7342 Prefix mode names with E_ in case statements.
7343 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7344 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
7345 (aarch64_split_simd_move): Likewise.
7346 (aarch64_gen_storewb_pair): Likewise.
7347 (aarch64_gen_loadwb_pair): Likewise.
7348 (aarch64_gen_store_pair): Likewise.
7349 (aarch64_gen_load_pair): Likewise.
7350 (aarch64_get_condition_code_1): Likewise.
7351 (aarch64_constant_pool_reload_icode): Likewise.
7352 (get_rsqrte_type): Likewise.
7353 (get_rsqrts_type): Likewise.
7354 (get_recpe_type): Likewise.
7355 (get_recps_type): Likewise.
7356 (aarch64_gimplify_va_arg_expr): Likewise.
7357 (aarch64_simd_container_mode): Likewise.
7358 (aarch64_emit_load_exclusive): Likewise.
7359 (aarch64_emit_store_exclusive): Likewise.
7360 (aarch64_expand_compare_and_swap): Likewise.
7361 (aarch64_gen_atomic_cas): Likewise.
7362 (aarch64_emit_bic): Likewise.
7363 (aarch64_emit_atomic_swap): Likewise.
7364 (aarch64_emit_atomic_load_op): Likewise.
7365 (aarch64_evpc_trn): Likewise.
7366 (aarch64_evpc_uzp): Likewise.
7367 (aarch64_evpc_zip): Likewise.
7368 (aarch64_evpc_ext): Likewise.
7369 (aarch64_evpc_rev): Likewise.
7370 (aarch64_evpc_dup): Likewise.
7371 (aarch64_gen_ccmp_first): Likewise.
7372 (aarch64_gen_ccmp_next): Likewise.
7373 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
7374 (alpha_emit_xfloating_libcall): Likewise.
7375 (emit_insxl): Likewise.
7376 (alpha_arg_type): Likewise.
7377 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
7378 (arc_preferred_simd_mode): Likewise.
7379 (arc_secondary_reload): Likewise.
7380 (get_arc_condition_code): Likewise.
7381 (arc_print_operand): Likewise.
7382 (arc_legitimate_constant_p): Likewise.
7383 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7384 * config/arc/arc.md (casesi_load): Likewise.
7385 (casesi_compact_jump): Likewise.
7386 * config/arc/predicates.md (proper_comparison_operator): Likewise.
7387 (cc_use_register): Likewise.
7388 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7389 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
7390 (arm_init_iwmmxt_builtins): Likewise.
7391 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
7392 (neon_expand_vector_init): Likewise.
7393 (arm_attr_length_move_neon): Likewise.
7394 (maybe_get_arm_condition_code): Likewise.
7395 (arm_emit_vector_const): Likewise.
7396 (arm_preferred_simd_mode): Likewise.
7397 (arm_output_iwmmxt_tinsr): Likewise.
7398 (thumb1_output_casesi): Likewise.
7399 (thumb2_output_casesi): Likewise.
7400 (arm_emit_load_exclusive): Likewise.
7401 (arm_emit_store_exclusive): Likewise.
7402 (arm_expand_compare_and_swap): Likewise.
7403 (arm_evpc_neon_vuzp): Likewise.
7404 (arm_evpc_neon_vzip): Likewise.
7405 (arm_evpc_neon_vrev): Likewise.
7406 (arm_evpc_neon_vtrn): Likewise.
7407 (arm_evpc_neon_vext): Likewise.
7408 (arm_validize_comparison): Likewise.
7409 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
7410 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
7411 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
7412 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
7413 (c6x_preferred_simd_mode): Likewise.
7414 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
7415 (epiphany_rtx_costs): Likewise.
7416 * config/epiphany/predicates.md (proper_comparison_operator):
7418 * config/frv/frv.c (condexec_memory_operand): Likewise.
7419 (frv_emit_move): Likewise.
7420 (output_move_single): Likewise.
7421 (output_condmove_single): Likewise.
7422 (frv_hard_regno_mode_ok): Likewise.
7423 (frv_matching_accg_mode): Likewise.
7424 * config/h8300/h8300.c (split_adds_subs): Likewise.
7425 (h8300_rtx_costs): Likewise.
7426 (h8300_print_operand): Likewise.
7427 (compute_mov_length): Likewise.
7428 (output_logical_op): Likewise.
7429 (compute_logical_op_length): Likewise.
7430 (compute_logical_op_cc): Likewise.
7431 (h8300_shift_needs_scratch_p): Likewise.
7432 (output_a_shift): Likewise.
7433 (compute_a_shift_length): Likewise.
7434 (compute_a_shift_cc): Likewise.
7435 (expand_a_rotate): Likewise.
7436 (output_a_rotate): Likewise.
7437 * config/i386/i386.c (classify_argument): Likewise.
7438 (function_arg_advance_32): Likewise.
7439 (function_arg_32): Likewise.
7440 (function_arg_64): Likewise.
7441 (function_value_64): Likewise.
7442 (ix86_gimplify_va_arg): Likewise.
7443 (ix86_legitimate_constant_p): Likewise.
7444 (put_condition_code): Likewise.
7445 (split_double_mode): Likewise.
7446 (ix86_avx256_split_vector_move_misalign): Likewise.
7447 (ix86_expand_vector_logical_operator): Likewise.
7448 (ix86_split_idivmod): Likewise.
7449 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
7450 (ix86_build_const_vector): Likewise.
7451 (ix86_build_signbit_mask): Likewise.
7452 (ix86_match_ccmode): Likewise.
7453 (ix86_cc_modes_compatible): Likewise.
7454 (ix86_expand_branch): Likewise.
7455 (ix86_expand_sse_cmp): Likewise.
7456 (ix86_expand_sse_movcc): Likewise.
7457 (ix86_expand_int_sse_cmp): Likewise.
7458 (ix86_expand_vec_perm_vpermi2): Likewise.
7459 (ix86_expand_vec_perm): Likewise.
7460 (ix86_expand_sse_unpack): Likewise.
7461 (ix86_expand_int_addcc): Likewise.
7462 (ix86_split_to_parts): Likewise.
7463 (ix86_vectorize_builtin_gather): Likewise.
7464 (ix86_vectorize_builtin_scatter): Likewise.
7465 (avx_vpermilp_parallel): Likewise.
7466 (inline_memory_move_cost): Likewise.
7467 (ix86_tieable_integer_mode_p): Likewise.
7468 (x86_maybe_negate_const_int): Likewise.
7469 (ix86_expand_vector_init_duplicate): Likewise.
7470 (ix86_expand_vector_init_one_nonzero): Likewise.
7471 (ix86_expand_vector_init_one_var): Likewise.
7472 (ix86_expand_vector_init_concat): Likewise.
7473 (ix86_expand_vector_init_interleave): Likewise.
7474 (ix86_expand_vector_init_general): Likewise.
7475 (ix86_expand_vector_set): Likewise.
7476 (ix86_expand_vector_extract): Likewise.
7477 (emit_reduc_half): Likewise.
7478 (ix86_emit_i387_round): Likewise.
7479 (ix86_mangle_type): Likewise.
7480 (ix86_expand_round_sse4): Likewise.
7481 (expand_vec_perm_blend): Likewise.
7482 (canonicalize_vector_int_perm): Likewise.
7483 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
7484 (expand_vec_perm_1): Likewise.
7485 (expand_vec_perm_interleave3): Likewise.
7486 (expand_vec_perm_even_odd_pack): Likewise.
7487 (expand_vec_perm_even_odd_1): Likewise.
7488 (expand_vec_perm_broadcast_1): Likewise.
7489 (ix86_vectorize_vec_perm_const_ok): Likewise.
7490 (ix86_expand_vecop_qihi): Likewise.
7491 (ix86_expand_mul_widen_hilo): Likewise.
7492 (ix86_expand_sse2_abs): Likewise.
7493 (ix86_expand_pextr): Likewise.
7494 (ix86_expand_pinsr): Likewise.
7495 (ix86_preferred_simd_mode): Likewise.
7496 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
7497 * config/i386/sse.md (*andnot<mode>3): Likewise.
7498 (<mask_codefor><code><mode>3<mask_name>): Likewise.
7499 (*<code><mode>3): Likewise.
7500 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
7501 (ia64_expand_atomic_op): Likewise.
7502 (ia64_arg_type): Likewise.
7503 (ia64_mode_to_int): Likewise.
7504 (ia64_scalar_mode_supported_p): Likewise.
7505 (ia64_vector_mode_supported_p): Likewise.
7506 (expand_vec_perm_broadcast): Likewise.
7507 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
7508 (iq2000_function_arg_advance): Likewise.
7509 (iq2000_function_arg): Likewise.
7510 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
7511 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
7512 (m68k_libcall_value): Likewise.
7513 (m68k_function_value): Likewise.
7514 (sched_attr_op_type): Likewise.
7515 * config/mcore/mcore.c (mcore_output_move): Likewise.
7516 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
7518 (microblaze_function_arg): Likewise.
7519 * config/mips/mips.c (mips16_build_call_stub): Likewise.
7520 (mips_print_operand): Likewise.
7521 (mips_mode_ok_for_mov_fmt_p): Likewise.
7522 (mips_vector_mode_supported_p): Likewise.
7523 (mips_preferred_simd_mode): Likewise.
7524 (mips_expand_vpc_loongson_even_odd): Likewise.
7525 (mips_expand_vec_unpack): Likewise.
7526 (mips_expand_vi_broadcast): Likewise.
7527 (mips_expand_vector_init): Likewise.
7528 (mips_expand_vec_reduc): Likewise.
7529 (mips_expand_msa_cmp): Likewise.
7530 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
7531 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
7532 (cc_flags_for_mode): Likewise.
7533 * config/msp430/msp430.c (msp430_print_operand): Likewise.
7534 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
7535 (nds32_output_casesi_pc_relative): Likewise.
7536 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7537 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
7538 (nvptx_gen_unpack): Likewise.
7539 (nvptx_gen_pack): Likewise.
7540 (nvptx_gen_shuffle): Likewise.
7541 (nvptx_gen_wcast): Likewise.
7542 (nvptx_preferred_simd_mode): Likewise.
7543 * config/pa/pa.c (pa_secondary_reload): Likewise.
7544 * config/pa/predicates.md (base14_operand): Likewise.
7545 * config/powerpcspe/powerpcspe-c.c
7546 (altivec_resolve_overloaded_builtin): Likewise.
7547 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
7549 (rs6000_preferred_simd_mode): Likewise.
7550 (output_vec_const_move): Likewise.
7551 (rs6000_expand_vector_extract): Likewise.
7552 (rs6000_split_vec_extract_var): Likewise.
7553 (reg_offset_addressing_ok_p): Likewise.
7554 (rs6000_legitimate_offset_address_p): Likewise.
7555 (rs6000_legitimize_address): Likewise.
7556 (rs6000_emit_set_const): Likewise.
7557 (rs6000_const_vec): Likewise.
7558 (rs6000_emit_move): Likewise.
7559 (spe_build_register_parallel): Likewise.
7560 (rs6000_darwin64_record_arg_recurse): Likewise.
7561 (swap_selector_for_mode): Likewise.
7562 (spe_init_builtins): Likewise.
7563 (paired_init_builtins): Likewise.
7564 (altivec_init_builtins): Likewise.
7565 (do_load_for_compare): Likewise.
7566 (rs6000_generate_compare): Likewise.
7567 (rs6000_expand_float128_convert): Likewise.
7568 (emit_load_locked): Likewise.
7569 (emit_store_conditional): Likewise.
7570 (rs6000_output_function_epilogue): Likewise.
7571 (rs6000_handle_altivec_attribute): Likewise.
7572 (rs6000_function_value): Likewise.
7573 (emit_fusion_gpr_load): Likewise.
7574 (emit_fusion_p9_load): Likewise.
7575 (emit_fusion_p9_store): Likewise.
7576 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
7577 (fusion_gpr_mem_load): Likewise.
7578 (fusion_addis_mem_combo_load): Likewise.
7579 (fusion_addis_mem_combo_store): Likewise.
7580 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
7581 (fusion_gpr_mem_load): Likewise.
7582 (fusion_addis_mem_combo_load): Likewise.
7583 (fusion_addis_mem_combo_store): Likewise.
7584 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7586 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
7587 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
7588 (rs6000_preferred_simd_mode): Likewise.
7589 (output_vec_const_move): Likewise.
7590 (rs6000_expand_vector_extract): Likewise.
7591 (rs6000_split_vec_extract_var): Likewise.
7592 (reg_offset_addressing_ok_p): Likewise.
7593 (rs6000_legitimate_offset_address_p): Likewise.
7594 (rs6000_legitimize_address): Likewise.
7595 (rs6000_emit_set_const): Likewise.
7596 (rs6000_const_vec): Likewise.
7597 (rs6000_emit_move): Likewise.
7598 (rs6000_darwin64_record_arg_recurse): Likewise.
7599 (swap_selector_for_mode): Likewise.
7600 (paired_init_builtins): Likewise.
7601 (altivec_init_builtins): Likewise.
7602 (rs6000_expand_float128_convert): Likewise.
7603 (emit_load_locked): Likewise.
7604 (emit_store_conditional): Likewise.
7605 (rs6000_output_function_epilogue): Likewise.
7606 (rs6000_handle_altivec_attribute): Likewise.
7607 (rs6000_function_value): Likewise.
7608 (emit_fusion_gpr_load): Likewise.
7609 (emit_fusion_p9_load): Likewise.
7610 (emit_fusion_p9_store): Likewise.
7611 * config/rx/rx.c (rx_gen_move_template): Likewise.
7612 (flags_from_mode): Likewise.
7613 * config/s390/predicates.md (s390_alc_comparison): Likewise.
7614 (s390_slb_comparison): Likewise.
7615 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
7616 (s390_vector_mode_supported_p): Likewise.
7617 (s390_cc_modes_compatible): Likewise.
7618 (s390_match_ccmode_set): Likewise.
7619 (s390_canonicalize_comparison): Likewise.
7620 (s390_emit_compare_and_swap): Likewise.
7621 (s390_branch_condition_mask): Likewise.
7622 (s390_rtx_costs): Likewise.
7623 (s390_secondary_reload): Likewise.
7624 (__SECONDARY_RELOAD_CASE): Likewise.
7625 (s390_expand_cs): Likewise.
7626 (s390_preferred_simd_mode): Likewise.
7627 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
7628 * config/sh/sh.c (sh_print_operand): Likewise.
7629 (dump_table): Likewise.
7630 (sh_secondary_reload): Likewise.
7631 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7632 * config/sh/sh.md (casesi_worker_1): Likewise.
7633 (casesi_worker_2): Likewise.
7634 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
7635 (fcc_comparison_operator): Likewise.
7636 * config/sparc/sparc.c (sparc_expand_move): Likewise.
7637 (emit_soft_tfmode_cvt): Likewise.
7638 (sparc_preferred_simd_mode): Likewise.
7639 (output_cbranch): Likewise.
7640 (sparc_print_operand): Likewise.
7641 (sparc_expand_vec_perm_bmask): Likewise.
7642 (vector_init_bshuffle): Likewise.
7643 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
7644 (spu_vector_mode_supported_p): Likewise.
7645 (spu_expand_insv): Likewise.
7646 (spu_emit_branch_or_set): Likewise.
7647 (spu_handle_vector_attribute): Likewise.
7648 (spu_builtin_splats): Likewise.
7649 (spu_builtin_extract): Likewise.
7650 (spu_builtin_promote): Likewise.
7651 (spu_expand_sign_extend): Likewise.
7652 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
7653 (tilegx_simd_int): Likewise.
7654 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
7655 (tilepro_simd_int): Likewise.
7656 * config/v850/v850.c (const_double_split): Likewise.
7657 (v850_print_operand): Likewise.
7658 (ep_memory_offset): Likewise.
7659 * config/vax/vax.c (vax_rtx_costs): Likewise.
7660 (vax_output_int_move): Likewise.
7661 (vax_output_int_add): Likewise.
7662 (vax_output_int_subtract): Likewise.
7663 * config/visium/predicates.md (visium_branch_operator): Likewise.
7664 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
7665 (visium_print_operand_address): Likewise.
7666 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7667 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
7668 (xtensa_expand_conditional_branch): Likewise.
7669 (xtensa_copy_incoming_a7): Likewise.
7670 (xtensa_output_literal): Likewise.
7671 * dfp.c (decimal_real_maxval): Likewise.
7672 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
7674 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7675 Alan Hayward <alan.hayward@arm.com>
7676 David Sherwood <david.sherwood@arm.com>
7678 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
7679 (mode_nunits_inline): Likewise.
7680 (mode_inner_inline): Likewise.
7681 (mode_unit_size_inline): Likewise.
7682 (mode_unit_precision_inline): Likewise.
7683 (emit_insn_modes_h): Likewise. Also emit a #define of the
7685 (emit_mode_wider): Add an E_ prefix to mode names.
7686 (emit_mode_complex): Likewise.
7687 (emit_mode_inner): Likewise.
7688 (emit_mode_adjustments): Likewise.
7689 (emit_mode_int_n): Likewise.
7690 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
7691 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
7692 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
7693 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
7694 (CRC32_BUILTIN, ENTRY): Likewise.
7695 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
7696 (aarch64_pop_regs): Likewise.
7697 (aarch64_process_components): Likewise.
7698 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
7699 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
7700 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
7701 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
7702 * config/arm/arm.c (arm_init_libfuncs): Likewise.
7703 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
7705 * config/i386/i386-builtin.def (pcmpestr): Likewise.
7706 (pcmpistr): Likewise.
7707 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
7708 * config/mmix/mmix.c (mmix_output_condition): Likewise.
7709 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
7711 * config/rl78/rl78.c (mduc_regs): Likewise.
7712 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
7713 (htm_expand_builtin): Likewise.
7714 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
7715 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
7716 * config/xtensa/xtensa.c (print_operand): Likewise.
7717 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
7718 (NUM_MODE_VECTOR_INT): Likewise.
7719 * genoutput.c (null_operand): Likewise.
7720 (output_operand_data): Likewise.
7721 * genrecog.c (print_parameter_value): Likewise.
7722 * lra.c (debug_operand_data): Likewise.
7724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7726 * dwarf2out.c (scompare_loc_descriptor_wide)
7727 (scompare_loc_descriptor_narrow): New functions, split out from...
7728 (scompare_loc_descriptor): ...here.
7729 * expmed.c (emit_store_flag_int): New function, split out from...
7730 (emit_store_flag): ...here.
7732 2017-08-30 Richard Biener <rguenther@suse.de>
7734 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
7735 (dwarf2out_early_finish): Move setting of AT_pubnames from
7736 early debug output to early finish.
7738 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
7740 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
7741 and -mdata-region to the assembler.
7743 2017-08-30 Richard Biener <rguenther@suse.de>
7745 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
7747 (gen_subprogram_die): Add DW_AT_object_pointer only early.
7748 (dwarf2out_early_global_decl): Only generate a DIE for the
7749 abstract origin if it doesn't already exist or is a declaration DIE.
7750 (resolve_addr): Do not add the linkage name twice when
7751 generating a stub DIE for the DW_TAG_GNU_call_site target.
7753 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7755 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7756 Use machine_mode rather than int for arg1_mode.
7758 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
7761 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
7762 that the second argument of the built-in functions to unpack
7763 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
7764 switch statement instead a lot of if statements.
7765 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
7766 Allow 64-bit values to be in Altivec registers as well as
7767 traditional floating point registers.
7768 (pack<mode>, FMOVE128_VSX iterator): Likewise.
7770 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
7772 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
7773 MAX_REGS_PER_ADDRESS == 1.
7775 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
7777 * config/i386/i386.opt (flag_fentry): Do not init to -1.
7778 * config/i386/i386.c (ix86_option_override_internal): Simplify
7779 setting of opts->x_flag_entry.
7781 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7782 Jakub Jelinek <jakub@redhat.com>
7783 Richard Biener <rguenther@suse.de>
7785 PR tree-optimization/81503
7786 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
7787 folded constant fits in the target type; reorder tests for clarity.
7789 2017-08-29 Martin Liska <mliska@suse.cz>
7791 * passes.def: Include pass_lower_switch.
7792 * stmt.c (dump_case_nodes): Remove and move to
7793 tree-switch-conversion.
7794 (case_values_threshold): Likewise.
7795 (expand_switch_as_decision_tree_p): Likewise.
7796 (emit_case_decision_tree): Likewise.
7797 (expand_case): Likewise.
7798 (balance_case_nodes): Likewise.
7799 (node_has_low_bound): Likewise.
7800 (node_has_high_bound): Likewise.
7801 (node_is_bounded): Likewise.
7802 (emit_case_nodes): Likewise.
7803 (struct simple_case_node): New struct.
7804 (add_case_node): Remove.
7805 (emit_case_dispatch_table): Use vector instead of case_list.
7806 (reset_out_edges_aux): Remove.
7807 (compute_cases_per_edge): Likewise.
7808 (expand_case): Build list of simple_case_node.
7809 (expand_sjlj_dispatch_table): Use it.
7810 * tree-switch-conversion.c (struct case_node): Moved from
7811 stmt.c and adjusted.
7812 (emit_case_nodes): Likewise.
7813 (node_has_low_bound): Likewise.
7814 (node_has_high_bound): Likewise.
7815 (node_is_bounded): Likewise.
7816 (case_values_threshold): Likewise.
7817 (reset_out_edges_aux): Likewise.
7818 (compute_cases_per_edge): Likewise.
7819 (add_case_node): Likewise.
7820 (dump_case_nodes): Likewise.
7821 (balance_case_nodes): Likewise.
7822 (expand_switch_as_decision_tree_p): Likewise.
7823 (emit_jump): Likewise.
7824 (emit_case_decision_tree): Likewise.
7825 (try_switch_expansion): Likewise.
7826 (do_jump_if_equal): Likewise.
7827 (emit_cmp_and_jump_insns): Likewise.
7828 (fix_phi_operands_for_edge): New function.
7829 (record_phi_operand_mapping): Likewise.
7830 (class pass_lower_switch): New pass.
7831 (pass_lower_switch::execute): New function.
7832 (make_pass_lower_switch): Likewise.
7833 (conditional_probability):
7834 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
7835 * tree-pass.h: Add make_pass_lower_switch.
7837 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
7840 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
7843 2017-08-29 Richard Biener <rguenther@suse.de>
7845 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
7846 we do not add a DW_AT_inline attribute twice.
7847 (gen_subprogram_die): Remove code setting DW_AT_inline on
7848 DECL_ABSTRACT_P nodes.
7850 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
7852 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
7853 calls to internal functions.
7854 (gimplify_modify_expr): Likewise.
7855 * tree-call-cdce.c (use_internal_fn): Likewise.
7856 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
7857 (convert_to_divmod): Set the nothrow flag.
7858 * tree-if-conv.c (predicate_mem_writes): Likewise.
7859 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7860 (vectorizable_call): Likewise.
7861 (vectorizable_store): Likewise.
7862 (vectorizable_load): Likewise.
7863 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
7864 (vect_recog_mask_conversion_pattern): Likewise.
7866 2017-08-29 Martin Liska <mliska@suse.cz>
7869 * gcc.c (driver_handle_option): Add new argument.
7870 * opts-common.c (handle_option): Pass
7871 target_option_override_hook.
7872 * opts-global.c (lang_handle_option): Add new option.
7873 (set_default_handlers): Add new argument.
7874 (decode_options): Likewise.
7875 * opts.c (target_handle_option): Likewise.
7876 (common_handle_option): Call target_option_override_hook.
7877 * opts.h (struct cl_option_handler_func): Add hook for
7878 target option override.
7879 (struct cl_option_handlers): Likewise.
7880 (set_default_handlers): Add new argument.
7881 (decode_options): Likewise.
7882 (common_handle_option): Likewise.
7883 (target_handle_option): Likewise.
7884 * toplev.c (toplev::main): Pass targetm.target_option.override
7887 2017-08-29 Richard Biener <rguenther@suse.de>
7888 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7890 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
7891 life to the active subtree.
7893 2017-08-28 Jeff Law <law@redhat.com>
7895 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
7896 derive_equivalences.
7897 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
7899 (edge_info::derive_equivalences): New private member function
7901 * tree-ssa-dom.c (class edge_info): Changed from a struct
7902 to a class. Add ctor/dtor, methods and data members.
7903 (edge_info::edge_info): Renamed from allocate_edge_info.
7904 Initialize additional members.
7905 (edge_info::~edge_info): New.
7906 (free_dom_edge_info): Delete the edge info.
7907 (record_edge_info): Use new class & associated member functions.
7908 Tighten forms for testing for edge equivalences.
7909 (record_temporary_equivalences): Iterate over the simple
7910 equivalences rather than assuming there's only one per edge.
7911 (cprop_into_successor_phis): Iterate over the simple
7912 equivalences rather than assuming there's only one per edge.
7913 (optimize_stmt): Use operand_equal_p rather than pointer
7914 equality for mini-DSE code.
7916 2017-08-28 Nathan Sidwell <nathan@acm.org>
7918 * gcc.c (execute): Fold SIGPIPE handling into switch
7919 statement. Adjust internal error message.
7921 2017-08-28 Richard Biener <rguenther@suse.de>
7924 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
7925 Do nothing for removed DIEs.
7927 2017-08-28 Richard Biener <rguenther@suse.de>
7929 PR tree-optimization/81977
7930 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
7933 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
7936 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
7937 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
7938 using targetm.gen_mem_thread_fence.
7940 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
7943 * config/i386/i386.md (*<btsc><mode>): Change operand 2
7944 predicate to register_operand. Reorder operands.
7945 (*btr<mode>): Ditto.
7946 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
7947 (*btr<mode>_mask): Ditto.
7949 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
7951 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
7952 * config/rs6000/xmmintrin.h: New file.
7953 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
7955 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7958 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
7959 parameter and_insn and return it.
7960 (recombine_lvx_pattern): Insert a copy to ensure availability of
7961 the base register of the copied masking operation at the point of
7962 the instruction replacement.
7963 (recombine_stvx_pattern): Likewise.
7965 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
7967 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
7968 undocumented switches.
7969 (-mpower9-dform-vector): Likewise.
7970 (-mpower9-dform): Likewise.
7971 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
7972 comments to delete references to -mpower9-dform* switches.
7973 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
7974 Delete reference to -mpower9-dform* switches, test for
7975 -mpower9-vector instead.
7976 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
7977 (OTHER_P9_VECTOR_MASKS): Likewise.
7978 (POWERPC_MASKS): Likewise.
7979 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
7980 tests against -mpower9-dform* to -mpower9-vector. Delete code
7981 that checked for -mpower9-dform* consistancy with other options.
7982 Add test for -mpower9-misc to enable other power9 switches.
7983 (rs6000_init_hard_regno_mode_ok): Likewise.
7984 (rs6000_option_override_internal): Likewise.
7985 (rs6000_emit_prologue): Likewise.
7986 (rs6000_emit_epilogue): Likewise.
7987 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
7988 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
7989 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
7991 (emit_fusion_p9_store): Likewise.
7992 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
7993 resetting these macros if the assembler does not support ISA 3.0
7995 (TARGET_P9_DFORM_VECTOR): Likewise.
7996 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
7997 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
7999 2017-08-25 Alan Modra <amodra@gmail.com>
8001 PR rtl-optimization/81747
8002 * cse.c (cse_extended_basic_block): Don't attempt to record
8003 equivalences for degenerate conditional branches that branch
8004 to their fall-through.
8006 2017-08-24 Martin Sebor <msebor@redhat.com>
8009 * gimple-fold.c (size_must_be_zero_p): New function.
8010 (gimple_fold_builtin_memory_op): Call it.
8012 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
8014 * config/rs6000/mm_malloc.h: New file.
8016 2017-08-24 Bin Cheng <bin.cheng@arm.com>
8018 PR tree-optimization/81913
8019 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
8020 analysis when either IVs in condition can wrap.
8022 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
8024 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
8025 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
8027 2017-08-24 Richard Biener <rguenther@suse.de>
8030 * targhooks.c (default_target_can_inline_p): Properly
8031 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
8032 is present and always compare.
8033 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
8034 infer -mfpmath=sse from TARGET_SSE_P.
8035 (ix86_can_inline_p): Properly use target_option_default_node when
8036 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
8038 2017-08-24 Richard Biener <rguenther@suse.de>
8041 * dwarf2out.c (output_die): Handle flag_generate_offload like
8043 (output_comp_unit): Likewise.
8044 (gen_array_type_die): Likewise.
8045 (dwarf2out_early_finish): Likewise.
8046 (note_variable_value_in_expr): Likewise.
8047 (dwarf2out_finish): Likewise. Adjust assert.
8048 * cgraphunit.c (symbol_table::compile): Move setting of
8049 flag_generate_offload earlier ...
8050 (symbol_table::finalize_compilation_unit): ... here, before
8051 early debug finalization.
8053 2017-08-24 Richard Biener <rguenther@suse.de>
8055 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
8056 and ipa-fnsummary.h.
8057 (ix86_can_inline_p): When ix86_fpmath flags do not match
8058 check whether the callee uses FP math at all.
8060 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
8063 * tree-ssanames.c (get_nonzero_bits): Use element_precision
8064 instead of TYPE_PRECISION.
8066 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
8067 Alan Hayward <alan.hayward@arm.com>
8068 David Sherwood <david.sherwood@arm.com>
8070 * combine.c (make_extraction): Use subreg_offset_from_lsb.
8072 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
8074 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
8076 (ix86_frame::stack_realign_allocate): New field.
8077 (struct machine_frame_state): Modify comments.
8078 (machine_frame_state::sp_realigned_fp_end): New field.
8079 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
8081 (sp_valid_at): Add assertion to assure no attempt to access invalid
8082 offset of a realigned stack.
8083 (fp_valid_at): Likewise.
8084 (choose_baseaddr): Modify comments.
8085 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
8086 ix86_expand_prologue.
8087 (ix86_expand_prologue): Modify stack realignment and allocation.
8088 (ix86_expand_epilogue): Modify comments.
8089 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
8090 avx2_runtime, avx512f, and avx512f_runtime.
8092 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
8094 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
8095 (mstackrealign): Do not init to -1.
8096 * config/i386/i386.c (ix86_option_override_internal):
8097 Check opts_set, not opts when setting default value of
8098 opts->x_ix86_force_align_arg_pointer.
8100 2017-08-23 Richard Biener <rguenther@suse.de>
8102 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
8103 lang_hooks.decl_printable_name.
8104 * print-rtl-function.c (print_rtx_function): Likewise.
8105 * tree-pretty-print.c (dump_function_header): Likewise.
8107 2017-08-23 Richard Biener <rguenther@suse.de>
8110 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
8111 -g0 at compile-time.
8113 2017-08-23 Tamar Christina <tamar.christina@arm.com>
8116 * doc/sourcebuild.texi (Other hardware attributes):
8119 2017-08-23 Tamar Christina <tamar.christina@arm.com>
8122 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
8123 Fix single-use check.
8125 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
8127 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
8129 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
8131 * doc/install.texi: Modify to add more details on running selected
8134 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
8136 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
8137 is combined with -mabi=ms.
8138 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
8141 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8143 PR tree-optimization/81488
8144 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
8145 and cached_basis fields.
8146 (MAX_SPREAD): New constant.
8147 (alloc_cand_and_find_basis): Initialize new fields.
8148 (clear_visited): New function.
8149 (create_phi_basis_1): Rename from create_phi_basis, set visited
8150 and cached_basis fields.
8151 (create_phi_basis): New wrapper function.
8152 (phi_add_costs_1): Rename from phi_add_costs, add spread
8153 parameter, set visited field, short-circuit when limits reached.
8154 (phi_add_costs): New wrapper function.
8155 (record_phi_increments_1): Rename from record_phi_increments, set
8157 (record_phi_increments): New wrapper function.
8158 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
8159 (phi_incr_cost): New wrapper function.
8160 (all_phi_incrs_profitable_1): Rename from
8161 all_phi_incrs_profitable, set visited field.
8162 (all_phi_incrs_profitable): New wrapper function.
8164 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
8165 Alan Hayward <alan.hayward@arm.com>
8166 David Sherwood <david.sherwood@arm.com>
8168 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
8169 that takes the outer and inner modes.
8170 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
8171 comparison as the canonical test for a paradoxical subreg.
8172 * combine.c (simplify_set): Use paradoxical_subreg_p.
8173 (make_extraction): Likewise.
8174 (force_to_mode): Likewise.
8175 (rtx_equal_for_field_assignment_p): Likewise.
8176 (gen_lowpart_for_combine): Likewise.
8177 (simplify_comparison): Likewise.
8178 * cse.c (equiv_constant): Likewise.
8179 * expmed.c (store_bit_field_1): Likewise.
8180 * final.c (alter_subreg): Likewise.
8181 * fwprop.c (propagate_rtx): Likewise.
8182 (forward_propagate_subreg): Likewise.
8183 * ira-conflicts.c (ira_build_conflicts): Likewise.
8184 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
8185 * lra-constraints.c (curr_insn_transform): Likewise.
8186 (split_reg): Likewise.
8187 * lra-eliminations.c (move_plus_up): Likewise.
8188 (lra_eliminate_regs_1): Likewise.
8189 * recog.c (general_operand): Likewise.
8190 * ree.c (combine_reaching_defs): Likewise.
8191 * reload.c (push_reload): Likewise.
8192 (find_reloads): Likewise.
8193 * reload1.c (elimination_effects): Likewise.
8194 (compute_reload_subreg_offset): Likewise.
8195 (choose_reload_regs): Likewise.
8196 * rtlanal.c (subreg_lsb_1): Likewise.
8197 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8198 (simplify_subreg): Likewise.
8199 * var-tracking.c (track_loc_p): Likewise.
8200 * emit-rtl.c (byte_lowpart_offset): Likewise.
8201 (paradoxical_subreg_p): Delete out-of-line definition.
8203 2017-08-22 Jeff Law <law@redhat.com>
8205 PR tree-optimization/81741
8206 PR tree-optimization/71947
8207 * tree-ssa-dom.c: Include tree-inline.h.
8208 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
8209 equivalences if one is more expensive to compute than the other.
8210 * tree-ssa-scopedtables.h (class const_or_copies): Make
8211 record_const_or_copy_raw method private.
8212 (class avail_exprs_stack): New method simplify_binary_operation.
8213 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
8214 avail_exprs_stack::simplify_binary_operation as needed.
8215 (avail_exprs_stack::simplify_binary_operation): New function.
8217 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
8219 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
8220 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
8221 (DOT_SYMBOLS): Likewise.
8222 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
8223 (RELOCATABLE_NEEDS_FIXUP): Likewise.
8224 (RS6000_ABI_NAME): Likewise.
8225 (TARGET_CMODEL): Likewise.
8226 (TOC_SECTION_ASM_OP): Likewise.
8227 (SET_CMODEL): New macro.
8228 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
8230 2017-08-22 Richard Biener <rguenther@suse.de>
8232 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
8233 to allow for free-lang-data replacements similar to verify_type_variant.
8235 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
8236 Alan Hayward <alan.hayward@arm.com>
8237 David Sherwood <david.sherwood@arm.com>
8239 * config/aarch64/aarch64.md (casesi): Use DImode rather than
8240 VOIDmode for the LABEL_REF.
8242 2017-08-22 Richard Biener <rguenther@suse.de>
8244 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
8246 2017-08-22 Richard Biener <rguenther@suse.de>
8248 * common.opt (feliminate-dwarf2-dups): Ignore.
8249 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
8250 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
8251 same_die_p_wrap, compute_section_prefix,
8252 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
8253 (comdat_symbol_id, comdat_symbol_number): Likewise.
8254 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
8256 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
8257 (output_die): Mark unreachable path unreachable.
8258 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
8259 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
8260 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
8261 (dwarf2out_early_finish): Likewise.
8263 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
8265 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
8267 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
8270 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
8271 not VAR_P. Filter attribute warnings with OPT_Wattributes.
8272 (avr_attribute_table) <io, io_low, address>: Initialize
8273 .decl_required with true.
8275 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
8277 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
8278 undocumented debugging options.
8279 (-mvsx-scalar-double): Likewise.
8280 (-mallow-df-permute): Likewise.
8281 (-mvectorize-builtins): Likewise.
8282 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8283 (rs6000_builtin_vectorized_function): Likewise.
8284 (rs6000_builtin_md_vectorized_function): Likewise.
8285 (rs6000_opt_vars): Likewise.
8287 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
8290 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
8291 (*btrq_imm): Rename from *btrq.
8292 (*btcq_imm): Rename from *btcq.
8293 (btsc): New code attribute.
8294 (*<btsc><mode>): New insn pattern.
8295 (*btr<mode>): Ditto.
8296 (*<btsc><mode>_mask): New insn_and_split pattern.
8297 (*btr<mode>_mask): Ditto.
8299 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8300 Alan Hayward <alan.hayward@arm.com>
8301 David Sherwood <david.sherwood@arm.com>
8303 * function.c (pad_below): Simplify padding calculation.
8305 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8306 Alan Hayward <alan.hayward@arm.com>
8307 David Sherwood <david.sherwood@arm.com>
8309 * target.def (function_prologue): Remove frame size argument.
8310 (function_epilogue): Likewise.
8311 * doc/tm.texi: Regenerate.
8312 * final.c (final_start_function): Update call to function_prologue.
8313 (final_end_function): Update call to function_epilogue.
8314 (default_function_pro_epilogue): Remove frame size argument.
8315 * output.h (default_function_pro_epilogue): Likewise.
8316 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
8317 (arm_output_function_prologue): Likewise.
8318 * config/frv/frv.c (frv_function_prologue): Likewise.
8319 (frv_function_epilogue): Likewise.
8320 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
8321 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
8322 (ia64_output_function_epilogue): Likewise.
8323 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
8324 (m32r_output_function_epilogue): Likewise.
8325 * config/microblaze/microblaze.c (microblaze_function_prologue)
8326 (microblaze_function_epilogue): Likewise.
8327 * config/mips/mips.c (mips_output_function_prologue): Likewise.
8328 (mips_output_function_epilogue): Likewise.
8329 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
8330 (mmix_target_asm_function_epilogue): Likewise.
8331 * config/msp430/msp430.c (msp430_start_function): Likewise.
8332 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
8333 (nds32_asm_function_epilogue): Likewise.
8334 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
8335 * config/pa/pa.c (pa_output_function_prologue): Likewise.
8336 (pa_output_function_epilogue): Likewise.
8337 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
8338 (rs6000_output_function_epilogue): Likewise.
8339 * config/rl78/rl78.c (rl78_start_function): Likewise.
8340 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
8341 (rs6000_output_function_epilogue): Likewise.
8342 * config/rx/rx.c (rx_output_function_prologue): Likewise.
8343 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
8344 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
8345 (sparc_asm_function_epilogue): Likewise.
8347 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8349 * tree.h (type_has_mode_precision_p): New function.
8350 * convert.c (convert_to_integer_1): Use it.
8351 * expr.c (expand_expr_real_2): Likewise.
8352 (expand_expr_real_1): Likewise.
8353 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
8354 * match.pd: Likewise.
8355 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
8356 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
8357 * tree-tailcall.c (process_assignment): Likewise.
8358 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8359 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
8360 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
8361 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8362 (vectorizable_assignment): Likewise.
8363 (vectorizable_shift): Likewise.
8364 (vectorizable_operation): Likewise.
8365 * tree-vrp.c (register_edge_assert_for_2): Likewise.
8367 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
8369 * match.pd: Add pow (C, x) simplification.
8371 2017-08-21 Richard Biener <rguenther@suse.de>
8373 PR tree-optimization/81900
8374 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
8375 for blocks with abnormal predecessors.
8376 (compute_antic): Do not set visited flag prematurely.
8378 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
8381 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
8383 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8385 * stor-layout.h (vector_type_mode): Move to...
8386 * tree.h (vector_type_mode): ...here.
8387 * stor-layout.c (vector_type_mode): Move to...
8388 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
8390 2017-08-21 Richard Biener <rguenther@suse.de>
8392 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
8393 register_external_die hooks.
8394 (debug_false_tree_charstarstar_uhwistar): Declare.
8395 (debug_nothing_tree_charstar_uhwi): Likewise.
8396 * debug.c (do_nothing_debug_hooks): Adjust.
8397 (debug_false_tree_charstarstar_uhwistar): New do nothing.
8398 (debug_nothing_tree_charstar_uhwi): Likewise.
8399 * dbxout.c (dbx_debug_hooks): Adjust.
8400 (xcoff_debug_hooks): Likewise.
8401 * sdbout.c (sdb_debug_hooks): Likewise.
8402 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
8403 * dwarf2out.c (macinfo_label_base): New global.
8404 (dwarf2out_register_external_die): New function for the
8405 register_external_die hook.
8406 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
8407 (dwarf2_debug_hooks): Use them.
8408 (dwarf2_lineno_debug_hooks): Adjust.
8409 (struct die_struct): Add with_offset flag.
8410 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
8411 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
8412 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
8413 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
8414 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
8415 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
8416 defining section names for the early LTO debug variants.
8417 (reset_indirect_string): New helper.
8418 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
8419 (print_dw_val): Add support for offsetted symbol references.
8420 (get_ultimate_context): Split out from is_cxx.
8421 (is_cxx): Use get_ultimate_context.
8422 (is_fortran): Add decl overload.
8423 (compute_comp_unit_symbol): Split out worker from
8424 compute_section_prefix.
8425 (compute_section_prefix): Call compute_comp_unit_symbol and
8426 set comdat_type_p here.
8427 (output_die): Skip DIE symbol output for the LTO added one.
8428 Handle DIE symbol references with offset.
8429 (output_comp_unit): Guard section name mangling properly.
8430 For LTO debug sections emit a symbol at the section beginning
8431 which we use to refer to its DIEs.
8432 (add_abstract_origin_attribute): For DIEs registered via
8433 dwarf2out_register_external_die directly refer to the early
8434 DIE rather than indirectly through the shadow one we created.
8435 Remove obsolete call to dwarf2out_abstract_function for
8436 non-function/block origins.
8437 (gen_array_type_die): When generating early LTO debug do
8438 not emit DW_AT_string_length.
8439 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
8440 late when in LTO. As suggested place a gcc_unreachable for
8441 the DECL_ABSTRACT_P case.
8442 (gen_subprogram_die): Avoid another specification DIE
8443 for early built declarations/definitions for the late LTO case.
8444 (gen_variable_die): Add type references for late duplicated VLA dies
8446 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
8447 we have the abstract instance already.
8448 (process_scope_var): Adjust decl DIE contexts in LTO which
8449 first puts them in limbo.
8450 (gen_decl_die): Do not generate type DIEs late apart from
8451 types for VLAs or for decls we do not yet have a DIE. Do not
8452 call dwarf2out_abstract_function late.
8453 (dwarf2out_early_global_decl): Make sure to create DIEs
8454 for abstract instances of a decl first.
8455 (dwarf2out_late_global_decl): Adjust comment.
8456 (output_macinfo_op): With multiple macro sections use
8457 macinfo_label_base to distinguish labels.
8458 (output_macinfo): Likewise. Update macinfo_label_base.
8459 Pass in the line info label.
8460 (note_variable_value_in_expr): When generating LTO resolve
8461 all variable values here by generating DIEs as needed.
8462 (init_sections_and_labels): Add early LTO debug flag parameter
8463 and generate different sections and names if set. Add generation
8464 counter for the labels so we can have multiple of them.
8465 (reset_dies): Helper to allow DIEs to be output multiple times.
8466 (dwarf2out_finish): When outputting DIEs to the fat part of an
8467 LTO object first reset DIEs.
8468 (dwarf2out_early_finish): Output early DIEs when generating LTO.
8469 (modified_type_die): Check for decl_ultimate_origin being self
8471 (gen_type_die_with_usage): Likewise.
8472 (gen_typedef_die): Allow decl_ultimate_origin being self.
8473 (set_decl_abstract_flags): Remove.
8474 (set_block_abstract_flags): Likewise.
8475 (dwarf2out_abstract_function): Treat the early generated DIEs
8476 as the abstract copy and only add DW_AT_inline and
8477 DW_AT_artificial here and call set_decl_origin_self.
8478 If the DIE has an abstract origin don't do anything.
8479 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
8480 if we have none yet (Go fails to build one, PR78628).
8481 (variably_modified_type_p): Prevent endless recursion for Ada
8482 cyclic pointer types.
8483 * lto-streamer-in.c: Include debug.h.
8484 (dref_queue): New global.
8485 (lto_read_tree_1): Stream in DIE references.
8486 (lto_input_tree): Register DIE references.
8487 (input_function): Stream DECL_DEBUG_ARGS.
8488 * lto-streamer-out.c: Include debug.h.
8489 (lto_write_tree_1): Output DIE references.
8490 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
8491 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
8492 (output_function): Stream DECL_DEBUG_ARGS.
8493 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
8494 Stream DECL_ABSTRACT_ORIGIN.
8495 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
8496 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
8497 DECL_CONTEXT for file-scope decls.
8498 * lto-streamer.h (struct dref_entry): Declare.
8499 (dref_queue): Likewise.
8500 * cfgexpand.c (pass_expand::execute): Do not call the
8501 outlining_inline_function hook here.
8502 * lto-wrapper.c (debug_obj): New global.
8503 (tool_cleanup): Unlink it if required.
8504 (debug_objcopy): New function.
8505 (run_gcc): Handle early debug sections in the IL files by
8506 extracting them to separate files, partially linkin them and
8507 feeding the result back as result to the linker.
8508 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
8509 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
8510 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
8511 sections into a separate segment.
8512 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
8514 (darwin_asm_dwarf_section): Likewise.
8515 (darwin_asm_output_dwarf_offset): Likewise.
8516 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
8518 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8519 Alan Hayward <alan.hayward@arm.com>
8520 David Sherwood <david.sherwood@arm.com>
8522 * read-md.h (md_reader::record_potential_iterator_use): Replace
8523 pointer argument with an rtx and an index.
8524 * read-rtl.c (iterator_group::apply_iterator): Likewise.
8525 (apply_mode_iterator): Likewise.
8526 (apply_code_iterator): Likewise.
8527 (apply_int_iterator): Likewise.
8528 (apply_subst_iterator): Likewise.
8529 (record_iterator_use): Likewise.
8530 (record_attribute_use): Likewise.
8531 (md_reader::record_potential_iterator_use): Likewise. Update calls
8532 to record_iterator_use and apply_iterator.
8533 (iterator_use): Replace ptr with x and index.
8534 (attribute_use): Likewise.
8535 (apply_attribute_uses): Update calls to apply_iterator.
8536 (apply_iterators): Likewise. Update initialization of iterator_use.
8537 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
8538 and record_potential_iterator_use.
8539 (rtx_reader::read_rtx_operand): Likewise.
8541 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8542 Alan Hayward <alan.hayward@arm.com>
8543 David Sherwood <david.sherwood@arm.com>
8545 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
8548 2017-08-21 Richard Biener <rguenther@suse.de>
8551 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
8552 at struct end conservatively when comparing common bases.
8554 2017-08-21 Richard Biener <rguenther@suse.de>
8556 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
8557 (mem_ref_in_stmt): Remove.
8558 (determine_max_movement): Use ref index to get at the reference.
8559 (invariantness_dom_walker::before_dom_children): Deal with
8560 lim data already initialized.
8561 (gather_mem_refs_stmt): Initialize lim data and record ref index.
8563 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
8565 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
8566 (TARGET_ISA_ROUND): Ditto.
8567 (TARGET_ROUND): Ditto.
8568 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
8569 * config/i386/i386.md: Ditto.
8570 * config/i386/sse.md: Ditto.
8571 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
8572 with OPTION_MASK_ISA_SSE4_1.
8574 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
8577 * doc/extend.texi (x86 Built-in Functions): Correct the name of
8578 __builtin_ia32_lzcnt_u16.
8580 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
8583 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
8584 (rs6000_set_current_function): Rewrite function to use it.
8586 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
8589 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
8590 and TYPE_WARN_IF_NOT_ALIGN.
8591 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
8592 (handle_warn_if_not_align): New.
8593 (place_union_field): Call handle_warn_if_not_align.
8594 (place_field): Call handle_warn_if_not_align.
8595 Copy TYPE_WARN_IF_NOT_ALIGN.
8596 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
8597 (layout_type): Likewise.
8598 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
8600 (tree_decl_common): Add warn_if_not_align.
8601 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
8602 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
8603 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
8604 (DECL_WARN_IF_NOT_ALIGN): Likewise.
8605 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
8606 * doc/extend.texi: Document warn_if_not_aligned attribute.
8607 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
8609 2017-08-17 Martin Liska <mliska@suse.cz>
8612 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
8613 (get_data_dependence): Use it as pointer type.
8614 (distribute_loop): Likewise.
8616 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8618 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
8619 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
8620 (p8_vmrgow_<mode>_direct): New define_insn.
8621 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
8622 handle endianness for vmrgew and vmrgow permute patterns.
8624 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
8626 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
8627 * config/rs6000/rs6000-cpus.def: Remove comment.
8628 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
8629 (POWERPC_MASKS): Likewise.
8630 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
8631 use of TARGET_VSX_TIMODE.
8632 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
8633 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
8634 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
8635 (rs6000_option_override_internal): Remove dead code.
8636 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
8637 (rs6000_legitimize_reload_address): Likewise.
8638 (rs6000_legitimate_address_p): Likewise.
8639 (rs6000_opt_masks): Delete "vsx-timode".
8640 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
8641 from function comment.
8642 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
8643 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
8644 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
8646 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
8647 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
8648 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
8651 (define_peephole2): Likewise.
8653 2017-08-17 Martin Sebor <msebor@redhat.com>
8656 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
8657 past the end of an array.
8658 (test_pp_format): Add test cases.
8660 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
8662 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
8663 missing ECF_NOTHROW flags.
8665 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
8668 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
8669 operands residing in integer registers.
8670 (*vsx_le_perm_load_<mode>): Likewise.
8671 (*vsx_le_perm_store_<mode>): Likewise.
8672 (define_peephole2): Add peepholes to optimize the above.
8674 2017-08-17 Marek Polacek <polacek@redhat.com>
8677 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
8678 to mimic what shorten_compare did. Change the return type to bool.
8679 (fold_cond_expr_with_comparison): Update call to
8680 operand_equal_for_comparison_p.
8681 (fold_ternary_loc): Likewise.
8683 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
8685 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
8687 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
8689 2017-08-17 Richard Biener <rguenther@suse.de>
8691 * tree-ssa-structalias.c (solve_graph): When propagating
8692 to successors update the graphs succ edges and avoid duplicate work.
8694 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
8697 * config/i386/i386.c (ix86_option_override_internal): Save target
8698 specific options after ix86_stack_protector_guard_reg was changed.
8700 2017-08-17 Richard Biener <rguenther@suse.de>
8702 PR tree-optimization/81827
8703 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
8704 (new_var_info): Initialize it conservatively.
8705 (get_call_vi): Mark register vars.
8706 (new_scalar_tmp_constraint_exp): Likewise.
8707 (handle_rhs_call): Likewise.
8708 (handle_const_call): Likewise.
8709 (create_function_info_for): Likewise.
8710 (solve_constraints): Sort varinfos to separate register from
8711 non-register vars to pack points-to solution bitmaps during
8714 2017-08-17 Marek Polacek <polacek@redhat.com>
8716 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
8718 2017-08-17 Richard Biener <rguenther@suse.de>
8720 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
8721 to true when overflow is undefined and we saturated the result.
8723 2017-08-17 Alan Modra <amodra@gmail.com>
8726 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
8727 Don't use store multiple if only one reg needs saving.
8728 (interesting_frame_related_regno): New function.
8729 (rs6000_frame_related): Don't emit frame info for regs that
8731 (rs6000_emit_epilogue): Likewise.
8733 2017-08-16 Nathan Sidwell <nathan@acm.org>
8735 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
8736 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
8737 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
8738 * tree.c (free_lang_data_in_type): Use else-if chain. Always
8739 clear TYPE_LANG_1. Remove obsolete member-function stripping.
8740 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
8741 (verify_type): Adjust for TYPE_BINFO move.
8742 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
8743 process TYPE_BINFO directly.
8744 (hash_tree): Likewise.
8745 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
8747 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
8750 2017-08-16 David Malcolm <dmalcolm@redhat.com>
8752 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
8754 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
8757 * config/i386/i386.md (*anddi_1_btr): Change predicates of
8758 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
8759 Add ix86_binary_operator_ok to insn constraint.
8760 (*iordi_1_bts): Ditto.
8761 (*xordi_1_btc): Ditto.
8762 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
8763 Update corresponding peephole2 pattern.
8767 2017-08-16 Bin Cheng <bin.cheng@arm.com>
8769 PR tree-optimization/81832
8770 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
8771 copy loop header which has IFN_LOOP_DIST_ALIAS call.
8773 2017-08-16 Marek Polacek <polacek@redhat.com>
8776 * fold-const.c (fold_indirect_ref_1): Restore original behavior
8777 regarding size_zero_node.
8779 2017-08-16 Martin Liska <mliska@suse.cz>
8782 * config.gcc: Respect previously set extra_objs in case
8785 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
8787 PR tree-optimization/81835
8788 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
8789 the phi SSA_NAME. Check that the condition in a COND_EXPR does
8790 not depend on the phi.
8792 2017-08-16 Alan Modra <amodra@gmail.com>
8794 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
8797 2017-08-16 Alan Modra <amodra@gmail.com>
8799 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
8800 (save_reg_p): ..into this. Update all callers.
8801 (first_reg_to_save): Simplify.
8803 2017-08-16 Alan Modra <amodra@gmail.com>
8805 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
8808 2017-08-15 Joseph Myers <joseph@codesourcery.com>
8812 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
8813 constant count if that count is less than 32.
8815 2017-08-15 Nathan Sidwell <nathan@acm.org>
8817 * gcc.c (execute): Emit friendlier message if inferior is killed
8818 by an external cause.
8820 2017-08-15 Richard Biener <rguenther@suse.de>
8822 PR tree-optimization/81790
8823 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
8824 CONSTRUCTORs from simplifying and VN.
8826 2017-08-14 Martin Sebor <msebor@redhat.com>
8828 * builtin-attrs.def: Add comments.
8830 2017-08-14 Martin Sebor <msebor@redhat.com>
8833 * doc/extend.texi (attribute nonstring): Document new attribute.
8835 2017-08-14 Martin Sebor <msebor@redhat.com>
8838 * tree-diagnostic.c (default_tree_printer): Handle %G.
8839 * gimple-pretty-print.h (percent_G_format): Declare new function.
8840 * gimple-pretty-print.c (percent_G_format): Define.
8841 * tree-pretty-print.c (percent_K_format): Add argument.
8843 2017-08-14 Martin Sebor <msebor@redhat.com>
8845 PR translation/79998
8846 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
8847 Remove a stray space.
8849 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
8852 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
8853 (*iordi_1_bts): Ditto.
8854 (*xordi_1_btc): Ditto.
8856 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8859 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
8860 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8862 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
8863 quoted strings, and make more translator-friendly.
8864 (darwin_rs6000_override_options): Likewise.
8865 (rs6000_option_override_internal): Likewise.
8866 (rs6000_return_in_memory): Fix overlong line.
8867 (init_cmulative_args): Use quoted strings, and make more
8868 translator-friendly.
8869 (rs6000_pass_by_reference): Fix overlong line.
8870 (def_builtin): Use quoted strings.
8871 (altivec_expand_predicate_builtin): Use quoted strings, and make
8872 more translator-friendly.
8873 (htm_expand_builtin): Use quoted strings.
8874 (cpu_expand_builtin): Use quoted strings, and make more
8875 translator-friendly.
8876 (altivec_expand_builtin): Likewise.
8877 (paired_expand_predicate_builtin): Likewise.
8878 (rs6000_invalid_builtin): Likewise.
8879 (builtin_function_type): Use quoted strings.
8880 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
8881 more translator-friendly.
8882 (rs6000_trampoline_init): Likewise.
8883 (rs6000_handle_altivec_attribute): Likewise.
8884 (rs6000_inner_target_options): Use quoted strings.
8885 (rs6000_disable_incompatible_switches): Likewise.
8886 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
8887 strings, and make more translator-friendly.
8888 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
8890 2017-08-14 Bin Cheng <bin.cheng@arm.com>
8892 PR tree-optimization/81799
8893 * tree-loop-distribution.c (version_loop_by_alias_check): Force
8894 cond_expr to simple gimple operand.
8896 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
8899 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
8901 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
8905 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
8906 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
8907 TARGET_GASISR_PROLOGUES.
8908 * config/avr/avr.c (avr_option_override): Same.
8909 (avr_pass_pre_proep::execute): Same.
8911 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
8914 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
8915 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
8917 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
8919 * config/i386/i386.md (*load_tp_<mode>): Redefine as
8920 define_insn_and_split. Split to a memory load from 0 in
8921 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
8922 using PTR mode iterator.
8923 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
8924 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
8925 (*add_tp_<mode>): Redefine as define_insn_and_split.
8926 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
8927 address space. Merge with *add_tp_x32 using PTR mode iterator.
8928 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
8929 Split to an add with a memory load from 0 in
8930 DEFAULT_TLS_SEG_REG address space.
8932 2017-08-12 Andrew Pinski <apinski@cavium.com>
8934 * config/aarch64/aarch64-option-extensions.def (rdma):
8935 Fix feature string to what Linux prints out in /proc/cpuinfo.
8937 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
8940 * dwarf2out.c (modified_type_die): For C typedef types that have
8941 an ultimate origin, process the ultimate origin instead of the
8943 (gen_typedef_die): Assert that input DECLs have no ultimate
8945 (gen_type_die_with_usage): For typedef variants that have an
8946 ultimate origin, just call gen_decl_die on the original DECL.
8947 (process_scope_var): Avoid creating DIEs for local typedefs and
8948 concrete static variables.
8950 2017-08-12 Alan Modra <amodra@gmail.com>
8954 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
8955 match gnu-user.h startfile.
8956 (ENDFILE_LINUX_SPEC): Similarly.
8958 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
8961 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
8963 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
8965 2017-08-11 Tamar Christina <tamar.christina@arm.com>
8966 * config/aarch64/aarch64.md (mov<mode>): Change.
8967 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
8968 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
8969 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
8971 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
8973 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
8974 for storage order barriers.
8976 2017-08-11 Martin Liska <mliska@suse.cz>
8979 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
8980 variables of void type.
8982 2017-08-11 Martin Liska <mliska@suse.cz>
8984 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
8985 TARGET_SUPPORTS_ALIASES.
8986 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
8987 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
8988 (optimize_weakref): Likewise.
8989 * symtab.c (symtab_node::noninterposable_alias): Likewise.
8990 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
8991 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
8993 2017-08-11 Martin Liska <mliska@suse.cz>
8996 * config/i386/i386.c (make_resolver_func): Do complete
8997 refactoring of the function.
8999 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
9002 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
9003 * config/i386/i386.c (ix86_stack_protect_guard): Use
9004 ix86_stack_protect_guard_symbol_str to generate varible declaration.
9005 * doc/invoke.texi (x86 Options): Document
9006 -mstack-protector-guard-symbol= option.
9008 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
9010 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
9011 * config/i386/i386.c (ix86_split_stack_guard): New function.
9012 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
9013 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
9014 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
9015 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
9016 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
9017 (split_stack_space_check): Call ix86_split_stack_guard.
9019 2017-08-10 Martin Sebor <msebor@redhat.com>
9021 * print-tree.c (print_node): Print location using the established
9023 Replace spaces with colons.
9024 (debug_raw, debug): Ditto.
9026 2017-08-10 Martin Sebor <msebor@redhat.com>
9029 * pretty-print.c (pp_format): Correct the handling of %s precision.
9031 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
9034 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
9036 (ix86_finalize_stack_frame_flags): This. Also clear
9037 frame_pointer_needed if -fno-omit-frame-pointer is used without
9039 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
9040 with ix86_finalize_stack_frame_flags.
9041 (ix86_expand_epilogue): Likewise.
9042 (ix86_expand_split_stack_prologue): Likewise.
9043 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
9045 2017-08-10 Martin Liska <mliska@suse.cz>
9048 * c-attribs.c (handle_target_attribute):
9049 Report warning for an empty string argument of target attribute.
9051 2017-08-09 Jakub Jelinek <jakub@redhat.com>
9054 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
9056 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
9057 or DECL_NONLOCAL labels.
9058 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
9059 or DECL_NONLOCAL labels here.
9061 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
9063 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
9064 to indicate when early gimple folding has been disabled.
9065 (rs6000_gimple_fold_builtin): Add debug content.
9066 (rs6000_invalid_builtin): Fix whitespace.
9067 (rs6000_expand_builtin): Fix whitespace.
9068 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
9070 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
9073 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
9074 SAVE_MULTIPLE if not all the registers that saves, should be saved.
9076 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
9078 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
9079 (qdf24xx): Likewise.
9080 * config/aarch64/aarch64.md: Include falkor.md.
9081 * config/aarch64/falkor.md: New.
9083 2017-08-09 Marek Polacek <polacek@redhat.com>
9086 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
9087 * diagnostic.c (emit_diagnostic): Add a comment.
9088 (emit_diagnostic_valist): New function.
9090 2017-08-09 Marek Polacek <polacek@redhat.com>
9093 * input.c (make_location): New overload.
9094 * input.h (make_location): Declare.
9096 2017-08-08 Alan Modra <amodra@gmail.com>
9097 H.J. Lu <hongjiu.lu@intel.com>
9100 * gcc.c (NO_PIE_SPEC): Delete.
9101 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
9103 (LINK_PIE_SPEC): ..to here.
9104 (LINK_COMMAND_SPEC): Support -no-pie.
9105 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
9106 chain of crtbegin*.o selection, update for PIE_SPEC changes and
9108 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
9109 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
9110 (ENDFILE_CRTEND_SPEC): Similarly.
9112 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
9115 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
9116 (mstack-protector-guard-offset=): Ditto.
9117 * config/i386/i386.c (ix86_option_override): Handle
9118 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
9120 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
9121 ix86_stack_protect_guard_offset variables.
9122 (TARGET_STACK_PROTECT_GUARD): Always define.
9123 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
9124 and -mstack-protector-guard-offset= options.
9126 2017-08-08 Bin Cheng <bin.cheng@arm.com>
9128 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
9129 boundary case for the last candidate.
9131 2017-08-08 Bin Cheng <bin.cheng@arm.com>
9133 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
9134 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
9136 2017-08-08 Tamar Christina <tamar.christina@arm.com>
9139 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
9140 * config/aarch64/aarch64-builtins.c
9141 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
9142 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
9143 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
9145 2017-08-08 Tamar Christina <tamar.christina@arm.com>
9146 Andrew Pinski <pinskia@gmail.com>
9149 * internal-fn.def (XORSIGN): New.
9150 * optabs.def (xorsign_optab): New.
9151 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
9152 (convert_expand_mult_copysign): New.
9153 (pass_optimize_widening_mul::execute): Call
9154 convert_expand_mult_copysign.
9156 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9158 PR tree-optimization/81354
9159 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
9160 Insert on edges rather than explicitly creating landing pads.
9161 (analyze_candidates_and_replace): Commit edge inserts.
9163 2017-08-08 Richard Biener <rguenther@suse.de>
9166 * tree-ssa-loop-niter.c: Include tree-dfa.h.
9167 (expand_simple_operations): Also look through ADDR_EXPRs with
9168 MEM_REF bases treating them as POINTER_PLUS_EXPR.
9170 2017-08-08 Richard Biener <rguenther@suse.de>
9172 PR tree-optimization/81723
9173 * tree-vect-slp.c (struct bst_traits): New hash traits.
9174 (bst_fail): New global.
9175 (vect_build_slp_tree_2): New worker, split out from ...
9176 (vect_build_slp_tree): ... this now wrapping it with using
9177 bst_fail set to cache SLP tree build fails. Properly handle
9179 (vect_analyze_slp_instance): Allocate and free bst_fail.
9181 2017-08-08 Martin Liska <mliska@suse.cz>
9184 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
9185 LABEL_DECLs that can be from a different function.
9187 2017-08-08 Bin Cheng <bin.cheng@arm.com>
9189 PR tree-optimization/81744
9190 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
9191 loop's number of iterations.
9193 2017-08-08 Martin Liska <mliska@suse.cz>
9195 * asan.c: Include header files.
9196 * attribs.c (build_decl_attribute_variant): New function moved
9198 (build_type_attribute_qual_variant): Likewise.
9199 (cmp_attrib_identifiers): Likewise.
9200 (simple_cst_list_equal): Likewise.
9201 (omp_declare_simd_clauses_equal): Likewise.
9202 (attribute_value_equal): Likewise.
9203 (comp_type_attributes): Likewise.
9204 (build_type_attribute_variant): Likewise.
9205 (lookup_ident_attribute): Likewise.
9206 (remove_attribute): Likewise.
9207 (merge_attributes): Likewise.
9208 (merge_type_attributes): Likewise.
9209 (merge_decl_attributes): Likewise.
9210 (merge_dllimport_decl_attributes): Likewise.
9211 (handle_dll_attribute): Likewise.
9212 (attribute_list_equal): Likewise.
9213 (attribute_list_contained): Likewise.
9214 * attribs.h (lookup_attribute): New function moved from tree.[ch].
9215 (lookup_attribute_by_prefix): Likewise.
9216 * bb-reorder.c: Include header files.
9217 * builtins.c: Likewise.
9218 * calls.c: Likewise.
9219 * cfgexpand.c: Likewise.
9220 * cgraph.c: Likewise.
9221 * cgraphunit.c: Likewise.
9222 * convert.c: Likewise.
9223 * dwarf2out.c: Likewise.
9224 * final.c: Likewise.
9225 * fold-const.c: Likewise.
9226 * function.c: Likewise.
9227 * gimple-expr.c: Likewise.
9228 * gimple-fold.c: Likewise.
9229 * gimple-pretty-print.c: Likewise.
9230 * gimple.c: Likewise.
9231 * gimplify.c: Likewise.
9232 * hsa-common.c: Likewise.
9233 * hsa-gen.c: Likewise.
9234 * internal-fn.c: Likewise.
9235 * ipa-chkp.c: Likewise.
9236 * ipa-cp.c: Likewise.
9237 * ipa-devirt.c: Likewise.
9238 * ipa-fnsummary.c: Likewise.
9239 * ipa-inline.c: Likewise.
9240 * ipa-visibility.c: Likewise.
9242 * lto-cgraph.c: Likewise.
9243 * omp-expand.c: Likewise.
9244 * omp-general.c: Likewise.
9245 * omp-low.c: Likewise.
9246 * omp-offload.c: Likewise.
9247 * omp-simd-clone.c: Likewise.
9248 * opts-global.c: Likewise.
9249 * passes.c: Likewise.
9250 * predict.c: Likewise.
9251 * sancov.c: Likewise.
9252 * sanopt.c: Likewise.
9253 * symtab.c: Likewise.
9254 * toplev.c: Likewise.
9255 * trans-mem.c: Likewise.
9256 * tree-chkp.c: Likewise.
9257 * tree-eh.c: Likewise.
9258 * tree-into-ssa.c: Likewise.
9259 * tree-object-size.c: Likewise.
9260 * tree-parloops.c: Likewise.
9261 * tree-profile.c: Likewise.
9262 * tree-ssa-ccp.c: Likewise.
9263 * tree-ssa-live.c: Likewise.
9264 * tree-ssa-loop.c: Likewise.
9265 * tree-ssa-sccvn.c: Likewise.
9266 * tree-ssa-structalias.c: Likewise.
9267 * tree-ssa.c: Likewise.
9268 * tree-streamer-in.c: Likewise.
9269 * tree-vectorizer.c: Likewise.
9270 * tree-vrp.c: Likewise.
9272 * ubsan.c: Likewise.
9273 * varasm.c: Likewise.
9274 * varpool.c: Likewise.
9275 * tree.c: Remove functions moved to attribs.[ch].
9277 * config/aarch64/aarch64.c: Add attrs.h header file.
9278 * config/alpha/alpha.c: Likewise.
9279 * config/arc/arc.c: Likewise.
9280 * config/arm/arm.c: Likewise.
9281 * config/avr/avr.c: Likewise.
9282 * config/bfin/bfin.c: Likewise.
9283 * config/c6x/c6x.c: Likewise.
9284 * config/cr16/cr16.c: Likewise.
9285 * config/cris/cris.c: Likewise.
9286 * config/darwin.c: Likewise.
9287 * config/epiphany/epiphany.c: Likewise.
9288 * config/fr30/fr30.c: Likewise.
9289 * config/frv/frv.c: Likewise.
9290 * config/ft32/ft32.c: Likewise.
9291 * config/h8300/h8300.c: Likewise.
9292 * config/i386/winnt.c: Likewise.
9293 * config/ia64/ia64.c: Likewise.
9294 * config/iq2000/iq2000.c: Likewise.
9295 * config/lm32/lm32.c: Likewise.
9296 * config/m32c/m32c.c: Likewise.
9297 * config/m32r/m32r.c: Likewise.
9298 * config/m68k/m68k.c: Likewise.
9299 * config/mcore/mcore.c: Likewise.
9300 * config/microblaze/microblaze.c: Likewise.
9301 * config/mips/mips.c: Likewise.
9302 * config/mmix/mmix.c: Likewise.
9303 * config/mn10300/mn10300.c: Likewise.
9304 * config/moxie/moxie.c: Likewise.
9305 * config/msp430/msp430.c: Likewise.
9306 * config/nds32/nds32-isr.c: Likewise.
9307 * config/nds32/nds32.c: Likewise.
9308 * config/nios2/nios2.c: Likewise.
9309 * config/nvptx/nvptx.c: Likewise.
9310 * config/pa/pa.c: Likewise.
9311 * config/pdp11/pdp11.c: Likewise.
9312 * config/powerpcspe/powerpcspe.c: Likewise.
9313 * config/riscv/riscv.c: Likewise.
9314 * config/rl78/rl78.c: Likewise.
9315 * config/rx/rx.c: Likewise.
9316 * config/s390/s390.c: Likewise.
9317 * config/sh/sh.c: Likewise.
9318 * config/sol2.c: Likewise.
9319 * config/sparc/sparc.c: Likewise.
9320 * config/spu/spu.c: Likewise.
9321 * config/stormy16/stormy16.c: Likewise.
9322 * config/tilegx/tilegx.c: Likewise.
9323 * config/tilepro/tilepro.c: Likewise.
9324 * config/v850/v850.c: Likewise.
9325 * config/vax/vax.c: Likewise.
9326 * config/visium/visium.c: Likewise.
9327 * config/xtensa/xtensa.c: Likewise.
9329 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9332 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
9333 constraints since the -mupper-regs-* switches have been
9335 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
9336 into a vector from a double word element that was extracted from
9337 another vector, and eliminate extra XXPERMDI instructions.
9338 (vsx_concat_<mode>_2): Likewise.
9339 (vsx_concat_<mode>_3): Likewise.
9340 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
9341 concat to allow optimizing inserts from previous extracts.
9343 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
9345 * config/i386/i386.c (ix86_stack_protect_guard): Generate
9346 memory reference to a SSP offset in TLS address space.
9347 (ix86_print_operand) <case '@'>: Remove.
9348 (ix86_print_operand_punct_valid_p): Remove '@' code.
9349 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
9351 (stack_tls_protect_set_<mode>): Remove.
9352 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
9353 (stack_tls_protect_test_<mode>): Remove.
9354 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
9356 2017-08-07 Olivier Hainque <hainque@adacore.com>
9359 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
9361 2017-08-07 Douglas Rupp <rupp@adacore.com>
9363 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
9364 variable was referenced as multidir in command.
9366 2017-08-07 Jakub Jelinek <jakub@redhat.com>
9369 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
9372 2017-08-07 Martin Liska <mliska@suse.cz>
9374 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
9375 * config/rl78/rl78.c: Add include of attribs.h.
9376 * config/sh/sh.c: Likewise.
9377 * config/v850/v850.c: Likewise.
9379 2017-08-07 Tom de Vries <tom@codesourcery.com>
9382 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
9384 2017-08-07 Martin Liska <mliska@suse.cz>
9386 * config/mips/mips.c: Include attribs.h.
9388 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
9391 * doc/invoke.texi: Document change in behvaior for -Ofast for
9394 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
9396 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
9398 (aarch64_pop_regs): Likewise.
9399 (aarch64_gen_load_pair): Likewise.
9400 (aarch64_save_callee_saves): Likewise.
9401 (aarch64_restore_callee_saves): Likewise.
9403 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
9405 * config/i386/i386.c: Revert the last change.
9407 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
9410 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
9412 (ix86_finalize_stack_frame_flags): This. Also clear
9413 frame_pointer_needed if -fno-omit-frame-pointer is used without
9415 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
9416 with ix86_finalize_stack_frame_flags.
9417 (ix86_expand_epilogue): Likewise.
9418 (ix86_expand_split_stack_prologue): Likewise.
9420 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
9423 * config/i386/i386.c (get_builtin_code_for_version): Set priority
9424 to P_AES for Westmere.
9426 2017-08-07 Jonathan Yong <10walls@gmail.com>
9428 * config/i386/mingw.opt (fset-stack-executable): Removed.
9429 * config/i386/cygming.opt (fset-stack-executable): Moved
9431 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
9433 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
9435 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
9437 2017-08-07 Marek Polacek <polacek@redhat.com>
9440 * fold-const.c (fold_indirect_ref_1): Check type_domain.
9442 2017-08-07 Martin Liska <mliska@suse.cz>
9444 * attribs.h (canonicalize_attr_name): New function.
9445 (cmp_attribs): Move from c-format.c and adjusted.
9446 (is_attribute_p): Moved from tree.h.
9447 * tree-inline.c: Add new includes.
9448 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
9449 (private_is_attribute_p): Remove.
9450 (private_lookup_attribute): Likewise.
9451 (private_lookup_attribute_by_prefix): Simplify.
9452 (remove_attribute): Use is_attribute_p.
9453 * tree.h: Remove removed declarations.
9455 2017-08-07 Jakub Jelinek <jakub@redhat.com>
9458 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
9459 instead of computing it in the function. Formatting fix.
9460 (expand_case): Don't rely on default_edge being the first edge,
9461 clear it if removing it, pass default_edge to
9462 emit_case_dispatch_table.
9463 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
9466 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
9468 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
9469 insn in the function, emit NOP after the insn.
9471 2017-08-06 Tom de Vries <tom@codesourcery.com>
9473 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
9476 2017-08-06 Tom de Vries <tom@codesourcery.com>
9478 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
9481 2017-08-04 Yury Gribov <tetra2005@gmail.com>
9483 PR tree-optimization/57371
9484 * match.pd: New pattern.
9486 2017-08-04 Marek Polacek <polacek@redhat.com>
9489 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
9490 perform the computation in offset_int.
9492 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
9494 PR tree-optimization/81136
9495 * tree-vectorizer.h: Include tree-hash-traits.h.
9496 (vec_base_alignments): New typedef.
9497 (vec_info): Add a base_alignments field.
9498 (vect_record_base_alignments): Declare.
9499 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
9501 (DR_IS_CONDITIONAL_IN_STMT): New macro.
9502 (create_data_ref): Add an is_conditional_in_stmt argument.
9503 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
9504 the is_conditional_in_stmt field.
9505 (data_ref_loc): Add an is_conditional_in_stmt field.
9506 (get_references_in_stmt): Set the is_conditional_in_stmt field.
9507 (find_data_references_in_stmt): Update call to create_data_ref.
9508 (graphite_find_data_references_in_stmt): Likewise.
9509 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
9510 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
9511 (vect_record_base_alignment): New function.
9512 (vect_record_base_alignments): Likewise.
9513 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
9514 for nested statements even if we fail to compute a misalignment.
9515 Use pooled base alignments for unconditional references.
9516 (vect_find_same_alignment_drs): Compare base addresses instead
9518 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
9519 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
9521 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
9523 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
9524 Add an explicit name for the enum. Use auto_vec for slp_instances
9526 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
9528 (_bb_vec_info): Add a constructor and destructor.
9529 (vinfo_for_stmt): Return NULL for uids of -1 as well.
9530 (destroy_loop_vec_info): Delete.
9531 (vect_destroy_datarefs): Likewise.
9532 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
9533 (vec_info::vec_info): New function.
9534 (vec_info::~vec_info): Likewise.
9535 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
9536 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
9537 destroy_loop_vec_info.
9538 * tree-vect-loop.c (new_loop_vec_info): Replace with...
9539 (_loop_vec_info::_loop_vec_info): ...this.
9540 (destroy_loop_vec_info): Replace with...
9541 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
9542 the stmt_vec_infos. Leave handling of vec_info information to its
9543 destructor. Remove explicit vector releases.
9544 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
9545 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
9546 * tree-vect-slp.c (new_bb_vec_info): Replace with...
9547 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
9548 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
9549 (destroy_bb_vec_info): Replace with...
9550 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
9551 information to its destructor.
9552 (vect_slp_analyze_bb_1): Use new and delete instead of
9553 new_bb_vec_info and destroy_bb_vec_info.
9554 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
9557 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
9559 * tree-data-ref.h (subscript): Add access_fn field.
9560 (data_dependence_relation): Add could_be_independent_p.
9561 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
9562 (same_access_functions): Move to tree-data-ref.c.
9563 * tree-data-ref.c (ref_contains_union_access_p): New function.
9564 (access_fn_component_p): Likewise.
9565 (access_fn_components_comparable_p): Likewise.
9566 (dr_analyze_indices): Add a reference to access_fn_component_p.
9567 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
9569 (constant_access_functions): Likewise.
9570 (add_other_self_distances): Likewise.
9571 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
9572 (initialize_data_dependence_relation): Use XCNEW and remove
9573 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
9574 of access functions that have the same type. Allow the
9575 subsequence to end with different bases in some circumstances.
9576 Record the chosen access functions in SUB_ACCESS_FN.
9577 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
9578 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
9579 (subscript_dependence_tester_1): Likewise dra and drb.
9580 (build_classic_dist_vector): Update calls accordingly.
9581 (subscript_dependence_tester): Likewise.
9582 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
9583 DDR_COULD_BE_INDEPENDENT_P.
9584 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
9585 comp_alias_ddrs instead of may_alias_ddrs.
9586 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
9588 (vect_analyze_data_ref_dependence): Use it if
9589 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
9590 distance vectors if that fails.
9591 (dependence_distance_ge_vf): New function.
9592 (vect_prune_runtime_alias_test_list): Use it. Don't clear
9593 LOOP_VINFO_MAY_ALIAS_DDRS.
9595 2017-08-04 Richard Biener <rguenther@suse.de>
9598 * fold-const.c (fold_binary_loc): Properly restrict
9599 minus_var0 && minus_var1 case when associating undefined overflow
9602 2017-08-04 Tom de Vries <tom@codesourcery.com>
9604 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
9606 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9608 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9609 Don't start diagnostic messages with a capital letter.
9610 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9612 (rs6000_invalid_builtin): Likewise.
9613 (rs6000_trampoline_init): Likewise.
9615 2017-08-03 Jakub Jelinek <jakub@redhat.com>
9618 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
9619 after setting changeable df flags.
9621 2017-08-03 Richard Biener <rguenther@suse.de>
9623 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
9624 up if the use is in USE - X.
9626 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
9628 * toplev.c (dumpfile.h): New include.
9629 (internal_error_reentered): New static function. Use it...
9630 (internal_error_function): ...here to handle reentered internal_error.
9632 2017-08-03 Richard Biener <rguenther@suse.de>
9635 * fold-const.c (split_tree): Add minus_var and minus_con
9636 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
9637 here but always use minus_*.
9638 (associate_trees): Assert we never associate with MINUS_EXPR
9639 and NULL first operand. Do not recurse for PLUS_EXPR operands
9640 when associating as MINUS_EXPR either.
9641 (fold_binary_loc): Track minus_var and minus_con.
9643 2017-08-03 Tom de Vries <tom@codesourcery.com>
9646 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
9647 ACCEL_COMPILER, apply finish_options on
9648 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
9650 2017-08-03 Tom de Vries <tom@codesourcery.com>
9653 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
9654 function_entry_patch_area_size > 0.
9656 2017-08-03 Jakub Jelinek <jakub@redhat.com>
9659 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
9660 instead of 10??LU, perform unit multiplication in wide_int,
9661 don't change alloc_object_size_limit if the limit is larger
9664 PR tree-optimization/81655
9665 PR tree-optimization/81588
9666 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
9667 the case when ranges[i].low and high are 1 for unsigned type with
9671 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
9672 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
9674 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9676 * tree-vrp.h: Add include guard.
9678 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
9681 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
9682 (ud2): New insn pattern.
9683 * config/i386/i386.c (ix86_expand_epilogue):
9684 For naked functions, generate ud2 instead of trap insn.
9686 2017-08-02 Marek Polacek <polacek@redhat.com>
9689 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
9691 2017-08-02 Tom de Vries <tom@codesourcery.com>
9692 Cesar Philippidis <cesar@codesourcery.com>
9694 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
9695 Add missing edge probabilities.
9697 2017-08-02 Tamar Christina <tamar.christina@arm.com>
9699 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
9702 2017-08-02 Jakub Jelinek <jakub@redhat.com>
9705 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
9706 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
9707 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
9708 prologue_seq sequences - if any.
9710 2017-08-02 Richard Biener <rguenther@suse.de>
9712 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
9713 via vectors if supported, integer extracts via punning if supported
9714 or otherwise vector extracts.
9716 2017-08-02 Richard Biener <rguenther@suse.de>
9718 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
9720 (bitmap_insert_into_set): ... this.
9722 2017-08-02 Richard Biener <rguenther@suse.de>
9724 PR tree-optimization/81633
9726 2015-08-17 Alan Hayward <alan.hayward@arm.com>
9728 PR tree-optimization/71752
9729 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
9731 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
9733 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
9734 (machine_function::call_ms2sysv_pad_out): Remove field.
9735 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
9736 (ix86_compute_frame_layout): Likewise.
9738 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
9741 * config/i386/i386.c (ix86_set_func_type): Disallow naked
9742 attribute with interrupt attribute.
9744 2017-08-01 Andrew Pinski <apinski@cavium.com>
9746 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
9747 BIT_INSERT_EXPR's operand 1
9748 to see if the types precision matches.
9750 2017-08-01 Martin Liska <mliska@suse.cz>
9753 * builtins.c (expand_builtin_memcpy_args): Remove.
9754 (expand_builtin_memcpy): Call newly added function
9755 expand_builtin_memory_copy_args.
9756 (expand_builtin_memcpy_with_bounds): Likewise.
9757 (expand_builtin_mempcpy): Remove last argument.
9758 (expand_builtin_mempcpy_with_bounds): Likewise.
9759 (expand_builtin_memory_copy_args): New function created from
9760 expand_builtin_mempcpy_args with small modifications.
9761 (expand_builtin_mempcpy_args): Remove.
9762 (expand_builtin_stpcpy): Remove unused argument.
9763 (expand_builtin): Likewise.
9764 (expand_builtin_with_bounds): Likewise.
9766 2017-08-01 Martin Liska <mliska@suse.cz>
9769 Make mempcpy more optimal (PR middle-end/70140).
9771 2017-08-01 Jakub Jelinek <jakub@redhat.com>
9774 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9775 __builtin_vec_cmpne verify both arguments are compatible vectors
9776 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
9777 verify arg1_type is a pointer or array type. For __builtin_vec_st,
9778 move computation of aligned to after checking the argument types.
9782 * config/rs6000/vsx.md (vextract_fp_from_shorth,
9783 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
9786 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
9787 Doug Rupp <rupp@adacore.com>
9788 Olivier Hainque <hainque@adacore.com>
9790 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
9791 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
9793 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
9794 for TARGET_OS_CPP_BUILTIN.
9795 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
9796 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
9798 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
9799 passing required abi options to the assembler for EABI configurations.
9800 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
9801 of .text.hot and .text.unlikely sections for kernel modules when
9802 using ARM style exceptions.
9803 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
9804 options. Add EXTRA_CC1_SPEC.
9805 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
9807 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
9809 (ARM_TARGET2_DWARF_FORMAT): Define.
9810 * config/arm/t-vxworks: Adjust multilib control to removal of the
9811 Diab command line options.
9813 2017-08-01 Martin Liska <mliska@suse.cz>
9815 PR gcov-profile/81561
9816 * gcov.c (unblock): Make unblocking safe as we need to preserve
9817 index correspondence of blocks and block_lists.
9819 2017-08-01 Richard Biener <rguenther@suse.de>
9821 PR tree-optimization/81181
9822 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9823 (compute_antic): ... end of iteration here.
9825 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
9827 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
9828 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
9829 (ftree-slp-vectorize): Likewise.
9830 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
9831 can no longer be set independent of flag_tree_loop_vectorize.
9832 * omp-general.c (emp_max_vf): Likewise.
9833 * opts.c (enable_fdo_optimizations): Remove references to
9834 flag_tree_vectorize, these are now implicit.
9835 (common_handle_option): Remove handling for OPT_ftree_vectorize,
9836 and leave it for the options machinery.
9838 2017-08-01 Martin Liska <mliska@suse.cz>
9841 * builtins.c (expand_builtin_memcpy_args): Remove.
9842 (expand_builtin_memcpy): Call newly added function
9843 expand_builtin_memory_copy_args.
9844 (expand_builtin_memcpy_with_bounds): Likewise.
9845 (expand_builtin_mempcpy): Remove last argument.
9846 (expand_builtin_mempcpy_with_bounds): Likewise.
9847 (expand_builtin_memory_copy_args): New function created from
9848 expand_builtin_mempcpy_args with small modifications.
9849 (expand_builtin_mempcpy_args): Remove.
9850 (expand_builtin_stpcpy): Remove unused argument.
9851 (expand_builtin): Likewise.
9852 (expand_builtin_with_bounds): Likewise.
9854 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
9857 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
9858 print "ds:" only for immediates in generic address space.
9860 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
9863 * config/i386/i386.c (ix86_funciton_naked): New prototype.
9864 (ix86_function_ok_for_sibcall): Return false for naked functions.
9866 2017-08-01 Richard Biener <rguenther@suse.de>
9868 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
9869 (compute_antic): Seed worklist with exit block predecessors.
9870 * cfganal.c (dfs_find_deadend): For a cycle return the source
9871 of the edge closing it.
9873 2017-08-01 Tamar Christina <tamar.christina@arm.com>
9875 * config/aarch64/aarch64.c
9876 (aarch64_can_const_movi_rtx_p): Move 0 check.
9878 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9880 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
9881 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
9883 * match.pd: Ditto in address comparison pattern.
9885 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9887 PR tree-optimization/81627
9888 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
9889 closed ssa form for store-store chain.
9891 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9893 PR tree-optimization/81620
9894 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
9895 for store-store chain.
9897 2017-08-01 Jakub Jelinek <jakub@redhat.com>
9899 PR tree-optimization/81588
9900 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
9901 ranges[i].in_p, invert comparison code ccode. For >/>=,
9902 swap rhs1 and rhs2 and comparison code unconditionally,
9903 for </<= don't do that. Don't swap rhs1/rhs2 again if
9904 ranges[i].in_p, instead invert comparison code ccode if
9905 opcode or oe->rank is BIT_IOR_EXPR.
9908 * optabs.def (vec_extract_optab, vec_init_optab): Change from
9909 a direct optab to conversion optab.
9910 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
9911 with GET_MODE_INNER as last argument instead of optab_handler.
9912 * expmed.c (extract_bit_field_1): Likewise. Use vector from
9913 vector extraction if possible and optab is available.
9914 * expr.c (store_constructor): Use convert_optab_handler instead
9915 of optab_handler. Use vector initialization from smaller
9916 vectors if possible and optab is available.
9917 * tree-vect-stmts.c (vectorizable_load): Likewise.
9918 * doc/md.texi (vec_extract, vec_init): Document that the optabs
9920 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
9921 of vec_init from half-sized vectors with the same element mode.
9922 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
9923 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
9924 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
9925 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
9926 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
9927 after mode in gen_vec_extract* calls.
9928 (vec_extract<mode>): Renamed to ...
9929 (vec_extract<mode><ssescalarmodelower>): ... this.
9930 (vec_extract<mode><ssehalfvecmodelower>): New expander.
9931 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
9932 element mode after mode in gen_vec_init* calls.
9933 (VEC_INIT_HALF_MODE): New mode iterator.
9934 (vec_init<mode>): Renamed to ...
9935 (vec_init<mode><ssescalarmodelower>): ... this.
9936 (vec_init<mode><ssehalfvecmodelower>): New expander.
9937 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
9938 (vec_extractv2sfsf): ... this.
9939 (vec_initv2sf): Renamed to ...
9940 (vec_initv2sfsf): ... this.
9941 (vec_extractv2si): Renamed to ...
9942 (vec_extractv2sisi): ... this.
9943 (vec_initv2si): Renamed to ...
9944 (vec_initv2sisi): ... this.
9945 (vec_extractv4hi): Renamed to ...
9946 (vec_extractv4hihi): ... this.
9947 (vec_initv4hi): Renamed to ...
9948 (vec_initv4hihi): ... this.
9949 (vec_extractv8qi): Renamed to ...
9950 (vec_extractv8qiqi): ... this.
9951 (vec_initv8qi): Renamed to ...
9952 (vec_initv8qiqi): ... this.
9953 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
9954 (vec_init<mode>): Renamed to ...
9955 (vec_init<mode><VEC_base_l>): ... this.
9956 (vec_extract<mode>): Renamed to ...
9957 (vec_extract<mode><VEC_base_l>): ... this.
9958 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
9959 (vec_initv2sfsf): ... this.
9960 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
9961 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
9962 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
9963 element mode after mode in gen_vec_init* calls.
9964 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
9965 (vec_init<mode><Vel>): ... this.
9966 (vec_extract<mode>): Renamed to ...
9967 (vec_extract<mode><Vel>): ... this.
9968 * config/aarch64/iterators.md (Vel): New mode attribute.
9969 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
9970 Add element mode after mode in gen_vec_extract* calls.
9971 * config/s390/vector.md (non_vec_l): New mode attribute.
9972 (vec_extract<mode>): Renamed to ...
9973 (vec_extract<mode><non_vec_l>): ... this.
9974 (vec_init<mode>): Renamed to ...
9975 (vec_init<mode><non_vec_l>): ... this.
9976 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
9977 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
9979 * config/arm/iterators.md (V_elem_l): New mode attribute.
9980 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
9981 (vec_extract<mode><V_elem_l>): ... this.
9982 (vec_extractv2di): Renamed to ...
9983 (vec_extractv2didi): ... this.
9984 (vec_init<mode>): Renamed to ...
9985 (vec_init<mode><V_elem_l>): ... this.
9986 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
9987 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
9988 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
9989 Add element mode after gen_vec_extract* calls.
9990 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
9991 (vec_init<mode><unitmode>): ... this.
9992 (vec_extract<mode>): Renamed to ...
9993 (vec_extract<mode><unitmode>): ... this.
9994 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
9995 (vec_init<mode><unitmode>): ... this.
9996 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
9997 (vec_initv2sfsf): ... this.
9998 (vec_extractv2sf): Renamed to ...
9999 (vec_extractv2sfsf): ... this.
10000 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
10001 Add element mode after gen_vec_extract* calls.
10002 * config/mips/mips.md (unitmode): New mode iterator.
10003 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
10004 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
10005 * config/spu/spu.md (inner_l): New mode attribute.
10006 (vec_init<mode>): Renamed to ...
10007 (vec_init<mode><inner_l>): ... this.
10008 (vec_extract<mode>): Renamed to ...
10009 (vec_extract<mode><inner_l>): ... this.
10010 * config/sparc/sparc.md (veltmode): New mode iterator.
10011 (vec_init<VMALL:mode>): Renamed to ...
10012 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
10013 * config/ia64/vect.md (vec_initv2si): Renamed to ...
10014 (vec_initv2sisi): ... this.
10015 (vec_initv2sf): Renamed to ...
10016 (vec_initv2sfsf): ... this.
10017 (vec_extractv2sf): Renamed to ...
10018 (vec_extractv2sfsf): ... this.
10019 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
10020 (vec_init<mode>): Renamed to ...
10021 (vec_init<mode><VEC_base_l>): ... this.
10022 (vec_extract<mode>): Renamed to ...
10023 (vec_extract<mode><VEC_base_l>): ... this.
10024 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
10025 (vec_initv2sfsf): ... this.
10026 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
10027 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
10028 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
10029 gen_vec_init* calls.
10031 2017-08-01 Richard Biener <rguenther@suse.de>
10033 PR tree-optimization/81297
10034 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
10035 TREE_OVERFLOW from INTEGER_CSTs.
10037 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
10039 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
10041 2017-07-31 Carl Love <cel@us.ibm.com>
10043 * config/rs6000/rs6000-c: Add support for built-in functions
10044 vector signed char vec_xl_be (signed long long, signed char *);
10045 vector unsigned char vec_xl_be (signed long long, unsigned char *);
10046 vector signed int vec_xl_be (signed long long, signed int *);
10047 vector unsigned int vec_xl_be (signed long long, unsigned int *);
10048 vector signed long long vec_xl_be (signed long long, signed long long *);
10049 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
10050 vector signed short vec_xl_be (signed long long, signed short *);
10051 vector unsigned short vec_xl_be (signed long long, unsigned short *);
10052 vector double vec_xl_be (signed long long, double *);
10053 vector float vec_xl_be (signed long long, float *);
10054 * config/rs6000/altivec.h (vec_xl_be): Add #define.
10055 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
10056 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
10058 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
10059 (altivec_expand_builtin): Add switch statement to call
10060 altivec_expand_xl_be for each builtin.
10061 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
10062 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
10063 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
10064 __builtin_vsx_le_be_v16qi.
10065 * doc/extend.texi: Update the built-in documentation file for the
10066 new built-in functions.
10068 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
10071 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
10073 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
10075 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10077 * config.gcc: Add z14.
10078 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
10079 CPU model numbers for z13s and z14.
10080 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
10082 * config/s390/s390-opts.h (enum processor_type): Rename
10083 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
10084 * config/s390/s390.c (processor_table): Add field for CPU name to
10085 be passed to Binutils.
10086 (s390_asm_output_machine_for_arch): Use the new field in
10087 processor_table for Binutils.
10088 (s390_expand_builtin): Replace arch12 with z14.
10089 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
10090 (s390_get_sched_attrmask): Likewise.
10091 (s390_get_unit_mask): Likewise.
10092 * config/s390/s390.opt: Add z14 to processor_type enum.
10094 2017-07-31 Martin Jambor <mjambor@suse.cz>
10097 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
10098 regardless of optimization level.
10100 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
10101 Martin Liska <mliska@suse.cz>
10103 * predict.def: Remove old comment and adjust probability.
10104 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
10105 PREDICT statements.
10107 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
10110 * config/i386/i386.c (ix86_function_naked): New function.
10111 (ix86_can_use_return_insn_p): Return false for naked functions.
10112 (ix86_expand_prologue): Skip prologue for naked functions.
10113 (ix86_expand_epilogue): Skip epilogue for naked functions
10114 and emit trap instruction.
10115 (ix86_warn_func_return): New function.
10116 (ix86_attribute_table): Add "naked" attribute specification.
10117 (TARGET_WARN_FUNC_RETURN): Define.
10118 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
10120 2017-07-31 Martin Liska <mliska@suse.cz>
10122 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
10123 (dump_gimple_bb_header): Always dump BB info.
10124 (pp_cfg_jump): Do not append info about BB when dumping a jump.
10126 2017-07-31 Martin Liska <mliska@suse.cz>
10129 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
10130 also with current_function_decl non-null equality.
10132 2017-07-31 Jakub Jelinek <jakub@redhat.com>
10135 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
10136 change type to the element type, instead add eltype variable and
10137 use it where we are interested in the element type.
10139 PR tree-optimization/81603
10140 * ipa-polymorphic-call.c
10141 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
10142 offset arithmetic in offset_int, bail out if the resulting bit offset
10143 doesn't fit into shwi.
10145 2017-07-31 Martin Liska <mliska@suse.cz>
10147 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
10148 (gimplify_save_expr): Fix comment.
10150 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
10153 * config/i386/i386.c (ix86_function_arg): Update arguments for
10155 (ix86_compute_frame_layout): Set the initial stack offset to
10156 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
10157 INCOMING_FRAME_SP_OFFSET.
10158 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
10159 stack before exception handler returns.
10160 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
10161 the 'ERROR_CODE' for exception handler.
10163 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
10165 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
10166 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
10167 (ASM_OUTPUT_REG_POP): Ditto.
10168 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
10169 instead of asm_fprintf to output pure string.
10171 2017-07-29 Jakub Jelinek <jakub@redhat.com>
10173 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
10174 to imported_module_or_decl hook.
10175 (debug_nothing_tree_tree_tree_bool): Remove.
10176 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
10177 * debug.c (do_nothing_debug_hooks): Use
10178 debug_nothing_tree_tree_tree_bool_bool instead of
10179 debug_nothing_tree_tree_tree_bool.
10180 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
10181 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
10182 * sdbout.c (sdb_debug_hooks): Likewise.
10183 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
10184 (gen_namespace_die): Add DW_AT_export_symbols attribute if
10186 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
10187 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
10188 attribute, don't add anything.
10190 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10192 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
10193 (fold_build2_stat_loc): Likewise.
10194 (fold_build3_stat_loc): Likewise.
10195 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
10196 (fold_build1_loc): Remove macro.
10197 (fold_build2_loc): Likewise.
10198 (fold_build3_loc): Likewise.
10200 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10202 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
10203 (gimple_build_debug_bind_source_stat): Likewise.
10204 * gimple.h (gimple_build_debug_bind): Remove macro.
10205 (gimple_build_debug_bind_source): Likewise.
10207 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10209 * bitmap.c (bitmap_alloc): Adjust.
10210 (bitmap_gc_alloc): Likewise.
10211 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
10213 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10215 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
10216 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
10217 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
10218 (bitmap_gc_alloc_stat): Likewise.
10219 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
10221 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10223 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
10224 * rtl.h (shallow_copy_rtx): Remove macro.
10226 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10228 * emit-rtl.c (gen_raw_REG): Adjust.
10229 * gengenrtl.c (gendef): Likewise.
10230 * rtl.c (rtx_alloc_stat): Remove _stat from name.
10231 * rtl.h (rtx_alloc): Remove macro.
10233 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10235 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
10236 (build_tree_list_stat): Likewise.
10237 * tree.h (build_tree_list): Remove macro.
10238 (build_tree_list_vec): Likewise.
10240 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10242 * tree.c (make_vector_stat): Remove _stat from name.
10243 (build_vector_stat): Likewise.
10244 * tree.h (make_vector_stat): Remove macro.
10245 (build_vector_stat): Likewise.
10247 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10249 * tree.h (build_var_debug_value): Remove prototype.
10251 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10253 * tree.c (tree_cons_stat): Remove _stat from name.
10254 * tree.h (tree_cons): Remove macro.
10256 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10258 * tree.c (build_vl_exp_stat): Remove _stat from name.
10259 * tree.h (build_vl_exp): Remove macro.
10261 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10263 * tree.c (build_decl_stat): Remove _stat from name.
10264 * tree.h (build_decl): Remove macro.
10266 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10268 * gimple.c (gimple_build_with_ops_stat): Adjust.
10269 (gimple_alloc_stat): Remove _stat from name.
10270 * gimple.h (gimple_alloc): Remove macro.
10272 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10274 * tree.c (make_tree_vec_stat): Remove _stat from name.
10275 (grow_tree_vec_stat): Likewise.
10276 * tree.h (make_tree_vec_stat): Adjust prototype.
10277 (grow_tree_vec_stat): Likewise.
10278 (make_tree_vec): Remove macro.
10279 (grow_tree_vec): Likewise.
10281 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10283 * fold-const.c (fold_build1_stat_loc): Adjust.
10284 (fold_build2_stat_loc): Likewise.
10285 (fold_build3_stat_loc): Likewise.
10286 * tree.c (build0_stat): Remove _stat from name.
10287 (build1_stat): Likewise.
10288 (build2_stat): Likewise.
10289 (build3_stat): Likewise.
10290 (build4_stat): Likewise.
10291 (build5_stat): Likewise.
10292 * tree.h (build1_loc): Remove macro, and rename _stat function
10294 (build2_loc): Likewise.
10295 (build3_loc): Likewise.
10296 (build4_loc): Likewise.
10297 (build5_loc): Likewise.
10299 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10301 * tree.c (make_int_cst_stat): Remove _stat from name.
10302 * tree.h (make_int_cst_stat): Adjust prototype.
10303 (make_int_cst): Remove macro.
10305 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10307 * tree.c (make_tre_binfo_stat): Remove _stat from name.
10308 * tree.h (make_tree_binfo_stat): Adjust prototype.
10309 (make_tree_binfo): Remove.
10311 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10313 * tree.c (copy_node_stat): Rename to copy_node.
10314 (build_distinct_type_copy): Adjust.
10315 * tree.h (copy_node_stat): Adjust prototype.
10316 (copy_node): Remove macro.
10318 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10320 * tree.c (make_node_stat): rename to make_node.
10321 (build_tree_list_stat): Adjust.
10322 (build0_stat): Likewise.
10323 (build2_stat): Likewise.
10324 (build3_stat): Likewise.
10325 (build4_stat): Likewise.
10326 (build5_stat): Likewise.
10327 (build_decl_stat): Likewise.
10328 * tree.h (make_node_stat): Adjust prototype.
10329 (make_node): remove macro.
10331 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
10333 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
10334 (PPC_FEATURE2_SCV): Likewise.
10335 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
10337 2017-07-28 Tamar Christina <tamar.christina@arm.com>
10339 * config/aarch64/aarch64.c
10340 (aarch64_internal_mov_immediate): Add new special pattern.
10341 * config/aarch64/aarch64.md (*movdi_aarch64):
10342 Add reg/32bit const mov case.
10344 2017-07-28 Tamar Christina <tamar.christina@arm.com>
10345 Richard Sandiford <richard.sandiford@linaro.org>
10347 * config/aarch64/aarch64.md (mov<mode>): Generalize.
10348 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
10349 Add integer and movi cases.
10350 (movi-split-hf-df-sf split, fp16): New.
10351 (enabled): Added TARGET_FP_F16INST.
10352 * config/aarch64/iterators.md (GPF_HF): New.
10353 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
10355 2017-07-28 Tamar Christina <tamar.christina@arm.com>
10357 * config/aarch64/aarch64.c
10358 (aarch64_simd_container_mode): Add prototype.
10359 (aarch64_expand_mov_immediate): Add HI support.
10360 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
10361 (aarch64_can_const_movi_rtx_p): New.
10362 (aarch64_preferred_reload_class):
10363 Remove restrictions of using FP registers for certain SIMD operations.
10364 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
10365 (aarch64_valid_floating_const): Add integer move validation.
10366 (aarch64_simd_imm_scalar_p): Remove.
10367 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
10368 (aarch64_legitimate_constant_p): Expand list of supported cases.
10369 * config/aarch64/aarch64-protos.h
10370 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
10371 (aarch64_reinterpret_float_as_int): New.
10372 (aarch64_simd_imm_scalar_p): Remove.
10373 * config/aarch64/constraints.md (Uvi): New.
10374 (Dd): Split into Ds and new Dd.
10375 * config/aarch64/aarch64.md (*movsi_aarch64):
10377 (*movdi_aarch64): Add SIMD mov case.
10379 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10381 * tree-predcom.c: (struct chain): Handle store-store chain in which
10382 stores for elimination only store loop invariant values.
10383 (execute_pred_commoning_chain): Ditto.
10384 (prepare_initializers_chain_store_elim): Ditto.
10385 (prepare_finalizers): Ditto.
10386 (is_inv_store_elimination_chain): New function.
10387 (initialize_root_vars_store_elim_1): New function.
10389 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10391 * tree-predcom.c: Revise general description of the pass.
10392 (enum chain_type): New enum type for store elimination.
10393 (struct chain): New field supporting store elimination.
10394 (struct component): Ditto.
10395 (dump_chain): Dump store-stores chain.
10396 (release_chain): Release resources.
10397 (split_data_refs_to_components): Compute and create component
10398 contains only stores for elimination.
10399 (get_chain_last_ref_at): New function.
10400 (make_invariant_chain): Initialization.
10401 (make_rooted_chain): Specify chain type in parameter and record it.
10402 (add_looparound_copies): Skip for store-stores chain.
10403 (determine_roots_comp): Compute type of chain and pass it to
10405 (initialize_root_vars_store_elim_2): New function.
10406 (finalize_eliminated_stores): New function.
10407 (remove_stmt): Handle store for elimination.
10408 (execute_pred_commoning_chain): Execute predictive commoning on
10409 store-store chains.
10410 (determine_unroll_factor): Skip unroll for store-stores chain.
10411 (prepare_initializers_chain_store_elim): New function.
10412 (prepare_initializers_chain): Hanlde store-store chain.
10413 (prepare_finalizers_chain, prepare_finalizers): New function.
10414 (tree_predictive_commoning_loop): Return integer value indicating
10415 if loop is unrolled or lcssa form is corrupted.
10416 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
10418 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10420 * tree-predcom.c (initialize_root): Delete.
10421 (execute_pred_commoning_chain): Initialize root vars and replace
10422 reference of non-combined chain directly, rather than call above
10425 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10427 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
10428 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
10430 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10432 * tree-predcom.c (struct chain): New field init_seq.
10433 (release_chain): Release init_seq.
10434 (prepare_initializers_chain): Record intialization stmts in above
10436 (insert_init_seqs): New function.
10437 (tree_predictive_commoning_loop): Call insert_init_seqs.
10439 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10441 * tree-predcom.c (determine_roots_comp): Skip trivial components.
10443 2017-07-28 Richard Biener <rguenther@suse.de>
10445 * match.pd: Remove superfluous :c.
10446 * genmatch.c (simplify::id): Add member.
10447 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
10449 (current_id): New global.
10450 (dt_node::parent): Move from ...
10451 (dt_operand::parent): ... here. Add for_id member.
10452 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
10453 (decision_tree::find_node): Relax order requirement when
10454 merging DT_TRUE nodes to ones inbetween the current simplify
10455 and the one we try to merge with. Add diagnostic whenever
10456 we need to enforce pattern order by not merging.
10457 (decision_tree::insert): Set current_id.
10458 (decision_tree::print_node): Dump parent node and for_id.
10459 (parser::last_id): Add member.
10460 (parser::push_simplify): Assign unique id.
10461 (parser::parser): Initialize last_id.
10463 2017-07-28 Martin Liska <mliska@suse.cz>
10466 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
10467 gimple_build_debug_bind.
10469 2017-07-28 Richard Biener <rguenther@suse.de>
10471 PR tree-optimization/81502
10472 * match.pd: Add pattern combining BIT_INSERT_EXPR with
10474 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
10476 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
10477 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
10478 for BIT_FIELD_REF args.
10479 * fold-const.c (make_bit_field_ref): Likewise.
10480 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
10482 2017-07-28 Jakub Jelinek <jakub@redhat.com>
10485 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
10486 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
10487 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
10488 Or it into SANITIZER_UNDEFINED.
10489 * ubsan.c: Include gimple-fold.h and varasm.h.
10490 (ubsan_expand_ptr_ifn): New function.
10491 (instrument_pointer_overflow): New function.
10492 (maybe_instrument_pointer_overflow): New function.
10493 (instrument_object_size): Formatting fix.
10494 (pass_ubsan::execute): Call instrument_pointer_overflow
10495 and maybe_instrument_pointer_overflow.
10496 * internal-fn.c (expand_UBSAN_PTR): New function.
10497 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
10498 * sanitizer.def (__ubsan_handle_pointer_overflow,
10499 __ubsan_handle_pointer_overflow_abort): New builtins.
10500 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
10501 * internal-fn.def (UBSAN_PTR): New internal function.
10502 * opts.c (sanitizer_opts): Add pointer-overflow.
10503 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
10504 * fold-const.c (build_range_check): Compute pointer range check in
10505 integral type if pointer arithmetics would be needed. Formatting
10508 2017-07-28 Martin Liska <mliska@suse.cz>
10511 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
10512 parameters that are of a variable-length.
10514 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
10516 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
10518 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
10519 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
10520 (CRT_CALL_STATIC_FUNCTION): Likewise.
10521 (ASM_DEFAULT_SPEC): New define.
10522 (ASM_SPEC32): Likewise.
10523 (ASM_SPEC64): Likewise.
10524 (ASM_SPEC_COMMON): Likewise.
10525 (ASM_SPEC): Likewise.
10526 (INVALID_64BIT): Likewise.
10527 (LINK_OS_DEFAULT_SPEC): Likewise.
10528 (LINK_OS_SPEC32): Likewise.
10529 (LINK_OS_SPEC64): Likewise.
10530 (POWERPC_LINUX): Likewise.
10531 (PTRDIFF_TYPE): Likewise.
10532 (RESTORE_FP_PREFIX): Likewise.
10533 (RESTORE_FP_SUFFIX): Likewise.
10534 (SAVE_FP_PREFIX): Likewise.
10535 (SAVE_FP_SUFFIX): Likewise.
10536 (SIZE_TYPE): Likewise.
10537 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
10538 (TARGET_64BIT): Likewise.
10539 (TARGET_64BIT): Likewise.
10540 (TARGET_AIX): Likewise.
10541 (WCHAR_TYPE_SIZE): Likewise.
10542 (WCHAR_TYPE): Undefine.
10543 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
10544 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
10545 (CPP_OS_RTEMS_SPEC): Delete.
10546 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
10547 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
10549 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
10551 2017-07-28 Jakub Jelinek <jakub@redhat.com>
10553 PR tree-optimization/81578
10554 * tree-parloops.c (build_new_reduction): Bail out if
10555 reduction_code isn't one of the standard OpenMP reductions.
10556 Move the details printing after that decision.
10558 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
10560 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
10561 related to reload_in_progress.
10562 (splat_input_operand): Likewise.
10563 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
10565 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
10567 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
10568 (TARGET_INSTANTIATE_DECLS): Likewise.
10569 (legitimate_indexed_address_p): Delete reload_in_progress code.
10570 (rs6000_debug_legitimate_address_p): Likewise.
10571 (rs6000_eliminate_indexed_memrefs): Likewise.
10572 (rs6000_emit_le_vsx_store): Likewise.
10573 (rs6000_emit_move_si_sf_subreg): Likewise.
10574 (rs6000_emit_move): Likewise.
10575 (register_to_reg_type): Likewise.
10576 (rs6000_pre_atomic_barrier): Likewise.
10577 (rs6000_machopic_legitimize_pic_address): Likewise.
10578 (rs6000_allocate_stack_temp): Likewise.
10579 (rs6000_address_for_fpconvert): Likewise.
10580 (rs6000_address_for_altivec): Likewise.
10581 (rs6000_secondary_memory_needed_rtx): Delete function.
10582 (rs6000_check_sdmode): Likewise.
10583 (rs6000_alloc_sdmode_stack_slot): Likewise.
10584 (rs6000_instantiate_decls): Likewise.
10585 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
10586 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
10587 Delete reload_in_progress.
10588 (*vec_reload_and_plus_<mptrsize>): Likewise.
10589 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
10590 (vsx_div_v2di): Likewise.
10591 (vsx_udiv_v2di): Likewise.
10593 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
10595 * config/rs6000/rs6000.opt (mlra): Replace with stub.
10596 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
10597 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
10598 (rs6000_debug_reg_global): Delete print of LRA status.
10599 (rs6000_option_override_internal): Delete dead LRA related code.
10600 (rs6000_lra_p): Delete function.
10601 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
10603 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
10605 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
10606 * config/riscv/rtems.h: New file.
10608 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10609 Sudakshina Das <sudi.das@arm.com>
10611 * config/aarch64/aarch64.md
10612 (define_split for and<mode>3nr_compare): Move
10613 non aarch64_logical_operand to a register.
10614 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
10615 register immediate operand to a register.
10616 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
10618 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
10620 PR middle-end/81564
10621 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
10623 2017-07-27 Richard Biener <rguenther@suse.de>
10625 PR tree-optimization/81573
10626 PR tree-optimization/81494
10627 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
10628 multi defuse cycle case.
10630 2017-07-27 Richard Biener <rguenther@suse.de>
10632 PR tree-optimization/81571
10633 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
10636 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
10638 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
10639 earlier and only if MASK_FPU is set. Adjust formatting.
10641 2017-07-27 Martin Liska <mliska@suse.cz>
10643 * opt-functions.awk: Add validation of value of Init.
10644 * optc-gen.awk: Pass new argument.
10646 2017-07-27 Martin Liska <mliska@suse.cz>
10648 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
10649 Fix wrong condition.
10651 2017-07-27 Martin Liska <mliska@suse.cz>
10653 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
10654 BBs and edges seen by autoFDO.
10656 2017-07-27 Richard Biener <rguenther@suse.de>
10658 PR tree-optimization/81502
10659 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
10660 with incompatible but same sized type.
10661 (execute_update_addresses_taken): Likewise.
10663 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
10665 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
10666 flag_tree_loop_vectorize rather than flag_tree_vectorize.
10668 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10671 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
10672 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
10673 Change s_operand to memory_operand.
10675 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
10677 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
10678 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
10679 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
10680 Emit instructions rather than returning an expression. Handle TFmode
10681 and KFmode by casting to TImode.
10682 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
10683 (rs6000_emit_le_vsx_store): Likewise.
10684 * config/rs6000/vsx.md (VSX_TI): New iterator.
10685 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
10686 (*vsx_le_undo_permute_<mode>): Likewise.
10687 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
10688 emit the split sequence.
10689 (*vsx_le_perm_store_<mode>): Likewise.
10691 2017-07-27 Jakub Jelinek <jakub@redhat.com>
10693 PR tree-optimization/81555
10694 PR tree-optimization/81556
10695 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
10696 if true, force CHANGED for the recursive invocation.
10697 (reassociate_bb): Remember original length of ops array, pass
10698 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
10700 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
10701 attributes for noipa attribute. For naked attribute use
10702 lookup_attribute first before lookup_attribute_spec.
10703 * final.c (rest_of_handle_final): Disable IPA RA for functions with
10705 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
10706 for functions with noipa attribute.
10707 (cgraph_externally_visible_p): Return true for functions with noipa
10709 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
10710 for functions with noipa attribute.
10711 * doc/extend.texi: Document noipa function attribute.
10712 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
10713 also for functions with noipa attribute.
10714 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
10716 2017-07-26 Andrew Pinski <apinski@cavium.com>
10718 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
10719 vec_unalign_load_cost and vec_unalign_store_cost.
10721 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10723 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
10724 -mvsx-small-integer option.
10725 (ISA_3_0_MASKS_IEEE): Likewise.
10726 (OTHER_VSX_VECTOR_MASKS): Likewise.
10727 (POWERPC_MASKS): Likewise.
10728 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
10729 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
10730 code, only testing for DImode being allowed in non-VSX floating
10732 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
10733 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
10735 (rs6000_option_override_internal): Delete -mvsx-small-integer.
10736 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
10737 TARGET_P8_VECTOR test.
10738 (rs6000_secondary_reload_simple_move): Likewise.
10739 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
10740 since TARGET_P9_VECTOR was already tested.
10741 (rs6000_opt_masks): Remove -mvsx-small-integer.
10742 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
10743 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
10745 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
10746 test for TARGET_VEXTRACTUB was used, and that uses
10748 (p9 extract splitter): Likewise.
10749 (vsx_extract_<mode>_di_p9): Likewise.
10750 (vsx_extract_<mode>_store_p9): Likewise.
10751 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
10752 for TARGET_P9_VECTOR was used. Delete code that is now dead with
10753 the elimination of TARGET_VSX_SMALL_INTEGER.
10754 (vsx_extract_<mode>_p8): Likewise.
10755 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
10756 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
10757 (vsx_set_<mode>_p9): Likewise.
10758 (vsx_set_v4sf_p9): Likewise.
10759 (vsx_set_v4sf_p9_zero): Likewise.
10760 (vsx_insert_extract_v4sf_p9): Likewise.
10761 (vsx_insert_extract_v4sf_p9_2): Likewise.
10762 * config/rs6000/rs6000.md (sign extend splitter): Change
10763 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
10764 (floatsi<mode>2_lfiwax_mem): Likewise.
10765 (floatunssi<mode>2_lfiwzx_mem): Likewise.
10766 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
10767 since a test for TARGET_P9_VECTOR was used.
10768 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10769 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
10770 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10771 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
10772 TARGET_P8_VECTOR test.
10773 (fix_trunc<mode>si2_stfiwx): Likewise.
10774 (fix_trunc<mode>si2_internal): Likewise.
10775 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
10776 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
10778 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
10779 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
10780 TARGET_P8_VECTOR test.
10781 (fixuns_trunc<mode>si2_stfiwx): Likewise.
10782 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
10783 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
10785 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
10786 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
10787 since a test for TARGET_P9_VECTOR was used.
10788 (splitter for loading small constants): Likewise.
10790 2017-07-26 Andrew Pinski <apinski@cavium.com>
10792 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
10795 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
10798 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
10799 (fp_valid_at): Likewise.
10801 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
10803 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
10804 (qdf24xx_addrcost_table): Likewise.
10805 (cortexa57_tunings): Update to use generic_branch_cost.
10806 (cortexa72_tunings): Likewise.
10807 (cortexa73_tunings): Likewise.
10808 (qdf24xx_tunings): Likewise.
10810 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
10812 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
10813 (thunderx2t99_branch_cost): Likewise.
10814 (cortexa35_tunings): Update to use generic_branch_cost.
10815 (cortexa53_tunings): Likewise.
10816 (cortexa57_tunings): Likewise.
10817 (cortexa72_tunings): Likewise.
10818 (cortexa73_tunings): Likewise.
10819 (thunderx2t99_tunings): Likewise.
10821 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
10823 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
10824 (sparc_option_override): Honour MASK_FSMULD.
10825 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
10826 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
10827 * config/sparc/sparc.opt (mfsmuld): New option.
10828 * doc/invoke.texi (mfsmuld): Document option.
10830 2017-07-26 Marek Polacek <polacek@redhat.com>
10832 PR middle-end/70992
10833 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
10835 2017-07-26 Richard Biener <rguenther@suse.de>
10837 * gimple-match-head.c (do_valueize): Return OP if valueize
10839 (get_def): New helper to get at the def stmt of a SSA name
10840 if valueize allows.
10841 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
10842 do_valueize to get at the def stmt.
10843 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
10845 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
10847 PR middle-end/46932
10848 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
10850 2017-07-26 Martin Liska <mliska@suse.cz>
10853 * function.c (expand_function_start): Make expansion of
10854 nonlocal_goto_save_area after parm_birth_insn.
10856 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
10858 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
10859 from all CPU target flags enable members.
10861 2017-07-26 Richard Biener <rguenther@suse.de>
10863 * genmatch.c (dt_simplify::gen): Make iterator vars const.
10864 (decision_tree::gen): Make 'type' const.
10865 (write_predicate): Likewise.
10867 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
10869 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
10870 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
10871 (rs6000_option_override_internal): Likewise.
10872 (rs6000_expand_vector_set): Likewise.
10873 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
10874 (TARGET_UPPER_REGS_SF): Likewise.
10875 (TARGET_UPPER_REGS_DI): Likewise.
10876 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
10877 (TARGET_DIRECT_MOVE_64BIT): Likewise.
10878 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
10879 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10880 (Splitters for DI constants in Altivec registers): Likewise.
10881 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
10882 (vsx_set_v4sf_p9): Likewise.
10883 (vsx_set_v4sf_p9_zero): Likewise.
10884 (vsx_insert_extract_v4sf_p9): Likewise.
10885 (vsx_insert_extract_v4sf_p9_2): Likewise.
10887 2017-07-25 Carl Love <cel@us.ibm.com>
10889 * doc/extend.texi: Update the built-in documentation file for the
10890 existing built-in functions
10891 vector signed char vec_cnttz (vector signed char);
10892 vector unsigned char vec_cnttz (vector unsigned char);
10893 vector signed short vec_cnttz (vector signed short);
10894 vector unsigned short vec_cnttz (vector unsigned short);
10895 vector signed int vec_cnttz (vector signed int);
10896 vector unsigned int vec_cnttz (vector unsigned int);
10897 vector signed long long vec_cnttz (vector signed long long);
10898 vector unsigned long long vec_cnttz (vector unsigned long long);
10900 2017-07-25 Andrew Pinski <apinski@cavium.com>
10902 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
10903 accesses where the use is for the first operand of a BIT_INSERT.
10905 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
10908 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
10909 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
10911 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
10913 * config/i386/gstabs.h: Delete.
10914 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
10916 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
10918 * config/i386/i386.c (ix86_decompose_address): Do not check for
10919 register RTX when looking at index_reg or base_reg.
10920 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
10922 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
10924 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
10925 to update EH info here.
10927 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10929 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
10931 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10933 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
10935 2017-07-25 Torsten Duwe <duwe@suse.de>
10937 * common.opt: Introduce -fpatchable-function-entry
10938 command line option, and its variables function_entry_patch_area_size
10939 and function_entry_patch_area_start.
10940 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
10941 including a two-value parser.
10942 * target.def (print_patchable_function_entry): New target hook.
10943 * targhooks.h (default_print_patchable_function_entry): New function.
10944 * targhooks.c (default_print_patchable_function_entry): Likewise.
10945 * toplev.c (process_options): Switch off IPA-RA if
10946 patchable function entries are being generated.
10947 * varasm.c (assemble_start_function): Look at the
10948 patchable-function-entry command line switch and current
10949 function attributes and maybe generate NOP instructions by
10950 calling the print_patchable_function_entry hook.
10951 * doc/extend.texi: Document patchable_function_entry attribute.
10952 * doc/invoke.texi: Document -fpatchable_function_entry
10953 command line option.
10954 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
10956 * doc/tm.texi: Re-generate.
10958 2017-07-25 Jakub Jelinek <jakub@redhat.com>
10961 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
10962 TARGET_AVX512DQ rather than TARGET_AVX512BW.
10964 2017-07-25 Tamar Christina <tamar.christina@arm.com>
10966 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
10968 2017-07-25 Richard Biener <rguenther@suse.de>
10970 PR tree-optimization/81455
10971 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
10972 not walk in cycles when looking for guards.
10974 2017-07-25 Richard Biener <rguenther@suse.de>
10976 PR tree-optimization/81529
10977 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
10978 when optimizing backedge uses.
10980 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
10982 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
10984 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
10985 to dl_section_ref. On AIX, append an expression to subtract
10986 the size of the section length to dl_section_ref.
10988 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
10990 * configure.ac: If any of the config.* scripts fail, exit 1.
10991 * configure: Regenerate.
10993 2017-07-25 Richard Biener <rguenther@suse.de>
10995 PR middle-end/81546
10996 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
10997 of immediate uses, be more verbose on errors.
10999 2017-07-25 Richard Biener <rguenther@suse.de>
11001 PR tree-optimization/81510
11002 * tree-vect-loop.c (vect_is_simple_reduction): When the
11003 reduction stmt is not inside the loop bail out.
11005 2017-07-25 Richard Biener <rguenther@suse.de>
11007 PR tree-optimization/81303
11008 * tree-vect-loop-manip.c (vect_loop_versioning): Build
11009 profitability check against LOOP_VINFO_NITERSM1.
11011 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
11013 * domwalk.c (cmp_bb_postorder): Simplify.
11014 (sort_bbs_postorder): New function. Use it...
11015 (dom_walker::walk): ...here to optimize common cases.
11017 2017-07-25 Martin Liska <mliska@suse.cz>
11020 * ipa-visibility.c (function_and_variable_visibility): Make the
11021 redirection just on target that supports aliasing.
11022 Fix GNU coding style.
11024 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
11027 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
11029 * config/arm/rtems.h: Likewise.
11030 * config/bfin/rtems.h: Likewise.
11031 * config/i386/rtemself.h: Likewise.
11032 * config/lm32/rtems.h: Likewise.
11033 * config/m32c/rtems.h: Likewise.
11034 * config/m68k/rtemself.h: Likewise.
11035 * config/microblaze/rtems.h: Likewise.
11036 * config/mips/rtems.h: Likewise.
11037 * config/moxie/rtems.h: Likewise.
11038 * config/nios2/rtems.h: Likewise.
11039 * config/powerpcspe/rtems.h: Likewise.
11040 * config/rs6000/rtems.h: Likewise.
11041 * config/rtems.h: Likewise.
11042 * config/sh/rtems.h: Likewise.
11043 * config/sh/rtemself.h: Likewise.
11044 * config/sparc/rtemself.h: Likewise.
11046 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
11049 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
11050 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
11051 * tree-ssa-structalias.c (alias_get_name): Same.
11053 2017-07-25 Bin Cheng <bin.cheng@arm.com>
11056 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
11057 instructions if no du chain is found.
11059 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
11061 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
11063 2017-07-25 Richard Biener <rguenther@suse.de>
11065 PR middle-end/81505
11066 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
11069 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
11071 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
11072 upper-regs options.
11073 (ISA_2_7_MASKS_SERVER): Likewise.
11074 (ISA_3_0_MASKS_IEEE): Likewise.
11075 (OTHER_P8_VECTOR_MASKS): Likewise.
11076 (OTHER_VSX_VECTOR_MASKS): Likewise.
11077 (POWERPC_MASKS): Likewise.
11078 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
11079 duplicate list of options.
11080 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
11081 explicit -mupper-regs options.
11082 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
11083 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
11084 alias for -mupper-regs-df.
11085 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
11086 (rs6000_init_hard_regno_mode_ok): Likewise.
11087 (rs6000_option_override_internal): Likewise.
11088 (rs6000_opt_masks): Likewise.
11089 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
11090 options in terms of whether -mvsx or -mpower8-vector was used.
11091 (TARGET_UPPER_REGS_DI): Likewise.
11092 (TARGET_UPPER_REGS_SF): Likewise.
11093 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
11094 -mupper-regs-* options.
11096 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
11098 * passes.c (emergency_dump_function): Print some empty lines and a
11099 header before the RTL dump.
11101 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
11103 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
11105 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
11108 * config/aarch64/aarch64.c (aarch64_classify_symbol):
11109 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
11111 2017-07-24 Carl Love <cel@us.ibm.com>
11113 * config/rs6000/rs6000-c.c: Add support for built-in functions
11114 vector float vec_extract_fp32_from_shorth (vector unsigned short);
11115 vector float vec_extract_fp32_from_shortl (vector unsigned short);
11116 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
11117 vec_extract_fp_from_shortl): Add defines for the two builtins.
11118 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
11119 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
11121 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
11122 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
11123 * doc/extend.texi: Update the built-in documentation file for the
11124 new built-in function.
11126 2017-07-24 Jakub Jelinek <jakub@redhat.com>
11129 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
11131 * doc/generic.texi: Likewise.
11132 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
11133 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
11135 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
11137 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
11138 (aarch64_mls_elt_merge<mode>): Likewise.
11140 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
11142 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
11143 having __cxa_atexit.
11145 2017-07-23 Michael Collison <michael.collison@arm.com>
11147 * config/arm/arm.c (arm_option_override): Deprecate
11148 use of -mstructure-size-boundary.
11149 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
11150 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
11152 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11155 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
11156 Reduce cost estimate for direct moves.
11158 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
11161 * config/i386/i386.c (ix86_option_override_internal): Disable
11162 BMI, BMI2 and TBM instructions for -m16.
11164 2017-07-21 Carl Love <cel@us.ibm.com>
11166 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11167 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11168 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11169 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11170 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11171 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11172 VMULOSW): New enum "unspec" values.
11173 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11174 altivec_vmulosw): New patterns.
11175 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11176 VMULOSW): Add definitions.
11178 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
11180 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
11181 (qdf24xx): Likewise.
11182 * config/aarch64/aarch64-options-extensions.def (rdma); New.
11183 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
11184 (AARCH64_FL_V8_1): Renumber.
11185 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
11186 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
11187 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
11188 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
11189 rdma to feature modifiers list.
11191 2017-07-21 Yury Gribov <tetra2005@gmail.com>
11193 PR middle-end/56727
11194 * ipa-visibility (function_and_variable_visibility): Convert
11195 recursive PLT call to direct call if appropriate.
11197 2017-07-21 Andrew Pinski <apinski@cavium.com>
11199 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
11200 operand 1 to see if the types precision matches.
11201 * fold-const.c (operand_equal_p): Likewise.
11203 2017-07-21 Richard Biener <rguenther@suse.de>
11205 PR tree-optimization/81303
11206 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
11207 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
11208 (vect_peeling_hash_get_lowest_cost): Adjust.
11209 (vect_enhance_data_refs_alignment): Likewise. Use
11210 vect_get_peeling_costs_all_drs to compute the penalty for no
11211 peeling to match up costs.
11213 2017-07-21 Richard Biener <rguenther@suse.de>
11215 PR tree-optimization/81500
11216 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
11217 we didn't identify a reduction path.
11219 2017-07-21 Tom de Vries <tom@codesourcery.com>
11220 Cesar Philippidis <cesar@codesourcery.com>
11222 PR gcov-profile/81442
11223 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
11226 2017-07-21 Tom de Vries <tom@codesourcery.com>
11229 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
11231 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
11232 nvptx_override_options_after_change.
11234 2017-07-21 Ulrich Drepper <drepper@redhat.com>
11236 * dwarf2out.c (output_file_names): Avoid double testing for
11237 dwarf_version >= 5.
11239 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
11241 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
11243 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
11245 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
11247 (try_crossjump_to_edge): Do not punt on partitioned functions.
11249 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
11251 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
11252 Put all BBs reachable only via paths crossing cold region to cold
11254 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
11256 2016-07-21 Richard Biener <rguenther@suse.de>
11258 PR tree-optimization/81303
11259 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
11260 into account prologue and epilogue iterations when raising
11261 min_profitable_iters to sth at least covering one vector iteration.
11263 2017-07-21 Tamar Christina <tamar.christina@arm.com>
11265 * config/arm/arm.c (arm_test_cpu_arch_dat):
11268 2017-07-20 Nathan Sidwell <nathan@acm.org>
11270 Remove TYPE_METHODS.
11271 * tree.h (TYPE_METHODS): Delete.
11272 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
11273 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
11274 (dbxout_type_methods): Scan TYPE_FIELDS.
11275 (dbxout_type): Don't check TYPE_METHODS here.
11276 * function.c (use_register_for_decl): Always ignore register for
11277 class types when not optimizing.
11278 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
11279 * tree.c (free_lang_data_in_type): Stitch out member functions and
11280 templates from TYPE_FIELDS.
11281 (build_distinct_type_copy, verify_type_variant,
11282 verify_type): Member fns are on TYPE_FIELDS.
11283 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
11284 * tree-pretty-print.c (dump_generic_node): Likewise.
11286 2017-07-20 Jakub Jelinek <jakub@redhat.com>
11289 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
11290 V2TImode and V4TImode.
11291 (ix86_expand_vector_extract): Likewise.
11292 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
11293 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
11294 (ssescalarmode): Handle V4TImode and V2TImode.
11295 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
11296 (*vec_extractv2ti, *vec_extractv4ti): New insns.
11297 (VEXTRACTI128_MODE): New mode iterator.
11298 (splitter for *vec_extractv?ti first element): New.
11299 (VEC_INIT_MODE): New mode iterator.
11300 (vec_init<mode>): Consolidate 3 expanders into one using
11301 VEC_INIT_MODE mode iterator.
11303 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
11305 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
11306 non_spilled_static_chain_regno_p.
11308 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
11310 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
11312 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
11314 * bb-reorder.c (connect_traces): Allow copying of blocks within
11317 2017-07-20 Richard Biener <rguenther@suse.de>
11319 * gimple.h (gimple_phi_result): Add gphi * overload.
11320 (gimple_phi_result_ptr): Likewise.
11321 (gimple_phi_arg): Likewise. Adjust index assert to only
11322 allow actual argument accesses rather than all slots available
11324 (gimple_phi_arg_def): Add gphi * overload.
11325 * tree-phinodes.c (make_phi_node): Initialize only actual
11327 (resize_phi_node): Clear memory not covered by old node,
11328 do not initialize excess argument slots.
11329 (reserve_phi_args_for_new_edge): Initialize new argument slot
11332 2017-07-20 Bin Cheng <bin.cheng@arm.com>
11334 PR tree-optimization/81388
11336 2016-07-21 Bin Cheng <bin.cheng@arm.com>
11338 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
11339 by removing computation of may_be_zero.
11341 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
11342 Tom de Vries <tom@codesourcery.com>
11344 PR middle-end/81030
11345 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
11346 when gimple level profile disagrees with what RTL expander did.
11348 2017-07-20 Richard Biener <rguenther@suse.de>
11350 PR tree-optimization/61171
11351 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
11352 (vect_analyze_stmt): Add slp instance parameter.
11353 (vectorizable_reduction): Likewise.
11354 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
11355 (vect_is_simple_reduction): Deal with chains not detected
11356 as SLP reduction chain, specifically not properly associated
11357 chains containing a mix of plus/minus.
11358 (get_reduction_op): Remove.
11359 (get_initial_defs_for_reduction): Simplify, pass in whether
11360 this is a reduction chain, pass in the SLP node for the PHIs.
11361 (vect_create_epilog_for_reduction): Get the SLP instance as
11363 (vectorizable_reduction): Get the SLP instance as arg.
11364 During analysis remember the SLP node with the PHIs in the
11365 instance. Simplify getting at the vectorized reduction PHIs.
11366 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
11367 through SLP instance.
11368 (vect_slp_analyze_operations): Likewise.
11369 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
11370 (vect_transform_stmt): Likewise.
11372 2017-07-20 Tom de Vries <tom@codesourcery.com>
11374 PR tree-optimization/81489
11375 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
11376 read of phi arg location to before loop that modifies phi.
11378 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11380 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
11383 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
11385 PR middle-end/81331
11386 * except.c (execute): Fix ordering issue.
11388 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
11390 PR rtl-optimization/81423
11391 * combine.c (make_compound_operation_int): Don't try to optimize
11392 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
11394 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
11396 PR rtl-optimization/81423
11397 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
11398 with a constant that is -1 in the truncated to mode.
11400 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
11402 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
11403 (determine_unlikely_bbs): ... here.
11404 * predict.h (propagate_unlikely_bbs_forward): Declare.
11405 * cfgexpand.c (pass_expand::execute): Use it.
11406 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
11408 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
11409 propagate_unlikely_bbs_forward.
11411 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
11413 PR middle-end/81331
11414 * except.c (maybe_add_nop_after_section_switch): New function.
11417 2017-07-19 Tom de Vries <tom@codesourcery.com>
11419 * gimple.h (gimple_phi_set_arg): Make assert more strict.
11421 2017-07-19 Tom de Vries <tom@codesourcery.com>
11423 * gimple.h (gimple_phi_arg): Make assert more strict.
11425 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
11427 * config.gcc (powerpc*-*-*): Add mmintrin.h.
11428 * config/rs6000/mmintrin.h: New file.
11429 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
11431 2017-07-19 Jakub Jelinek <jakub@redhat.com>
11433 PR tree-optimization/81346
11434 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
11436 2017-07-19 Tom de Vries <tom@codesourcery.com>
11438 * config/nvptx/nvptx.md (VECIM): Add V2DI.
11440 2017-07-19 Tom de Vries <tom@codesourcery.com>
11442 * config/nvptx/nvptx-modes.def: Add V2DImode.
11443 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
11444 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
11445 (nvptx_output_mov_insn): Handle lack of mov.b128.
11446 (nvptx_print_operand): Handle 'H' and 'L' codes.
11447 (nvptx_vector_mode_supported): Allow V2DImode.
11448 (nvptx_preferred_simd_mode): New function.
11449 (nvptx_data_alignment): New function.
11450 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
11451 nvptx_preferred_simd_mode.
11452 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
11454 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
11456 2017-07-19 Tom de Vries <tom@codesourcery.com>
11458 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
11459 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
11460 (nvptx_vector_mode_supported): New function. Allow V2SImode.
11461 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
11462 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
11463 (mov<VECIM>_insn): New define_insn.
11464 (define_expand "mov<VECIM>): New define_expand.
11466 2017-07-19 Tom de Vries <tom@codesourcery.com>
11468 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
11470 2017-07-19 Jakub Jelinek <jakub@redhat.com>
11472 PR tree-optimization/81346
11473 * fold-const.h (fold_div_compare, range_check_type): Declare.
11474 * fold-const.c (range_check_type): New function.
11475 (build_range_check): Use range_check_type.
11476 (fold_div_compare): No longer static, rewritten into
11477 a match.pd helper function.
11478 (fold_comparison): Don't call fold_div_compare here.
11479 * match.pd (X / C1 op C2): New optimization using fold_div_compare
11480 as helper function.
11482 2017-07-19 Nathan Sidwell <nathan@acm.org>
11484 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
11485 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
11486 * tree.c (find_decls_types_r, verify_type): Use
11487 TYPE_{MIN,MAX}_VALUE_RAW.
11488 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
11489 (hash_tree): Likewise.
11490 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
11492 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
11495 2017-07-18 Tom de Vries <tom@codesourcery.com>
11497 PR middle-end/81464
11498 * omp-expand.c (expand_omp_for_static_chunk): Handle
11499 equal-argument loop exit phi.
11501 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
11504 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
11505 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
11506 operand 2 predicate.
11507 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
11508 operand 2 predicate.
11509 (ror,rol -> rorx splitters): Use const_int_operand as
11510 operand 2 predicate.
11512 2017-06-18 Richard Biener <rguenther@suse.de>
11514 PR tree-optimization/81410
11515 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
11516 the gap in the ! slp_perm SLP case after each group.
11518 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
11520 PR middle-end/81463
11521 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
11524 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
11526 PR middle-end/81462
11527 * predict.c (set_even_probabilities): Cleanup; do not affect
11528 probabilities that are already known.
11529 (combine_predictions_for_bb): Call even when count is set.
11531 2017-07-18 Nathan Sidwell <nathan@acm.org>
11533 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
11536 2017-07-18 Bin Cheng <bin.cheng@arm.com>
11539 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
11540 optimization for loop niter analysis.
11542 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
11545 * config/avr/avr.c (avr_optimize_casesi): Don't use
11546 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
11548 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11550 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
11551 body_cost_vec from _vect_peel_extended_info.
11552 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
11553 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
11556 2017-07-18 Bin Cheng <bin.cheng@arm.com>
11558 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
11560 2017-07-18 Richard Biener <rguenther@suse.de>
11562 PR tree-optimization/80620
11563 PR tree-optimization/81403
11564 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
11565 info when re-using a VN table entry.
11567 2017-07-18 Richard Biener <rguenther@suse.de>
11569 PR tree-optimization/81418
11570 * tree-vect-loop.c (vectorizable_reduction): Properly compute
11571 vectype_in. Verify that with lane-reducing reduction operations
11572 we have a single def-use cycle.
11574 2017-07-17 Carl Love <cel@us.ibm.com>
11576 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
11578 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11579 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11580 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11581 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11582 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11583 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11584 VMULOSW): New enum "unspec" values.
11585 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
11586 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
11587 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11588 altivec_vmulosw): New patterns.
11589 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11590 VMULOSW): Add definitions.
11591 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
11593 * config/alpha/alpha.c: Include predict.h.
11595 2017-07-17 Yury Gribov <tetra2005@gmail.com>
11597 * tree-vrp.c (compare_assert_loc): Fix comparison function
11598 to return predictable results.
11600 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11602 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
11604 (subdi3): Likewise.
11605 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
11606 * doc/invoke.texi (mexpand-adddi): Update text.
11608 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11610 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
11611 that also clobbers the CC register. The old expand code is moved
11613 (*arc_clzsi2): ... here.
11614 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
11615 the CC register. The old expand code is moved to ...
11616 (arc_ctzsi2): ... here.
11618 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11620 * config/arc/arc.opt (mindexed-loads): Use initial value
11621 TARGET_INDEXED_LOADS_DEFAULT.
11622 (mauto-modify-reg): Use initial value
11623 TARGET_AUTO_MODIFY_REG_DEFAULT.
11624 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
11625 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
11626 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
11627 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
11629 2017-07-17 Martin Liska <mliska@suse.cz>
11632 * opts.c (finish_options): Do not allow -fgnu-tm
11633 w/ -fsanitize={kernel-,}address. Say sorry.
11635 2017-07-17 Bin Cheng <bin.cheng@arm.com>
11638 * tree-loop-distribution.c (classify_partition): Only assert on
11639 numer of iterations.
11640 (merge_dep_scc_partitions): Delete prameter. Update function call.
11641 (distribute_loop): Remove code handling loop with unknown niters.
11642 (pass_loop_distribution::execute): Skip loop with unknown niters.
11644 2017-07-17 Bin Cheng <bin.cheng@arm.com>
11647 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
11648 function sort_partitions_by_post_order.
11650 2017-07-17 Bin Cheng <bin.cheng@arm.com>
11652 PR tree-optimization/81374
11653 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
11654 the max index of basic blocks, rather than number of basic blocks.
11656 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11658 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
11660 (arc_legitimate_pic_operand_p): Likewise.
11661 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
11663 (arc_needs_pcl_p): Likewise.
11664 (arc_legitimate_pc_offset_p): Likewise.
11665 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
11666 function is also used in constrains.md.
11667 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
11668 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
11669 PLUS. Only return true/false in known cases, otherwise assert.
11670 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
11671 is already called in arc_legitimate_constant_p.
11672 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
11674 (LEGITIMATE_PIC_OPERAND_P): Use
11675 arc_raw_symbolic_reference_mentioned_p function.
11676 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
11681 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11682 Andrew Burgess <andrew.burgess@embecosm.com>
11684 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
11685 (arc_return_address_register): New function.
11686 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
11687 (arc_handle_fndecl_attribute): Add naked attribute.
11688 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
11689 (TARGET_WARN_FUNC_RETURN): Likewise.
11690 (arc_allocate_stack_slots_for_args): New function.
11691 (arc_warn_func_return): Likewise.
11692 (machine_function): Change type fn_type.
11693 (arc_compute_function_type): Consider new naked function type,
11694 change function return type.
11695 (arc_must_save_register): Adapt to handle new
11696 arc_compute_function_type's return type.
11697 (arc_expand_prologue): Likewise.
11698 (arc_expand_epilogue): Likewise.
11699 (arc_return_address_regs): Delete.
11700 (arc_return_address_register): New function.
11701 (arc_epilogue_uses): Use above function.
11702 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
11703 (arc_function_type): Change encoding, add naked type.
11704 (ARC_INTERRUPT_P): Change to handle the new encoding.
11705 (ARC_FAST_INTERRUPT_P): Likewise.
11706 (ARC_NORMAL_P): Define.
11707 (ARC_NAKED_P): Likewise.
11708 (arc_compute_function_type): Delete prototype.
11709 * config/arc/arc.md (in_ret_delay_slot): Use
11710 arc_return_address_register function.
11711 (simple_return): Likewise.
11712 (p_return_i): Likewise.
11714 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11716 PR tree-optimization/81428
11717 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
11718 can't be built for those types.
11720 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
11722 Remove stuff dead since r239246.
11724 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
11725 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
11726 (avr_inform_devices): Remove dead stuff.
11728 2017-07-17 Tamar Christina <tamar.christina@arm.com>
11730 * config/arm/arm_neon.h: Fix softp typo.
11732 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11734 PR tree-optimization/81365
11735 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
11736 aggregate moves onto bb predecessor edges, make sure there are no
11737 loads that could alias the lhs in between the start of bb and the
11740 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
11743 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
11744 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
11745 [LSHIFTRT, outer_code = TRUNCATE]: Same.
11747 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11749 PR tree-optimization/81396
11750 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
11751 (init_symbolic_number): Initialize it to 1.
11752 (perform_symbolic_merge): Add n_ops from both operands into the new
11754 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
11755 without base_addr as useless if they need more than one operation.
11756 (bswap_replace): Handle !bswap case for NULL base_addr.
11758 2017-07-17 Tom de Vries <tom@codesourcery.com>
11761 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
11764 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
11766 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
11767 conditional builtin define __FIX_LEON3FT_B2BST.
11769 2017-07-17 Daniel Cederman <cederman@gaisler.com>
11771 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
11772 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
11775 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
11777 PR rtl-optimization/81424
11778 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
11779 to remove potential trapping from operands if -fnon-call-exceptions.
11781 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11783 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
11784 profile_proability for scalling.
11785 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
11787 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11789 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
11791 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11793 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
11794 fixpoint arithmetics.
11796 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11798 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
11799 fixpoint arithmetics.
11801 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11803 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
11804 fixpoint arithmetics.
11806 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11808 * profile-count.h (profile_probability::from_reg_br_prob_note,
11809 profile_probability::to_reg_br_prob_note): New functions.
11810 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
11811 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
11812 * predict.c (probability_reliable_p): Update.
11813 (edge_probability_reliable_p): Update.
11814 (br_prob_note_reliable_p): Update.
11815 (invert_br_probabilities): Update.
11816 (add_reg_br_prob_note): New function.
11817 (combine_predictions_for_insn): Update.
11818 * asan.c (asan_clear_shadow): Update.
11819 * cfgbuild.c (compute_outgoing_frequencies): Update.
11820 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
11821 (update_br_prob_note): Update.
11822 (rtl_verify_edges): Update.
11823 (purge_dead_edges): Update.
11824 (fixup_reorder_chain): Update.
11825 * emit-rtl.c (try_split): Update.
11826 * ifcvt.c (cond_exec_process_insns): Update.
11827 (cond_exec_process_if_block): Update.
11828 (dead_or_predicable): Update.
11829 * internal-fn.c (expand_addsub_overflow): Update.
11830 (expand_neg_overflow): Update.
11831 (expand_mul_overflow): Update.
11832 * loop-doloop.c (doloop_modify): Update.
11833 * loop-unroll.c (compare_and_jump_seq): Update.
11834 * optabs.c (emit_cmp_and_jump_insn_1): Update.
11835 * predict.h: Update.
11836 * reorg.c (mostly_true_jump): Update.
11838 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
11839 * config/alpha/alpha.c (emit_unlikely_jump): Update.
11840 * config/arc/arc.c: (emit_unlikely_jump): Update.
11841 * config/arm/arm.c: (emit_unlikely_jump): Update.
11842 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
11843 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
11844 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
11845 (ix86_print_operand): Update.
11846 (ix86_split_fp_branch): Update.
11847 (predict_jump): Update.
11848 * config/ia64/ia64.c (ia64_print_operand): Update.
11849 * config/mmix/mmix.c (mmix_print_operand): Update.
11850 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
11851 (rs6000_expand_split_stack_prologue): Update.
11852 * config/rs6000/rs6000.c: Update.
11853 * config/s390/s390.c (s390_expand_vec_strlen): Update.
11854 (s390_expand_vec_movstr): Update.
11855 (s390_expand_cs_tdsi): Update.
11856 (s390_expand_split_stack_prologue): Update.
11857 * config/sh/sh.c (sh_print_operand): Update.
11858 (expand_cbranchsi4): Update.
11859 (expand_cbranchdi4): Update.
11860 * config/sparc/sparc.c (output_v9branch): Update.
11861 * config/spu/spu.c (get_branch_target): Update.
11862 (ea_load_store_inline): Update.
11863 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
11864 * config/tilepro/tilepro.c: Update.
11866 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
11868 * gimplify.c (mostly_copy_tree_r): Revert latest change.
11869 (gimplify_save_expr): Likewise.
11871 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11873 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
11875 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11877 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
11879 * timevar.def (TV_IPA_FNSUMMARY): Define.
11881 2017-07-16 Daniel Cederman <cederman@gaisler.com>
11883 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
11884 to back store errata sensitive sequence from being generated.
11885 (sqrtdf2_fix): Likewise.
11887 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11889 * tree-ssa-threadupdate.c (compute_path_counts,
11890 update_joiner_offpath_counts): Use profile_probability.
11892 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
11895 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11897 * config/arm/arm-c.c (arm_cpu_builtins): Define
11898 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
11900 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
11902 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11903 array entries to represent __ieee128 versions of the
11904 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
11905 scalar_extract_sig, and scalar_insert_exp built-in functions.
11906 (altivec_resolve_overloaded_builtin): Add special case handling
11907 for the __builtin_scalar_insert_exp function, as represented by
11908 the P9V_BUILTIN_VEC_VSIEDP constant.
11909 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
11910 exponent support for __ieee128 argument.
11911 (VSESQP): Add scalar extract signature support for __ieee128
11913 (VSTDCNQP): Add scalar test negative support for __ieee128
11915 (VSIEQP): Add scalar insert exponent support for __int128 argument
11916 with __ieee128 result.
11917 (VSIEQPF): Add scalar insert exponent support for __ieee128
11918 argument with __ieee128 result.
11919 (VSTDCQP): Add scalar test data class support for __ieee128
11921 (VSTDCNQP): Add overload support for scalar test negative with
11922 __ieee128 argument.
11923 (VSTDCQP): Add overload support for scalar test data class
11924 __ieee128 argument.
11925 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
11926 UNSPEC_VSX_SXSIGDP.
11927 (UNSPEC_VSX_SIEXPQP): New constant.
11928 (xsxexpqp): New insn for VSX scalar extract exponent quad
11930 (xsxsigqp): New insn for VSX scalar extract significand quad
11932 (xsiexpqpf): New insn for VSX scalar insert exponent quad
11933 precision with floating point argument.
11934 (xststdcqp): New expand for VSX scalar test data class quad
11936 (xststdcnegqp): New expand for VSX scalar test negative quad
11938 (xststdcqp): New insn to match expansions for VSX scalar test data
11939 class quad precision and VSX scalar test negative quad precision.
11940 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
11941 special case operand checking to enforce that second operand of
11942 VSX scalar test data class with quad precision argument is a 7-bit
11944 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
11945 prototypes and descriptions of __ieee128 versions of
11946 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
11947 scalar_test_data_class, and scalar_test_neg built-in functions.
11949 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11951 PR tree-optimization/81162
11952 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
11953 replace a negate with an add.
11955 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
11957 * doc/invoke.texi (arm/-mcpu): Document +crypto.
11959 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11961 * config/arm/arm-c.c (arm_cpu_builtins): Define
11962 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
11964 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11966 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
11967 (armv8-r): Set ARM Cortex-R52 as default CPU.
11968 * config/arm/arm-tables.opt: Regenerate.
11969 * config/arm/arm-tune.md: Regenerate.
11970 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
11972 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
11973 extension for -mcpu=cortex-r52.
11975 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11977 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
11978 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
11979 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
11980 (fp-armv8): Define it as FP_ARMv8 only.
11981 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
11982 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
11984 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
11985 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
11986 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
11987 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
11988 than TARGET_FPU_ARMV8.
11989 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
11990 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
11991 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
11993 * config/arm/neon.md (neon_vrint): Likewise.
11994 (neon_vcvt): Likewise.
11995 (neon_<fmaxmin_op><mode>): Likewise.
11996 (<fmaxmin><mode>3): Likewise.
11997 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
11998 * config/arm/predicates.md (arm_cond_move_operator): Check against
11999 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
12001 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
12003 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
12004 to top of function.
12006 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12008 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
12009 loop in comment with memset.
12011 2017-07-14 Martin Liska <mliska@suse.cz>
12013 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
12014 * dwarf2out.c (is_java): Remove the function.
12015 (output_pubname): Remove usage of the function.
12016 (lower_bound_default): Remove usage of DW_LANG_Java.
12017 (gen_compile_unit_die): Likewise.
12018 * gcc.c: Remove compiler defaults for .java and .zip files.
12019 * gimple-expr.c (remove_suffix): Change as there's no longer
12020 extension than 4-letter one.
12021 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
12022 (gimplify_save_expr): Likewise.
12023 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
12024 as it's possible even for other languages than Java.
12025 * langhooks.h (struct lang_hooks): Remove Java from a comment.
12026 * lto-opts.c (lto_write_options): Remove reference to Java.
12027 * opts.c (strip_off_ending): Update file extension handling.
12028 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
12029 * tree-eh.c (lower_resx): Likewise.
12030 * tree.c (free_lang_data_in_type): Remove dead code.
12031 (find_decls_types_r): Likewise.
12032 (build_common_builtin_nodes): Remove Java from a comment.
12033 (verify_type): Remove dead code.
12034 * varasm.c (assemble_external): Remove Java from a comment.
12036 2017-07-14 Martin Liska <mliska@suse.cz>
12038 * opts.c (finish_options): Add quotes.
12039 (common_handle_option): Likewise.
12041 2017-07-14 Martin Liska <mliska@suse.cz>
12043 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
12044 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
12045 Remove N_SO_PASCAL.
12046 * dwarf2out.c (lower_bound_default): Do not handle
12048 (gen_compile_unit_die): Likewise.
12049 * gcc.c: Remove default extension binding for GNU Pascal.
12050 * stmt.c: Remove Pascal language from a comment.
12051 * xcoffout.c: Likewise.
12053 2017-07-13 David Malcolm <dmalcolm@redhat.com>
12056 * diagnostic-show-locus.c (fixit_cmp): New function.
12057 (layout::layout): Sort m_fixit_hints.
12058 (column_range::column_range): Assert that the values are valid.
12059 (struct char_span): New struct.
12060 (correction::overwrite): New method.
12061 (struct source_line): New struct.
12062 (line_corrections::add_hint): Add assertions. Reimplement memcpy
12063 calls in terms of classes source_line and char_span, and
12064 correction::overwrite.
12065 (selftest::test_overlapped_fixit_printing_2): New function.
12066 (selftest::diagnostic_show_locus_c_tests): Call it.
12068 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
12070 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
12071 early if there is no lhs.
12073 2017-07-13 Martin Liska <mliska@suse.cz>
12075 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
12076 (gen_reference_type_die): Likewise.
12077 * stor-layout.c: Remove Pascal-related comment.
12079 2017-07-13 Martin Liska <mliska@suse.cz>
12081 * opts.c (finish_options): Add quotes to error messages.
12082 (parse_sanitizer_options): Likewise.
12084 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
12086 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
12088 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
12090 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
12092 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
12094 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
12096 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
12098 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
12101 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
12102 provides the hardware capability bits, define the macro
12103 __BUILTIN_CPU_SUPPORTS__.
12104 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
12105 if GLIBC does not provide the hardware capability bits. Add a
12106 gcc_unreachable call if the built-in cpu function is neither
12107 __builtin_cpu_is nor __builtin_cpu_supports.
12108 (rs6000_get_function_versions_dispatcher): Change the warning
12109 that an old GLIBC is used which does not export the capability
12110 bits to be an error.
12111 * doc/extend.texi (target_clones attribute): Document the
12112 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
12113 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
12114 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
12115 the macros defined by GCC if the newer GLIBC is available.
12117 2017-07-12 Jeff Law <law@redhat.com>
12119 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
12120 remaining includes slightly.
12121 * config/riscv/riscv-builtins.c: Include profile-count.h.
12123 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
12126 * config/avr/avr.c (avr_set_current_function): In diagnostic
12127 messages: Quote keywords and (parts of) identifiers.
12128 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
12131 2017-07-12 Carl Love <cel@us.ibm.com>
12133 * config/rs6000/rs6000-c.c: Add support for built-in functions
12134 vector bool char vec_revb (vector bool char);
12135 vector bool short vec_revb (vector short char);
12136 vector bool int vec_revb (vector bool int);
12137 vector bool long long vec_revb (vector bool long long);
12138 * doc/extend.texi: Update the built-in documentation file for the
12139 new built-in functions.
12141 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12143 * config/s390/s390.md: Remove movcc splitter.
12145 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12147 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
12148 load/store on condition.
12150 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
12153 * config/avr/avr.c (avr_encode_section_info)
12154 [progmem && !TREE_READONLY]: Error if progmem object needs
12157 2017-07-11 Michael Collison <michael.collison@arm.com>
12159 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
12162 2017-07-11 Carl Love <cel@us.ibm.com>
12164 * config/rs6000/rs6000-c.c: Add support for builtins
12165 vector unsigned int vec_parity_lsbb (vector signed int);
12166 vector unsigned int vec_parity_lsbb (vector unsigned int);
12167 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
12168 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
12169 vector unsigned long long vec_parity_lsbb (vector signed long long);
12170 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
12171 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
12172 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
12173 * doc/extend.texi: Update the built-in documentation file for the
12174 new built-in functions.
12176 2017-07-11 David Malcolm <dmalcolm@redhat.com>
12178 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
12179 (layout::m_primary_loc): New field.
12180 (layout::layout): Initialize new field. Move location filtering
12181 logic from here to...
12182 (layout::maybe_add_location_range): ...this new method. Add
12183 support for filtering to just the lines already specified by other
12185 (layout::will_show_line_p): New method.
12186 (gcc_rich_location::add_location_if_nearby): New method.
12187 (selftest::test_add_location_if_nearby): New test function.
12188 (selftest::diagnostic_show_locus_c_tests): Call it.
12189 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
12192 2017-07-11 Tom de Vries <tom@codesourcery.com>
12194 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
12195 (bb_first_real_insn): New function.
12196 (nvptx_single): Add extra initialization of broadcasted condition
12199 2017-07-11 Nathan Sidwell <nathan@acm.org>
12201 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
12203 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
12205 * doc/extend.texi (AVR Function Attributes): Remove weblink to
12206 Binutils doc as TEXI will mess them up.
12207 * doc/invoke.texi (AVR Options): Same here.
12209 2017-07-11 Daniel Cederman <cederman@gaisler.com>
12211 * config/sparc/sparc.opt (mfix-ut700): New option.
12212 (mfix-gr712rc): Likewise.
12213 (sparc_fix_b2bst): New variable.
12214 * doc/invoke.texi (SPARC options): Document them.
12215 (ARM options): Fix warnings.
12216 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
12217 instructions to prevent sequences that can trigger the store-store
12218 errata for certain LEON3FT processors.
12219 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
12220 (sparc_option_override): Set sparc_fix_b2bst appropriately.
12221 * config/sparc/sparc.md (fix_b2bst): New attribute.
12222 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
12224 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
12227 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
12229 (*rsqrtsf2_sse): Ditto.
12231 (div<mode>3): Macroize insn from divdf3 and divsf3
12232 using MODEF mode iterator.
12234 2017-07-10 Martin Sebor <msebor@redhat.com>
12236 PR tree-optimization/80397
12237 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
12238 instead of testing for equality to INTEGER_TYPE.
12240 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
12242 * config.gcc: Remove uclibc from arc target spec.
12244 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
12246 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
12248 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12251 * lto-wrapper.c (remove_option): New function.
12252 (merge_and_complain): Merge PIC/PIE options more realistically.
12254 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
12256 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
12260 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
12261 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
12262 * config.in: Regenerate.
12263 * configure: Regenerate.
12264 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
12265 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
12266 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
12267 (TARGET_GASISR_PROLOGUES): ...target mask.
12268 * common/config/avr/avr-common.c
12269 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
12270 Set -mgas-isr-prologues.
12271 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
12272 INSERT_PASS_BEFORE for it.
12273 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
12274 * config/avr/avr.c (avr_option_override)
12275 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
12276 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
12277 (avr_attribute_table) <no_gccisr>: Add new function attribute.
12278 (avr_set_current_function) <is_no_gccisr>: Init machine field.
12279 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
12281 (make_avr_pass_pre_proep): New function.
12282 (emit_push_sfr) <treg>: Add argument to function and use it
12283 instead of TMP_REG.
12284 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
12285 and set machine->gasisr.yes.
12286 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
12287 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
12288 __gcc_isr.n_pushed to .L__stack_usage.
12289 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
12290 (avr_asm_final_postscan_insn): ...this new static function.
12291 * config/avr/avr.h (machine_function)
12292 <is_no_gccisr, use_L__stack_usage>: New fields.
12293 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
12294 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
12295 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
12296 (gasisr, *gasisr): New expander and insn.
12297 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
12298 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
12299 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
12301 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
12303 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
12306 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
12308 Move jump-tables out of .text again.
12311 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
12312 (ASM_OUTPUT_ADDR_VEC): New function.
12313 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
12314 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
12315 INSN_ADDRESSes as asm comment.
12316 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
12317 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
12318 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
12319 * config/avr/avr.md (*tablejump): Adjust comment.
12320 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
12321 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
12323 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
12324 (avr_output_addr_vec): New proto.
12325 (avr_log_t) <insn_addresses>: New field.
12327 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
12330 * config/i386/i386.c (ix86_function_arg_advance): Set
12331 outgoing_args_on_stack to true if there are outgoing arguments
12333 (ix86_function_arg): Likewise.
12334 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
12335 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
12336 * config/i386/i386.h (machine_function): Add
12337 outgoing_args_on_stack.
12339 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
12341 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
12342 supporting pthreds.
12343 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
12345 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
12347 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
12348 (REAL_H): Remove $(MACHMODE_H).
12349 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
12351 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
12352 $(MACHMODE_H) and double-int.h.
12353 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
12355 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
12358 2017-07-07 Andrew Pinski <apinski@cavium.com>
12360 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
12361 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
12363 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
12365 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
12366 Add warning if GCC was not configured to link against a GLIBC that
12367 exports the hardware capability bits.
12368 (make_resolver_func): Make resolver function private and not a
12369 COMDAT function. Create the name with clone_function_name instead
12370 of make_unique_name.
12373 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
12374 correct operand in doing the split.
12376 2017-07-07 Carl Love <cel@us.ibm.com>
12378 * config/rs6000/rs6000-c: Add support for built-in function
12379 vector unsigned short vec_pack_to_short_fp32 (vector float,
12381 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
12382 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
12383 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
12384 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
12385 (convert_4f32_8i16): Add define_expand.
12386 * doc/extend.texi: Update the built-in documentation file for the
12387 new built-in function.
12389 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12391 * config/sparc/m8.md: New file.
12392 * config/sparc/sparc.md: Include m8.md.
12394 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12396 * config/sparc/sparc.opt: New option -mvis4b.
12397 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
12398 (sparc_option_override): Handle VIS4B.
12399 (enum sparc_builtins): Define
12400 SPARC_BUILTIN_DICTUNPACK{8,16,32},
12401 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
12402 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
12403 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
12404 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
12405 (check_constant_argument): New function.
12406 (sparc_vis_init_builtins): Define builtins
12407 __builtin_vis_dictunpack{8,16,32},
12408 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
12409 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
12410 __builtin_vis_fpcmpde{8,16,32}shl and
12411 __builtin_vis_fpcmpur{8,16,32}shl.
12412 (sparc_expand_builtin): Check that the constant operands to
12413 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
12414 constant and in range.
12415 * config/sparc/sparc-c.c (sparc_target_macros): Handle
12417 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
12418 (SPARC_IMM5_P): Likewise.
12419 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
12420 (enabled): Handle vis4b.
12421 (UNSPEC_DICTUNPACK): New unspec.
12422 (UNSPEC_FPCMPSHL): Likewise.
12423 (UNSPEC_FPUCMPSHL): Likewise.
12424 (UNSPEC_FPCMPDESHL): Likewise.
12425 (UNSPEC_FPCMPURSHL): Likewise.
12426 (cpu_feature): New CPU feature `vis4b'.
12427 (dictunpack{8,16,32}): New insns.
12428 (FPCSMODE): New mode iterator.
12429 (fpcscond): New code iterator.
12430 (fpcsucond): Likewise.
12431 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
12432 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
12433 (fpcmpde{8,16,32}{si,di}shl): Likewise.
12434 (fpcmpur{8,16,32}{si,di}shl): Likewise.
12435 * config/sparc/constraints.md: Define constraints `q' for unsigned
12436 2-bit integer constants and `t' for unsigned 5-bit integer
12438 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
12440 (imm5_operand_dictunpack16): Likewise.
12441 (imm5_operand_dictunpack32): Likewise.
12442 (imm2_operand): Likewise.
12443 * doc/invoke.texi (SPARC Options): Document -mvis4b.
12444 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
12445 ditunpack* and fpcmp*shl builtins.
12447 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12449 * config.gcc: Handle m8 in --with-{cpu,tune} options.
12450 * config.in: Add HAVE_AS_SPARC6 define.
12451 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
12453 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
12455 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
12456 (CPP_CPU_SPEC): Handle m8.
12457 (ASM_CPU_SPEC): Likewise.
12458 * config/sparc/sparc-opts.h (enum processor_type): Add
12460 * config/sparc/sparc.c (m8_costs): New struct.
12461 (sparc_option_override): Handle TARGET_CPU_m8.
12462 (sparc32_initialize_trampoline): Likewise.
12463 (sparc64_initialize_trampoline): Likewise.
12464 (sparc_issue_rate): Likewise.
12465 (sparc_register_move_cost): Likewise.
12466 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
12467 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
12468 (ASM_CPU64_DEFAULT_SPEC): Likewise.
12469 (CPP_CPU_SPEC): Handle M8.
12470 (ASM_CPU_SPEC): Likewise.
12471 (AS_M8_FLAG): Define.
12472 * config/sparc/sparc.md: Add m8 to the cpu attribute.
12473 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
12474 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
12476 * configure: Regenerate.
12477 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
12480 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12482 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
12484 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
12485 ("*movdi_insn_sp32"): Do not set v3pipe.
12486 ("*movsi_insn"): Likewise.
12487 ("*movdi_insn_sp64"): Likewise.
12488 ("*movsf_insn"): Likewise.
12489 ("*movdf_insn_sp32"): Likewise.
12490 ("*movdf_insn_sp64"): Likewise.
12491 ("*zero_extendsidi2_insn_sp64"): Likewise.
12492 ("*sign_extendsidi2_insn"): Likewise.
12493 ("*mov<VM32:mode>_insn"): Likewise.
12494 ("*mov<VM64:mode>_insn_sp64"): Likewise.
12495 ("*mov<VM64:mode>_insn_sp32"): Likewise.
12496 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
12497 ("<vlop:code><VL:mode>3"): Likewise.
12498 ("*not_<vlop:code><VL:mode>3"): Likewise.
12499 ("*nand<VL:mode>_vis"): Likewise.
12500 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
12501 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
12502 ("one_cmpl<VL:mode>2"): Likewise.
12503 ("faligndata<VM64:mode>_vis"): Likewise.
12504 ("alignaddrsi_vis"): Likewise.
12505 ("alignaddrdi_vis"): Likweise.
12506 ("alignaddrlsi_vis"): Likewise.
12507 ("alignaddrldi_vis"): Likewise.
12508 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
12509 ("bmaskdi_vis"): Likewise.
12510 ("bmasksi_vis"): Likewise.
12511 ("bshuffle<VM64:mode>_vis"): Likewise.
12512 ("cmask8<P:mode>_vis"): Likewise.
12513 ("cmask16<P:mode>_vis"): Likewise.
12514 ("cmask32<P:mode>_vis"): Likewise.
12515 ("pdistn<P:mode>_vis"): Likewise.
12516 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
12518 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12520 * config/sparc/sparc.md ("subtype"): New insn attribute.
12521 ("*wrgsr_sp64"): Set insn subtype.
12522 ("*rdgsr_sp64"): Likewise.
12523 ("alignaddrsi_vis"): Likewise.
12524 ("alignaddrdi_vis"): Likewise.
12525 ("alignaddrlsi_vis"): Likewise.
12526 ("alignaddrldi_vis"): Likewise.
12527 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
12528 ("fexpand_vis"): Likewise.
12529 ("fpmerge_vis"): Likewise.
12530 ("faligndata<VM64:mode>_vis"): Likewise.
12531 ("bshuffle<VM64:mode>_vis"): Likewise.
12532 ("cmask8<P:mode>_vis"): Likewise.
12533 ("cmask16<P:mode>_vis"): Likewise.
12534 ("cmask32<P:mode>_vis"): Likewise.
12535 ("fchksm16_vis"): Likewise.
12536 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
12537 ("fmean16_vis"): Likewise.
12538 ("fp<plusminus_insn>64_vis"): Likewise.
12539 ("<plusminus_insn>v8qi3"): Likewise.
12540 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
12541 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
12542 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
12543 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
12544 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
12545 ("*movqi_insn"): Likewise.
12546 ("*movhi_insn"): Likewise.
12547 ("*movsi_insn"): Likewise.
12548 ("movsi_pic_gotdata_op"): Likewise.
12549 ("*movdi_insn_sp32"): Likewise.
12550 ("*movdi_insn_sp64"): Likewise.
12551 ("movdi_pic_gotdata_op"): Likewise.
12552 ("*movsf_insn"): Likewise.
12553 ("*movdf_insn_sp32"): Likewise.
12554 ("*movdf_insn_sp64"): Likewise.
12555 ("*zero_extendhisi2_insn"): Likewise.
12556 ("*zero_extendqihi2_insn"): Likewise.
12557 ("*zero_extendqisi2_insn"): Likewise.
12558 ("*zero_extendqidi2_insn"): Likewise.
12559 ("*zero_extendhidi2_insn"): Likewise.
12560 ("*zero_extendsidi2_insn_sp64"): Likewise.
12561 ("ldfsr"): Likewise.
12562 ("prefetch_64"): Likewise.
12563 ("prefetch_32"): Likewise.
12564 ("tie_ld32"): Likewise.
12565 ("tie_ld64"): Likewise.
12566 ("*tldo_ldub_sp32"): Likewise.
12567 ("*tldo_ldub1_sp32"): Likewise.
12568 ("*tldo_ldub2_sp32"): Likewise.
12569 ("*tldo_ldub_sp64"): Likewise.
12570 ("*tldo_ldub1_sp64"): Likewise.
12571 ("*tldo_ldub2_sp64"): Likewise.
12572 ("*tldo_ldub3_sp64"): Likewise.
12573 ("*tldo_lduh_sp32"): Likewise.
12574 ("*tldo_lduh1_sp32"): Likewise.
12575 ("*tldo_lduh_sp64"): Likewise.
12576 ("*tldo_lduh1_sp64"): Likewise.
12577 ("*tldo_lduh2_sp64"): Likewise.
12578 ("*tldo_lduw_sp32"): Likewise.
12579 ("*tldo_lduw_sp64"): Likewise.
12580 ("*tldo_lduw1_sp64"): Likewise.
12581 ("*tldo_ldx_sp64"): Likewise.
12582 ("*mov<VM32:mode>_insn"): Likewise.
12583 ("*mov<VM64:mode>_insn_sp64"): Likewise.
12584 ("*mov<VM64:mode>_insn_sp32"): Likewise.
12586 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12588 * config/sparc/sparc.md ("type"): New insn type viscmp.
12589 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
12591 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
12592 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
12593 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
12594 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
12596 ("n7_vis_logical_11cycle"): Likewise.
12597 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
12598 * config/sparc/niagara2.md ("niag3_vis": Likewise.
12599 * config/sparc/niagara.md ("niag_vis"): Likewise.
12600 * config/sparc/ultra3.md ("us3_fga"): Likewise.
12601 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
12603 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12605 * config/sparc/sparc.md: New instruction type `bmask'.
12606 (bmaskdi_vis): Use the `bmask' type.
12607 (bmasksi_vis): Likewise.
12608 * config/sparc/ultra3.md (us3_array): Likewise.
12609 * config/sparc/niagara7.md (n7_array): Likewise.
12610 * config/sparc/niagara4.md (n4_array): Likewise.
12611 * config/sparc/niagara2.md (niag2_vis): Likewise.
12612 (niag3_vis): Likewise.
12613 * config/sparc/niagara.md (niag_vis): Likewise.
12615 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12617 * ipa-comdats.c: Remove optimize check from gate.
12618 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
12619 for functions not optimized.
12620 (ipa_fn_summary_read): Skip optimize check.
12621 (ipa_fn_summary_write): Likewise.
12622 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
12624 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
12626 (can_inline_edge_p): Check flag_pcc_struct_return for match.
12627 (check_callers): Give up on caller which is not optimized.
12628 (inline_small_functions): Likewise.
12629 (ipa_inline): Do not give up when not optimizing.
12630 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
12631 away unoptimizes cdtors.
12632 (whole_program_function_and_variable_visibility): Do
12633 ipa_discover_readonly_nonaddressable_vars in LTO mode.
12634 * ipa.c (process_references): Do not check optimize.
12635 (symbol_table::remove_unreachable_nodes): Update optimize check.
12636 (set_writeonly_bit): Update optimize check.
12637 (pass_ipa_cdtor_merge::gate): Do not check optimize.
12638 (pass_ipa_single_use::gate): Remove.
12640 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12642 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
12643 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
12644 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
12645 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
12646 permute_load, permute_store, adjust_extract, adjust_splat,
12647 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
12648 replace_swap_with_copy, dump_swap_insn_table,
12649 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
12650 recombine_lvx_pattern, recombine_stvx_pattern,
12651 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
12652 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
12653 to file rs6000-p8swap.c.
12654 * config/rs6000/rs6000-p8swap.c: New file.
12655 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
12656 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
12657 and rs6000*-*-* targets.
12659 2017-07-06 David Malcolm <dmalcolm@redhat.com>
12661 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
12663 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12665 * lto-wrapper.c (merge_and_complain): Do not merge
12666 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
12667 fsigned_zeros, ftrapping_math, fwrapv.
12668 (append_compiler_options): Do not track these options.
12669 (append_linker_options): Likewie
12671 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12673 * cgraphunit.c (cgraph_node::finalize_function): When
12674 !flag_toplevel_reorde set no_reorder flag.
12675 (varpool_node::finalize_decl): Likewise.
12676 (symbol_table::compile): Drop no toplevel reorder path.
12678 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12680 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
12681 edges; zero probability is not better than uninitialized.
12683 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
12685 * asan.h (asan_sanitize_allocas_p): Declare.
12686 * asan.c (asan_sanitize_allocas_p): New function.
12687 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
12688 (handle_builtin_alloca): Likewise.
12689 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
12690 if !asan_sanitize_allocas_p.
12691 * params.def (asan-instrument-allocas): Add new option.
12692 * params.h (ASAN_PROTECT_ALLOCAS): Define.
12693 * opts.c (common_handle_option): Disable allocas sanitization for
12696 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
12698 * asan.c: Include gimple-fold.h.
12699 (get_last_alloca_addr): New function.
12700 (handle_builtin_stackrestore): Likewise.
12701 (handle_builtin_alloca): Likewise.
12702 (asan_emit_allocas_unpoison): Likewise.
12703 (get_mem_refs_of_builtin_call): Add new parameter, remove const
12704 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
12705 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
12706 (instrument_builtin_call): Pass gimple iterator to
12707 get_mem_refs_of_builtin_call.
12708 (last_alloca_addr): New global.
12709 * asan.h (asan_emit_allocas_unpoison): Declare.
12710 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
12711 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
12712 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
12713 if function calls alloca.
12714 * gimple-fold.c (replace_call_with_value): Remove static keyword.
12715 * gimple-fold.h (replace_call_with_value): Declare.
12716 * internal-fn.c: Include asan.h.
12717 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
12718 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
12720 2017-07-06 David Malcolm <dmalcolm@redhat.com>
12722 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
12723 (C_SELFTEST_FLAGS): New.
12724 (CPP_SELFTEST_FLAGS): New.
12725 (SELFTEST_DEPS): New, from deps of s-selftest.
12726 (C_SELFTEST_DEPS): New, from deps of s-selftest.
12727 (CPP_SELFTEST_DEPS): New.
12728 (selftest): Add dependency on s-selftest-c++.
12729 (s-selftest): Rename to...
12730 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
12731 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
12732 than SELFTEST_FLAGS.
12733 (selftest-gdb): Rename to...
12734 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
12736 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
12737 (selftest-valgrind): Rename to...
12738 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
12740 (selftest-valgrind): Reintroduce as an alias for
12741 selftest-c-valgrind.
12742 (s-selftest-c++): New.
12743 (selftest-c++-gdb): New.
12744 (selftest-c++-valgrind): New.
12746 2017-07-06 Olivier Hainque <hainque@adacore.com>
12748 * gcc.c (process_command): When deciding if undefined variables
12749 should be ignored when processing specs, accept "gcc -v" as well.
12751 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12753 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
12754 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
12756 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
12758 * config/arm/arm-cpus.in (armv8-r): Add new entry.
12759 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
12760 * config/arm/arm-tables.opt: Regenerate.
12761 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
12763 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
12765 2017-07-06 Carl Love <cel@us.ibm.com>
12767 * ChangeLog: Clean up from mid air collision
12769 2017-07-06 Carl Love <cel@us.ibm.com>
12771 * config/rs6000/rs6000-c.c: Add support for built-in functions
12772 vector signed int vec_subc (vector signed int, vector signed int);
12773 vector signed __int128 vec_subc (vector signed __int128,
12774 vector signed __int128);
12775 vector unsigned __int128 vec_subc (vector unsigned __int128,
12776 vector unsigned __int128);
12777 vector signed int vec_sube (vector signed int, vector signed int,
12778 vector signed int);
12779 vector unsigned int vec_sube (vector unsigned int,
12780 vector unsigned int,
12781 vector unsigned int);
12782 vector signed __int128 vec_sube (vector signed __int128,
12783 vector signed __int128,
12784 vector signed__int128);
12785 vector unsigned __int128 vec_sube (vector unsigned __int128,
12786 vector unsigned __int128,
12787 vector unsigned __int128);
12788 vector signed int vec_subec (vector signed int, vector signed int,
12789 vector signed int);
12790 vector unsigned int vec_subec (vector unsigned int,
12791 vector unsigned int,
12792 vector unsigned int);
12793 vector signed __int128 vec_subec (vector signed __int128,
12794 vector signed __int128,
12795 vector signed__int128);
12796 vector unsigned __int128 vec_subec (vector unsigned __int128,
12797 vector unsigned __int128,
12798 vector unsigned __int128);
12799 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
12800 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
12801 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
12802 BU_ALTIVEC_OVERLOAD_X definitions.
12803 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
12804 * doc/extend.texi: Update the built-in documentation file for the new
12805 built-in functions.
12807 2017-07-06 David Malcolm <dmalcolm@redhat.com>
12810 * diagnostic-show-locus.c (layout::layout): Use start and finish
12811 spelling location for the start and finish of each range.
12812 * genmatch.c (linemap_client_expand_location_to_spelling_point):
12813 Add unused aspect param.
12814 * input.c (expand_location_1): Add "aspect" param, and use it
12815 to access the correct part of the location.
12816 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
12818 (expand_location_to_spelling_point): Likewise.
12819 (linemap_client_expand_location_to_spelling_point): Add "aspect"
12820 param, and pass it to expand_location_1.
12822 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
12824 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
12825 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
12826 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
12827 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
12828 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
12829 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
12830 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
12831 _mm_maskz_getmant_ss): New intrinsics.
12832 (__builtin_ia32_getexpss128_mask): Changed to ...
12833 __builtin_ia32_getexpss128_round ... this.
12834 (__builtin_ia32_getexpsd128_mask): Changed to ...
12835 __builtin_ia32_getexpsd128_round ... this.
12836 * config/i386/i386-builtin-types.def
12837 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
12838 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
12839 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
12840 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
12841 __builtin_ia32_getmantss_mask_round): New builtins.
12842 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
12843 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
12844 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
12845 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
12846 * config/i386/sse.md
12847 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
12848 avx512f_sgetexp<mode><mask_scalar_name>
12849 <round_saeonly_scalar_name> ... this.
12850 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
12851 %0, %1, %2<round_saeonly_op3>}): Changed to ...
12852 vgetexp<ssescalarmodesuffix>
12853 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12854 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
12855 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
12856 avx512f_vgetmant<mode><mask_scalar_name>
12857 <round_saeonly_scalar_name> ... this.
12858 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
12859 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
12860 vgetmant<ssescalarmodesuffix>
12861 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
12862 %0<mask_scalar_operand4>, %1, %2
12863 <round_saeonly_scalar_mask_op4>, %3} ... this.
12864 * config/i386/subst.md (mask_scalar_operand4,
12865 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
12866 round_saeonly_scalar_nimm_predicate): New subst attributes.
12868 2017-07-06 Julia Koval <julia.koval@intel.com>
12870 * config/i386/i386.c (ix86_erase_embedded_rounding):
12871 Remove code for old rounding pattern.
12873 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
12875 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
12877 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
12879 * doc/sourcebuild.texi (Test Directives, Variants of
12880 dg-require-support): Add documentation for dg-require-stack-check.
12882 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
12884 * config/i386/subst.md (mask_scalar, round_scalar,
12885 round_saeonly_scalar): New meta-templates.
12886 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
12887 round_scalar_mask_operand3, round_scalar_mask_op3,
12888 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
12889 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
12890 round_saeonly_scalar_constraint,
12891 round_saeonly_scalar_prefix): New subst attribute.
12892 * config/i386/sse.md
12893 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
12894 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
12895 <round_scalar_name> ... this.
12896 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
12897 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
12898 <round_scalar_name> ... this.
12899 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
12900 <sse>_vm<code><mode>3<mask_scalar_name>
12901 <round_saeonly_scalar_name> ... this.
12902 (v<plusminus_mnemonic><ssescalarmodesuffix>
12903 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
12904 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
12905 v<plusminus_mnemonic><ssescalarmodesuffix>
12906 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12907 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
12908 (v<multdiv_mnemonic><ssescalarmodesuffix>
12909 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
12910 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
12911 v<multdiv_mnemonic><ssescalarmodesuffix>
12912 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12913 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
12914 (v<maxmin_float><ssescalarmodesuffix>
12915 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
12916 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
12917 v<maxmin_float><ssescalarmodesuffix>
12918 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12919 %0<mask_scalar_operand3>, %1, %<iptr>2
12920 <round_saeonly_scalar_mask_op3>} ... this.
12922 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
12924 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
12925 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
12927 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
12928 Alan Hayward <alan.hayward@arm.com>
12929 David Sherwood <david.sherwood@arm.com>
12931 * combine.c (simplify_if_then_else): Remove "enum" before
12933 * compare-elim.c (can_eliminate_compare): Likewise.
12934 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
12936 (aarch64_lookup_simd_builtin_type): Likewise.
12937 (aarch64_simd_builtin_type): Likewise.
12938 (aarch64_init_simd_builtin_types): Likewise.
12939 (aarch64_simd_expand_args): Likewise.
12940 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
12942 (aarch64_reverse_mask): Likewise.
12943 (aarch64_simd_emit_reg_reg_move): Likewise.
12944 (aarch64_gen_adjusted_ldpstp): Likewise.
12945 (aarch64_ccmp_mode_to_code): Likewise.
12946 (aarch64_operands_ok_for_ldpstp): Likewise.
12947 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12948 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
12950 (aarch64_min_divisions_for_recip_mul): Likewise.
12951 (aarch64_reassociation_width): Likewise.
12952 (aarch64_get_condition_code_1): Likewise.
12953 (aarch64_simd_emit_reg_reg_move): Likewise.
12954 (aarch64_simd_attr_length_rglist): Likewise.
12955 (aarch64_reverse_mask): Likewise.
12956 (aarch64_operands_ok_for_ldpstp): Likewise.
12957 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12958 (aarch64_gen_adjusted_ldpstp): Likewise.
12959 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
12961 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
12962 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
12963 (arm_lookup_simd_builtin_type): Likewise.
12964 (arm_simd_builtin_type): Likewise.
12965 (arm_init_simd_builtin_types): Likewise.
12966 (arm_expand_builtin_args): Likewise.
12967 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
12968 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
12969 (ft32_setup_incoming_varargs): Likewise.
12970 (ft32_function_arg): Likewise.
12971 (ft32_function_arg_advance): Likewise.
12972 (ft32_pass_by_reference): Likewise.
12973 (ft32_arg_partial_bytes): Likewise.
12974 (ft32_valid_pointer_mode): Likewise.
12975 (ft32_addr_space_pointer_mode): Likewise.
12976 (ft32_addr_space_legitimate_address_p): Likewise.
12977 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
12979 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
12980 (ix86_emit_outlined_ms2sysv_restore): Likewise.
12981 (iamcu_alignment): Likewise.
12982 (canonicalize_vector_int_perm): Likewise.
12983 (ix86_noce_conversion_profitable_p): Likewise.
12984 (ix86_mpx_bound_mode): Likewise.
12985 (ix86_operands_ok_for_move_multiple): Likewise.
12986 * config/microblaze/microblaze-protos.h
12987 (microblaze_expand_conditional_branch_reg): Likewise.
12988 * config/microblaze/microblaze.c
12989 (microblaze_expand_conditional_branch_reg): Likewise.
12990 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12992 (rs6000_reassociation_width): Likewise.
12993 (rs6000_invalid_binary_op): Likewise.
12994 (fusion_p9_p): Likewise.
12995 (emit_fusion_p9_load): Likewise.
12996 (emit_fusion_p9_store): Likewise.
12997 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
12999 (riscv_hard_regno_mode_ok_p): Likewise.
13000 (riscv_address_insns): Likewise.
13001 (riscv_split_symbol): Likewise.
13002 (riscv_legitimize_move): Likewise.
13003 (riscv_function_value): Likewise.
13004 (riscv_hard_regno_nregs): Likewise.
13005 (riscv_expand_builtin): Likewise.
13006 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
13007 (riscv_build_integer): Likewise.
13008 (riscv_split_integer): Likewise.
13009 (riscv_legitimate_constant_p): Likewise.
13010 (riscv_cannot_force_const_mem): Likewise.
13011 (riscv_regno_mode_ok_for_base_p): Likewise.
13012 (riscv_valid_base_register_p): Likewise.
13013 (riscv_valid_offset_p): Likewise.
13014 (riscv_valid_lo_sum_p): Likewise.
13015 (riscv_classify_address): Likewise.
13016 (riscv_legitimate_address_p): Likewise.
13017 (riscv_address_insns): Likewise.
13018 (riscv_load_store_insns): Likewise.
13019 (riscv_force_binary): Likewise.
13020 (riscv_split_symbol): Likewise.
13021 (riscv_force_address): Likewise.
13022 (riscv_legitimize_address): Likewise.
13023 (riscv_move_integer): Likewise.
13024 (riscv_legitimize_const_move): Likewise.
13025 (riscv_legitimize_move): Likewise.
13026 (riscv_address_cost): Likewise.
13027 (riscv_subword): Likewise.
13028 (riscv_output_move): Likewise.
13029 (riscv_canonicalize_int_order_test): Likewise.
13030 (riscv_emit_int_order_test): Likewise.
13031 (riscv_function_arg_boundary): Likewise.
13032 (riscv_pass_mode_in_fpr_p): Likewise.
13033 (riscv_pass_fpr_single): Likewise.
13034 (riscv_pass_fpr_pair): Likewise.
13035 (riscv_get_arg_info): Likewise.
13036 (riscv_function_arg): Likewise.
13037 (riscv_function_arg_advance): Likewise.
13038 (riscv_arg_partial_bytes): Likewise.
13039 (riscv_function_value): Likewise.
13040 (riscv_pass_by_reference): Likewise.
13041 (riscv_setup_incoming_varargs): Likewise.
13042 (riscv_print_operand): Likewise.
13043 (riscv_elf_select_rtx_section): Likewise.
13044 (riscv_save_restore_reg): Likewise.
13045 (riscv_for_each_saved_reg): Likewise.
13046 (riscv_register_move_cost): Likewise.
13047 (riscv_hard_regno_mode_ok_p): Likewise.
13048 (riscv_hard_regno_nregs): Likewise.
13049 (riscv_class_max_nregs): Likewise.
13050 (riscv_memory_move_cost): Likewise.
13051 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
13052 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
13053 (rl78_addr_space_address_mode): Likewise.
13054 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13056 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13057 (rs6000_reassociation_width): Likewise.
13058 (rs6000_invalid_binary_op): Likewise.
13059 (fusion_p9_p): Likewise.
13060 (emit_fusion_p9_load): Likewise.
13061 (emit_fusion_p9_store): Likewise.
13062 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
13063 (ok_for_simple_move_operands): Likewise.
13064 (ok_for_simple_move_strict_operands): Likewise.
13065 (ok_for_simple_arith_logic_operands): Likewise.
13066 (visium_legitimize_reload_address): Likewise.
13067 (visium_select_cc_mode): Likewise.
13068 (output_cbranch): Likewise.
13069 (visium_split_double_move): Likewise.
13070 (visium_expand_copysign): Likewise.
13071 (visium_expand_int_cstore): Likewise.
13072 (visium_expand_fp_cstore): Likewise.
13073 * config/visium/visium.c (visium_pass_by_reference): Likewise.
13074 (visium_function_arg): Likewise.
13075 (visium_function_arg_advance): Likewise.
13076 (visium_libcall_value): Likewise.
13077 (visium_setup_incoming_varargs): Likewise.
13078 (visium_legitimate_constant_p): Likewise.
13079 (visium_legitimate_address_p): Likewise.
13080 (visium_legitimize_address): Likewise.
13081 (visium_secondary_reload): Likewise.
13082 (visium_register_move_cost): Likewise.
13083 (visium_memory_move_cost): Likewise.
13084 (prepare_move_operands): Likewise.
13085 (ok_for_simple_move_operands): Likewise.
13086 (ok_for_simple_move_strict_operands): Likewise.
13087 (ok_for_simple_arith_logic_operands): Likewise.
13088 (visium_function_value_1): Likewise.
13089 (rtx_ok_for_offset_p): Likewise.
13090 (visium_legitimize_reload_address): Likewise.
13091 (visium_split_double_move): Likewise.
13092 (visium_expand_copysign): Likewise.
13093 (visium_expand_int_cstore): Likewise.
13094 (visium_expand_fp_cstore): Likewise.
13095 (visium_split_cstore): Likewise.
13096 (visium_select_cc_mode): Likewise.
13097 (visium_split_cbranch): Likewise.
13098 (output_cbranch): Likewise.
13099 (visium_print_operand_address): Likewise.
13100 * expmed.c (flip_storage_order): Likewise.
13101 * expmed.h (emit_cstore): Likewise.
13102 (flip_storage_order): Likewise.
13103 * genrecog.c (validate_pattern): Likewise.
13104 * hsa-gen.c (gen_hsa_addr): Likewise.
13105 * internal-fn.c (expand_arith_overflow): Likewise.
13106 * ira-color.c (allocno_copy_cost_saving): Likewise.
13107 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13108 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
13109 (process_invariant_for_inheritance): Likewise.
13110 * lra-eliminations.c (move_plus_up): Likewise.
13111 * omp-low.c (lower_oacc_reductions): Likewise.
13112 * simplify-rtx.c (simplify_subreg): Likewise.
13113 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
13114 (TARGET_CHKP_BOUND_MODE): Likewise..
13115 * targhooks.c (default_chkp_bound_mode): Likewise.
13116 (default_setup_incoming_vararg_bounds): Likewise.
13117 * targhooks.h (default_chkp_bound_mode): Likewise.
13118 (default_setup_incoming_vararg_bounds): Likewise.
13119 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
13120 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
13121 (have_whole_vector_shift): Likewise.
13122 * tree-vect-stmts.c (vectorizable_load): Likewise.
13123 * doc/tm.texi: Regenerate.
13125 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
13127 Graceful degrade if Binutils PR21472 is not available.
13130 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
13131 .rodata in flash test fails.
13132 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
13133 * confgure: Regenerate.
13134 * config.in: Regenerate.
13135 * config/avr/avr.c (avr_asm_named_section)
13136 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
13137 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
13138 (avr_asm_init_sections): Same.
13140 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13142 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
13143 (fma<VH:mode>4_intrinsic): Likewise.
13144 (*fmsub<VCVTF:mode>4): Likewise.
13145 (*fmsub<VH:mode>4_intrinsic): Likewise.
13147 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
13150 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
13151 Don't depend on "optimize > 0".
13152 (out_movhi_r_mr, out_movqi_mr_r): Same.
13153 (out_movhi_mr_r, out_movqi_r_mr): Same.
13154 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
13155 io_address_operand on "optimize > 0".
13157 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13159 * tree-loop-distribution.c: Add general explanantion on the pass.
13160 (generate_loops_for_partition): Mark distributed loop.
13161 (pg_add_dependence_edges): New parameter. Handle alias data
13162 dependence specially and record it in the parameter if asked.
13163 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
13164 (init_partition_graph_vertices, add_partition_graph_edge): New.
13165 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
13166 (free_partition_graph_vdata, build_partition_graph): New.
13167 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
13168 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
13169 (data_ref_segment_size, latch_dominated_by_data_ref): New.
13170 (compute_alias_check_pairs, version_loop_by_alias_check): New.
13171 (version_for_distribution_p, finalize_partitions): New.
13172 (distribute_loop): Handle alias data dependence specially. Factor
13173 out loop fusion code as functions and call these functions.
13175 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13177 * tree-loop-distribution.c (classify_partition): New parameter and
13178 better handle reduction statement.
13179 (rdg_build_partitions): Revise comment.
13180 (distribute_loop): Compute statements in all partitions and pass it
13181 to classify_partition.
13183 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13185 * tree-loop-distribution.c (enum partition_type): New.
13186 (struct partition): New field type.
13187 (partition_merge_into): Add parameter. Update partition type.
13188 (data_dep_in_cycle_p, update_type_for_merge): New functions.
13189 (build_rdg_partition_for_vertex): Compute partition type.
13190 (rdg_build_partitions): Dump partition type.
13191 (distribute_loop): Update calls to partition_merge_into.
13193 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13195 * tree-loop-distribution.c (struct ddr_hasher): New.
13196 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
13198 (classify_partition): Call get_data_dependence.
13199 (pg_add_dependence_edges): Ditto.
13200 (distribute_loop): Release data dependence hash table.
13202 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13204 * tree-loop-distribution.c (ref_base_address): Delete.
13205 (similar_memory_accesses): Rename ...
13206 (share_memory_accesses): ... to this. Check if partitions access
13207 the same memory reference.
13208 (distribute_loop): Call share_memory_accesses.
13210 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13212 * tree-loop-distribution.c (struct partition): New field recording
13213 its data reference.
13214 (partition_alloc, partition_free): Init and release data refs.
13215 (partition_merge_into): Merge data refs.
13216 (build_rdg_partition_for_vertex): Collect data refs for partition.
13217 (pg_add_dependence_edges): Change parameters from vector to bitmap.
13219 (distribute_loop): Remve data refs from vertice data of partition
13222 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13224 * tree-loop-distribution.c (params.h): Include header file.
13225 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
13226 (datarefs_vec): New global var.
13227 (create_rdg_vertices): Use datarefs_vec directly.
13228 (free_rdg): Don't free data references.
13229 (build_rdg): Update use. Don't free data references.
13230 (distribute_loop): Compute global variable for data references.
13231 Bail out if there are too many data references.
13233 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13235 * tree-loop-distribution.c (loop_nest): New global var.
13236 (build_rdg): Use loop directly, rather than loop nest.
13237 (pg_add_dependence_edges): Remove loop nest parameter. Use global
13239 (distribute_loop): Compute global variable loop nest. Update use.
13241 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13243 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
13244 (partition_merge_into): New parameter. Dump reason for fusion.
13245 (distribute_loop): Update use of partition_merge_into.
13247 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13249 * tree-loop-distribution.c (bb_top_order_index): New.
13250 (bb_top_order_index_size, bb_top_order_cmp): New.
13251 (stmts_from_loop): Use topological order.
13252 (pass_loop_distribution::execute): Compute and release topological
13253 order for basic blocks.
13255 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13257 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
13260 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13262 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
13263 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
13264 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
13265 * internal-fn.def (LOOP_DIST_ALIAS): New.
13266 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
13267 (fold_loop_internal_call): ... this.
13268 (vect_loop_dist_alias_call): New function.
13269 (set_uid_loop_bbs): Call fold_loop_internal_call.
13270 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
13273 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
13276 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
13277 Require dead FLAGS_REG at the beginning of a peephole.
13279 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
13282 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
13283 arguments in the call to __builtin_ia32_sbb_u32.
13284 (_subborrow_u64): Swap _X and _Y arguments in the call to
13285 __builtin_ia32_sbb_u64.
13287 2017-07-04 Jakub Jelinek <jakub@redhat.com>
13290 * tree-vrp.c (compare_assert_loc): Turn into a function template
13291 with stable template parameter. Only test if a->e is NULL,
13292 !a->e == !b->e has been verified already. Use e == NULL or
13293 e != NULL instead of e or ! e tests. If stable is true, don't use
13294 iterative_hash_expr, on the other side allow a or b or both NULL
13295 and sort the NULLs last.
13296 (process_assert_insertions): Sort using compare_assert_loc<false>
13297 instead of compare_assert_loc, later sort using
13298 compare_assert_loc<true> before calling process_assert_insertions_for
13299 in a loop. Use break instead of continue once seen NULL pointer.
13301 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
13303 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
13304 Cortex-R7 and Cortex-R8 processors.
13306 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
13308 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
13309 uninitialized while src is not.
13311 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
13313 * common/config/arm/arm-common.c: Adjust include path for
13315 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
13316 (arm-cpu.h): Create in build directory. Adjust dependency rules.
13317 (arm-cpu-data.h): Likewise.
13318 (arm-cpu-cdata.h): Likewise.
13319 * config/arm/arm-cpu.h: Delete.
13320 * config/arm/arm-cpu-cdata.h: Delete.
13321 * config/arm/arm-cpu-data.h: Delete.
13323 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
13325 * config/arm/arm-cpus.in (cortex-a55): New.
13326 (cortex-a75): Likewise.
13327 (cortex-a75.cortex-a55): Likewise.
13328 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
13330 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
13331 * config/arm/arm-cpu-cdata.h: Regenerate.
13332 * config/arm/arm-cpu-data.h: Regenerate.
13333 * config/arm/arm-cpu.h: Regenerate.
13334 * config/arm/arm-tables.opt: Regenerate.
13335 * config/arm/arm-tune.md: Regenerate.
13337 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
13339 * haifa-sched.c (sched_create_recovery_edges): Update profile.
13341 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
13343 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
13346 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
13348 PR tree-optimization/81292
13349 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
13350 full_string_p, also call adjust_related_strinfos if the adjustment
13351 is simple, otherwise invalidate related strinfos.
13353 2017-07-04 Martin Liska <mliska@suse.cz>
13356 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
13357 newly created variable as DECL_IGNORED_P.
13359 2017-07-04 Martin Liska <mliska@suse.cz>
13362 * ipa-inline.c (inline_small_functions):
13363 Use xstrdup_for_dump.
13365 2017-07-04 Tom de Vries <tom@codesourcery.com>
13367 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
13369 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
13372 * config/darwin.c (darwin_function_switched_text_sections):
13375 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
13377 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
13379 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
13381 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
13383 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13385 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
13386 min_profitable_iters, and th as inclusive lower bounds.
13387 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
13388 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
13389 for min_profitable_iters and min_profitable_estimate.
13390 (vect_transform_loop): Treat th as an inclusive lower bound.
13391 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
13393 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
13396 * config/darwin.c (darwin_function_switched_text_sections):
13397 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
13398 in two pieces, and suppress the use of buf.
13400 2017-07-03 Nathan Sidwell <nathan@acm.org>
13402 * hash-table.h (hash_table_mod1): Fix indentation.
13404 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13406 PR middle-end/81290
13407 * predict.c (force_edge_cold): Be more careful about propagation
13409 * profile-count.h (profile_probability::guessed,
13410 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
13412 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
13414 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
13416 * doc/invoke.texi (rcpc architecture extension): Document it.
13418 2017-07-03 Richard Biener <rguenther@suse.de>
13420 PR tree-optimization/60510
13421 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
13422 the scalar reduction PHI and use it.
13423 (vectorizable_reduction): Properly guard the single_defuse_cycle
13424 path for non-SLP reduction chains where we cannot use it.
13425 Rework reduc_def/index and vector type deduction. Rework
13426 vector operand gathering during reduction op code-gen.
13427 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
13428 chains dissolve the chain and leave it to non-SLP reduction
13431 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13433 * tree-data-ref.h (dr_alignment): Declare.
13434 * tree-data-ref.c (dr_alignment): New function.
13435 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
13436 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
13438 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
13440 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13442 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
13443 and base_misalignment fields.
13444 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
13445 * tree-data-ref.c: Include builtins.h.
13446 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
13447 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
13448 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
13449 * tree-vect-data-refs.c: Include tree-cfg.h.
13450 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
13451 fields instead of calculating an alignment here.
13452 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
13453 innermost_loop_behavior fields.
13455 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13457 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
13459 (DR_STEP_ALIGNMENT): New macro.
13460 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
13461 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
13462 (create_data_ref): Print it.
13463 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
13464 to tell whether the step preserves vector (mis)alignment.
13465 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
13466 Move the check for an integer step and generalise to all INTEGER_CST.
13467 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
13468 Print the outer step alignment.
13470 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13472 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
13473 with offset_alignment.
13474 (DR_ALIGNED_TO): Delete.
13475 (DR_OFFSET_ALIGNMENT): New macro.
13476 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
13477 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
13478 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
13479 (create_data_ref): Likewise.
13480 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
13481 (vect_analyze_data_refs): Likewise.
13482 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
13483 creating dummy innermost behavior.
13485 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13487 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
13488 with a "innermost_loop_behavior *" and refeence tree.
13489 * tree-data-ref.c (dr_analyze_innermost): Likewise.
13490 (create_data_ref): Update call accordingly.
13491 * tree-predcom.c (find_looparound_phi): Likewise.
13493 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13495 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
13496 fields with dr_wrt_vec_loop.
13497 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
13498 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
13499 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
13500 (vect_dr_behavior): New function.
13501 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
13502 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
13503 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
13504 track whether the step preserves the misalignment.
13505 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
13506 Use vect_dr_behavior.
13507 (vect_setup_realignment): Update call accordingly.
13508 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
13509 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
13510 call to vect_create_addr_base_for_vector_ref.
13511 (vect_create_cond_for_align_checks): Likewise.
13512 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
13513 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
13514 (vect_recog_mask_conversion_pattern): Likewise.
13515 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
13516 (new_stmt_vec_info): Remove redundant zeroing.
13518 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
13520 * common/config/arm/arm-common.c (arm_be8_option): New function.
13521 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
13522 (ISA_ARMv6): Add isa_bit_be8.
13523 * config/arm/arm.h (arm_be8_option): Add prototype.
13524 (BE8_SPEC_FUNCTION): New define.
13525 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
13526 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
13527 (mlittle-endian): Similarly.
13528 (mbe8, mbe32): New options.
13529 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
13530 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
13532 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13534 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
13536 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13538 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
13539 (cleanup_tree_cfg_bb): Use it.
13540 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
13542 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
13544 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13547 * loop-doloop.c (add_test): Update profile.
13549 2017-07-03 Martin Liska <mliska@suse.cz>
13552 * sanopt.c (rewrite_usage_of_param): New function.
13553 (sanitize_rewrite_addressable_params): Likewise.
13554 (pass_sanopt::execute): Call rewrite_usage_of_param.
13556 2017-07-03 Richard Biener <rguenther@suse.de>
13558 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
13559 back to using VIEW_CONVERT_EXPR.
13561 2017-07-03 Martin Liska <mliska@suse.cz>
13564 * doc/extend.texi: Document when a resolver function is
13565 generated for target_clones.
13567 2017-07-03 Martin Liska <mliska@suse.cz>
13569 * asan.c (asan_emit_stack_protection): Unpoison just red zones
13570 and shadow memory of auto variables which are subject of
13571 use-after-scope sanitization.
13572 (asan_expand_mark_ifn): Add do set only when is_poison.
13574 2016-07-03 Richard Biener <rguenther@suse.de>
13576 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
13578 (vect_force_simple_reduction): Record reduction def -> phi mapping.
13579 (vectorizable_reduction): Perform reduction PHI creation when
13580 visiting a reduction PHI and adjust and simplify code generation
13581 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
13582 (vect_transform_loop): Visit reduction PHIs.
13583 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
13584 defs into the SLP tree.
13585 (vect_build_slp_tree): Reduction defs terminate the recursion.
13586 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
13588 (vect_get_vec_defs_for_stmt_copy): Export.
13589 (vect_get_vec_defs): Likewise.
13590 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
13592 (vect_get_vec_defs_for_stmt_copy): Declare.
13593 (vect_get_vec_defs): Likewise.
13595 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13597 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
13598 parameter with a "loop" parameter and use it instead of the
13599 loop containing DR_STMT. Don't check simple_iv when doing
13600 BB analysis. Describe the two analysis modes in the comment.
13602 2017-07-03 Tom de Vries <tom@codesourcery.com>
13604 PR tree-optimization/69468
13605 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
13606 (find_same_succ_bb): Handle ignore_edge_flags.
13608 2017-07-03 Tom de Vries <tom@codesourcery.com>
13610 PR tree-optimization/81192
13611 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
13613 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
13615 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
13617 2017-07-03 Tom de Vries <tom@codesourcery.com>
13619 PR tree-optimization/81192
13620 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
13621 BB_SAME_SUCC (bb) == NULL.
13623 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13625 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
13628 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13630 * dumpfile.c: Include profile-count.h
13631 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
13633 (insert_cond_bb): Update profile.
13634 * tree-cfg.h (insert_cond_bb): Update prototype.
13635 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
13636 * tree-dump.c: Do not include tree-cfg.
13638 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13640 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
13642 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13644 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
13647 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13649 * tree-complex.c (expand_complex_div_wide): update profile.
13651 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13652 Alan Hayward <alan.hayward@arm.com>
13653 David Sherwood <david.sherwood@arm.com>
13655 * Makefile.in (MACHMODE_H): Remove insn-modes.h
13656 (CORETYPES_H): New define.
13657 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
13658 (insn-modes-inline.h, s-modes-inline-h): New rules.
13659 (generated_files): Add insn-modes-inline.h.
13660 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
13661 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
13662 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
13663 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
13664 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
13665 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
13666 (build/gencodes.o, build/genconditions.o): Likewise.
13667 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
13668 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
13669 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
13670 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
13671 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
13672 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
13673 * coretypes.h: Include everything up to real.h for generators.
13674 Include insn-modes.h first. Include wide-int-print.h after
13675 wide-int.h. Include insn-modes-inline.h and then machmode.h.
13676 * machmode.h: Don't include insn-modes.h here.
13677 * function-tests.c: Remove includes of signop.h, machmode.h,
13678 double-int.h and wide-int.h.
13680 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
13682 * optc-save-gen.awk: Likewise.
13683 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
13684 * godump.c: Remove include of wide-int-print.h.
13685 * pretty-print.h: Likewise.
13686 * wide-int-print.cc: Likewise.
13687 * wide-int.cc: Likewise.
13688 * hash-map-tests.c: Remove include of signop.h.
13689 * hash-set-tests.c: Likewise.
13690 * rtl-tests.c: Likewise.
13691 * mkconfig.sh: Remove include of machmode.h.
13692 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
13694 (emit_insn_modes_inline_h): ...this new function. Emit the code
13695 into an insn-modes-inline.h header file, adding appropriate
13696 include guards and end comments.
13697 (emit_insn_modes_c_header): Remove include of machmode.h.
13698 (emit_min_insn_modes_c_header): Include coretypes.h rather than
13700 (main): Handle -i flag and call emit_insn_modes_inline_h when
13703 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13705 * tree-ssa-strlen.c (strinfo): Rename the length field to
13706 nonzero_chars. Add a full_string_p field.
13707 (compare_nonzero_chars, zero_length_string_p): New functions.
13708 (get_addr_stridx): Add an offset_out parameter.
13709 Use compare_nonzero_chars.
13710 (get_stridx): Update accordingly. Use compare_nonzero_chars.
13711 (new_strinfo): Update after above changes to strinfo.
13712 (set_endptr_and_length): Set full_string_p.
13713 (get_string_length): Update after above changes to strinfo.
13714 (unshare_strinfo): Update call to new_strinfo.
13715 (maybe_invalidate): Likewise.
13716 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
13717 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
13718 as a uhwi instead of an shwi. Update after above changes to
13719 strinfo and new_strinfo.
13720 (zero_length_string): Assert that chainsi contains full strings.
13721 Use zero_length_string_p. Update call to new_strinfo.
13722 (adjust_related_strinfos): Update after above changes to strinfo.
13723 Copy full_string_p from origsi.
13724 (adjust_last_stmt): Use zero_length_string_p.
13725 (handle_builtin_strlen): Update after above changes to strinfo and
13726 new_strinfo. Install the lhs as the string length if the previous
13727 entry didn't describe a full string.
13728 (handle_builtin_strchr): Update after above changes to strinfo
13730 (handle_builtin_strcpy): Likewise.
13731 (handle_builtin_strcat): Likewise.
13732 (handle_builtin_malloc): Likewise.
13733 (handle_pointer_plus): Likewise.
13734 (handle_builtin_memcpy): Likewise. Track nonzero characters
13735 that aren't necessarily followed by a nul terminator.
13736 (handle_char_store): Likewise.
13738 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13740 PR tree-optimization/80769
13741 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
13742 for malloc and calloc. Document the new invariant that all related
13743 strinfos have delayed lengths or none do.
13744 (verify_related_strinfos): Move earlier in file.
13745 (set_endptr_and_length): New function, split out from...
13746 (get_string_length): ...here. Also set the lengths of related
13748 (zero_length_string): Assert that chainsi has known (rather than
13750 (adjust_related_strinfos): Likewise.
13752 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13754 PR tree-optimization/81136
13755 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
13756 assert that two references with the same misalignment have the same
13757 compile-time misalignment if those compile-time misalignments
13760 2017-07-01 Andi Kleen <ak@linux.intel.com>
13762 * print-tree.c (print_node): Print all attributes.
13764 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13766 * cfg.c (scale_bbs_frequencies): New function.
13767 * cfg.h (scale_bbs_frequencies): Declare it.
13768 * cfgloopanal.c (single_likely_exit): Cleanup.
13769 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
13771 (scale_loop_profile): Likewise.
13772 (loop_version): Likewise.
13773 (create_empty_loop_on_edge): Update.
13774 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
13775 scale_loop_frequencies, scale_loop_profile, loopify,
13776 loop_version): Update prototypes.
13777 * modulo-sched.c (sms_schedule): Update.
13778 * predict.c (unlikely_executed_edge_p): Also check probability.
13779 (probably_never_executed_edge_p): Fix typo.
13780 * tree-if-conv.c (version_loop_for_if_conversion): Update.
13781 * tree-parloops.c (gen_parallel_loop): Update.
13782 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
13783 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13784 * tree-ssa-loop-split.c (split_loop): Update.
13785 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
13786 * tree-vect-loop-manip.c (vect_do_peeling): Update.
13787 (vect_loop_versioning): Update.
13788 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13790 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13792 * trans-mem.c (split_bb_make_tm_edge): Update profile.
13794 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13796 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
13797 to keep profile consistent.
13799 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13801 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
13802 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
13803 * profile-count.h (max_safe_multiplier): Make unsigned.
13804 (profile_count::guessed_zero): New.
13806 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13808 * bb-reorder.c (fix_up_crossing_landing_pad,
13809 fix_crossing_conditional_branches): Use make_single_succ_edge
13810 to keep profile consistent.
13812 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13814 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
13817 2017-07-01 Jakub Jelinek <jakub@redhat.com>
13820 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
13821 the right scopes, make sure cond_jump isn't preserved between multiple
13822 iterations. Search for fallthru edge whenever there are 3+ edges and
13823 use find_fallthru_edge for it.
13825 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13827 Patch by Alexander Monakov <amonakov@ispras.ru>
13828 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
13829 probabilities consistently.
13831 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13833 * pa.c (pa_expand_compare_and_swap_loop): Update call of
13834 emit_cmp_and_jump_insns.
13836 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13839 * tree-inline.c (expand_call_inline): Combine profile statuses.
13841 2017-06-30 Andrew Pinski <apinski@cavium.com>
13843 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
13844 fold_stmt returned true.
13846 2017-06-30 Nathan Sidwell <nathan@acm.org>
13848 * ggc.h (empty_string): Delete.
13849 * cfgexpand.c (expand_asm_stmt): Use plain "".
13850 * optabs.c (expand_asm_memory_barrier): Likewise.
13851 * stringpool.c (empty_string): Delete.
13852 (digit_vector, digit_string): Delete.
13853 (ggc_alloc_string): Use plain "", don't optimize single digit
13854 strings. Use ggc_alloc_atomic.
13856 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
13858 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
13859 comparison set and one other set, use the cost of the non-comparison
13862 2017-06-30 Nathan Sidwell <nathan@acm.org>
13864 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
13867 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
13869 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
13870 loops. Remove now unneeded calls to gimple_switch_set_label() that
13871 just set removed labels to NULL_TREE.
13873 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
13875 * tree-ssanames.c (set_range_info_raw): Abstract from ...
13876 (set_range_info): ...here. Only call set_range_info_raw if domain
13878 (set_nonzero_bits): Call set_range_info_raw.
13879 * tree-ssanames.h (set_range_info_raw): New.
13881 2017-06-30 Jakub Jelinek <jakub@redhat.com>
13884 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
13885 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
13886 of nonimmediate_operand and <store_mask_constraint> instead of m
13887 for the input operand. For V8FI iterator, always split if input
13888 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
13889 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
13890 <store_mask_predicate> instead of register_operand and
13891 <store_mask_constraint> instead of v for the input operand. Make
13892 sure both operands aren't MEMs for if not <mask_applied>.
13894 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
13896 * lto-wrapper.c (copy_file) Close both file descriptors before
13899 2017-06-30 Martin Liska <mliska@suse.cz>
13902 * multiple_target.c (create_dispatcher_calls): Make ifunc
13903 also for function that don't have calls or are not referenced.
13905 2017-06-30 Richard Biener <rguenther@suse.de>
13907 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
13908 analyze the first scalar stmt. Move vector type computation
13909 for the BB case here from ...
13910 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
13911 live operation processing in the SLP case properly.
13913 2017-06-30 Richard Biener <rguenther@suse.de>
13915 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
13917 2017-06-30 Martin Liska <mliska@suse.cz>
13920 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
13921 before BUILT_IN_UNWIND_RESUME when ASAN is used.
13923 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
13925 * doc/invoke.texi (AArch64): Add missing options and remove redundant
13928 2017-06-30 Richard Biener <rguenther@suse.de>
13930 PR tree-optimization/81249
13931 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
13932 condition reduction result to original scalar type.
13934 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13936 * profile-count.h (enum profile_quality): Fix typos and whitespace
13939 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13941 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
13942 type for branch probabilities.
13944 2017-06-29 Julian Brown <julian@codesourcery.com>
13945 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13947 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
13948 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
13949 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
13950 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
13952 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13954 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
13955 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
13956 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
13957 CC usage from generic code to here.
13958 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
13959 CC usage into the target macros.
13961 2017-06-29 Maya Rashish <coypu@sdf.org>
13963 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
13966 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13968 * arm/arm-builtins.c: Include profile-count.h
13969 * except.c (sjlj_emit_function_enter): Use
13970 profile_probability::unlikely.
13972 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13974 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
13975 and tocrel_offset be pointer args rather than implicitly using
13977 (legitimate_constant_pool_address_p, rs6000_emit_move,
13978 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
13979 tocrel_offset and use in toc_relative_expr_p call.
13980 (print_operand, print_operand_address): Use static tocrel_base_oac
13981 and tocrel_offset_oac.
13982 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
13985 2017-06-29 Maya Rashish <coypu@sdf.org>
13987 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
13989 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
13991 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
13992 objects, take into account only the alignment of 'op0' and 'mode1' if
13995 2017-06-29 Steve Ellcey <sellcey@cavium.com>
13997 * ccmp.c (ccmp_tree_comparison_p): New function.
13998 (ccmp_candidate_p): Update to use above function.
13999 (get_compare_parts): New function.
14000 (expand_ccmp_next): Update to use new functions.
14001 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
14003 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
14004 take mode as argument.
14005 * ccmp.h (expand_ccmp_expr): Add mode as argument.
14006 * expr.c (expand_expr_real_1): Pass mode as argument.
14008 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
14010 * combine.c (combine_instructions): Print insns to dump_file, together
14013 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14015 * asan.c (asan_emit_stack_protection): Update.
14016 (create_cond_insert_point): Update.
14017 * auto-profile.c (afdo_propagate_circuit): Update.
14018 * basic-block.h (struct edge_def): Turn probability to
14019 profile_probability.
14020 (EDGE_FREQUENCY): Update.
14021 * bb-reorder.c (find_traces_1_round): Update.
14022 (better_edge_p): Update.
14023 (sanitize_hot_paths): Update.
14024 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
14025 (make_single_succ_edge): Update.
14026 (check_bb_profile): Update.
14027 (dump_edge_info): Update.
14028 (update_bb_profile_for_threading): Update.
14029 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
14030 probabilitycount to 0.
14031 * cfgbuild.c (compute_outgoing_frequencies): Update.
14032 * cfgcleanup.c (try_forward_edges): Update.
14033 (outgoing_edges_match): Update.
14034 (try_crossjump_to_edge): Update.
14035 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
14036 (expand_gimple_tailcall): Update.
14037 (construct_init_block): Use make_single_succ_edge.
14038 (construct_exit_block): Use make_single_succ_edge.
14039 * cfghooks.c (verify_flow_info): Update.
14040 (redirect_edge_succ_nodup): Update.
14041 (split_edge): Update.
14042 (account_profile_record): Update.
14043 * cfgloopanal.c (single_likely_exit): Update.
14044 * cfgloopmanip.c (scale_loop_profile): Update.
14045 (set_zero_probability): Remove.
14046 (duplicate_loop_to_header_edge): Update.
14047 * cfgloopmanip.h (loop_version): Update prototype.
14048 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
14049 (force_nonfallthru_and_redirect): Update.
14050 (update_br_prob_note): Update.
14051 (rtl_verify_edges): Update.
14052 (purge_dead_edges): Update.
14053 (rtl_lv_add_condition_to_bb): Update.
14054 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
14055 * cgraphunit.c (init_lowered_empty_function): Update.
14056 (cgraph_node::expand_thunk): Update.
14057 * cilk-common.c: Include profile-count.h
14058 * dojump.c (inv): Remove.
14059 (jumpifnot): Update.
14060 (jumpifnot_1): Update.
14061 (do_jump_1): Update.
14063 (do_jump_by_parts_greater_rtx): Update.
14064 (do_compare_rtx_and_jump): Update.
14065 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
14066 do_jump_1. do_compare_rtx_and_jump): Update prototype.
14067 * dwarf2cfi.c: Include profile-count.h
14068 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
14069 (sjlj_emit_dispatch_table): Likewise.
14070 * explow.c: Include profile-count.h
14071 * expmed.c (emit_store_flag_force): Update.
14072 (do_cmp_and_jump): Update.
14073 * expr.c (compare_by_pieces_d::generate): Update.
14074 (compare_by_pieces_d::finish_mode): Update.
14075 (emit_block_move_via_loop): Update.
14076 (store_expr_with_bounds): Update.
14077 (store_constructor): Update.
14078 (expand_expr_real_2): Update.
14079 (expand_expr_real_1): Update.
14080 * expr.h (try_casesi, try_tablejump): Update prototypes.
14081 * gimple-pretty-print.c (dump_probability): Update.
14082 (dump_profile): New.
14083 (dump_gimple_label): Update.
14084 (dump_gimple_bb_header): Update.
14085 * graph.c (draw_cfg_node_succ_edges): Update.
14086 * hsa-gen.c (convert_switch_statements): Update.
14087 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
14088 (find_if_case_1): Update.
14089 (find_if_case_2): Update.
14090 * internal-fn.c (expand_arith_overflow_result_store): Update.
14091 (expand_addsub_overflow): Update.
14092 (expand_neg_overflow): Update.
14093 (expand_mul_overflow): Update.
14094 (expand_vector_ubsan_overflow): Update.
14095 * ipa-cp.c (good_cloning_opportunity_p): Update.
14096 * ipa-split.c (split_function): Use make_single_succ_edge.
14097 * ipa-utils.c (ipa_merge_profiles): Update.
14098 * loop-doloop.c (add_test): Update.
14099 (doloop_modify): Update.
14100 * loop-unroll.c (compare_and_jump_seq): Update.
14101 (unroll_loop_runtime_iterations): Update.
14102 * lra-constraints.c (lra_inheritance): Update.
14103 * lto-streamer-in.c (input_cfg): Update.
14104 * lto-streamer-out.c (output_cfg): Update.
14105 * mcf.c (adjust_cfg_counts): Update.
14106 * modulo-sched.c (sms_schedule): Update.
14107 * omp-expand.c (expand_omp_for_init_counts): Update.
14108 (extract_omp_for_update_vars): Update.
14109 (expand_omp_ordered_sink): Update.
14110 (expand_omp_for_ordered_loops): Update.
14111 (expand_omp_for_generic): Update.
14112 (expand_omp_for_static_nochunk): Update.
14113 (expand_omp_for_static_chunk): Update.
14114 (expand_cilk_for): Update.
14115 (expand_omp_simd): Update.
14116 (expand_omp_taskloop_for_outer): Update.
14117 (expand_omp_taskloop_for_inner): Update.
14118 * omp-simd-clone.c (simd_clone_adjust): Update.
14119 * optabs.c (expand_doubleword_shift): Update.
14120 (expand_abs): Update.
14121 (emit_cmp_and_jump_insn_1): Update.
14122 (expand_compare_and_swap_loop): Update.
14123 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
14124 * predict.c (predictable_edge_p): Update.
14125 (edge_probability_reliable_p): Update.
14126 (set_even_probabilities): Update.
14127 (combine_predictions_for_insn): Update.
14128 (combine_predictions_for_bb): Update.
14129 (propagate_freq): Update.
14130 (estimate_bb_frequencies): Update.
14131 (force_edge_cold): Update.
14132 * profile-count.c (profile_count::dump): Add missing space into dump.
14133 (profile_count::debug): Add newline.
14134 (profile_count::differs_from_p): Explicitly convert to unsigned.
14135 (profile_count::stream_in): Update.
14136 (profile_probability::dump): New member function.
14137 (profile_probability::debug): New member function.
14138 (profile_probability::differs_from_p): New member function.
14139 (profile_probability::differs_lot_from_p): New member function.
14140 (profile_probability::stream_in): New member function.
14141 (profile_probability::stream_out): New member function.
14142 * profile-count.h (profile_count_quality): Rename to ...
14143 (profile_quality): ... this one.
14144 (profile_probability): New.
14145 (profile_count): Update.
14146 * profile.c (compute_branch_probabilities): Update.
14147 * recog.c (peep2_attempt): Update.
14148 * sched-ebb.c (schedule_ebbs): Update.
14149 * sched-rgn.c (find_single_block_region): Update.
14150 (compute_dom_prob_ps): Update.
14151 (schedule_region): Update.
14152 * sel-sched-ir.c (compute_succs_info): Update.
14153 * stmt.c (struct case_node): Update.
14154 (do_jump_if_equal): Update.
14155 (get_outgoing_edge_probs): Update.
14156 (conditional_probability): Update.
14157 (emit_case_dispatch_table): Update.
14158 (expand_case): Update.
14159 (expand_sjlj_dispatch_table): Update.
14160 (emit_case_nodes): Update.
14161 * targhooks.c: Update.
14162 * tracer.c (better_p): Update.
14163 (find_best_successor): Update.
14164 * trans-mem.c (expand_transaction): Update.
14165 * tree-call-cdce.c: Update.
14166 * tree-cfg.c (gimple_split_edge): Upate.
14167 (move_sese_region_to_fn): Upate.
14168 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
14169 * tree-eh.c (lower_resx): Upate.
14170 (cleanup_empty_eh_move_lp): Upate.
14171 * tree-if-conv.c (version_loop_for_if_conversion): Update.
14172 * tree-inline.c (copy_edges_for_bb): Update.
14173 (copy_cfg_body): Update.
14174 * tree-parloops.c (gen_parallel_loop): Update.
14175 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
14176 (gimple_gen_time_profiler): Update.
14177 * tree-ssa-dce.c (remove_dead_stmt): Update.
14178 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
14179 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
14180 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
14181 (unloop_loops): Update.
14182 (try_peel_loop): Update.
14183 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14184 * tree-ssa-loop-split.c (connect_loops): Update.
14185 (split_loop): Update.
14186 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
14187 (hoist_guard): Update.
14188 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
14189 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
14190 (value_replacement): Update.
14191 * tree-ssa-reassoc.c (branch_fixup): Update.
14192 * tree-ssa-tail-merge.c (replace_block_by): Update.
14193 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
14194 (create_edge_and_update_destination_phis): Update.
14195 (compute_path_counts): Update.
14196 (recompute_probabilities): Update.
14197 (update_joiner_offpath_counts): Update.
14198 (freqs_to_counts_path): Update.
14199 (duplicate_thread_path): Update.
14200 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
14201 (struct switch_conv_info): Update.
14202 (gen_inbound_check): Update.
14203 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
14204 (vect_do_peeling): Update.
14205 (vect_loop_versioning): Update.
14206 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14207 (optimize_mask_stores): Update.
14208 * ubsan.c (ubsan_expand_null_ifn): Update.
14209 * value-prof.c (gimple_divmod_fixed_value): Update.
14210 (gimple_divmod_fixed_value_transform): Update.
14211 (gimple_mod_pow2): Update.
14212 (gimple_mod_pow2_value_transform): Update.
14213 (gimple_mod_subtract): Update.
14214 (gimple_mod_subtract_transform): Update.
14215 (gimple_ic): Update.
14216 (gimple_stringop_fixed_value): Update.
14217 (gimple_stringops_transform): Update.
14218 * value-prof.h: Update.
14220 2017-06-29 Carl Love <cel@us.ibm.com>
14222 * config/rs6000/rs6000-c.c: Add support for built-in functions
14223 vector signed int vec_signed (vector float);
14224 vector signed long long vec_signed (vector double);
14225 vector signed int vec_signed2 (vector double, vector double);
14226 vector signed int vec_signede (vector double);
14227 vector signed int vec_signedo (vector double);
14228 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
14229 instruction generator.
14230 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
14231 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
14232 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
14234 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
14235 vunsignede_v2df): Add define_expands.
14236 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
14237 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
14238 VEC_UNSIGNEDO): Add definitions.
14239 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
14240 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
14241 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
14242 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
14243 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
14244 * config/rs6000/altivec.h (vec_signed, vec_signed2,
14245 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
14246 vec_unsignede, vec_unsignedo): Add builtin defines.
14247 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
14249 * doc/extend.texi: Update the built-in documentation file for the
14250 new built-in functions.
14252 2017-06-29 Richard Biener <rguenther@suse.de>
14254 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
14255 reduction chains to LOOP_VINFO_REDUCTIONS.
14256 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
14257 SLP reductions after processing reduction chains.
14259 2017-06-29 Nathan Sidwell <nathan@acm.org>
14261 * builtins.c (fold_builtin_FUNCTION): Use
14262 lang_hooks.decl_printable_name.
14264 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
14266 PR middle-end/81194
14267 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
14268 with only one label.
14269 * stmt.c (expand_case): Assert NCASES is greater than one.
14271 2017-06-29 Richard Biener <rguenther@suse.de>
14273 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
14275 (group_case_labels): Likewise.
14276 (find_taken_edge): Push sanity checking on val to workers...
14277 (find_taken_edge_cond_expr): ... here
14278 (find_taken_edge_switch_expr): ... and here, handle cases
14279 with just a default label.
14280 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
14281 (group_case_labels): Likewise.
14282 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
14283 group_case_labels does anything cleanup the CFG again.
14285 2017-06-29 Bin Cheng <bin.cheng@arm.com>
14287 PR tree-optimization/81196
14288 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
14289 exit condition comparing two IVs.
14291 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
14293 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
14294 profile to the dummy entry at the end of the list of architectures.
14295 * config/arm/arm-cpu-cdata.h: Regenerated.
14297 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14298 Michael Collison <michael.collison@arm.com>
14301 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
14303 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
14304 (*aarch64_reg_<mode>3_minus_mask): New pattern.
14305 (*aarch64_<optab>_reg_di3_mask2): New pattern.
14306 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
14307 of shift when the shift amount is masked with constant equal to
14308 the size of the mode.
14309 * config/aarch64/predicates.md (subreg_lowpart_operator): New
14312 2017-06-29 Martin Liska <mliska@suse.cz>
14314 * config/i386/i386.opt: Change range from [1,5] to [0,5].
14316 2017-06-29 Yury Gribov <tetra2005@gmail.com>
14319 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
14322 (edge_growth_cache_entry::edge_growth_cache_entry): New
14324 (reset_edge_growth_cache): Update to use constructor.
14326 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14328 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
14329 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
14330 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
14332 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
14334 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
14335 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
14337 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
14339 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
14340 (*-linux-uclibc*): Add t-uclibc tmake_file.
14341 * config/t-musl: New.
14342 * config/t-uclibc: New.
14344 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
14346 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
14348 (gen_comm_data): Emit architectural setting of arch_prof.
14349 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
14351 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
14352 (armv8-m.base, armv8-m.main): Likewise.
14353 * arm-protos.h (arm_build_target): Add profile field.
14354 (arch_option): Likewise.
14355 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
14357 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
14358 arm_active_target.profile.
14360 2017-06-28 Richard Biener <rguenther@suse.de>
14362 PR middle-end/81227
14363 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
14364 TYPE_OVERFLOW_WRAPS.
14365 * match.pd (negate_expr_p): Likewise.
14366 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
14367 fold_build2, not fold_binary.
14369 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
14371 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
14372 Convert memory address to Pmode.
14373 (aarch64_print_operand): Assert MEM operands are always Pmode.
14375 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
14378 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
14379 Remove redundant if.
14380 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
14381 * config/arm/aarch-common-protos.h
14382 (aarch_forward_to_shift_is_not_shifted_re): Remove.
14383 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
14385 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
14388 * multiple_target.c (create_dispatcher_calls): Set the default
14389 clone to be static, not public.
14391 2017-06-28 Richard Biener <rguenther@suse.de>
14393 * tree-vect-loop.c (vectorizable_reduction): Move special
14394 cond reduction IV var creation ...
14395 (vect_create_epilog_for_reduction): ... here. Remove induction_index
14396 parameter. Use STMT_VINFO_VECTYPE.
14397 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
14400 2017-06-28 Martin Liska <mliska@suse.cz>
14403 * ipa-visibility.c (non_local_p): Handle visibility.
14405 2017-06-28 Martin Liska <mliska@suse.cz>
14408 * common.opt: Add IntegerRange to various options.
14409 * opt-functions.awk (integer_range_info): New function.
14410 * optc-gen.awk: Add integer_range_info to cl_options struct.
14411 * opts-common.c (decode_cmdline_option): Handle
14412 CL_ERR_INT_RANGE_ARG.
14413 (cmdline_handle_error): Likewise.
14414 * opts.c (print_filtered_help): Show valid interval in
14415 when --help is provided.
14416 * opts.h (struct cl_option): Add range_min and range_max fields.
14417 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
14419 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
14421 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
14422 (x * C EQ/NE y * C): New transformation.
14424 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
14426 * genmultilib (combination_space): Accept '+' in option names.
14428 2017-06-28 Martin Liska <mliska@suse.cz>
14431 * asan.c (instrument_derefs): Bail out inner references
14432 that are hard register variables.
14434 2017-06-28 Jakub Jelinek <jakub@redhat.com>
14437 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
14438 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
14440 2017-06-28 Richard Biener <rguenther@suse.de>
14442 * tree-vectorizer.h (vect_get_vec_defs): Remove.
14443 (vect_get_slp_defs): Adjust.
14444 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
14446 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
14448 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
14449 get_initial_defs_for_reduction instead of vect_get_vec_defs.
14450 (vectorizable_reduction): Adjust.
14451 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
14453 (vect_get_slp_defs): Likewise.
14454 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
14455 (vectorizable_bswap): Adjust.
14456 (vectorizable_call): Likewise.
14457 (vectorizable_conversion): Likewise.
14458 (vectorizable_assignment): Likewise.
14459 (vectorizable_shift): Likewise.
14460 (vectorizable_operation): Likewise.
14461 (vectorizable_store): Likewise.
14462 (vectorizable_condition): Likewise.
14463 (vectorizable_comparison): Likewise.
14465 2017-06-28 Michael Collison <michael.collison@arm.com>
14468 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
14470 (arm_gen_movmemqi): Removed unused variable 'i'.
14471 Convert 'for' loop into 'while' loop.
14472 (arm_expand_prologue): Remove last unnecessary set of insn.
14473 (thumb_pop): Remove unused variable 'pushed_words'.
14474 (thumb_exit): Remove last unnecessary set of regs_to_pop.
14476 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14478 * config/s390/predicates.md: Use s390_rel_address_ok_p.
14479 * config/s390/s390-protos.h: Add prototype of
14480 s390_rel_address_ok_p.
14481 * config/s390/s390.c (s390_got_symbol): New function.
14482 (s390_rel_address_ok_p): New function.
14483 (legitimize_pic_address): Use s390_rel_address_ok_p.
14484 (s390_load_got): Use s390_got_symbol.
14485 (s390_option_override): Issue error if
14486 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
14487 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
14489 * config/s390/s390.opt: New option mpic-data-is-text-relative.
14491 2017-06-27 Andrew Pinski <apinski@cavium.com>
14493 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
14494 (X * copysign (1.0, X)): New pattern.
14495 (X * copysign (1.0, -X)): New pattern.
14496 (copysign (-1.0, CST)): New pattern.
14498 2017-06-27 Joseph Myers <joseph@codesourcery.com>
14500 * genmultilib (combination_space): Remove variable.
14501 Validate reuse rules against regular expression for any sequence
14502 of multilib options in any order.
14504 2017-06-27 Michael Collison <michael.collison@arm.com>
14506 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
14507 call aarch64_split_simd_combine.
14508 * (aarch64_combine_internal<mode>): Delete pattern.
14509 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
14510 Allow register and subreg operands.
14512 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14514 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
14515 specific need, just fallback on defaults.
14516 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
14518 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14519 Olivier Hainque <hainque@adacore.com>
14521 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
14523 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
14524 targets. Pick a default if no particular attempt applied.
14525 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
14528 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14530 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
14531 (x86_64-wrs-vxworks7): Likewise.
14533 2017-06-27 Marek Polacek <polacek@redhat.com>
14536 * ubsan.c (instrument_null): Check get_base_address's result for null.
14538 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
14540 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
14542 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
14544 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
14545 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
14546 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
14547 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
14548 New function types.
14549 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
14550 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
14551 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
14552 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
14553 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
14554 BUILT_IN_FEUPDATEENV): New builtins.
14555 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
14556 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
14557 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
14558 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
14560 (builtin_structptr_types): Adjust size.
14561 * tree.c (builtin_structptr_types): Add four entries.
14563 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14564 Olivier Hainque <hainque@adacore.com>
14566 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
14567 (TLS_SYM): New local macro, forcing reference to __tls__ on
14568 link command lines for VxWorks 7 RTPs, triggering initialization
14570 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
14571 OS features TLS support, true for RTPs on VxWorks 7.
14572 * config/vxworks.c (vxworks_override_options): Setup emutls
14575 2017-06-27 Jakub Jelinek <jakub@redhat.com>
14577 * predict.c (test_prediction_value_range): Use -1U instead of -1
14578 to avoid narrowing conversion warning.
14579 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
14580 to avoid narrowing conversion warning.
14581 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
14583 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
14585 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14587 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
14588 64bit configurations.
14589 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
14590 (SIZE_TYPE): Likewise.
14591 * config/vxworks.c (vxworks_emutls_var_fields): Use
14592 long_unsigned_type_node instead of unsigned_type_node as the offset
14593 field type, which is "pointer" mode in emutls.c.
14595 2017-06-27 Jakub Jelinek <jakub@redhat.com>
14598 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
14600 PR middle-end/81207
14601 * gimple-fold.c (replace_call_with_call_and_fold): Handle
14602 gimple_vuse copying separately from gimple_vdef copying.
14604 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14606 * value-prof.c (free_hist): Remove call to memset and the enclosing if
14609 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
14610 Olivier Hainque <hainque@adacore.com>
14612 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
14613 for all vxworks7 targets.
14614 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
14615 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
14616 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
14617 variations for VX6/VX7 and 32/64bits later on in ...
14618 (VXWORKS_LIB_SPEC): Leverage new macros.
14619 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
14620 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
14622 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
14624 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
14625 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
14627 2017-06-26 Carl Love <cel@us.ibm.com>
14629 * config/rs6000/rs6000-c.c: Add support for built-in functions
14630 vector bool char vec_reve (vector bool char);
14631 vector signed char vec_reve (vector signed char);
14632 vector unsigned char vec_reve (vector unsigned char);
14633 vector bool int vec_reve (vector bool int);
14634 vector signed int vec_reve (vector signed int);
14635 vector unsigned int vec_reve (vector unsigned int);
14636 vector bool long long vec_reve (vector bool long long);
14637 vector signed long long vec_reve (vector signed long long);
14638 vector unsigned long long vec_reve (vector unsigned long long);
14639 vector bool short vec_reve (vector bool short);
14640 vector signed short vec_reve (vector signed short);
14641 vector double vec_reve (vector double);
14642 vector float vec_reve (vector float);
14643 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
14644 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
14645 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
14646 (altivec_vreve): New pattern.
14647 * config/rs6000/altivec.h (vec_reve): New define.
14648 * doc/extend.texi (vec_rev): Update the built-in documentation file
14649 for the new built-in functions.
14651 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14653 PR tree-optimization/71815
14654 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
14656 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
14658 (slsr_process_phi): Likewise.
14659 (replace_uncond_cands_and_profitable_phis): Don't replace a
14660 multiply candidate with a stride of 1 (copy or cast).
14661 (phi_incr_cost): Call uses_consumed_by_stmt rather than
14663 (lowest_cost_path): Likewise.
14664 (total_savings): Likewise.
14666 2017-06-26 Richard Biener <rguenther@suse.de>
14669 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
14670 Use def_builtin_pure for all gather builtins.
14672 2017-06-26 Richard Biener <rguenther@suse.de>
14674 PR tree-optimization/81203
14675 * tree-tailcall.c (find_tail_calls): Do not move stmts into
14676 non-dominating BBs.
14678 2017-06-26 Marek Polacek <polacek@redhat.com>
14681 * doc/invoke.texi: Document -Wmultistatement-macros.
14683 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
14685 * doc/sourcebuild.texi (ARM-specific attributes): Document new
14686 arm_neon_ok_no_float_abi effective target.
14688 2017-06-26 Richard Biener <rguenther@suse.de>
14690 PR tree-optimization/80928
14691 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
14692 (copy_bbs): Set BB_DUPLICATED flag early.
14693 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
14695 (execute_on_shrinking_pred): Likewise.
14696 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
14697 BB_DUPLICATED blocks.
14698 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
14699 iterate over all PHIs considering removal of *gsi.
14701 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
14703 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
14706 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14708 * config/rs6000/rs6000-string.c: (expand_block_clear,
14709 do_load_for_compare, select_block_compare_mode,
14710 compute_current_alignment, expand_block_compare,
14711 expand_strncmp_align_check, expand_strn_compare,
14712 expand_block_move, rs6000_output_load_multiple)
14713 Move functions related to string/block move/compare
14714 to a separate file.
14715 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
14716 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
14717 for this function which is now used in two files.
14718 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
14719 * config.gcc: Add rs6000-string.o to extra_objs for
14720 targets powerpc*-*-* and rs6000*-*-*.
14722 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
14725 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
14726 32-bit, since indexed is not valid for DImode.
14727 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
14728 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
14729 (define_peephole2 for Altivec d-form load): Add 32-bit support.
14730 (define_peephole2 for Altivec d-form store): Likewise.
14733 * multiple_target.c (create_dispatcher_calls): Only create the
14734 dispatcher call if the function is the default clone of a
14735 versioned function.
14737 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
14739 PR middle-end/80902
14740 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
14741 a call, force the call to not be a tail call.
14743 2017-06-23 Jeff Law <law@redhat.com>
14745 * doc/contrib.texi: Add entry for Steven Pemberton's work on
14748 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
14750 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
14751 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
14752 handling for early expansion of vector shifts (sl,sr,sra,rl).
14753 (builtin_function_type): Add vector shift right instructions
14754 to the unsigned argument list.
14756 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
14758 rtl-optimizatoin/79286
14759 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
14760 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
14761 trap. PIC register plus a const unspec without offset can never trap.
14763 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
14765 * tree.h (builtin_structptr_type): New type.
14766 (builtin_structptr_types): Declare new array.
14767 * tree.c (builtin_structptr_types): New array.
14768 (free_lang_data, build_common_tree_nodes): Use it.
14770 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
14773 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
14776 2017-06-22 Matt Turner <mattst88@gmail.com>
14778 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
14779 Lake models to skylake case. Assume skylake for unknown
14780 models with clflushopt.
14782 2017-06-22 Jeff Law <law@redhat.com>
14784 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
14785 frame sizes that do not satisfy aarch64_uimm12_shift.
14787 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
14789 * profile-count.h (apply_probability,
14790 apply_scale, probability_in): Fix checks for zero.
14792 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14794 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
14795 * doc/cppdiropts.texi (-I @var{dir}): Document it.
14797 2016-06-22 Richard Biener <rguenther@suse.de>
14799 * tree-vect-loop.c (vect_model_reduction_cost): Handle
14800 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
14801 REDUC_MAX_EXPR support.
14802 (vectorizable_reduction): Likewise.
14803 (vect_create_epilog_for_reduction): Likewise.
14805 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
14807 * match.pd (A / (1 << B) -> A >> B): New.
14808 * generic-match-head.c: Include optabs-tree.h.
14809 * gimple-match-head.c: Likewise.
14810 * optabs-tree.h (target_supports_op_p): New.
14811 * optabs-tree.c (target_supports_op_p): New.
14813 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14815 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
14816 $gcc_cv_ld --help output.
14817 (gcc_cv_ld_demangle): Likewise.
14818 (gcc_cv_ld_eh_frame_hdr): Likewise.
14819 (gcc_cv_ld_pie): Likewise.
14820 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
14821 (gcc_cv_ld_buildid): Likewise.
14822 (gcc_cv_ld_sysroot): Likewise.
14823 (ld_bndplt_support): Likewise.
14824 (ld_pushpopstate_support): Likewise.
14825 * configure: Regenerate.
14826 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
14828 2017-06-21 Jakub Jelinek <jakub@redhat.com>
14831 * config/i386/sse.md (round<mode>2): Renumber match_dup and
14832 operands indexes to avoid gap between operands and match_dups.
14834 2017-06-21 Andrew Pinski <apinski@cavium.com>
14836 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
14837 Increment Arith_shift and Arith_shift_reg by 1.
14838 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
14840 * config/aarch64/aarch64.c (thunderx_tunings): Enable
14841 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
14842 (aarch64_strip_extend): Add new argument and test for it.
14843 (aarch64_cheap_mult_shift_p): New function.
14844 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
14845 add a cost if it is true.
14846 Update calls to aarch64_strip_extend.
14847 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
14849 2017-06-21 Andrew Pinski <apinski@cavium.com>
14851 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
14853 (thunderxt88): Likewise.
14854 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
14855 (thunderx_prefetch_tune): New variable.
14856 (thunderx2t99_prefetch_tune): Update for the correct values.
14857 (thunderxt88_tunings): New variable.
14858 (thunderx_tunings): Use thunderx_prefetch_tune instead of
14859 generic_prefetch_tune.
14860 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
14862 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14864 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
14865 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
14866 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
14867 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
14868 (aarch64_atomic_cas<mode>, GPI): Likewise.
14870 2017-06-21 Martin Liska <mliska@suse.cz>
14872 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
14873 statements on cold and hot labels.
14874 * predict.c (tree_estimate_probability_bb): Remove the
14875 prediction from this place.
14877 2017-06-21 Martin Liska <mliska@suse.cz>
14879 PR tree-optimization/79489
14880 * gimplify.c (maybe_add_early_return_predict_stmt): New
14882 (gimplify_return_expr): Call the function.
14883 * predict.c (tree_estimate_probability_bb): Remove handling
14885 * predict.def: Update comment about early return predictor.
14886 * gimple-predict.h (is_gimple_predict): New function.
14887 * predict.def: Change default value of early return to 66.
14888 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
14890 * passes.def: Put pass_strip_predict_hints to the beginning of
14893 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
14895 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
14896 FUNCTION_DECL declarations.
14897 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
14899 (dwaf2out_decl): Likewise.
14900 * godump.c (go_early_global_decl): Skip call to the real debug hook
14901 for FUNCTION_DECL declarations.
14902 * passes.c (rest_of_decl_compilation): Skip call to the
14903 early_global_decl debug hook for FUNCTION_DECL declarations, unless
14904 -fdump-go-spec is passed.
14906 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
14908 * config/i386/i386.c (struct builtin_isa): New field pure_p.
14909 Reorder for compactness.
14910 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
14911 (def_builtin_pure, def_builtin_pure2): New functions.
14912 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
14914 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
14916 * match.pd (nop_convert): New predicate.
14917 ((A +- CST1) +- CST2): Allow some NOP conversions.
14919 2017-06-21 Jakub Jelinek <jakub@redhat.com>
14922 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
14923 with ctors/dtors if GOVD_SHARED is set.
14925 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14927 * config/aarch64/aarch64.md (movti_aarch64):
14928 Emit mov rather than orr.
14929 (movtf_aarch64): Likewise.
14930 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
14931 Emit mov rather than orr.
14933 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14935 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
14936 Swap alternatives, make integer dup more expensive.
14938 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14940 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
14941 Return true for non-tls symbols.
14943 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
14945 * config/aarch64/aarch64-cores.def (cortex-a55): New.
14946 (cortex-a75): Likewise.
14947 (cortex-a75.cortex-a55): Likewise.
14948 * config/aarch64/aarch64-tune.md: Regenerate.
14949 * doc/invoke.texi (-mtune): Document new values for -mtune.
14951 2017-06-21 Tom de Vries <tom@codesourcery.com>
14953 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
14954 stack_size feature.
14955 (Effective-Target Keywords, Other attributes): Suggest using
14956 dg-add-options stack_size feature to get stack limit in stack_size
14957 effective target documentation.
14959 2017-06-21 Julian Brown <julian@codesourcery.com>
14960 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14962 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
14963 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
14964 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
14966 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
14967 attribute type list for neon_multiply.
14968 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
14969 attribute type list for neon_multiply.
14970 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
14971 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
14972 attribute type list for neon_multiply.
14973 * config/arm/types.md (crypto_pmull): Add.
14974 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
14975 attribute type list.
14977 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
14979 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
14982 2017-06-20 Jakub Jelinek <jakub@redhat.com>
14984 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
14985 to make sure not to dereference a NULL cost_classes_ptr pointer.
14987 2017-06-20 Carl Love <cel@us.ibm.com>
14989 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14990 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14991 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14992 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14993 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14994 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
14995 VMULOSW): New enum "unspec" values.
14996 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
14997 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
14998 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
14999 altivec_vmulosw): New patterns.
15000 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15001 VMULOSW): Add definitions.
15003 2017-06-20 Julia Koval <julia.koval@intel.com>
15005 * config/i386/i386.c: Fix rounding expand for new pattern.
15006 * config/i386/subst.md: Fix pattern (parallel -> unspec).
15008 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
15010 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
15011 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
15013 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
15015 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
15018 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
15020 * config/aarch64/aarch64-cores.def: Rearrange to sort by
15021 architecture, then by implementer ID.
15022 * config/aarch64/aarch64-tune.md: Regenerate.
15024 2017-06-20 Richard Biener <rguenther@suse.de>
15026 PR middle-end/81097
15027 * fold-const.c (split_tree): Fold to type before negating.
15029 2017-06-20 David Malcolm <dmalcolm@redhat.com>
15031 * diagnostic-show-locus.c
15032 (selftest::test_fixit_deletion_affecting_newline): New function.
15033 (selftest::diagnostic_show_locus_c_tests): Call it.
15035 2017-06-20 Andreas Schwab <schwab@suse.de>
15038 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
15041 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15043 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
15044 __ARM_FEATURE_COPROC according to support.
15046 2017-06-20 Jakub Jelinek <jakub@redhat.com>
15048 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
15049 Rewritten to avoid overflow for > 32-bit pointers.
15052 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
15053 by removing enum keyword.
15054 (ubsan_type_descriptor): Likewise. Formatting fix.
15057 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
15058 splitter): Require TARGET_SSE2 in the condition.
15060 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
15063 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
15064 for doing vector set of SFmode on ISA 3.0.
15065 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
15066 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
15068 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
15069 SFmode value into a V4SF variable that was extracted from another
15070 V4SF variable without converting the element to double precision
15071 and back to single precision vector format.
15072 (vsx_insert_extract_v4sf_p9_2): Likewise.
15074 2017-06-19 Jakub Jelinek <jakub@redhat.com>
15076 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
15077 in UWHI to avoid undefined overflow.
15080 * ubsan.h (enum ubsan_encode_value_phase): New.
15081 (ubsan_encode_value): Change second argument to
15082 enum ubsan_encode_value_phase with default value of
15083 UBSAN_ENCODE_VALUE_GENERIC.
15084 * ubsan.c (ubsan_encode_value): Change second argument to
15085 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
15086 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
15087 create_tmp_var_raw instead of create_tmp_var and use a
15089 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
15090 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
15091 ubsan_encode_value callers.
15094 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
15095 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
15096 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
15098 2017-06-19 Richard Biener <rguenther@suse.de>
15100 PR middle-end/81118
15101 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
15102 estimates if we changed anything.
15104 2017-06-19 Richard Biener <rguenther@suse.de>
15106 PR tree-optimization/80887
15107 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
15108 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
15109 simplified lookups, then reset mprts_hook.
15110 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
15112 (try_to_simplify): Likewise.
15114 2017-06-19 Martin Liska <mliska@suse.cz>
15117 * gimplify.c (gimplify_switch_expr):
15118 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
15120 2017-06-19 Martin Liska <mliska@suse.cz>
15122 * doc/install.texi: Document that PGO runs in 4 stages.
15124 2017-06-19 Martin Liska <mliska@suse.cz>
15127 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
15128 to dispatcher function name.
15129 * multiple_target.c (replace_function_decl): New function.
15130 (create_dispatcher_calls): Redirect both edges and references.
15132 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
15134 * profile-count.c (profile_count::dump): Dump quality.
15135 (profile_count::differs_from_p): Update for unsigned val.
15136 * profile-count.h (profile_count_quality): New enum.
15137 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
15139 2017-06-19 Richard Biener <rguenther@suse.de>
15141 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
15142 struct function as arg.
15143 (estimate_numbers_of_iterations): Export overload with loop arg.
15144 (free_numbers_of_iterations_estimates_loop): Use an overload of
15145 free_numbers_of_iterations_estimates instead.
15146 * tree-cfg.c (remove_bb): Adjust.
15147 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
15148 * tree-parloops.c (gen_parallel_loop): Likewise.
15149 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
15151 (tree_unroll_loops_completely): Likewise.
15152 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
15153 Use an overload instead and export.
15154 (estimated_loop_iterations): Adjust.
15155 (max_loop_iterations): Likewise.
15156 (likely_max_loop_iterations): Likewise.
15157 (estimate_numbers_of_iterations): Take struct function as arg
15159 (loop_exits_before_overflow): Adjust.
15160 (free_numbers_of_iterations_estimates_loop): Use an overload.
15161 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
15162 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
15164 2017-06-19 Richard Biener <rguenther@suse.de>
15167 * ipa-prop.c (find_constructor_constant_at_offset): Handle
15168 RANGE_EXPR conservatively.
15170 2017-06-16 Carl Love <cel@us.ibm.com>
15172 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15173 definitions for vec_float, vec_float2, vec_floato,
15174 vec_floate built-ins.
15175 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
15176 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
15178 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
15179 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
15180 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
15181 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
15182 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
15183 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
15184 vec_floato): Add builtin defines.
15185 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
15186 Update the built-in documentation file for the new built-in
15189 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15191 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
15192 (mthumb): Mark as the negative of -marm.
15194 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15196 * doc/invoke.texi (ARM Options, -mcpu): Document supported
15198 (ARM Options, -mtune): Document that this accepts the same
15199 extension options as -mcpu.
15200 (ARM Options, -mfpu): Document addition of -mfpu=auto.
15202 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15204 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
15205 permitted extensions.
15207 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15209 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
15210 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
15211 (armv8-m.main): Add option +nodsp.
15212 * config/arm/arm-cpu-cdata.h: Regenerated.
15214 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15216 * config/arm/t-fuchsia: New file.
15217 * config.gcc (arm*-*-fuchsia*): Use it.
15219 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15221 * config/arm/t-symbian: Rewrite for new option infrastructure.
15223 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15225 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
15226 (MULTILIB_REQUIRED): Likewise.
15228 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15230 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
15231 (MULTILIB_RESUE): Likewise.
15232 (MULTILIB_MATCHES): Likewise.
15233 (MULTLIB_REQUIRED): Likewise.
15235 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15237 * config/arm/t-rtems: Rewrite for new option framework.
15239 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15241 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
15242 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
15243 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
15244 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
15245 * config/arm/t-multilib: ... here.
15246 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
15247 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
15248 armv7-a and armv8*-a when A-profile libraries have not been built.
15249 * config/arm/t-rmprofile: Rewrite.
15251 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15253 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
15254 with a backslash. Remove the backslash after substituting unescaped
15256 * doc/fragments.texi (MULTILIB_REUSE): Document it.
15258 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15260 * config.gcc: (arm*-*-*): When building a-profile libraries, force
15261 the driver to pass through the default setting of -mfloat-abi.
15262 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
15264 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
15265 (all_feat_combs): New rule.
15266 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
15268 * config/arm/t-aprofile: Rewrite.
15270 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15272 * config/arm/arm.h (FPUTYPE_AUTO): Define.
15273 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
15274 fpu is not specified by the user/command-line.
15275 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
15276 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
15277 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
15278 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
15279 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
15280 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
15282 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15284 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
15285 * config/arm/t-arm-elf: Rewritten.
15287 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15289 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
15290 have some floating-point instructions.
15291 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
15292 (TARGET_MAYBE_HARD_FLOAT): New macro.
15293 * config/arm/arm-builtins.c (arm_init_builtins): Use
15294 TARGET_MAYBE_HARD_FLOAT.
15295 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
15297 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15299 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
15300 (configargs.h): Include it.
15301 (arm_print_hint_for_fpu_option): New function.
15302 (arm_parse_fpu_option): New function.
15303 (candidate_extension): New class.
15304 (arm_canon_for_multilib): New function.
15305 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
15306 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
15307 (ARCH_CANONICAL_SPECS): New macro.
15308 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
15310 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15312 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
15313 are set after handling multilib fragments. Set target_cpu_default2
15316 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15318 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
15320 (arm*-*-*): Set target_cpu_default2 to a quoted string.
15321 * config/arm/parsecpu.awk (check_cpu): Validate any extension
15323 (check_arch): Likewise.
15324 * config/arm/arm.c (arm_configure_build_target): Handle
15325 TARGET_CPU_DEFAULT being a string constant. Scan any feature
15326 options in the default.
15328 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15330 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
15331 when an option is an alias of another.
15332 * config/arm/parsecpu.awk (optalias): New parser token.
15333 (gen_comm_data): Mark non-alias options as such. Emit entries
15334 for extension aliases.
15335 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
15336 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
15337 (armv6kz, armv6zk, armv6t2): Likewise.
15338 (armv7): Make vfpv3-d16 an alias.
15339 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
15341 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
15342 Sort in canonical order.
15343 (armv8-a): Sort in canonical order.
15344 (armv8.1-a, armv8.2-a): Likewise.
15345 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
15347 (cortex-a9): Sort in canonical order.
15348 * config/arm/arm.c (selftests.h): Include it.
15349 (arm_test_cpu_arch_data): New function.
15350 (arm_run_self_tests): New function.
15351 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
15352 (targetm): Move declaration to the end of the file.
15353 * arm-cpu-cdata.h: Regenerated.
15355 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15357 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
15358 call to target_mode_check describing the type of option passed.
15359 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
15360 (arm_target_thumb_only): Use arm_parse_arch_option_name or
15361 arm_parse_cpu_option_name to match parameters against list of
15363 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
15364 arm_arch_core_flags data structure.
15365 * config/arm/arm-cpu_cdata.h: Regenerated.
15367 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15369 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
15371 (arm_print_hint_for_cpu_option): Likewise.
15372 (arm_print_hint_for_arch_option): Likewise.
15373 (arm_parse_cpu_option_name): Likewise.
15374 (arm_parse_arch_option_name): Likewise.
15375 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
15376 of entries in the all_fpus list.
15377 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
15378 (arm_parse_cpu_option_name): Declare.
15379 (arm_parse_arch_option_name): Declare.
15380 (arm_parse_option_features): Declare.
15381 (arm_intialize_isa): Declare.
15382 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
15384 (gen_comm_data): ... here. Make definitions non-static.
15385 * config/arm/arm-cpu-data.h: Regenerated.
15386 * config/arm/arm-cpu-cdata.h: Regenerated.
15388 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15390 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
15391 (cpu_arch_extension): New structure.
15392 (cpu_arch_option, arch_option, cpu_option): New structures.
15393 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
15394 architecture types.
15395 (gen_data): Generate new format data tables.
15396 * config/arm/arm.c (cpu_tune): New structure.
15397 (cpu_option, processors): Delete.
15398 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
15399 (arm_print_hint_for_cpu_option): ... this and ...
15400 (arm_print_hint_for_arch_option): ... this.
15401 (arm_parse_arch_cpu_name): Delete. Replace with ...
15402 (arm_parse_cpu_option_name): ... this and ...
15403 (arm_parse_arch_option_name): ... this.
15404 (arm_unrecognized_feature): Change type of target parameter to
15406 (arm_parse_arch_cpu_features): Delete. Replace with ...
15407 (arm_parse_option_features): ... this.
15408 (arm_configure_build_target): Rework to use new configuration data
15410 (arm_print_tune_info): Rework for new configuration data tables.
15411 * config/arm/arm-cpu-data.h: Regenerated.
15412 * config/arm/arm-cpu.h: Regenerated.
15414 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15416 * Makefile.in (OBJS): Move sbitmap.o from here ...
15417 (OBJS-libcommon): ... to here.
15419 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15421 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
15422 (ISA_ALL_CRYPTO): New macro.
15423 (ISA_ALL_SIMD): New macro
15424 (ISA_ALL_FP): New macro.
15425 * config/arm/arm.c (fpu_bitlist): Update initializer.
15426 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
15428 (arm9e): Add fpu. Add option for nofp
15429 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
15430 (arm926ej-s, arm1026ej-s): Likewise.
15431 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
15432 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
15433 neon-fp16, neon-vfpv4, nofp and nosimd.
15434 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
15435 (cortex-a8): Add fpu. Add option for nofp.
15436 (cortex-a9): Add fpu. Add options for nosimd and nofp.
15437 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
15438 (cortex-r4f): Add fpu.
15439 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
15440 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
15442 (cortex-r8): Likewise.
15443 (cortex-m4): Add fpu. Add option for nofp.
15444 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
15445 (cortex-a17.cortex-a7): Likewise.
15446 (cortex-a32): Add fpu. Add options for crypto and nofp.
15447 (cortex-a35, cortex-a53): Likewise.
15448 (cortex-a57): Add fpu. Add option for crypto.
15449 (cortex-a72, cortex-a73): Likewise.
15450 (exynos-m1): Likewise.
15451 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
15452 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
15453 (cortex-m33): Add fpu. Add option for nofp.
15454 * config/arm/arm-cpu-cdata.h: Regenerated
15455 * config/arm/arm-cpu-data.h: Regenerated.
15457 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15459 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
15460 (armv5te, armv5tej): Likewise.
15461 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
15462 (armv7): Add options fp and vfpv3-d16.
15463 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
15464 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
15466 (armv7ve): Likewise.
15467 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
15468 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
15469 (armv8-a): Add nocrypto option.
15470 (armv8.1-a, armv8.2-a): Likewise.
15471 (armv8-m.main): add options fp, fp.dp and nofp.
15473 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15475 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
15477 (armv8-a+crc): Delete.
15478 (armv8.1-a): Add options simd, crypto and nofp.
15479 (armv8.2-a): Add options fp16, simd, crypto and nofp.
15480 (armv8.2-a+fp16): Delete.
15481 (armv8-m.main): Add option dsp.
15482 (armv8-m.main+dsp): Delete.
15483 (cortex-a8): Add fpu. Add nofp option.
15484 (cortex-a9): Add fpu. Add nofp and nosimd options.
15485 * config/arm/parsecpu.awk (gen_data): Generate option tables and
15486 link to main cpu and architecture data structures.
15487 (gen_comm_data): Only put isa attributes from the main architecture
15489 (option): New statement for architecture and CPU entries.
15490 * arm.c (struct cpu_option): New structure.
15491 (struct processors): Add entry for options.
15492 (arm_unrecognized_feature): New function.
15493 (arm_parse_arch_cpu_name): Ignore any characters after the first
15495 (arm_parse_arch_cpu_feature): New function.
15496 (arm_configure_build_target): Separate out any CPU and architecture
15497 features and parse separately. Don't error out if -mfpu=auto is
15498 used with only an architecture string.
15499 (arm_print_asm_arch_directives): New function.
15500 (arm_file_start): Call it.
15501 * config/arm/arm-cpu-cdata.h: Regenerated.
15502 * config/arm/arm-cpu-data.h: Likewise.
15503 * config/arm/arm-tables.opt: Likewise.
15505 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15507 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
15508 assembler when it is not -mfpu=auto.
15510 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15512 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
15513 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
15514 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
15515 (ASM_CPU_SPEC): Rewrite.
15516 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
15517 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
15518 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
15519 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
15520 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
15521 copied string is NUL-terminated. Also strip any characters prefixed
15523 (arm_rewrite_selected_arch): New function.
15524 (arm_rewrite_march): New function.
15526 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15528 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
15529 (x_arm_cpu_string, x_arm_tune_string): Likewise.
15530 (march, mcpu, mtune): Convert to string-based options.
15531 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
15532 (arm_parse_arch_cpu_name): New function.
15533 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
15534 identify selected architecture or CPU.
15535 (arm_option_save): New function.
15536 (TARGET_OPTION_SAVE): Redefine.
15537 (arm_option_restore): Restore string options.
15538 (arm_option_print): Print string options.
15540 2017-06-16 Martin Sebor <msebor@redhat.com>
15542 PR tree-optimization/80933
15543 PR tree-optimization/80934
15544 * builtins.c (fold_builtin_3): Do not handle bcmp here.
15545 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
15546 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
15547 (gimple_fold_builtin): Call them.
15549 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15551 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
15552 as unlikely; update profile.
15554 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15556 * predict.c (force_edge_cold): Handle declaring edges impossible
15559 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15561 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
15563 (try_unroll_loop_completely): Fix reporting.
15565 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15567 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
15569 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
15572 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
15573 if given a non-constant argument for an intrinsic which requires a
15576 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15578 * profile.c (compare_freqs): New function.
15579 (branch_prob): Sort edge list.
15580 (find_spanning_tree): Assume that the list is priority sorted.
15582 2017-06-16 Richard Biener <rguenther@suse.de>
15584 PR tree-optimization/81090
15585 * passes.def (pass_record_bounds): Remove.
15586 * tree-pass.h (make_pass_record_bounds): Likewise.
15587 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
15588 make_pass_record_bounds): Likewise.
15589 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
15590 not free niter estimates at the beginning but at the end.
15591 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
15593 2017-06-16 Richard Biener <rguenther@suse.de>
15595 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
15596 initializer to workaround ICE in host GCC 4.8.
15598 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15600 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
15602 (clone_inlined_nodes): Update.
15604 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15606 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
15607 prefetch settings, and enable prefetching by default at -O3.
15609 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15611 * config/aarch64/aarch64.c (aarch64_override_options_internal):
15612 Set flag_prefetch_loop_arrays according to tuning data.
15614 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15616 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
15617 New tune structure.
15618 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
15619 [Unrelated to main purpose of the patch] Place the pointer field last
15620 to enable type checking errors when tune structure are wrongly merged.
15621 * config/aarch64/aarch64.c (generic_prefetch_tune,)
15622 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
15623 (thunderx2t99_prefetch_tune): New tune constants.
15624 (tune_params *_tunings): Update all tunings (no functional change).
15625 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
15626 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
15627 from tunings structures.
15629 2017-06-16 Jakub Jelinek <jakub@redhat.com>
15632 * ubsan.c (instrument_null): Add T argument, use it instead
15633 of computing it based on IS_LHS.
15634 (instrument_object_size): Likewise.
15635 (pass_ubsan::execute): Adjust instrument_null and
15636 instrument_object_size callers to pass gimple_get_lhs or
15637 gimple_assign_rhs1 result to it. Use instrument_null instead of
15638 calling get_base_address and instrument_mem_ref. Handle
15639 aggregate call arguments for object-size sanitization.
15641 2017-06-16 Yury Gribov <tetra2005@gmail.com>
15643 PR tree-optimization/81089
15644 * tree-vrp.c (is_masked_range_test): Validate operands of
15647 2017-06-15 Martin Sebor <msebor@redhat.com>
15650 * dumpfile.c (dump_register): Avoid calling memset to initialize
15651 a class with a default ctor.
15652 * gcc.c (struct compiler): Remove const qualification.
15653 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
15654 * hash-table.h: Ditto.
15655 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
15657 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
15658 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
15660 * params.h (struct param_info): Make struct members non-const.
15661 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
15662 with default initialization.
15663 * vec.h (vec_copy_construct, vec_default_construct): New helper
15665 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
15666 with vec_copy_construct.
15667 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
15668 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
15669 * doc/invoke.texi (-Wclass-memaccess): Document.
15671 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15673 * emit-rtl.h (is_leaf): Update comment about local
15674 register allocator.
15676 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15679 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
15680 for a variable to have a section before checking if the section has a
15682 Set section to.persistent if persistent attribute is set.
15683 Warn if .persistent attribute is used on an automatic variable.
15685 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
15687 PR rtl-optimization/80474
15688 * reorg.c (update_block): Do not ignore instructions in a delay slot.
15690 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
15692 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
15695 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
15697 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
15698 (casesi): Emit bounds checking as RTL.
15699 (casesi_internal_mips16_<mode>): Remove bounds checking.
15701 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
15703 * config/xtensa/xtensa.c (xtensa_option_override): Append
15704 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
15705 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
15706 xtensa_doloop_hooks): Define unconditionally.
15707 (xtensa_reorg_loops): Only call reorg_loops in the presence of
15709 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
15710 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
15711 for it in xtensa_option_override.
15712 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
15713 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
15715 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
15717 * doc/cppopts.texi: Document '-' special value to -MF.
15719 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
15721 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
15722 (cortex_a53_fconst): Likewise.
15723 (cortex_a53_fpmul): Likewise.
15724 (cortex_a53_f_load_64): Likewise.
15725 (cortex_a53_f_load_many): Likewise.
15726 (cortex_a53_advsimd_alu): Likewise.
15727 (cortex_a53_advsimd_alu_q): Likewise.
15728 (cortex_a53_advsimd_mul): Likewise.
15729 (cortex_a53_advsimd_mul_q): Likewise.
15730 (fpmac bypass): Add new bypass for fpmac-fpmac case.
15731 Add missing fmul, r2f_cvt and fconst cases.
15733 2017-06-14 Richard Biener <rguenther@suse.de>
15735 PR middle-end/81088
15736 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
15738 (fold_binary_loc): When associating do not treat pre-existing
15739 TREE_OVERFLOW on literal constants as a reason to allow
15740 TREE_OVERFLOW on associated literal constants.
15742 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15744 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
15745 (MASK_FEATURES): New macro.
15746 * config/sparc/sparc.c (sparc_option_override): Remove the special
15747 handling of -mfpu and generalize it to all MASK_FEATURES switches.
15749 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15751 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
15752 a division of 0 if non-call exceptions are enabled.
15754 2017-06-14 Andrew Pinski <apinski@cavium.com>
15755 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15758 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
15759 Improve vector initialization code gen for only variable case.
15761 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15763 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
15765 2017-06-14 Richard Biener <rguenther@suse.de>
15767 PR tree-optimization/81083
15768 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
15771 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15773 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
15774 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
15775 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
15776 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
15777 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
15778 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
15780 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15782 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
15783 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
15785 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15787 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
15789 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15791 * config/rs6000/t-rtems: Don't handle SPE.
15793 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15795 * config/rs6000/t-linux: Don't handle SPE.
15797 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15799 * config/rs6000/eabispe.h: Delete file.
15801 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15803 * config/rs6000/t-spe: Delete file.
15805 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15807 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
15808 (rs6000_legitimate_offset_address_p): Return false for anything in
15809 V2SImode or V2SFmode.
15811 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15813 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
15814 except V2SF and V2SI. Rearrange the vector modes, and add comments.
15815 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
15817 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
15818 (rs6000_legitimate_offset_address_p): Ditto.
15819 (rs6000_emit_move): Ditto.
15820 (rs6000_init_builtins): Remove V4HI_type_node.
15822 2017-06-13 Martin Liska <mliska@suse.cz>
15825 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
15826 (gate_asan): Likewise.
15827 * asan.h (asan_no_sanitize_address_p): Remove the function.
15828 (sanitize_flags_p): New function.
15829 * builtins.def: Fix coding style.
15830 * common.opt: Use renamed enum value.
15831 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
15832 * doc/extend.texi: Document no_sanitize attribute.
15833 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
15834 to SANITIZE_UNDEFINED_NONDEFAULT.
15835 * gcc.c (sanitize_spec_function): Use the renamed enum value.
15836 * gimple-fold.c (optimize_atomic_compare_exchange_p):
15837 Use sanitize_flags_p.
15838 * gimplify.c (gimplify_function_tree): Likewise.
15839 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
15840 * opts.c (parse_no_sanitize_attribute): New function.
15841 (common_handle_option): Use renamed enum value.
15842 * opts.h (parse_no_sanitize_attribute): Declare.
15843 * tree.c (sanitize_flags_p): New function.
15844 * tree.h: Declared here.
15845 * tsan.c: Use sanitize_flags_p.
15846 * ubsan.c (ubsan_expand_null_ifn): Likewise.
15847 (instrument_mem_ref): Likewise.
15848 (instrument_bool_enum_load): Likewise.
15849 (do_ubsan_in_current_function): Remove the function.
15850 (pass_ubsan::execute): Use sanitize_flags_p.
15851 * ubsan.h: Remove do_ubsan_in_current_function
15852 * tree-cfg.c (print_no_sanitize_attr_value): New function.
15853 (dump_function_to_file): Use it here.
15855 2017-06-13 Martin Jambor <mjambor@suse.cz>
15857 PR tree-optimization/80803
15858 PR tree-optimization/81063
15859 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
15860 (propagate_subaccesses_across_link): Enqueue subtree whenever
15861 necessary instead of relying on the caller.
15863 2017-06-13 Martin Jambor <mjambor@suse.cz>
15865 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
15866 that have a first_link.
15867 (sort_and_splice_var_accesses): Do not check first_link before
15869 (subtree_mark_written_and_enqueue): Likewise.
15870 (propagate_all_subaccesses): Likewise and do not stop at first
15871 parent with a first_link.
15873 2017-06-13 Martin Jambor <mjambor@suse.cz>
15875 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
15878 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15880 * match.pd: New pattern.
15882 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15884 * tree-vrp.c (is_masked_range_test): New function.
15885 (register_edge_assert_for): Determine ranges for
15888 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15890 PR tree-optimization/67328
15891 * fold-const.c (maskable_range_p): New function.
15892 (build_range_check): Generate bittests if possible.
15894 2017-06-13 Martin Liska <mliska@suse.cz>
15896 * gimple-pretty-print.c (dump_probability): Add new argument.
15897 (dump_edge_probability): Dump both probability and count.
15898 (dump_gimple_label): Likewise.
15899 (dump_gimple_bb_header): Likewise.
15901 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
15904 * config/avr/avr-devices.c: Fix indentation.
15905 * config/avr/gen-avr-mmcu-specs.c: Dito.
15907 2017-06-13 Richard Biener <rguenther@suse.de>
15909 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
15910 instead get vector type from stmt_info.
15911 (vectorizable_reduction): Adjust. Remove dead code.
15913 2017-06-13 Richard Biener <rguenther@suse.de>
15915 PR middle-end/81065
15916 * fold-const.c (extract_muldiv_1): Remove bogus distribution
15917 case of C * (x * C2 + C3).
15918 (fold_addr_of_array_ref_difference): Properly fold index difference.
15920 2017-06-12 David S. Miller <davem@davemloft.net>
15923 * config/sparc/sparc.md (return expander): Emit frame blockage if
15924 function uses alloca.
15926 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
15928 * combine.c (make_field_assignment): Check len rather than the mode
15929 precision when calling force_to_mode.
15931 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
15933 Support multilibs and devices that see flash in RAM address range.
15936 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
15937 (avr_mcu_t) <flash_pm_offset>: New field.
15938 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
15939 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
15940 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
15941 (AVR_TINY_PM_OFFSET): Remove macro.
15942 * config/avr/avr.opt (-mshort-calls): New option.
15943 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
15944 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
15945 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
15946 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
15947 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
15948 instead of avr_arch->have_jmp_call.
15949 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
15950 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
15951 avr_arch->flash_pm_offset to define.
15952 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
15953 new field flash_pm_offset. Add entry for avrxmega3.
15954 (avr_texinfo): Add entry for avrxmega3.
15955 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
15956 attiny212, attiny214,
15957 attiny412, attiny414, attiny416, attiny417,
15958 attiny814, attiny816, attiny817,
15959 attiny1614, attiny1616, attiny1617,
15960 attiny3214, attiny3216, attiny3217.
15961 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
15962 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
15963 (avr_print_operand_address) [AVR_TINY]: Same.
15964 (avr_asm_init_sections) <readonly_data_section>: Only patch
15965 callback if avr_arch->flash_pm_offset = 0.
15966 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
15967 for rodata if avr_arch->flash_pm_offset != 0.
15968 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
15969 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
15970 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
15971 (m_options): Append opt_rcall.
15972 (m_dirnames): Append dir_rcall.
15973 * config/avr/t-multilib: Regenerate.
15975 * configure.ac [target=avr]: Check whether avrxmega3 default
15976 linker description file works as needed.
15977 * configure: Regenerate.
15978 * doc/avr-mmcu.texi: Regenerate.
15979 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
15980 <__AVR_ARCH__>: Document avrxmega3 and 103.
15981 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
15982 <__AVR_SHORT_CALLS__>: Document it.
15983 <__AVR_PM_BASE_ADDRESS__>: Document it.
15984 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
15985 (AVR Variable Attributes) <progmem>: Document this is
15986 not needed for avrxmega3.
15987 (AVR Named Address Spaces) <__flash>: Dito.
15989 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
15991 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
15993 2017-06-12 Doug Rupp <rupp@adacore.com>
15995 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
15996 Append vxworks-stdint.h to the tm_file list.
15997 * config/vxworks-stdint.h: New file.
15999 2017-06-12 Martin Liska <mliska@suse.cz>
16001 PR tree-optimization/81041
16002 * tree-profile.c (gimple_gen_ic_func_profiler):
16003 Create an extra BB in profile-generate
16004 (gimple_gen_time_profiler): Likewise.
16006 2017-06-12 Jakub Jelinek <jakub@redhat.com>
16008 PR tree-optimization/81003
16009 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
16010 (update_range_test): Use it instead of force_gimple_operand_gsi.
16012 2017-06-12 Richard Biener <rguenther@suse.de>
16014 PR tree-optimization/81053
16015 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
16016 with backedge value not defined in loop. Simplify def stmt
16019 2017-06-11 Tom de Vries <tom@codesourcery.com>
16022 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
16024 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
16025 nvptx_cannot_force_const_mem.
16027 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16029 * opts.c (finish_options): Move test for flag_split_stack after
16030 it has been initialized.
16032 2017-06-11 Jason Merrill <jason@redhat.com>
16034 * tree.h (id_equal): New.
16035 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
16036 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
16037 instead of strcmp of IDENTIFIER_POINTER.
16039 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16041 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
16042 (mark_all_inlined_calls_cdtor): Fix formating.
16043 (inline_transform): Rescale profile before inlining.
16045 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16047 * cgraph.h (cgraph_edge::clone): Update prototype.
16048 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
16049 (cgraph_node::create_clone): Update.
16050 (cgraph_node::create_version_clone): Update.
16051 * tree-inline.c (copy_bb): Update.
16052 (expand_call_inline): Update.
16054 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
16056 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
16057 factored out from ...
16058 (rs6000_emit_prologue): ... here.
16060 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
16062 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
16063 factored out from ...
16064 (rs6000_emit_prologue): ... here.
16066 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16068 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
16070 (handle_missing_profiles): Fix computation of tp_first_run.
16071 (counts_to_freqs): Do not touch freqs when count is 0.
16073 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16075 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
16078 2017-06-10 Tom de Vries <tom@codesourcery.com>
16080 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
16081 attributes): Document signal effective target.
16083 2017-06-10 Tom de Vries <tom@codesourcery.com>
16085 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
16086 Document effective target stack_size.
16088 2017-06-09 David Malcolm <dmalcolm@redhat.com>
16090 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
16091 to the edit_context if they can be auto-applied.
16093 2017-06-9 Ian Lance Taylor <iant@golang.org>
16095 * opts.c (finish_options): If -fsplit-stack, disable implicit
16096 -forder-blocks-and-partition.
16097 * doc/invoke.texi (Optimize Options): Document that when using
16098 -fsplit-stack -forder-blocks-and-partition is not implicitly
16101 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16103 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
16104 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
16105 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
16106 * builtins.def (abort, trap, unreachable): Declare cold.
16107 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
16108 * tree-core.h (ECF_COLD): New.
16109 * tree.c (set_call_expr_flags): Handle ECF_COLD.
16110 (build_common_builtin_nodes): Mark unreachable and abort as cold.
16112 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16114 * predict.c (unlikely_executed_stmt_p): Cleanup.
16116 2017-06-09 Richard Biener <rguenther@suse.de>
16118 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
16119 model if the ref is always written to.
16121 2017-06-09 Tamar Christina <tamar.christina@arm.com>
16123 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
16125 2017-06-09 Tamar Christina <tamar.christina@arm.com>
16127 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
16130 2017-06-09 Tom de Vries <tom@codesourcery.com>
16133 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
16134 "target cannot support label values" when encountering LABEL_REF.
16136 2017-06-09 Martin Liska <mliska@suse.cz>
16138 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
16139 (gimple_gen_ic_func_profiler): Emit direct comparison
16140 of __gcov_indirect_call_callee with NULL.
16141 (gimple_gen_time_profiler): Change probability from
16142 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
16144 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16146 * profile.c (edge_gcov_counts): Turn to pointer.
16147 (compute_branch_probabilities, compute_branch_probabilities): Update.
16148 (branch_prob): Do not clear edge_gcov_count.
16149 * profile.h (edge_gcov_counts): Turn to pointer.
16150 (edge_gcov_count): Update.
16152 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16154 * gimple.h (gimple_check_failed): Mark cold.
16156 2017-06-09 Richard Biener <rguenther@suse.de>
16158 PR tree-optimization/66623
16159 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
16160 refactor check_reduction into two parts, properly computing
16161 whether we have to check reduction validity for outer loop
16164 2017-06-09 Richard Biener <rguenther@suse.de>
16166 PR tree-optimization/79483
16167 * graphite-scop-detection.c (order): New global.
16168 (get_order): Compute bb to order mapping that satisfies code
16169 generation constraints.
16170 (cmp_pbbs): New helper.
16171 (build_scops): Start domwalk at entry block, sort generated
16174 2017-06-09 Richard Biener <rguenther@suse.de>
16176 PR middle-end/81007
16177 * ipa-polymorphic-call.c
16178 (ipa_polymorphic_call_context::restrict_to_inner_class):
16179 Skip FIELD_DECLs with error_mark_node type.
16180 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
16183 2017-06-09 Martin Liska <mliska@suse.cz>
16185 * predict.c (struct branch_predictor): New struct.
16186 (test_prediction_value_range): New test.
16187 (predict_c_tests): New function.
16188 * selftest-run-tests.c (selftest::run_tests): Run the function.
16189 * selftest.h: Declare new tests.
16191 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
16194 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
16195 gen_add3_insn did not fail.
16196 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
16197 r0, construct that number in a temporary reg and add that reg to r0.
16198 If asked to put the result in r0 as well, fail.
16200 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
16202 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16203 for early expansion of vec_eqv.
16205 2017-06-08 Jakub Jelinek <jakub@redhat.com>
16207 PR middle-end/81005
16208 * ubsan.c (instrument_null): Avoid pointless code temporary.
16209 (pass_ubsan::execute): Instrument aggregate arguments of calls.
16211 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
16215 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
16218 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
16221 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16223 * predict.c (unlikely_executed_edge_p): Move ahead.
16224 (probably_never_executed_edge_p): Use it.
16226 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
16228 PR middle-end/79988
16229 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
16230 gimple_call_builtin_p call.
16232 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16234 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
16235 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
16236 rtl_check_failed_type2, rtl_check_failed_code1,
16237 rtl_check_failed_code2, rtl_check_failed_code_mode,
16238 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
16239 rtvec_check_failed_bounds, rtl_check_failed_flag,
16240 _fatal_insn_not_found, _fatal_insn): Likewise.
16241 * tree.h (tree_contains_struct_check_failed,
16242 tree_check_failed, tree_not_check_failed,
16243 tree_class_check_failed, tree_range_check_failed,
16244 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
16245 tree_vec_elt_check_failed, phi_node_elt_check_failed,
16246 tree_operand_check_failed, omp_clause_check_failed,
16247 omp_clause_operand_check_failed, omp_clause_range_check_failed):
16250 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16252 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
16253 flag_branch_probabilities.
16254 * ipa-inline.c (edge_badness): Likewise.
16255 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
16256 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
16257 * predict.c (maybe_hot_frequency_p): Likewise.
16258 (probably_never_executed): Likewise.
16259 * sched-ebb.c (schedule_ebbs): Likewise.
16260 * sched-rgn.c (find_single_block_region): Likewise.
16261 * tracer.c (tail_duplicate): Likewise.
16263 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16265 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
16266 longer requires x_flag_profile_use.
16268 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16270 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
16271 instead of flag_reorder_blocks_and_partition.
16272 * dbxout.c (dbxout_function_end): Likewise.
16273 * dwarf2out.c (gen_subprogram_die): Likewise.
16274 * haifa-sched.c (sched_create_recovery_edges): Likewise.
16275 * hw-doloop.c (reorg_loops): Likewise.
16276 * varasm.c (assemble_start_function,
16277 assemble_end_function): Likewise.
16278 (decide_function_section): Do not check for
16279 flag_reorder_blocks_and_partition.
16281 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
16283 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
16285 (chkp_get_hard_register_fake_addr_expr): Ditto.
16286 (chkp_build_addr_expr): Add check for hard reg case.
16287 (chkp_parse_array_and_component_ref): Ditto.
16288 (chkp_find_bounds_1): Ditto.
16289 (chkp_process_stmt): Don't generate bounds store for
16292 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16294 * predict.c (maybe_hot_bb_p): Do not check profile status.
16295 (maybe_hot_edge_p): Likewise.
16296 (probably_never_executed): Check for zero counts even if profile
16298 (unlikely_executed_edge_p): New function.
16299 (unlikely_executed_stmt_p): New function.
16300 (unlikely_executed_bb_p): New function.
16301 (set_even_probabilities): Use unlikely predicates.
16302 (combine_predictions_for_bb): Likewise.
16303 (predict_paths_for_bb): Likewise.
16304 (predict_paths_leading_to_edge): Likewise.
16305 (determine_unlikely_bbs): New function.
16306 (estimate_bb_frequencies): Use it.
16307 (compute_function_frequency): Use zero counts even if profile is
16309 * profile-count.h: Fix typo.
16311 2017-08-08 Julia Koval <julia.koval@intel.com>
16313 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
16314 _mm512_mask_cvtsepi16_storeu_epi8,
16315 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
16316 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
16317 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
16318 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
16319 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
16320 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
16321 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
16322 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
16323 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
16324 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
16325 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
16326 __builtin_ia32_pmovuswb256mem_mask,
16327 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
16328 __builtin_ia32_pmovwb512mem_mask): New builtins.
16330 2017-08-08 Julia Koval <julia.koval@intel.com>
16332 PR target/73350,80862
16333 * config/i386/subst.md (round): Fix round pattern.
16334 * config/i386/i386.c (ix86_erase_embedded_rounding):
16335 Fix erasing rounding for the fixed pattern.
16337 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16339 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
16341 2017-06-08 Martin Liska <mliska@suse.cz>
16343 PR gcov-profile/80911
16344 * gcov.c (block_info::block_info): New constructor.
16346 2017-06-07 Carl Love <cel@us.ibm.com>
16348 * config/rs6000/rs6000-c: The return type of the following
16349 built-in functions was implemented as int not long long. Fix sign
16350 of return value for the unsigned version of vec_mulo and vec_mule.
16351 vector unsigned long long vec_bperm (vector unsigned long long,
16352 vector unsigned char)
16353 vector signed long long vec_mule (vector signed int,
16355 vector unsigned long long vec_mule (vector unsigned int,
16356 vector unsigned int)
16357 vector signed long long vec_mulo (vector signed int,
16359 vector unsigned long long vec_mulo (vector unsigned int,
16360 vector unsigned int)
16361 * doc/extend.texi: Fix the documentation for the built-in
16364 2017-06-07 Carl Love <cel@us.ibm.com>
16367 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
16370 2017-06-07 Carl Love <cel@us.ibm.com>
16372 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
16373 support, Generate doublehv for signed int/float for BE case only.
16375 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
16377 * doc/invoke.texi (mcx16): Rewrite.
16379 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16381 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
16382 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
16383 *mov<mode>_softfloat, and an anonymous splitter): Use
16384 nonimmediate_operand instead of rs6000_nonimmediate_operand.
16386 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16388 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
16390 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
16391 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
16392 (rs6000_debug_reg_global): Adjust.
16393 (rs6000_init_hard_regno_mode_ok): Adjust.
16394 (rs6000_dbx_register_number): Adjust.
16395 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
16396 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
16397 Remove SPE_ACC and SPEFSCR.
16398 (REG_ALLOC_ORDER): Ditto.
16399 (FRAME_POINTER_REGNUM): Change to 111.
16400 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
16401 (REG_CLASS_NAMES): Ditto.
16402 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
16403 (REGISTER_NAMES): Ditto.
16404 (ADDITIONAL_REG_NAMES): Ditto.
16405 (rs6000_reg_names): Ditto.
16406 * config/rs6000/rs6000.md: Renumber some register number
16409 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16411 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
16413 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
16414 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
16416 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
16417 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
16418 Delete the SPE high registers.
16419 (REG_ALLOC_ORDER): Ditto.
16420 (enum reg_class): Remove SPE_HIGH_REGS.
16421 (REG_CLASS_NAMES): Ditto.
16422 (REG_CLASS_CONTENTS): Delete the SPE high registers.
16423 (REGISTER_NAMES): Ditto.
16424 (rs6000_reg_names): Ditto.
16425 * doc/tm.texi.in: Remove SPE as example.
16426 * doc/tm.texi: Regenerate.
16428 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16430 * config/rs6000/8540.md (ppc8540_brinc): Delete.
16431 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
16432 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
16433 * config/rs6000/rs6000.md (type): Remove "brinc".
16435 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16437 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
16438 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
16439 * config/rs6000/linuxspe.h: Delete file.
16440 * config/rs6000/rs6000.md: Don't include spe.md.
16441 * config/rs6000/spe.h: Delete file.
16442 * config/rs6000/spe.md: Delete file.
16443 * config/rs6000/t-rs6000: Remove spe.md.
16445 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16447 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
16448 (reg_or_none500mem_operand): Delete.
16449 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
16450 instead of reg_or_none500mem_operand.
16452 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16454 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
16455 handling of SPE flags.
16456 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
16458 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16460 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
16462 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
16463 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
16464 paired_divv2sf3): Similar.
16465 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
16466 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
16467 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
16469 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
16470 Rename the paired_* instruction patterns.
16471 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
16473 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
16474 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
16475 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
16476 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
16477 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
16478 PAIRED_VECTOR_MODE.
16479 (struct machine_function): Delete field spe_insn_chain_scanned_p.
16480 (spe_func_has_64bit_regs_p): Delete.
16481 (spe_expand_predicate_builtin): Delete.
16482 (spe_expand_evsel_builtin): Delete.
16483 (TARGET_DWARF_REGISTER_SPAN): Do not define.
16484 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
16485 (invalid_e500_subreg): Delete.
16486 (rs6000_legitimize_address): Always force_reg op2 as well, for
16487 paired single memory accesses.
16488 (rs6000_member_type_forces_blk): Delete.
16489 (rs6000_spe_function_arg): Delete.
16490 (rs6000_expand_unop_builtin): Delete SPE handling.
16491 (rs6000_expand_binop_builtin): Ditto.
16492 (spe_expand_stv_builtin): Delete.
16493 (bdesc_2arg_spe): Delete.
16494 (spe_expand_builtin): Delete.
16495 (spe_expand_predicate_builtin): Delete.
16496 (spe_expand_evsel_builtin): Delete.
16497 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
16498 (spe_init_builtins): Delete.
16499 (spe_func_has_64bit_regs_p): Delete.
16500 (savres_routine_name): Delete "info" parameter. Adjust callers.
16501 (rs6000_emit_stack_reset): Ditto.
16502 (rs6000_dwarf_register_span): Delete.
16503 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
16504 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
16505 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
16507 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
16509 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
16510 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
16511 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
16512 mulv2sf3, divv2sf3): Delete expanders.
16514 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16516 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
16518 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16520 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
16521 * config/rs6000/rs6000.c: Ditto.
16523 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16525 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
16526 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
16527 comparison_operator.
16529 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16531 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
16532 * config/rs6000/rs6000.opt: Ditto.
16533 * config/rs6000/t-rtems: Ditto.
16535 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16537 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
16538 TARGET_E500_SINGLE by 0, simplify.
16539 * config/rs6000/rs6000.c: Ditto.
16540 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
16541 (spe_build_register_parallel): Delete.
16542 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
16543 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
16544 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
16545 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
16546 (E500_CONVERT): Delete.
16547 * config/rs6000/spe.md: Remove many patterns and all define_constants.
16549 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16551 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
16552 * config/rs6000/dfp.md: Ditto.
16553 (negdd2, *negdd2_fpr): Merge.
16554 (absdd2, *absdd2_fpr): Merge.
16555 (negtd2, *negtd2_fpr): Merge.
16556 (abstd2, *abstd2_fpr): Merge.
16557 * config/rs6000/e500.h: Delete file.
16558 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
16559 TARGET_FPRS by 1 and simplify.
16560 * config/rs6000/rs6000-c.c: Ditto.
16561 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
16562 TARGET_DF_SPE by 0.
16563 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
16565 * config/rs6000/rs6000.md: Ditto.
16566 (floatdidf2, *floatdidf2_fpr): Merge.
16567 (move_from_CR_gt_bit): Delete.
16568 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
16569 (E500_CR_IOR_COMPARE): Delete.
16570 (All patterns that require !TARGET_FPRS): Delete.
16571 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
16573 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16575 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
16577 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16579 * graphds.c (add_edge): Intitialize edge's attached data.
16580 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
16581 pointer parameter. Call pointed function on each edge during
16582 graph traversing. Skip traversing the edge when the function
16584 (graphds_dfs, graphds_scc): Ditto.
16585 (for_each_edge): New parameter. Pass the new parameter to callback
16587 * graphds.h (skip_edge_callback): New function pointer type.
16588 (graphds_dfs, graphds_scc): New function pointer parameter.
16589 (graphds_edge_callback, for_each_edge): New parameter.
16591 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16593 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
16594 out code checking if runtime alias check is possible to below ...
16595 Call the new function.
16596 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
16597 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
16599 2017-06-07 Marek Polacek <polacek@redhat.com>
16602 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
16603 TYPE_OVERFLOW_WRAPS check.
16605 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16607 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
16608 if versioning is required.
16609 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
16610 peeling with the check for versioning.
16612 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16614 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
16615 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
16616 Set true to new parameter if new ssa variable is defined.
16617 (vect_gen_vector_loop_niters): Refactor. Set range information
16618 for the new vector loop bound variable.
16619 (vect_do_peeling): Ditto.
16621 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16623 * tree-affine.c (ssa.h): Include header file.
16624 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
16625 has wrapping overflow behavior.
16627 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16629 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
16631 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16633 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
16634 (tree_to_aff_combination): ... here.
16636 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16638 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
16639 reg_pressure model function.
16640 (ivopts_global_cost_for_size): Delete.
16641 (determine_set_costs, iv_ca_recount_cost): Call new model function
16642 ivopts_estimate_reg_pressure.
16644 2017-06-07 Tamar Christina <tamar.christina@arm.com>
16646 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
16647 expensive than udiv. Remove floating point cases from mod.
16649 2017-06-07 Tamar Christina <tamar.christina@arm.com>
16651 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
16652 Increase idiv cost.
16654 2017-06-07 Tamar Christina <tamar.christina@arm.com>
16656 * config/aarch64/aarch64.md
16657 (copysignsf3): Fix mask generation.
16659 2017-06-07 Jakub Jelinek <jakub@redhat.com>
16661 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
16663 (class dump_manager): Add register_dumps method.
16664 * dumpfile.c: Include langhooks.h.
16665 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
16666 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
16667 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
16668 (dump_manager::dump_register): Start with 512 entries instead of 32.
16669 (dump_manager::register_dumps): New method.
16670 * toplev.c (general_init): Instead of invoking register_dumps
16671 langhook, invoke register_dumps method on the dump manager.
16672 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
16675 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
16677 * doc/md.texi: Clarify the restrictions on a define_insn condition.
16678 Say that # requires an associated define_split to exist, and that
16679 the define_split must be suitable for use after register allocation.
16681 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16683 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
16684 (compute_outgoing_frequencies): Also initialize zero counts.
16685 (find_many_sub_basic_blocks): Do not produce uninitialized profile
16686 around loops; preserve more of profile when nothing changes.
16688 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
16690 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
16692 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
16693 * config/arm/arm-cpu-cdata.h: Regenerate.
16694 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
16695 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
16696 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
16697 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
16698 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
16700 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
16702 * config/arm/t-rmprofile: Likewise.
16703 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
16705 2017-06-06 David S. Miller <davem@davemloft.net>
16708 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
16709 blockage if function uses alloca.
16711 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16713 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
16714 New "uid" fields to hold pretty-print IDs of group and ref.
16715 Memory references are now identified as <group_id>:<ref_id>
16716 instead of using [random] addresses.
16717 (dump_mem_details): Simplify, no functional change.
16718 (dump_mem_ref): Simplify and make output more concise.
16719 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
16720 (find_or_create_group): Initialize group uid.
16721 (record_ref): Initialize ref uid. Improve debug output.
16722 (prune_group_by_reuse, should_issue_prefetch_p,)
16723 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
16724 (mark_nontemporal_store, determine_loop_nest_reuse):
16725 Improve debug output.
16727 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16729 * dbgcnt.def (prefetch): New debug counter.
16730 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
16731 (schedule_prefetches): Stop issueing prefetches if debug counter
16734 2017-06-06 Tom de Vries <tom@codesourcery.com>
16736 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
16737 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
16739 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16741 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
16742 Use aarch64_reg_or_zero predicate for operand 4.
16743 (aarch64_compare_and_swap<mode> define_insn_and_split):
16744 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
16745 (aarch64_store_exclusive<mode>): Likewise for operand 2.
16747 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
16749 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
16750 (arm_compute_save_core_reg_mask): This.
16751 (thumb1_compute_save_reg_mask): Rename into ...
16752 (thumb1_compute_save_core_reg_mask): This.
16753 (arm_compute_save_reg0_reg12_mask): Adapt comment.
16754 (arm_compute_frame_layout): Likewise.
16756 2017-06-06 Richard Biener <rguenther@suse.de>
16758 PR tree-optimization/80974
16759 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
16760 keep or clear leaders SSA info.
16762 2017-06-06 Tom de Vries <tom@codesourcery.com>
16764 * config/nvptx/nvptx.c (split_mode_p): New function.
16765 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
16767 2017-06-06 Tom de Vries <tom@codesourcery.com>
16769 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
16771 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16774 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
16777 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16779 * shrink-wrap.c (handle_simple_exit): Update profile.
16780 (try_shrink_wrapping): Upate profile.
16782 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16784 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
16785 (tree_guess_outgoing_edge_probabilities): New.
16786 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
16787 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
16789 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16791 * ipa-split.c (split_function): Initialize return bb profile.
16793 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16795 * profile.c (compute_branch_probabilities): Also initialize
16796 EXIT_BLOCK profile.
16798 2017-06-06 Richard Biener <rguenther@suse.de>
16800 PR tree-optimization/80928
16801 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
16802 (vect_analyze_loop_operations): Properly guard analysis for
16804 (vect_transform_loop): Likewise.
16805 (vect_analyze_loop_2): Also reset SLP type on PHIs.
16806 (vect_model_induction_cost): Do not cost for pure SLP.
16807 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
16808 of induction in inner loop vectorization.
16809 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
16810 (vect_get_and_check_slp_defs): Handle vect_induction_def.
16811 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
16813 (vect_analyze_slp_cost_1): Cost induction.
16814 (vect_detect_hybrid_slp_stmts): Handle PHIs.
16815 (vect_get_slp_vect_defs): Likewise.
16816 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
16817 (vect_transform_stmt): Handle SLP reductions.
16818 * tree-vectorizer.h (vectorizable_induction): Adjust.
16820 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16822 * config/rs6000/rs6000.c (make_resolver_func): Update
16823 init_lowered_empty_function call.
16825 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
16827 * doc/invoke.texi: Document the -fprofile-abs-path option.
16828 * common.opt (fprofile-abs-path): New option.
16829 * gcov-io.h (gcov_write_filename): Declare.
16830 * gcov-io.c (gcov_write_filename): New function.
16831 * coverage.c (coverage_begin_function): Use gcov_write_filename.
16832 * profile.c (output_location): Likewise.
16834 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16836 * shring-wrap.c: Revert accidental commit.
16838 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
16840 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
16842 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16844 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
16846 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
16847 profile in callgraph edge.
16848 * profile-count.h (apply_probability): If THIS is 0, then result is 0
16849 (apply_scale): Likewise.
16850 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
16851 Also scale profile when inlining function with zero profile.
16852 (initialize_cfun): Update exit block profile even when it is zero.
16853 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
16854 when profile is read.
16856 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16858 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
16859 (CLONE_*): New constants to define the processors we can generate
16860 code for with the target_clone attribute.
16861 (rs6000_clone_map): New array to identify which clone processors
16862 the current program is running on.
16863 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
16864 target_clone attribute.
16865 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
16866 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
16867 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
16868 (cpu_expand_builtin): Add support for target_clone attribute.
16869 (rs6000_valid_attribute_p): Allow "default" attribute.
16870 (get_decl_name): New debug function to simplify printing the
16871 current function name in debugging statements.
16872 (rs6000_clone_priority): New functions to support the target_clone
16873 attribute, and be able to generate code to switch between ISA 2.05
16874 through ISA 3.0 (power6 through power9).
16875 (rs6000_compare_version_priority): Likewise.
16876 (rs6000_get_function_versions_dispatcher): Likewise.
16877 (make_resolver_func): Likewise.
16878 (add_condition_to_bb): Likewise.
16879 (dispatch_function_versions): Likewise.
16880 (rs6000_generate_version_dispatcher_body): Likewise.
16881 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
16882 (fusion_gpr_load_p): Fix a spacing issue.
16883 * doc/extend.texi (Common Function Attributes): Document that the
16884 PowerPC supports the target_clone attribute.
16886 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
16888 * config/arm/arm.h: explain F symbol found in description of ARM
16889 register allocation in its legend.
16891 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16893 * config/mips/frame-header-opt.c: Include profile-count.h.
16894 * config/riscv/riscv.c: Include profile-count.h
16896 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16898 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
16900 (sm_set_flag_if_changed): Add bbs field.
16901 (execute_sm_if_changed_flag_set): Pass BBS.
16902 (execute_sm): Update.
16904 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16906 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
16909 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16911 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
16912 (peephole2): New peephole2 to emit the above.
16913 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
16915 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16917 * config/aarch64/aarch64.c (define_peephole2 above
16918 *sub_<shift>_<mode>): New peephole.
16920 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16922 * config/i386/i386.c (make_resolver_func): Update.
16923 * Makefile.in: Add profile-count.h and profile-count.o
16924 * auto-profile.c (afdo_indirect_call): Update to new API.
16925 (afdo_set_bb_count): Update.
16926 (afdo_propagate_edge): Update.
16927 (afdo_propagate_circuit): Update.
16928 (afdo_calculate_branch_prob): Update.
16929 (afdo_annotate_cfg): Update.
16930 * basic-block.h: Include profile-count.h
16931 (struct edge_def): Turn count to profile_count.
16932 (struct basic_block_def): Likewie.
16933 (REG_BR_PROB_BASE): Move to profile-count.h
16934 (RDIV): Move to profile-count.h
16935 * bb-reorder.c (max_entry_count): Turn to profile_count.
16936 (find_traces): Update.
16937 (rotate_loop):Update.
16938 (connect_traces):Update.
16939 (sanitize_hot_paths):Update.
16940 * bt-load.c (migrate_btr_defs): Update.
16941 * cfg.c (RDIV): Remove.
16942 (init_flow): Use alloc_block.
16943 (alloc_block): Uninitialize count.
16944 (unchecked_make_edge): Uninitialize count.
16945 (check_bb_profile): Update.
16946 (dump_edge_info): Update.
16947 (dump_bb_info): Update.
16948 (update_bb_profile_for_threading): Update.
16949 (scale_bbs_frequencies_int): Update.
16950 (scale_bbs_frequencies_gcov_type): Update.
16951 (scale_bbs_frequencies_profile_count): New.
16952 * cfg.h (update_bb_profile_for_threading): Update.
16953 (scale_bbs_frequencies_profile_count): Declare.
16954 * cfgbuild.c (compute_outgoing_frequencies): Update.
16955 (find_many_sub_basic_blocks): Update.
16956 * cfgcleanup.c (try_forward_edges): Update.
16957 (try_crossjump_to_edge): Update.
16958 * cfgexpand.c (expand_gimple_tailcall): Update.
16959 (construct_exit_block): Update.
16960 * cfghooks.c (verify_flow_info): Update.
16961 (dump_bb_for_graph): Update.
16962 (split_edge): Update.
16963 (make_forwarder_block): Update.
16964 (duplicate_block): Update.
16965 (account_profile_record): Update.
16966 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
16967 (get_estimated_loop_iterations): Update.
16968 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
16969 (single_likely_exit): Update.
16970 * cfgloopmanip.c (scale_loop_profile): Update.
16972 (set_zero_probability): Update.
16973 (lv_adjust_loop_entry_edge): Update.
16974 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16975 (purge_dead_edges): Update.
16976 (rtl_account_profile_record): Update.
16977 * cgraph.c (cgraph_node::create): Uninitialize count.
16978 (symbol_table::create_edge): Uninitialize count.
16979 (cgraph_update_edges_for_call_stmt_node): Update.
16980 (cgraph_edge::dump_edge_flags): Update.
16981 (cgraph_node::dump): Update.
16982 (cgraph_edge::maybe_hot_p): Update.
16983 * cgraph.h: Include profile-count.h
16984 (create_clone), create_edge, create_indirect_edge): Update.
16985 (cgraph_node): Turn count to profile_count.
16986 (cgraph_edge0: Likewise.
16987 (make_speculative, clone): Update.
16988 (create_edge): Update.
16989 (init_lowered_empty_function): Update.
16990 * cgraphclones.c (cgraph_edge::clone): Update.
16991 (duplicate_thunk_for_node): Update.
16992 (cgraph_node::create_clone): Update.
16993 * cgraphunit.c (cgraph_node::analyze): Update.
16994 (cgraph_node::expand_thunk): Update.
16995 * final.c (dump_basic_block_info): Update.
16996 * gimple-streamer-in.c (input_bb): Update.
16997 * gimple-streamer-out.c (output_bb): Update.
16998 * graphite.c (print_global_statistics): Update.
16999 (print_graphite_scop_statistics): Update.
17000 * hsa-brig.c: Include basic-block.h.
17001 * hsa-dump.c: Include basic-block.h.
17002 * hsa-gen.c (T sum_slice): Update.
17003 (convert_switch_statements):Update.
17004 * hsa-regalloc.c: Include basic-block.h.
17005 * ipa-chkp.c (chkp_produce_thunks): Update.
17006 * ipa-cp.c (struct caller_statistics): Update.
17007 (init_caller_stats): Update.
17008 (gather_caller_stats): Update.
17009 (ipcp_cloning_candidate_p): Update.
17010 (good_cloning_opportunity_p): Update.
17011 (get_info_about_necessary_edges): Update.
17012 (dump_profile_updates): Update.
17013 (update_profiling_info): Update.
17014 (update_specialized_profile): Update.
17015 (perhaps_add_new_callers): Update.
17016 (decide_about_value): Update.
17017 (ipa_cp_c_finalize): Update.
17018 * ipa-devirt.c (struct odr_type_warn_count): Update.
17019 (struct decl_warn_count): Update.
17020 (struct final_warning_record): Update.
17021 (possible_polymorphic_call_targets): Update.
17022 (ipa_devirt): Update.
17023 * ipa-fnsummary.c (redirect_to_unreachable): Update.
17024 * ipa-icf.c (sem_function::merge): Update.
17025 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
17026 * ipa-inline.c (compute_uninlined_call_time): Update.
17027 (compute_inlined_call_time): Update.
17028 (want_inline_small_function_p): Update.
17029 (want_inline_self_recursive_call_p): Update.
17030 (edge_badness): Update.
17031 (lookup_recursive_calls): Update.
17032 (recursive_inlining): Update.
17033 (inline_small_functions): Update.
17034 (dump_overall_stats): Update.
17035 (dump_inline_stats): Update.
17036 * ipa-profile.c (ipa_profile_generate_summary): Update.
17037 (ipa_propagate_frequency): Update.
17038 (ipa_profile): Update.
17039 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
17040 * ipa-utils.c (ipa_merge_profiles): Update.
17041 * loop-doloop.c (doloop_modify): Update.
17042 * loop-unroll.c (report_unroll): Update.
17043 (unroll_loop_runtime_iterations): Update.
17044 * lto-cgraph.c (lto_output_edge): Update.
17045 (lto_output_node): Update.
17046 (input_node): Update.
17047 (input_edge): Update.
17048 (merge_profile_summaries): Update.
17049 * lto-streamer-in.c (input_cfg): Update.
17050 * lto-streamer-out.c (output_cfg): Update.
17051 * mcf.c (create_fixup_graph): Update.
17052 (adjust_cfg_counts): Update.
17053 (sum_edge_counts): Update.
17054 * modulo-sched.c (sms_schedule): Update.
17055 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
17056 * predict.c (maybe_hot_count_p): Update.
17057 (probably_never_executed): Update.
17058 (dump_prediction): Update.
17059 (combine_predictions_for_bb): Update.
17060 (propagate_freq): Update.
17061 (handle_missing_profiles): Update.
17062 (counts_to_freqs): Update.
17063 (rebuild_frequencies): Update.
17064 (force_edge_cold): Update.
17065 * predict.h: Include profile-count.h
17066 (maybe_hot_count_p, counts_to_freqs): UPdate.
17067 * print-rtl-function.c: Do not include cfg.h
17068 * print-rtl.c: Include basic-block.h
17069 * profile-count.c: New file.
17070 * profile-count.h: New file.
17071 * profile.c (is_edge_inconsistent): Update.
17072 (correct_negative_edge_counts): Update.
17073 (is_inconsistent): Update.
17074 (set_bb_counts): Update.
17075 (read_profile_edge_counts): Update.
17076 (compute_frequency_overlap): Update.
17077 (compute_branch_probabilities): Update; Initialize and deinitialize
17079 (branch_prob): Update.
17080 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
17081 (edge_gcov_count): New.
17082 (bb_gcov_count): New.
17083 * shrink-wrap.c (try_shrink_wrapping): Update.
17084 * tracer.c (better_p): Update.
17085 * trans-mem.c (expand_transaction): Update.
17086 (ipa_tm_insert_irr_call): Update.
17087 (ipa_tm_insert_gettmclone_call): Update.
17088 * tree-call-cdce.c: Update.
17089 * tree-cfg.c (gimple_duplicate_sese_region): Update.
17090 (gimple_duplicate_sese_tail): Update.
17091 (gimple_account_profile_record): Update.
17092 (execute_fixup_cfg): Update.
17093 * tree-inline.c (copy_bb): Update.
17094 (copy_edges_for_bb): Update.
17095 (initialize_cfun): Update.
17096 (freqs_to_counts): Update.
17097 (copy_cfg_body): Update.
17098 (expand_call_inline): Update.
17099 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
17100 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
17101 (try_unroll_loop_completely): Update.
17102 (try_peel_loop): Update.
17103 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17104 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
17105 * tree-ssa-loop-split.c (connect_loops): Update.
17106 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
17107 * tree-ssa-reassoc.c (branch_fixup): Update.
17108 * tree-ssa-tail-merge.c (replace_block_by): Update.
17109 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
17110 (compute_path_counts): Update.
17111 (update_profile): Update.
17112 (recompute_probabilities): Update.
17113 (update_joiner_offpath_counts): Update.
17114 (estimated_freqs_path): Update.
17115 (freqs_to_counts_path): Update.
17116 (clear_counts_path): Update.
17117 (ssa_fix_duplicate_block_edges): Update.
17118 (duplicate_thread_path): Update.
17119 * tree-switch-conversion.c (case_bit_test_cmp): Update.
17120 (struct switch_conv_info): Update.
17121 * tree-tailcall.c (decrease_profile): Update.
17122 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
17123 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17124 * value-prof.c (check_counter): Update.
17125 (gimple_divmod_fixed_value): Update.
17126 (gimple_mod_pow2): Update.
17127 (gimple_mod_subtract): Update.
17128 (gimple_ic_transform): Update.
17129 (gimple_stringop_fixed_value): Update.
17130 * value-prof.h (gimple_ic): Update.
17132 2017-06-02 Carl Love <cel@us.ibm.com>
17134 * config/rs6000/rs6000-c: Add support for built-in functions
17135 vector double vec_doublee (vector signed int);
17136 vector double vec_doublee (vector unsigned int);
17137 vector double vec_doublee (vector float);
17138 vector double vec_doubleh (vector signed int);
17139 vector double vec_doubleh (vector unsigned int);
17140 vector double vec_doubleh (vector float);
17141 vector double vec_doublel (vector signed int);
17142 vector double vec_doublel (vector unsigned int);
17143 vector double vec_doublel (vector float);
17144 vector double vec_doubleo (vector signed int);
17145 vector double vec_doubleo (vector unsigned int);
17146 vector double vec_doubleo (vector float);.
17147 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
17148 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
17150 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
17151 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
17152 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
17154 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
17155 vec_doublel, vec_doubleh.
17156 * doc/extend.texi: Update the built-in documentation file for the
17157 new built-in functions.
17159 2017-06-02 David Malcolm <dmalcolm@redhat.com>
17162 * ipa-inline-analysis.c (free_growth_caches): Set
17163 edge_removal_hook_holder to NULL after removing it.
17165 2017-06-02 Sudakshina Das <sudi.das@arm.com>
17167 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
17168 comparision with zero.
17170 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
17171 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
17172 for early expansion of vec_min and vec_max builtins.
17173 (builtin_function_type): Add min/max unsigned variants to those
17174 identified as having unsigned arguments.
17176 2017-06-02 Olivier Hainque <hainque@adacore.com>
17178 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
17180 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17182 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
17183 Use VALL_F16 iterator rather than VALL.
17185 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17187 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
17188 Emit CBNZ inside loop when doing a strong exchange and comparing
17189 against zero. Generate the CC flags after the loop.
17191 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
17193 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
17194 (dl_section_ref): New.
17195 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
17196 On AIX, append an expression to subtract the size of the
17197 section length to dl_section_ref.
17199 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
17201 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
17202 for early expansion of vector absolute builtins.
17204 2017-06-02 Richard Biener <rguenther@suse.de>
17206 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
17207 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
17209 2017-06-02 Richard Biener <rguenther@suse.de>
17211 PR tree-optimization/80948
17212 * tree-tailcall.c (find_tail_calls): Track stmts to move in
17213 stmt order as well.
17215 2017-06-02 Richard Biener <rguenther@suse.de>
17217 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
17219 * tree-vect-stmts.c (process_use): Do not mark backedge defs
17220 for inductions as relevant.
17222 2017-06-02 Richard Biener <rguenther@suse.de>
17224 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
17225 (vectorizable_induction): ... this. Remove dead code.
17227 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
17229 * builtins. (expand_builtin_alloca): Remove second parameter and
17230 infer its value from the first parameter instead.
17231 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
17233 2017-06-02 Jakub Jelinek <jakub@redhat.com>
17235 PR rtl-optimization/80903
17236 * loop-doloop.c (add_test): Unshare sequence.
17238 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
17240 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
17242 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
17244 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
17246 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
17247 xlogue_layout::get_instance, logue_layout::xlogue_layout,
17248 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
17249 (xlogue_layout::get_stub_rtx): Make static.
17250 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
17251 (xlogue_layout::compute_stub_managed_regs): Rename to...
17252 (xlogue_layout::count_stub_managed_regs): ...this.
17253 (xlogue_layout::is_stub_managed_reg): New function.
17254 (xlogue_layout::m_stub_names): Rename to...
17255 (xlogue_layout::s_stub_names): ...this, make static.
17256 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
17257 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
17258 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
17259 xlogue_layout::s_stub_names): Instantiate statics.
17260 (stub_managed_regs): Remove.
17261 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
17262 (disable_call_ms2sysv_xlogues): Rename to...
17263 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
17264 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
17266 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
17267 change after reload_completed.
17268 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
17270 (ix86_expand_prologue): Likewise.
17271 (ix86_expand_epilogue): Likewise.
17272 (ix86_expand_split_stack_prologue): Likewise.
17273 (ix86_compute_frame_layout): Remove frame parameter ...
17274 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
17275 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
17277 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
17278 (ix86_frame): Move from here ...
17279 * config/i386/i386.h (ix86_frame): ... to here.
17280 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
17281 complete ix86_frame data structure instead. Remove some_ld_name.
17283 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
17285 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
17286 symbols that hold a DECL_VALUE_EXPR.
17288 2017-06-01 Martin Jambor <mjambor@suse.cz>
17290 PR tree-optimization/80898
17291 * tree-sra.c (process_subtree_disqualification): Removed.
17292 (disqualify_candidate): Do not acll
17293 process_subtree_disqualification.
17294 (subtree_mark_written_and_enqueue): New function.
17295 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
17296 RHS has been disqualified and re-queue LHS if necessary. Apart
17297 from that, ignore disqualified RHS.
17299 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17301 * config/s390/s390.c (s390_emit_epilogue): Disable early return
17302 address fetch for z10 or later.
17304 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17306 * config/arc/arc.md (tst_movb): Add guard when splitting.
17308 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17310 * config/arc/arc.c (arc_can_eliminate): Test against
17311 arc_frame_pointer_needed.
17313 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17315 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
17316 to prevent store reordering.
17317 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
17318 (type): Add block type.
17319 (stack_tie): Define special instruction to be used in
17322 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17324 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
17325 constraint. It is not valid for the pattern.
17326 (noncommutative_binary_comparison): Likewise.
17328 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17330 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
17333 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17335 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
17336 be used by the reg-alloc.
17338 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17340 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
17341 reg-alloc when having mul64 or mul32x16 instructions.
17342 (mulsidi3): Likewise.
17343 (umulsidi3): Likewise.
17344 (mulsi32x16): New pattern.
17345 (mulsi64): Likewise.
17346 (mulsidi64): Likewise.
17347 (umulsidi64): Likewise.
17348 (MUL32x16_REG): Define.
17349 (mul64_600): Use MUL32x16_REG.
17350 (mac64_600): Likewise.
17351 (umul64_600): Likewise.
17352 (umac64_600): Likewise.
17354 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17356 * config/arc/arc.md (mulsi3_700): Make it commutative.
17358 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
17360 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
17362 (*sign_extendsidi2_insn): Likewise for movstosw.
17364 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
17366 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
17367 the type of the input discriminant value. Convert the
17368 discriminant value of signedness vary.
17370 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
17372 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
17373 Add to -Wall section.
17375 2017-06-01 Richard Biener <rguenther@suse.de>
17377 PR middle-end/66313
17378 * fold-const.c (fold_plusminus_mult_expr): If the factored
17379 factor may be zero use a wrapping type for the inner operation.
17380 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
17381 and handle moved defs.
17382 (process_assignment): Properly guard the unary op case. Return a
17383 tri-state indicating that moving the stmt before the call may allow
17384 to continue. Pass through to_move.
17385 (find_tail_calls): Handle moving unrelated defs before
17388 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
17391 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
17392 splitter result in the canonical way.
17394 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
17396 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
17397 also for 32bit target. Update insn attributes.
17398 (zero-extendsidi2 splitter): Allow all registers for operand 1.
17400 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
17402 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
17403 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
17404 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
17405 (_mm_maskz_min_ss): New intrinsics.
17407 2017-05-31 Martin Liska <mliska@suse.cz>
17409 * tree-vect-loop.c (vect_create_epilog_for_reduction):
17410 Change comment style to one we normally use.
17411 (vectorizable_reduction): Likewise.
17412 (vectorizable_induction): Likewise.
17413 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
17414 (vectorizable_call): Likewise.
17415 (vectorizable_simd_clone_call): Likewise.
17416 (vectorizable_conversion): Likewise.
17417 (vectorizable_assignment): Likewise.
17418 (vectorizable_shift): Likewise.
17419 (vectorizable_operation): Likewise.
17420 (vectorizable_store): Likewise.
17421 (vectorizable_load): Likewise.
17422 * tree-vectorizer.h: Likewise.
17424 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
17426 * passes.c (emergency_dump_function): New.
17427 * tree-pass.h (emergency_dump_function): Declare.
17428 * plugin.c (plugins_internal_error_function): Remove.
17429 * plugin.h (plugins_internal_error_function): Remove declaration.
17430 * toplev.c (internal_error_function): New static function. Use it...
17431 (general_init): ...here.
17433 2017-05-31 Graham Markall <graham.markall@embecosm.com>
17435 * config/arc/arc.c (arc_print_operand): Handle constant operands.
17436 (arc_rtx_costs): Add costs for new patterns.
17437 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
17438 * config/arc/predicates.md: Add _1_2_3_operand predicate.
17440 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
17442 * tree-ssa-strlen.c (get_next_strinfo): New function.
17443 (get_stridx_plus_constant): Use it.
17444 (zero_length_string): Likewise.
17445 (adjust_related_strinfos): Likewise.
17446 (adjust_last_stmt): Likewise.
17448 2017-05-31 Richard Biener <rguenther@suse.de>
17451 * config/i386/i386.c (ix86_expand_builtin): Remove assert
17452 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
17454 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
17456 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
17457 loop_vinfo argument and use of dependence distance vectors.
17458 Check instead whether the two references differ only in their
17459 initial value and assume that they have the same alignment if the
17460 difference is a multiple of the vector alignment.
17461 (vect_analyze_data_refs_alignment): Update call accordingly.
17463 2017-05-31 Martin Liska <mliska@suse.cz>
17466 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
17468 2017-05-31 Bin Cheng <bin.cheng@arm.com>
17470 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
17471 (create_intersect_range_checks): Move from ...
17472 * tree-data-ref.c (create_intersect_range_checks_index)
17473 (create_intersect_range_checks): ... to here.
17474 (create_runtime_alias_checks): New function factored from ...
17475 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
17476 here. Call above function.
17477 * tree-data-ref.h (create_runtime_alias_checks): New function.
17479 2017-05-31 Bin Cheng <bin.cheng@arm.com>
17481 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
17482 segment length for dr_b and compute it in wide_int.
17484 2017-05-31 Richard Biener <rguenther@suse.de>
17486 PR tree-optimization/80906
17487 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
17488 and pass through iv_map.
17489 (copy_bb_and_scalar_dependences): Adjust.
17490 (translate_pending_phi_nodes): Likewise.
17491 (copy_loop_close_phi_args): Handle code-generating IVs instead
17494 2017-05-30 David Malcolm <dmalcolm@redhat.com>
17496 * diagnostic-color.c (color_dict): Add "type-diff".
17497 (parse_gcc_colors): Update comment.
17498 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
17499 -fdiagnostics-show-template-tree and -fno-elide-type.
17500 (GCC_COLORS): Add type-diff to example.
17502 (-fdiagnostics-show-template-tree): New.
17503 (-fno-elide-type): New.
17504 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
17505 the pp_format_decoder callback. Call any m_format_postprocessor's
17507 (pretty_printer::pretty_printer): Initialize
17508 m_format_postprocessor.
17509 (pretty_printer::~pretty_printer): Delete any
17510 m_format_postprocessor.
17511 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
17512 (class format_postprocessor): New class.
17513 (struct pretty_printer::format_decoder): Document the new parameters.
17514 (struct pretty_printer::m_format_postprocessor): New field.
17515 * tree-diagnostic.c (default_tree_printer): Update for new
17516 bool and const char ** params.
17517 * tree-diagnostic.h (default_tree_printer): Likewise.
17519 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
17521 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
17522 (lwa_operand): Delete rs6000_gen_cell_microcode test.
17523 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
17524 rs6000_gen_cell_microcode code.
17525 (rs6000_final_prescan_insn): Delete.
17526 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
17527 "warn-cell-microcode" entries.
17528 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
17529 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
17530 throughout. Change cc_reg_not_micro_cr0_operand to
17531 cc_reg_not_cr0_operand throughout.
17532 (*extendhi<mode>2_noload): Delete.
17533 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
17534 (mwarn-cell-microcode): Delete.
17535 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
17536 -mgen-cell-microcode and -mwarn-cell-microcode.
17538 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
17541 * config/i386/constraints.md (Yd): New constraint.
17543 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
17544 and (?Yd, r) alternatives. Update insn attributes.
17545 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
17546 and (?*Yd, r) alternatives. Update insn attributes.
17547 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
17549 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
17551 * gimplify.c (gimplify_modify_expr): Don't create a
17552 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
17555 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
17557 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
17559 2017-05-30 Richard Biener <rguenther@suse.de>
17561 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
17562 and reduc_def fields.
17563 (STMT_VINFO_REDUC_TYPE): New define.
17564 (STMT_VINFO_REDUC_DEF): Likewise.
17565 (vect_force_simple_reduction): Adjust prototype.
17566 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
17567 (vect_is_simple_reduction): Remove check_reduction argument.
17568 (vect_force_simple_reduction): Adjust and set
17569 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
17570 (vectorizable_reduction): Do not re-do reduction analysis
17571 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
17572 * tree-parloops.c (gather_scalar_reductions): Adjust.
17574 2017-05-30 Richard Biener <rguenther@suse.de>
17576 PR middle-end/80901
17577 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
17580 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
17582 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
17583 Introduce unknown_misalignment parameter and remove vf.
17584 (vect_peeling_hash_get_lowest_cost):
17585 Pass unknown_misalignment parameter.
17586 (vect_enhance_data_refs_alignment):
17587 Fix unsupportable data ref treatment.
17589 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17591 * tree-vect-data-refs.c (vect_get_data_access_cost):
17592 Workaround for SLP handling.
17593 (vect_enhance_data_refs_alignment):
17594 Compute costs for doing no peeling at all, compare to the best
17595 peeling costs so far and avoid peeling if cheaper.
17597 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17599 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
17600 Return peeling info and set costs to zero for unlimited cost
17602 (vect_enhance_data_refs_alignment): Also inspect all datarefs
17603 with unknown misalignment. Compute and costs for unknown
17604 misalignment, compare them to the costs for known misalignment
17605 and choose the cheapest for peeling.
17607 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17609 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
17610 (vect_get_peeling_costs_all_drs): Create function.
17611 (vect_peeling_hash_get_lowest_cost):
17612 Use vect_get_peeling_costs_all_drs.
17613 (vect_peeling_supportable): Create function.
17614 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
17616 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17618 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
17619 DR_HAS_NEGATIVE_STEP.
17620 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
17621 (vect_enhance_data_refs_alignment): Use.
17622 (vect_duplicate_ssa_name_ptr_info): Use.
17623 * tree-vectorizer.h (dr_misalignment): Use.
17624 (known_alignment_for_access_p): Use.
17626 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17629 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
17631 (has_section_name): New function.
17633 2017-05-30 Martin Liska <mliska@suse.cz>
17636 * auto-profile.c (get_function_decl_from_block): Fix
17639 2017-05-30 Richard Biener <rguenther@suse.de>
17641 PR middle-end/80876
17642 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
17644 2017-05-30 Martin Liska <mliska@suse.cz>
17646 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
17647 * dumpfile.h (struct dump_file_info): Remove ctors.
17649 2017-05-30 Martin Liska <mliska@suse.cz>
17651 * predict.def: Fix GNU coding style.
17653 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
17655 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
17656 Mark 'to' argument with ATTRIBUTE_UNUSED.
17658 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
17660 * config/xtensa/xtensa.c (xtensa_emit_call): Use
17661 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
17662 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
17665 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
17667 * doc/install.texi (Options specification): Restore entry of
17668 --enable-sjlj-exceptions.
17670 2017-05-27 Michael Eager <eager@eagercon.com>
17673 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
17675 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
17677 * config/microblaze/microblaze.h
17678 (FIXED_REGISTERS): Update in macro.
17679 (CALL_USED_REGISTERS): Update in macro.
17681 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
17683 * doc/install.texi: Add links to macOS binary distributions.
17685 2017-05-27 Jakub Jelinek <jakub@redhat.com>
17689 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17691 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
17693 2017-05-26 Martin Liska <mliska@suse.cz>
17695 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
17697 2017-05-26 Martin Liska <mliska@suse.cz>
17699 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
17700 always leading ';; '.
17701 (dump_bb_info): Likewise.
17702 (brief_dump_cfg): Likewise.
17703 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
17704 * dumpfile.c: Remove usage of TDF_VERBOSE.
17705 * dumpfile.h (enum dump_kind): Likewise.
17706 (dump_gimple_bb_header): Do not use TDF_COMMENT.
17707 * print-tree.c (debug_verbose): Remove.
17708 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
17709 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
17710 * tree-diagnostic.c (default_tree_printer): Replace
17711 TDF_DIAGNOSTIC with TDF_SLIM.
17713 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17715 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
17716 in parameter loop, rather than loop_vinfo.
17717 (create_intersect_range_checks): Ditto.
17718 (vect_create_cond_for_alias_checks): Update call to above functions.
17720 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17722 PR tree-optimization/80815
17723 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
17724 for merging runtime alias checks. Handle negative DR_STEPs.
17726 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17728 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
17730 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
17731 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
17732 out code pruning runtime alias checks.
17733 * tree-data-ref.c (prune_runtime_alias_test_list): New function
17734 factored out from above.
17735 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
17737 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
17739 (prune_runtime_alias_test_list): New decalaration.
17741 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17743 * tree-vect-data-refs.c (compare_tree): Rename and move ...
17744 * tree-data-ref.c (data_ref_compare_tree): ... to here.
17745 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
17746 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
17747 (operator==, comp_dr_with_seg_len_pair): Ditto.
17748 (vect_prune_runtime_alias_test_list): Ditto.
17750 2017-05-26 Martin Liska <mliska@suse.cz>
17753 * params.def: Bound partial-inlining-entry-probability param.
17755 2017-05-26 Marek Polacek <polacek@redhat.com>
17758 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
17761 2017-05-26 Richard Biener <rguenther@suse.de>
17763 PR tree-optimization/80842
17764 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
17767 2017-05-26 Richard Biener <rguenther@suse.de>
17769 PR tree-optimization/80844
17770 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
17772 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
17774 * doc/md.texi (Machine Constraints): Update x86 family
17775 machine constraints section to match 'config/i386/constraints.md'.
17777 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
17779 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
17781 2017-05-25 Nathan Sidwell <nathan@acm.org>
17783 * doc/invoke.texi (--enable-languages): Update documentation.
17785 2017-05-25 Martin Liska <mliska@suse.cz>
17787 * dumpfile.c: Add TDF_FOLDING.
17788 * dumpfile.h (enum dump_kind): Likewise.
17789 * genmatch.c (dt_simplify::gen_1): Use it.
17791 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17793 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
17795 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17797 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
17798 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
17800 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17802 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
17803 * match.pd (X == C): Rewrite it here.
17804 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
17805 with_certain_nonzero_bits2): New predicates.
17806 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
17808 2017-05-24 Nathan Sidwell <nathan@acm.org>
17810 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
17813 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
17816 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
17818 * config/powerpcspe: New port. Files are copied from the rs6000
17819 port, with "rs6000" in filenames replaced by "powerpcspe".
17821 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
17823 PR rtl-optimization/80754
17824 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
17826 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
17828 * config/sparc/sparc.md (length): Return the correct value for -mflat
17829 sibcalls to match output_sibcall.
17831 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
17835 * config/rs6000/rs6000.c (struct machine_function): Add new field
17837 (rs6000_get_separate_components): Init that field, use it.
17838 (rs6000_components_for_bb): Use the field.
17840 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
17842 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
17844 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
17846 PR middle-end/80823
17847 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
17849 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17852 * config/s390/s390.c (s390_check_qrst_address): Check incoming
17853 address against address_operand predicate.
17854 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
17856 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
17858 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
17859 parameters passed indirectly.
17861 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
17863 * config/i386/i386.md (*movdi_internal): Remove SSE4
17864 alternative 18 (?r, *v). Update insn attributes.
17865 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
17866 Update insn attributes.
17867 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
17868 Update insn attributes.
17869 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
17870 alternative 1 (r, v). Remove isa attribute.
17871 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
17872 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
17873 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
17875 2017-05-23 Tom de Vries <tom@codesourcery.com>
17877 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
17880 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17882 * cgraphunit.c (symbol_table::process_new_functions): Update.
17883 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
17884 (inline_generate_summary): Rename to ...
17885 (ipa_fn_summary_generate): ... this one.
17886 (inline_read_summary): Rename to ...
17887 (ipa_fn_summary_read): ... this one.
17888 (inline_write_summary): Rename to ...
17889 (ipa_fn_summary_write): ... this one.
17890 (inline_free_summary): Rename to ...
17891 (ipa_free_fn_summary): ... this one.
17892 (pass_data_local_fn_summary, pass_local_fn_summary,
17893 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
17894 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
17895 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
17896 make_pass_ipa_fn_summary): New.
17897 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
17898 inline_write_summary, inline_free_summary): Remove.
17899 (ipa_free_fn_summary) : New.
17900 * ipa-inline.c (ipa_inline): Update.
17901 (pass_ipa_inline): Do not generate summaries.
17902 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
17904 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
17905 and add pass_ipa_fn_summary.
17906 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
17908 (make_pass_inline_parameters): Remove.
17910 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
17912 * omp-low.c (struct omp_context): Remove "default_kind" member.
17915 * omp-offload.c (execute_oacc_device_lower): Remove the
17916 parallelism dimensions function attributes for unparallelized
17917 OpenACC kernels constructs.
17919 2017-05-23 Martin Liska <mliska@suse.cz>
17921 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
17923 (cgraph_edge::make_speculative): Likewise.
17924 (cgraph_edge::resolve_speculation): Likewise.
17925 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
17926 (cgraph_node::dump): Likewise.
17927 * cgraph.h: Likewise.
17928 * cgraphunit.c (analyze_functions): Likewise.
17929 (symbol_table::compile): Likewise.
17930 * ipa-cp.c (print_all_lattices): Likewise.
17931 (determine_versionability): Likewise.
17932 (initialize_node_lattices): Likewise.
17933 (ipcp_verify_propagated_values): Likewise.
17934 (estimate_local_effects): Likewise.
17935 (update_profiling_info): Likewise.
17936 (create_specialized_node): Likewise.
17937 (perhaps_add_new_callers): Likewise.
17938 (decide_about_value): Likewise.
17939 (decide_whether_version_node): Likewise.
17940 (identify_dead_nodes): Likewise.
17941 (ipcp_store_bits_results): Likewise.
17942 * ipa-devirt.c (dump_targets): Likewise.
17943 (ipa_devirt): Likewise.
17944 * ipa-icf.c (sem_item::dump): Likewise.
17945 (sem_function::equals): Likewise.
17946 (sem_variable::equals): Likewise.
17947 (sem_item_optimizer::read_section): Likewise.
17948 (sem_item_optimizer::execute): Likewise.
17949 (congruence_class::dump): Likewise.
17950 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
17951 (dump_inline_summary): Likewise.
17952 (estimate_node_size_and_time): Likewise.
17953 (inline_analyze_function): Likewise.
17954 * ipa-inline-transform.c (inline_call): Likewise.
17955 * ipa-inline.c (report_inline_failed_reason): Likewise.
17956 (want_early_inline_function_p): Likewise.
17957 (edge_badness): Likewise.
17958 (update_edge_key): Likewise.
17959 (inline_small_functions): Likewise.
17960 * ipa-profile.c (ipa_profile): Likewise.
17961 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
17962 (ipa_make_edge_direct_to_target): Likewise.
17963 (remove_described_reference): Likewise.
17964 (ipa_impossible_devirt_target): Likewise.
17965 (propagate_controlled_uses): Likewise.
17966 (ipa_print_node_params): Likewise.
17967 (ipcp_transform_function): Likewise.
17968 * ipa-pure-const.c (pure_const_read_summary): Likewise.
17969 (propagate_pure_const): Likewise.
17970 * ipa-reference.c (generate_summary): Likewise.
17971 (read_write_all_from_decl): Likewise.
17972 (propagate): Likewise.
17973 (ipa_reference_read_optimization_summary): Likewise.
17974 * ipa-utils.c (ipa_merge_profiles): Likewise.
17975 * ipa.c (walk_polymorphic_call_targets): Likewise.
17976 (symbol_table::remove_unreachable_nodes): Likewise.
17977 (ipa_single_use): Likewise.
17978 * passes.c (execute_todo): Likewise.
17979 * predict.c (drop_profile): Likewise.
17980 * symtab.c (symtab_node::get_dump_name): New function.
17981 (symtab_node::dump_name): Likewise.
17982 (symtab_node::dump_asm_name): Likewise.
17983 (symtab_node::dump_references): Likewise.
17984 (symtab_node::dump_referring): Likewise.
17985 (symtab_node::dump_base): Likewise.
17986 (symtab_node::debug_symtab): Likewise.
17987 * tree-sra.c (convert_callers_for_node): Likewise.
17988 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
17989 * value-prof.c (init_node_map): Likewise.
17991 2017-05-23 Martin Liska <mliska@suse.cz>
17993 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
17994 and symtab_node::debug_symtab to symbol_table::debug.
17995 * cgraphunit.c (analyze_functions): Use the renamed function.
17996 (symbol_table::compile): Likewise.
17997 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
17998 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
17999 * passes.c (execute_todo): Likewise.
18000 * symtab.c (symbol_table::dump): New function.
18001 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
18003 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18005 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
18006 that nonconst implies exec.
18008 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18010 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
18011 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
18012 (inline_edge_summary_vec): Turn into ...
18013 (ipa_call_summaries): ... this one.
18014 (redirect_to_unreachable, edge_set_predicate,
18015 evaluate_properties_for_edge, inline_summary_alloc,
18016 reset_ipa_call_summary, reset_inline_summary,
18017 inline_summary_t::duplicate): Update.
18018 (inline_edge_duplication_hook): Turn to ...
18019 (ipa_call_summary_t::duplicate): ... this one.
18020 (inline_edge_removal_hook): Turn to ...
18021 (ipa_call_summary_t::remove): ... this one.
18022 (dump_inline_edge_summary): Turn to ...
18023 (dump_ipa_call_summary): ... this one.
18024 (estimate_function_body_sizes): Update.
18025 (inline_update_callee_summaries): Update.
18026 (remap_edge_change_prob): Update.
18027 (remap_edge_summaries): Update.
18028 (inline_merge_summary): Update.
18029 (do_estimate_edge_time): Update.
18030 (inline_generate_summary): Update.
18031 (inline_read_section): Update.
18032 (inline_read_summary): Update.
18033 (inline_free_summary): Update.
18034 * ipa-inline.c (can_inline_edge_p): Update.
18035 (compute_inlined_call_time): Update.
18036 (want_inline_small_function_p): Update.
18037 (edge_badness): Update.
18038 (early_inliner): Update.
18039 * ipa-inline.h (inline_edge_summary): Turn to ...
18040 (ipa_call_summary): ... this one.
18041 (ipa_call_summary_t): New class.
18042 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
18043 (ipa_call_summaries): New.
18044 (inline_edge_summary): Remove.
18045 (estimate_edge_growth): Update.
18046 * ipa-profile.c (ipa_propagate_frequency_1): Update.
18047 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
18048 * ipa-split.c (execute_split_functions): Update.
18049 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
18051 2017-05-23 Tom de Vries <tom@codesourcery.com>
18053 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
18054 attributes): Document rdrand effective target.
18056 2017-05-23 Tom de Vries <tom@codesourcery.com>
18058 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
18059 attributes): Sort alphabetically.
18061 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
18063 * config/avr/genmultilib.awk: Use gsub instead of gensub.
18065 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
18068 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
18069 V2DF/V2DI splat into two separate patterns, one that handles
18070 registers, and the other that only handles memory. Drop support
18071 for splatting from a GPR on ISA 2.07 and then splitting the
18072 splat into direct move and splat.
18073 (vsx_splat_<mode>_reg): Likewise.
18074 (vsx_splat_<mode>_mem): Likewise.
18076 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
18078 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
18080 2017-05-22 Jakub Jelinek <jakub@redhat.com>
18082 PR middle-end/80809
18083 * omp-low.c (finish_taskreg_remap): New function.
18084 (finish_taskreg_scan): If unit size of ctx->record_type
18085 is non-constant, unshare the size expression and replace
18086 decls in it with possible outer var refs.
18088 PR middle-end/80809
18089 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
18090 GOVD_SHARED rather than GOVD_PRIVATE with it.
18091 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
18092 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
18094 PR middle-end/80853
18095 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
18096 as last argument to build_outer_var_ref for pointer bases of array
18097 section reductions.
18099 2017-05-19 Martin Sebor <msebor@redhat.com>
18101 * print-tree.c (print_node): Print DECL_READ_P flag.
18103 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18105 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
18106 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
18107 * cgraph.c: Likewise.
18108 * cgraphunit.c: Likewise.
18109 * gengtype.c: Likewise.
18110 * ipa-cp.c: Likewise.
18111 * ipa-devirt.c: Likewise.
18112 * ipa-icf.c: Likewise.
18113 * ipa-predicate.c: Likewise.
18114 * ipa-profile.c: Likewise.
18115 * ipa-prop.c: Likewise.
18116 * ipa-split.c: Likewise.
18118 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
18119 edge_predicate_pool, dump_inline_hints,
18120 inline_summary::account_size_time, redirect_to_unreachable,
18121 edge_set_predicate, set_hint_predicate,
18122 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
18123 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
18124 inline_summary_t::remove, remap_hint_predicate_after_duplication,
18125 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
18126 ipa_call_summary_t::remove, initialize_growth_caches,
18127 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
18128 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
18129 mark_modified, unmodified_parm_1, unmodified_parm,
18130 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
18131 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
18132 compute_bb_predicates, will_be_nonconstant_expr_predicate,
18133 will_be_nonconstant_predicate, record_modified_bb_info,
18134 get_minimal_bb, record_modified, param_change_prob,
18135 phi_result_unknown_predicate, predicate_for_phi_result,
18136 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
18137 estimate_function_body_sizes, compute_inline_parameters,
18138 compute_inline_parameters_for_curren, pass_data_inline_parameters,
18139 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
18140 inline_update_callee_summaries, remap_edge_change_prob,
18141 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
18142 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
18143 inline_analyze_function, inline_summary_t::insert,
18144 inline_generate_summary, read_ipa_call_summary, inline_read_section,
18145 inline_read_summary, write_ipa_call_summary, inline_write_summary,
18146 inline_free_summary): Move to ipa-fnsummary.h
18147 (predicate_t): Remove.
18148 * ipa-fnsummary.c: New file.
18149 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
18150 (enum inline_hints_vals, inline_hints, agg_position_info,
18151 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
18152 inline_summaries, ipa_call_summary, ipa_call_summary_t,
18153 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
18154 dump_inline_summary, dump_inline_hints, inline_generate_summary,
18155 inline_read_summary, inline_write_summary, inline_free_summary,
18156 inline_analyze_function, initialize_inline_failed,
18157 inline_merge_summary, inline_update_overall_summary,
18158 compute_inline_parameters): Move to ipa-fnsummary.h
18159 * ipa-fnsummary.h: New file.
18160 * ipa-inline-transform.h: Include ipa-inline.h.
18161 * ipa-inline.c: LIkewise.
18163 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18165 * ipa-inline.c (edge_badness): Use inlined_time instead of
18166 inline_summaries->get.
18168 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18170 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
18172 2017-05-22 Nathan Sidwell <nathan@acm.org>
18174 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
18175 (fdump-lang): Document 'raw' option.
18176 * dumpfile.h (TDI_tu): Delete.
18177 * dumpfile.c (dump_files): Remove translation-unit.
18178 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
18180 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
18182 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
18183 command option from $(AWK) call.
18184 * config/avr/genmultilib.awk: Simplify and rewrite so that it
18185 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
18186 [FORMAT]: Remove handling of variable.
18187 * config/avr/t-multilib: Regenerate.
18189 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18191 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
18193 (dump_inline_summary): Do not print self_time.
18194 (estimate_function_body_sizes): Do not set self_time.
18195 (compute_inline_parameters): Likewise.
18196 (inline_read_section, inline_write_summary): Do not stream self_time.
18197 * ipa-inline.h (inline_summary): Drop self_time.
18199 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18201 * ipa-inline-analysis.c (account_size_time): Rename to ...
18202 (inline_summary::account_size_time): ... this one.
18203 (reset_ipa_call_summary): Turn to ...
18204 (ipa_call_summary::reset): ... this one.
18205 (reset_inline_summary): Turn to ...
18206 (inline_summary::reset): ... this one.
18207 (inline_summary_t::remove): Update.
18208 (inline_summary_t::duplicate): Update.
18209 (ipa_call_summary_t::remove): Update.
18210 (dump_inline_summary): Update.
18211 (estimate_function_body_sizes): Update.
18212 (compute_inline_parameters): Update.
18213 (estimate_node_size_and_time): Update.
18214 (inline_merge_summary): Update.
18215 (inline_update_overall_summary): Update.
18216 (inline_read_section): Update.
18217 (inline_write_summary): Update.
18218 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
18219 add account_size_time and reset member functions.
18220 (ipa_call_summary): Add reset function.
18221 * ipa-predicate.h (predicate::operator &): Constify.
18223 2017-05-22 Richard Biener <rguenther@suse.de>
18225 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
18227 2017-05-19 Jason Merrill <jason@redhat.com>
18229 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
18231 2017-05-19 Marek Polacek <polacek@redhat.com>
18234 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
18235 TYPE_OVERFLOW_WRAPS checks.
18237 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
18239 * tree-core.h (enum omp_clause_default_kind): Add
18240 "OMP_CLAUSE_DEFAULT_PRESENT".
18241 * tree-pretty-print.c (dump_omp_clause): Handle it.
18242 * gimplify.c (enum gimplify_omp_var_data): Add
18243 "GOVD_MAP_FORCE_PRESENT".
18244 (gimplify_adjust_omp_clauses_1): Map it to
18245 "GOMP_MAP_FORCE_PRESENT".
18246 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
18248 * gimplify.c (oacc_default_clause): Clarify.
18250 2017-05-19 Nathan Sidwell <nathan@acm.org>
18252 LANG_HOOK_REGISTER_DUMPS
18253 * toplev.c (general_init): Call register dump lang hook.
18254 * doc/invoke.texi: Document -fdump-lang option family.
18255 * dumpfile.c (dump_files): Remove class dump here.
18256 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
18257 * dumpfile.h (tree_dump_index): Remove TDI_class.
18258 * langhooks-def.h (lhd_register_dumps): Declare.
18259 (LANG_HOOKS_REGISTER_DUMPS): Define.
18260 (LANG_HOOKS_INITIALIZER): Add it.
18261 * langhooks.c (lhd_register_dumps): Define.
18262 * langhooks.h (struct lang_hooks): Add register_dumps.
18264 2017-05-19 Nathan Sidwell <nathan@acm.org>
18266 * context.h (context::set_passes): New.
18267 * context.c (context::context): Do not create pass manager.
18268 * toplev.c (general_init): Create pass manager here.
18270 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
18272 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
18273 use this splitter if two add or or instructions would also work for
18274 the constant we want to generate.
18276 2017-05-19 Richard Biener <rguenther@suse.de>
18279 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
18280 predicate evaluation.
18282 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
18284 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
18286 * ipa-inline.c (want_inline_small_function_p): Do not cast to
18289 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
18291 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
18292 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
18293 (inline_edge_summary_vec): Turn into ...
18294 (ipa_call_summaries): ... this one.
18295 (redirect_to_unreachable, edge_set_predicate,
18296 evaluate_properties_for_edge, inline_summary_alloc,
18297 reset_ipa_call_summary, reset_inline_summary,
18298 inline_summary_t::duplicate): Update.
18299 (inline_edge_duplication_hook): Turn to ...
18300 (ipa_call_summary_t::duplicate): ... this one.
18301 (inline_edge_removal_hook): Turn to ...
18302 (ipa_call_summary_t::remove): ... this one.
18303 (dump_inline_edge_summary): Turn to ...
18304 (dump_ipa_call_summary): ... this one.
18305 (estimate_function_body_sizes): Update.
18306 (inline_update_callee_summaries): Update.
18307 (remap_edge_change_prob): Update.
18308 (remap_edge_summaries): Update.
18309 (inline_merge_summary): Update.
18310 (do_estimate_edge_time): Update.
18311 (inline_generate_summary): Update.
18312 (inline_read_section): Update.
18313 (inline_read_summary): Update.
18314 (inline_free_summary): Update.
18315 * ipa-inline.c (can_inline_edge_p): Update.
18316 (compute_inlined_call_time): Update.
18317 (want_inline_small_function_p): Update.
18318 (edge_badness): Update.
18319 (early_inliner): Update.
18320 * ipa-inline.h (inline_edge_summary): Turn to ...
18321 (ipa_call_summary): ... this one.
18322 (ipa_call_summary_t): New class.
18323 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
18324 (ipa_call_summaries): New.
18325 (inline_edge_summary): Remove.
18326 (estimate_edge_growth): Update.
18327 * ipa-profile.c (ipa_propagate_frequency_1): Update.
18328 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
18329 * ipa-split.c (execute_split_functions): Update.
18330 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
18332 2017-05-19 Richard Biener <rguenther@suse.de>
18334 PR middle-end/80764
18335 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
18337 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
18339 * config/rs6000/rs6000.c (struct machine_function): Add field
18340 fpr_is_wrapped_separately.
18341 (rs6000_get_separate_components): Use 64 components. Handle the
18342 new FPR components.
18343 (rs6000_components_for_bb): Handle the FPR components.
18344 (rs6000_emit_prologue_components): Handle the FPR components.
18345 (rs6000_emit_epilogue_components): Handle the FPR components.
18346 (rs6000_set_handled_components): Handle the FPR components.
18347 (rs6000_emit_prologue): Don't output prologue code for those FPRs
18348 that are already separately shrink-wrapped.
18349 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
18350 that are already separately shrink-wrapped.
18352 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
18355 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
18358 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
18359 (define_peephole2 for Altivec d-form load): Add peepholes to catch
18360 cases where the register allocator uses a move and an offsettable
18361 memory operation to/from a FPR register on ISA 2.06/2.07.
18362 (define_peephole2 for Altivec d-form store): Likewise.
18364 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
18367 * config/i386/mmx.md (*mov<mode>_internal): Enable
18368 alternatives 11, 12, 13 and 14 also for 32bit targets.
18369 Remove alternatives 15, 16, 17 and 18.
18370 * config/i386/sse.md (vec_concatv2di): Change
18371 alternative (!x, *y) to (x, ?!*Yn).
18373 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
18375 * dumpfile.h (enum dump_kind): Remove stray comma.
18377 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
18379 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
18380 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
18381 predicate::num_conditions
18382 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
18383 (CHANGED): turn into predicate::changed.
18384 (agg_position_info): Move to ipa-predicate.h
18385 (add_condition, predicate::add_clause, predicate::operator &=,
18386 predicate::or_with, predicate::evaluate, predicate::probability,
18387 dump_condition, dump_clause, predicate::dump,
18388 predicate::remap_after_duplication, predicate::remap_after_inlining,
18389 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
18390 (evaluate_conditions_for_known_args): Update.
18391 (set_cond_stmt_execution_predicate): Update.
18392 * ipa-inline.h: Include ipa-predicate.h
18393 (condition, inline_param_summary, conditions, agg_position_info,
18394 predicate): Move to ipa-predicate.h
18395 * ipa-predicate.c: New file.
18396 * ipa-predicate.h: New file.
18398 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
18400 * final.c (leaf_function_p): Check we are not in a sequence.
18402 2017-05-18 Martin Liska <mliska@suse.cz>
18404 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
18405 * dumpfile.c (dump_register): Use new enum dump_kind.
18406 (get_dump_file_name): Likewise.
18407 (dump_enable_all): Likewise.
18408 (dump_switch_p_1): Likewise.
18409 (enable_rtl_dump_file): Remove usage of TDF_RTL.
18410 * dumpfile.h (enum dump_kind): New enum type.
18411 (struct dump_file_info): Create constructor and
18412 format fields and comments.
18413 * passes.c (pass_manager::register_one_dump_file):
18415 * statistics.c (statistics_early_init): Likewise.
18416 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
18417 TDF_TREE with TDF_SLIM.
18418 (gather_memory_references_ref): Likewise.
18420 2017-05-18 Martin Liska <mliska@suse.cz>
18422 * vec.h (struct vnull): Use it.
18424 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
18426 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
18427 (true_predicate, false_predicate, true_predicate_p,
18428 false_predicate_p): Remove.
18429 (single_cond_predicate, not_inlined_predicate): Turn to member function
18431 (add_condition): Update.
18432 (add_clause): Turn to...
18433 (predicate::add_clause): ... this one; update; allow passing NULL
18435 (and_predicates): Turn to ...
18436 (predicate::operator &=): ... this one.
18437 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
18438 (or_predicates): Turn to ...
18439 (predicate::or_with): ... this one.
18440 (evaluate_predicate): Turn to ...
18441 (predicate::evaluate): ... this one.
18442 (predicate_probability): Turn to ...
18443 (predicate::probability): ... this one.
18444 (dump_condition): Update.
18445 (dump_predicate): Turn to ...
18446 (predicate::dump): ... this one.
18447 (account_size_time): Update.
18448 (edge_set_predicate): Update.
18449 (set_hint_predicate): UPdate.
18450 (evaluate_conditions_for_known_args): Update.
18451 (evaluate_properties_for_edge): Update.
18452 (remap_predicate_after_duplication): Turn to...
18453 (predicate::remap_after_duplication): ... this one.
18454 (remap_hint_predicate_after_duplication): Update.
18455 (inline_summary_t::duplicate): UPdate.
18456 (dump_inline_edge_summary): Update.
18457 (dump_inline_summary): Update.
18458 (set_cond_stmt_execution_predicate): Update.
18459 (set_switch_stmt_execution_predicate): Update.
18460 (compute_bb_predicates): Update.
18461 (will_be_nonconstant_expr_predicate): Update.
18462 (will_be_nonconstant_predicate): Update.
18463 (phi_result_unknown_predicate): Update.
18464 (predicate_for_phi_result): Update.
18465 (array_index_predicate): Update.
18466 (estimate_function_body_sizes): Update.
18467 (estimate_node_size_and_time): Update.
18468 (estimate_ipcp_clone_size_and_time): Update.
18469 (remap_predicate): Rename to ...
18470 (predicate::remap_after_inlining): ... this one.
18471 (remap_hint_predicate): Update.
18472 (inline_merge_summary): Update.
18473 (inline_update_overall_summary): Update.
18474 (estimate_size_after_inlining): Update.
18475 (read_predicate): Rename to ...
18476 (predicate::stream_in): ... this one.
18477 (read_inline_edge_summary): Update.
18478 (write_predicate): Rename to ...
18479 (predicate::stream_out): ... this one.
18480 (write_inline_edge_summary): Update.
18481 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
18482 (clause_t): Turn to uint32_t
18483 (predicate): Turn to class; implement constructor and operators
18485 (size_time_entry): Update.
18486 (inline_summary): Update.
18487 (inline_edge_summary): Update.
18489 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
18491 * fold-const.c (fold_binary_loc): Move transformation...
18492 * match.pd (C - X CMP X): ... here.
18494 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
18496 * config/sparc/sparc.c (sparc_option_override): Set function
18497 alignment for -mcpu=niagara7 to 64 to match the I$ line.
18498 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
18500 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
18502 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
18504 2017-05-18 Marek Polacek <polacek@redhat.com>
18507 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
18508 (pass_ubsan::execute): Call gimple_assign_single_p instead of
18509 gimple_assign_load_p.
18511 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
18513 PR middle-end/80692
18514 * real.c (do_compare): Give decimal_do_compare preference over
18515 comparing just the signs.
18517 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
18519 * doc/md.texi (Canonicalization of Instructions): Describe the
18520 canonical form of instructions that inherently set a condition
18523 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
18525 PR middle-end/80775
18526 * tree-cfg.c: Move deletion of unreachable case statements to after
18527 the merging of consecutive case labels.
18529 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
18531 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
18532 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
18533 restoring of callee-saved registers.
18535 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
18537 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
18538 * config/visium/visium.c (single_set_and_flags): Likewise.
18539 * config/visium/visium.md (Substitutions): Likewise.
18541 2017-05-17 Martin Liska <mliska@suse.cz>
18543 * cfg.c: Introduce dump_flags_t type and
18544 use it instead of int type.
18546 * cfghooks.c: Likewise.
18547 * cfghooks.h (struct cfg_hooks): Likewise.
18548 * cfgrtl.c: Likewise.
18549 * cfgrtl.h: Likewise.
18550 * cgraph.c (cgraph_node::get_body): Likewise.
18551 * coretypes.h: Likewise.
18552 * domwalk.c: Likewise.
18553 * domwalk.h: Likewise.
18554 * dumpfile.c (struct dump_option_value_info): Likewise.
18555 (dump_enable_all): Likewise.
18556 (dump_switch_p_1): Likewise.
18557 (opt_info_switch_p): Likewise.
18558 * dumpfile.h (enum tree_dump_index): Likewise.
18559 (struct dump_file_info): Likewise.
18560 * genemit.c: Likewise.
18561 * generic-match-head.c: Likewise.
18562 * gengtype.c (open_base_files): Likewise.
18563 * gimple-pretty-print.c: Likewise.
18564 * gimple-pretty-print.h: Likewise.
18565 * graph.c (print_graph_cfg): Likewise.
18566 * graphite-scop-detection.c (dot_all_sese): Likewise.
18567 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
18568 * loop-unroll.c (report_unroll): Likewise.
18569 * passes.c (pass_manager::register_one_dump_file): Likewise.
18570 * print-tree.c: Likewise.
18571 * statistics.c: Likewise.
18572 * tree-cfg.c: Likewise.
18573 * tree-cfg.h: Likewise.
18574 * tree-dfa.c: Likewise.
18575 * tree-dfa.h: Likewise.
18576 * tree-dump.c (dump_function): Likewise.
18577 * tree-dump.h (struct dump_info): Likewise.
18578 * tree-pretty-print.c: Likewise.
18579 * tree-pretty-print.h: Likewise.
18580 * tree-ssa-live.c: Likewise.
18581 * tree-ssa-live.h: Likewise.
18582 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
18583 * tree-vect-loop.c: Likewise.
18584 * tree-vect-slp.c: Likewise.
18586 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
18587 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18589 PR tree-optimization/80457
18590 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
18591 of all arguments to a statement as scalar_to_vec operations.
18592 (vectorizable_call): Adjust call to vect_model_simple_cost for
18594 (vectorizable_conversion): Likewise.
18595 (vectorizable_assignment): Likewise.
18596 (vectorizable_shift): Likewise.
18597 (vectorizable_operation): Likewise.
18598 (vectorizable_comparison): Likewise.
18599 (vect_is_simple_cond): Record the def types for operands.
18600 (vectorizable_condition): Likewise, call vect_model_simple_cost.
18601 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
18602 for statement argument count.
18604 2017-05-16 Carl Love <cel@us.ibm.com>
18606 * config/rs6000/rs6000-c: Add support for built-in functions
18607 vector unsigned long long vec_bperm (vector unsigned long long,
18608 vector unsigned char)
18609 vector signed long long vec_mule (vector signed int,
18611 vector unsigned long long vec_mule (vector unsigned int,
18612 vector unsigned int)
18613 vector signed long long vec_mulo (vector signed int,
18615 vector unsigned long long vec_mulo (vector unsigned int,
18616 vector unsigned int)
18617 vector signed char vec_sldw (vector signed char,
18618 vector signed char,
18620 vector unsigned char vec_sldw (vector unsigned char,
18621 vector unsigned char,
18623 vector signed short vec_sldw (vector signed short,
18624 vector signed short,
18626 vector unsigned short vec_sldw (vector unsigned short,
18627 vector unsigned short,
18629 vector signed int vec_sldw (vector signed int,
18632 vector unsigned int vec_sldw (vector unsigned int,
18633 vector unsigned int,
18635 vector signed long long vec_sldw (vector signed long long,
18636 vector signed long long,
18638 vector unsigned long long vec_sldw (vector unsigned long long,
18639 vector unsigned long long,
18641 * config/rs6000/rs6000-c: Add support for built-in functions
18642 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
18643 * config/rs6000/altivec.h: Add defintion for vec_sldw.
18644 * doc/extend.texi: Update the built-in documentation for the
18645 new built-in functions.
18647 2017-05-16 Marek Polacek <polacek@redhat.com>
18651 * tree.c (save_expr): Don't fold the expression.
18653 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
18655 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
18656 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
18657 and (?*y,m). Update insn attributes.
18659 2017-05-16 Martin Liska <mliska@suse.cz>
18661 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
18662 flags argument of print_gimple_stmt, print_gimple_expr,
18663 print_generic_stmt and print_generic_expr.
18664 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
18665 * coretypes.h: Likewise.
18666 * except.c (dump_eh_tree): Likewise.
18667 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
18668 * gimple-pretty-print.h: Likewise.
18669 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
18670 (backprop::push_to_worklist): Likewise.
18671 (backprop::pop_from_worklist): Likewise.
18672 (backprop::process_use): Likewise.
18673 (backprop::intersect_uses): Likewise.
18674 (note_replacement): Likewise.
18675 * gimple-ssa-store-merging.c
18676 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
18677 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
18678 (pass_store_merging::execute): Likewise.
18679 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
18680 (ssa_base_cand_dump_callback): Likewise.
18681 (dump_incr_vec): Likewise.
18682 (replace_refs): Likewise.
18683 (replace_mult_candidate): Likewise.
18684 (create_add_on_incoming_edge): Likewise.
18685 (create_phi_basis): Likewise.
18686 (insert_initializers): Likewise.
18687 (all_phi_incrs_profitable): Likewise.
18688 (introduce_cast_before_cand): Likewise.
18689 (replace_one_candidate): Likewise.
18690 * gimplify.c (gimplify_expr): Likewise.
18691 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
18692 (set_rename): Likewise.
18693 (rename_uses): Likewise.
18694 (copy_loop_phi_nodes): Likewise.
18695 (add_close_phis_to_merge_points): Likewise.
18696 (copy_loop_close_phi_args): Likewise.
18697 (copy_cond_phi_args): Likewise.
18698 (graphite_copy_stmts_from_block): Likewise.
18699 (translate_pending_phi_nodes): Likewise.
18700 * graphite-poly.c (print_pdr): Likewise.
18701 (dump_gbb_cases): Likewise.
18702 (dump_gbb_conditions): Likewise.
18703 (print_scop_params): Likewise.
18704 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
18705 (build_cross_bb_scalars_use): Likewise.
18706 (gather_bbs::before_dom_children): Likewise.
18707 * hsa-dump.c (dump_hsa_immed): Likewise.
18708 * ipa-cp.c (print_ipcp_constant_value): Likewise.
18709 (get_replacement_map): Likewise.
18710 * ipa-inline-analysis.c (dump_condition): Likewise.
18711 (estimate_function_body_sizes): Likewise.
18712 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
18713 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
18714 * ipa-prop.c (ipa_dump_param): Likewise.
18715 (ipa_print_node_jump_functions_for_edge): Likewise.
18716 (ipa_modify_call_arguments): Likewise.
18717 (ipa_modify_expr): Likewise.
18718 (ipa_dump_param_adjustments): Likewise.
18719 (ipa_dump_agg_replacement_values): Likewise.
18720 (ipcp_modif_dom_walker::before_dom_children): Likewise.
18721 * ipa-pure-const.c (check_stmt): Likewise.
18722 (pass_nothrow::execute): Likewise.
18723 * ipa-split.c (execute_split_functions): Likewise.
18724 * omp-offload.c (dump_oacc_loop_part): Likewise.
18725 (dump_oacc_loop): Likewise.
18726 * trans-mem.c (tm_log_emit): Likewise.
18727 (tm_memopt_accumulate_memops): Likewise.
18728 (dump_tm_memopt_set): Likewise.
18729 (dump_tm_memopt_transform): Likewise.
18730 * tree-cfg.c (gimple_verify_flow_info): Likewise.
18731 (print_loop): Likewise.
18732 * tree-chkp-opt.c (chkp_print_addr): Likewise.
18733 (chkp_gather_checks_info): Likewise.
18734 (chkp_get_check_result): Likewise.
18735 (chkp_remove_check_if_pass): Likewise.
18736 (chkp_use_outer_bounds_if_possible): Likewise.
18737 (chkp_reduce_bounds_lifetime): Likewise.
18738 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
18739 (chkp_mark_completed_bounds): Likewise.
18740 (chkp_register_incomplete_bounds): Likewise.
18741 (chkp_mark_invalid_bounds): Likewise.
18742 (chkp_maybe_copy_and_register_bounds): Likewise.
18743 (chkp_build_returned_bound): Likewise.
18744 (chkp_get_bound_for_parm): Likewise.
18745 (chkp_build_bndldx): Likewise.
18746 (chkp_get_bounds_by_definition): Likewise.
18747 (chkp_generate_extern_var_bounds): Likewise.
18748 (chkp_get_bounds_for_decl_addr): Likewise.
18749 * tree-chrec.c (chrec_apply): Likewise.
18750 * tree-data-ref.c (dump_data_reference): Likewise.
18751 (dump_subscript): Likewise.
18752 (dump_data_dependence_relation): Likewise.
18753 (analyze_overlapping_iterations): Likewise.
18754 * tree-inline.c (expand_call_inline): Likewise.
18755 (tree_function_versioning): Likewise.
18756 * tree-into-ssa.c (dump_defs_stack): Likewise.
18757 (dump_currdefs): Likewise.
18758 (dump_names_replaced_by): Likewise.
18759 (dump_update_ssa): Likewise.
18760 (update_ssa): Likewise.
18761 * tree-object-size.c (pass_object_sizes::execute): Likewise.
18762 * tree-parloops.c (build_new_reduction): Likewise.
18763 (try_create_reduction_list): Likewise.
18764 (ref_conflicts_with_region): Likewise.
18765 (oacc_entry_exit_ok_1): Likewise.
18766 (oacc_entry_exit_single_gang): Likewise.
18767 * tree-pretty-print.h: Likewise.
18768 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
18769 (get_scalar_evolution): Likewise.
18770 (add_to_evolution): Likewise.
18771 (get_loop_exit_condition): Likewise.
18772 (analyze_evolution_in_loop): Likewise.
18773 (analyze_initial_condition): Likewise.
18774 (analyze_scalar_evolution): Likewise.
18775 (instantiate_scev): Likewise.
18776 (number_of_latch_executions): Likewise.
18777 (gather_chrec_stats): Likewise.
18778 (final_value_replacement_loop): Likewise.
18779 (scev_const_prop): Likewise.
18780 * tree-sra.c (dump_access): Likewise.
18781 (disqualify_candidate): Likewise.
18782 (create_access): Likewise.
18783 (reject): Likewise.
18784 (maybe_add_sra_candidate): Likewise.
18785 (create_access_replacement): Likewise.
18786 (analyze_access_subtree): Likewise.
18787 (analyze_all_variable_accesses): Likewise.
18788 (sra_modify_assign): Likewise.
18789 (initialize_constant_pool_replacements): Likewise.
18790 (find_param_candidates): Likewise.
18791 (decide_one_param_reduction): Likewise.
18792 (replace_removed_params_ssa_names): Likewise.
18793 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
18794 * tree-ssa-copy.c (dump_copy_of): Likewise.
18795 (copy_prop_visit_cond_stmt): Likewise.
18796 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
18797 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
18798 (record_equivalences_from_stmt): Likewise.
18799 * tree-ssa-dse.c (compute_trims): Likewise.
18800 (delete_dead_call): Likewise.
18801 (delete_dead_assignment): Likewise.
18802 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
18803 (forward_propagate_into_cond): Likewise.
18804 (pass_forwprop::execute): Likewise.
18805 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
18806 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
18808 (move_computations_worker): Likewise.
18809 (execute_sm): Likewise.
18810 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
18811 (remove_exits_and_undefined_stmts): Likewise.
18812 (remove_redundant_iv_tests): Likewise.
18813 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
18814 (adjust_iv_update_pos): Likewise.
18815 * tree-ssa-math-opts.c (bswap_replace): Likewise.
18816 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
18817 (value_replacement): Likewise.
18818 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
18819 * tree-ssa-pre.c (print_pre_expr): Likewise.
18820 (get_representative_for): Likewise.
18821 (create_expression_by_pieces): Likewise.
18822 (insert_into_preds_of_block): Likewise.
18823 (eliminate_insert): Likewise.
18824 (eliminate_dom_walker::before_dom_children): Likewise.
18825 (eliminate): Likewise.
18826 (remove_dead_inserted_code): Likewise.
18827 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
18828 * tree-ssa-reassoc.c (get_rank): Likewise.
18829 (eliminate_duplicate_pair): Likewise.
18830 (eliminate_plus_minus_pair): Likewise.
18831 (eliminate_not_pairs): Likewise.
18832 (undistribute_ops_list): Likewise.
18833 (eliminate_redundant_comparison): Likewise.
18834 (update_range_test): Likewise.
18835 (optimize_range_tests_var_bound): Likewise.
18836 (optimize_vec_cond_expr): Likewise.
18837 (rewrite_expr_tree): Likewise.
18838 (rewrite_expr_tree_parallel): Likewise.
18839 (linearize_expr): Likewise.
18840 (break_up_subtract): Likewise.
18841 (linearize_expr_tree): Likewise.
18842 (attempt_builtin_powi): Likewise.
18843 (attempt_builtin_copysign): Likewise.
18844 (transform_stmt_to_copy): Likewise.
18845 (transform_stmt_to_multiply): Likewise.
18846 (dump_ops_vector): Likewise.
18847 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
18848 (print_scc): Likewise.
18849 (set_ssa_val_to): Likewise.
18850 (visit_reference_op_store): Likewise.
18851 (visit_use): Likewise.
18852 (sccvn_dom_walker::before_dom_children): Likewise.
18853 (run_scc_vn): Likewise.
18854 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
18856 (expr_hash_elt::print): Likewise.
18857 (const_and_copies::pop_to_marker): Likewise.
18858 (const_and_copies::record_const_or_copy_raw): Likewise.
18859 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
18860 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
18861 (dump_predicates): Likewise.
18862 (find_uninit_use): Likewise.
18863 (warn_uninitialized_phi): Likewise.
18864 (pass_late_warn_uninitialized::execute): Likewise.
18865 * tree-ssa.c (verify_vssa): Likewise.
18866 (verify_ssa): Likewise.
18867 (maybe_optimize_var): Likewise.
18868 * tree-vrp.c (dump_value_range): Likewise.
18869 (dump_all_value_ranges): Likewise.
18870 (dump_asserts_for): Likewise.
18871 (register_edge_assert_for_2): Likewise.
18872 (vrp_visit_cond_stmt): Likewise.
18873 (vrp_visit_switch_stmt): Likewise.
18874 (vrp_visit_stmt): Likewise.
18875 (vrp_visit_phi_node): Likewise.
18876 (simplify_cond_using_ranges_1): Likewise.
18877 (fold_predicate_in): Likewise.
18878 (evrp_dom_walker::before_dom_children): Likewise.
18879 (evrp_dom_walker::push_value_range): Likewise.
18880 (evrp_dom_walker::pop_value_range): Likewise.
18881 (execute_early_vrp): Likewise.
18883 2017-05-16 Richard Biener <rguenther@suse.de>
18885 * dwarf2out.c (loc_list_from_tree_1): Do not create
18886 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
18888 2017-05-16 Richard Biener <rguenther@suse.de>
18890 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
18892 (note_variable_value_in_expr): If we resolved the decl ref
18893 do not push to the stack.
18895 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
18897 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
18898 operations in fast-math mode.
18899 (vaddq_f16): Likewise.
18900 (vmul_f16): Likewise.
18901 (vmulq_f16): Likewise.
18902 (vsub_f16): Likewise.
18903 (vsubq_f16): Likewise.
18904 * config/arm/neon.md (add<mode>3): New.
18906 (fma:<VH:mode>3): New. Also remove outdated comment.
18909 2017-05-16 Martin Liska <mliska@suse.cz>
18913 * ipa-devirt.c (warn_types_mismatch): Fix typo.
18914 (odr_types_equivalent_p): Likewise.
18916 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
18918 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
18920 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
18923 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
18924 non-interunit SSE move alternatives with '?'.
18925 (zero-extendsidi peephole2): New peephole to skip intermediate
18926 general register in SSE zero-extend sequence.
18928 2017-05-15 Jeff Law <law@redhat.com>
18930 * reorg.c (relax_delay_slots): Create a new variable to hold
18931 the temporary target rather than clobbering TARGET_LABEL.
18933 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
18934 missing argument to extract_bit_field call.
18935 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
18937 2017-05-15 Martin Liska <mliska@suse.cz>
18940 * gcc.c (process_command): Do not allow empty argument of -o option.
18942 2017-05-15 Renlin Li <renlin.li@arm.com>
18944 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
18945 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
18946 * config/aarch64/constraints.md (Usf): Add long call check.
18947 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
18948 (call_value): Likewise.
18949 (sibcall): Likewise.
18950 (sibcall_value): Likewise.
18952 (call_value_insn): New.
18953 (sibcall_insn): Update rtx pattern.
18954 (sibcall_value_insn): Likewise.
18955 (call_internal): Remove.
18956 (call_value_internal): Likewise.
18957 (sibcall_internal): Likewise.
18958 (sibcall_value_internal): Likewise.
18959 (call_reg): Likewise.
18960 (call_symbol): Likewise.
18961 (call_value_reg): Likewise.
18962 (call_value_symbol): Likewise.
18964 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
18967 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
18969 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
18971 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
18972 compatible with CCGOCmode and with CCZmode.
18974 2017-05-14 Martin Sebor <msebor@redhat.com>
18976 PR middle-end/77671
18977 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
18978 (gimple_fold_builtin_snprintf): Same.
18979 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
18980 (gimple_fold_builtin_snprintf): Same.
18981 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
18982 of character types.
18983 (is_call_safe): New function.
18984 (try_substitute_return_value): Call it.
18985 (try_simplify_call): New function.
18986 (pass_sprintf_length::handle_gimple_call): Call it.
18988 2017-05-14 Martin Sebor <msebor@redhat.com>
18990 PR middle-end/80669
18991 * builtins.c (expand_builtin_stpncpy): Simplify.
18993 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
18995 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
18996 * config/i386/i386.h
18997 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
18998 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
18999 (struct machine_function): Add new members call_ms2sysv,
19000 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
19001 (struct machine_frame_state): New fields sp_realigned and
19002 sp_realigned_offset.
19003 * config/i386/i386.c
19004 (enum xlogue_stub): New enum.
19005 (enum xlogue_stub_sets): New enum.
19006 (class xlogue_layout): New class.
19007 (struct ix86_frame): New fields stack_realign_allocate_offset,
19008 stack_realign_offset and outlined_save_offset. Modify comments to
19009 detail stack layout when using out-of-line stubs.
19010 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
19011 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
19012 -mcall-ms2sysv-xlogues.
19013 (stub_managed_regs): New static variable.
19014 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
19015 registers managed by out-of-line stub.
19016 (disable_call_ms2sysv_xlogues): New function.
19017 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
19018 m->call_ms2sysv when appropriate and compute frame layout for
19020 (sp_valid_at, fp_valid_at): New inline functions.
19021 (choose_basereg): New function.
19022 (choose_baseaddr): Add align parameter, use choose_basereg and modify
19024 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
19025 Use align parameter of choose_baseaddr to generated aligned SSE movs
19027 (pro_epilogue_adjust_stack): Modify to track
19028 machine_frame_state::sp_realigned.
19029 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
19030 (ix86_nsaved_sseregs): Likewise.
19031 (ix86_emit_save_regs): Likewise.
19032 (ix86_emit_save_regs_using_mov): Likewise.
19033 (ix86_emit_save_sse_regs_using_mov): Likewise.
19034 (get_scratch_register_on_entry): Likewise.
19035 (gen_frame_set): New function.
19036 (gen_frame_load): Likewise.
19037 (gen_frame_store): Likewise.
19038 (emit_outlined_ms2sysv_save): Likewise.
19039 (emit_outlined_ms2sysv_restore): Likewise.
19040 (ix86_expand_prologue): Modify stack re-alignment code and call
19041 emit_outlined_ms2sysv_save when appropriate.
19042 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
19043 parameter rtx_insn *insn, which allows the function to be used to only
19044 generate the notes.
19045 (ix86_expand_epilogue): Modify validity checks of frame and stack
19046 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
19047 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
19048 * config/i386/predicates.md
19049 (save_multiple): New predicate.
19050 (restore_multiple): Likewise.
19051 * config/i386/sse.md
19052 (save_multiple<mode>): New pattern.
19053 (save_multiple_realign<mode>): Likewise.
19054 (restore_multiple<mode>): Likewise.
19055 (restore_multiple_and_return<mode>): Likewise.
19056 (restore_multiple_leave_return<mode>): Likewise.
19057 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
19059 2017-05-14 Julia Koval <julia.koval@intel.com>
19061 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
19062 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
19063 (__builtin_ia32_xsetbv): New builtins.
19064 * config/i386/i386.c (ix86_expand_special_args_builtin):
19066 (ix86_expand_builtin): Special expand for new intrinsics.
19067 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
19068 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
19069 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
19071 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19073 * cfganal.c (inverted_post_order_compute): Change argument type
19075 * cfganal.h (inverted_post_order_compute): Adjust prototype.
19076 * df-core.c (rest_of_handle_df_initialize): Adjust.
19077 (rest_of_handle_df_finish): Likewise.
19078 (df_analyze_1): Likewise.
19079 (df_analyze): Likewise.
19080 (loop_inverted_post_order_compute): Change argument to be a vec *.
19081 (df_analyze_loop): Adjust.
19082 (df_get_n_blocks): Likewise.
19083 (df_get_postorder): Likewise.
19084 * df.h (struct df_d): Change field to be a vec.
19085 * lcm.c (compute_laterin): Adjust.
19086 (compute_available): Likewise.
19087 * lra-lives.c (lra_create_live_ranges_1): Likewise.
19088 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
19089 * tree-ssa-pre.c (compute_antic): Likewise.
19091 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19093 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
19094 (depth_first_search::depth_first_search): Change structure init
19095 function to this constructor.
19096 (depth_first_search::add_bb): Rename function to this member.
19097 (depth_first_search::execute): Likewise.
19098 (flow_dfs_compute_reverse_finish): Adjust.
19100 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19102 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
19103 (longest_simple_path): Likewise.
19104 * shrink-wrap.c (spread_components): Likewise.
19105 (disqualify_problematic_components): Likewise.
19106 (emit_common_heads_for_components): Likewise.
19107 (emit_common_tails_for_components): Likewise.
19108 (insert_prologue_epilogue_for_components): Likewise.
19110 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19112 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
19115 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19117 * df-core.c (df_set_blocks): Start using auto_bitmap.
19118 (df_compact_blocks): Likewise.
19119 * df-problems.c (df_rd_confluence_n): Likewise.
19120 * df-scan.c (df_insn_rescan_all): Likewise.
19121 (df_process_deferred_rescans): Likewise.
19122 (df_update_entry_block_defs): Likewise.
19123 (df_update_exit_block_uses): Likewise.
19124 (df_entry_block_bitmap_verify): Likewise.
19125 (df_exit_block_bitmap_verify): Likewise.
19126 (df_scan_verify): Likewise.
19127 * lra-constraints.c (lra_constraints): Likewise.
19128 (undo_optional_reloads): Likewise.
19129 (lra_undo_inheritance): Likewise.
19130 * lra-remat.c (calculate_gen_cands): Likewise.
19131 (do_remat): Likewise.
19132 * lra-spills.c (assign_spill_hard_regs): Likewise.
19133 (spill_pseudos): Likewise.
19134 * tree-ssa-pre.c (bitmap_set_and): Likewise.
19135 (bitmap_set_subtract_values): Likewise.
19137 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19139 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
19140 management with auto_bitmap.
19141 (fix_inter_tick): Likewise.
19142 (fix_recovery_deps): Likewise.
19143 * ira.c (add_store_equivs): Likewise.
19144 (find_moveable_pseudos): Likewise.
19145 (split_live_ranges_for_shrink_wrap): Likewise.
19146 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
19147 (rtx_reuse_manager::seen_def_p): Likewise.
19148 (rtx_reuse_manager::set_seen_def): Likewise.
19149 * print-rtl.h (class rtx_reuse_manager): Likewise.
19151 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19153 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
19155 (migrate_btr_def): Likewise.
19156 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
19157 * df-core.c (loop_post_order_compute): Likewise.
19158 (loop_inverted_post_order_compute): Likewise.
19159 * hsa-common.h: Likewise.
19160 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
19161 * init-regs.c (initialize_uninitialized_regs): Likewise.
19162 * ipa-inline.c (resolve_noninline_speculation): Likewise.
19163 (inline_small_functions): Likewise.
19164 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
19165 * ira.c (combine_and_move_insns): Likewise.
19166 (build_insn_chain): Likewise.
19167 * loop-invariant.c (find_invariants): Likewise.
19168 * lower-subreg.c (propagate_pseudo_copies): Likewise.
19169 * predict.c (tree_predict_by_opcode): Likewise.
19170 (predict_paths_leading_to): Likewise.
19171 (predict_paths_leading_to_edge): Likewise.
19172 (estimate_loops_at_level): Likewise.
19173 (estimate_loops): Likewise.
19174 * shrink-wrap.c (try_shrink_wrapping): Likewise.
19175 (spread_components): Likewise.
19176 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
19177 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
19178 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
19179 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
19180 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
19181 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
19182 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
19183 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
19184 (mark_threaded_blocks): Likewise.
19185 (thread_through_all_blocks): Likewise.
19186 * tree-ssa.c (verify_ssa): Likewise.
19187 (execute_update_addresses_taken): Likewise.
19188 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
19190 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19192 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
19194 (post_order_compute): Likewise.
19195 (inverted_post_order_compute): Likewise.
19196 (pre_and_rev_post_order_compute_fn): Likewise.
19198 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19200 * genrecog.c (int_set::int_set): Explicitly construct our
19201 auto_vec base class.
19202 * vec.h (auto_vec::auto_vec): New constructor.
19204 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19206 * bitmap.h (class auto_bitmap): New constructor taking
19207 bitmap_obstack * argument.
19209 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19211 * bitmap.h (class auto_bitmap): Change type of m_bits to
19212 bitmap_head, and adjust ctor / dtor and member operators.
19214 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
19216 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
19217 when returned register mode doesn't match original mode.
19219 2017-05-12 Jeff Law <law@redhat.com>
19220 Jakub Jelinek <jakub@redhat.com>
19222 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
19223 we look for cc setter after the compare-elim changes.
19224 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
19225 within the vector to match what compare-elim now expects.
19226 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
19227 (xorsi3_flags, one_cmplsi2_flags): Likewise.
19229 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
19230 after the compare-elim changes.
19231 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
19232 the vector to match what compare-elim now expects.
19233 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
19234 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
19235 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
19236 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
19237 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
19239 * config/visium/visium.c (single_set_and_flags): Fix where
19240 we look for cc setter after the compare-elim changes.
19241 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
19242 with the vector to match what compare-elim now expects.
19243 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
19244 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
19245 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
19246 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
19247 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
19248 (neg<mode>2_insn_set_overflow): Likewise.
19250 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
19252 PR middle-end/79794
19253 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
19254 maybe_expand_insn call, set ops[0].target. If still set after call,
19255 set alt_rtl. Add extra arg to recursive calls.
19256 (extract_bit_field): Add alt_rtl argument. Pass to
19258 * expmed.h (extract_bit_field): Fix prototype.
19259 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
19260 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
19261 to extract_bit_field_calls.
19262 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
19263 Pass alt_rtl to extract_bit_field calls.
19264 * calls.c (store_unaligned_arguments_into_psuedos)
19265 load_register_parameters): Pass extra NULL to extract_bit_field calls.
19266 * optabs.c (maybe_legitimize_operand): Clear op->target when call
19268 * optabs.h (struct expand_operand): Add target bitfield.
19270 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
19272 * compare-elim.c (try_eliminate_compare): Canonicalize
19273 operation with embedded compare to
19274 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
19275 (set (reg) (operation)].
19277 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
19279 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
19282 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
19283 cost of adding a carry flag for ADC instruction.
19284 [case MINUS]: Ignore the cost of subtracting a carry flag
19285 for SBB instruction.
19287 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
19289 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
19291 * config/rs6000/bmiintrin.h: New file.
19292 * config/rs6000/bmi2intrin.h: New file.
19293 * config/rs6000/x86intrin.h: New file.
19295 2017-05-12 Jeff Law <law@redhat.com>
19297 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
19300 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
19302 PR middle-end/80707
19303 * tree-cfg.c: Remove cfg edges of unreachable case statements.
19305 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19307 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
19308 early expansion of vector divide builtins.
19309 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
19310 builtins identified as having unsigned arguments.
19312 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19314 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
19315 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
19316 expansion of vector logical operations (and, andc, or, xor,
19319 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19321 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
19322 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
19324 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19326 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
19327 early GIMPLE expansion of vector multiplies.
19329 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19331 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
19332 TARGET_HAVE_MOVT conditional.
19333 (movt splitter): Likewise.
19335 2017-05-12 Richard Biener <rguenther@suse.de>
19337 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
19338 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
19339 Fold all stmts not inplace.
19341 2017-05-12 Richard Biener <rguenther@suse.de>
19343 PR tree-optimization/80713
19344 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
19345 inserted_exprs bit for not removed stmts.
19347 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
19349 PR middle-end/69921
19350 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
19351 parallelized" attribute for parallelized OpenACC kernels.
19352 * omp-offload.c (execute_oacc_device_lower): Use it.
19354 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
19355 Set "oacc kernels" attribute.
19356 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
19357 parameter. Adjust all users.
19358 (oacc_fn_attrib_kernels_p): Remove function.
19359 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
19360 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
19361 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
19362 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
19363 assert "oacc kernels" attribute is set.
19365 2017-05-11 Carl Love <cel@us.ibm.com>
19367 * config/rs6000/rs6000-c: Add support for built-in functions
19368 vector unsigned char vec_popcnt (vector signed char)
19369 vector unsigned char vec_popcnt (vector unsigned char)
19370 vector unsigned short vec_popcnt (vector signed short)
19371 vector unsigned short vec_popcnt (vector unsigned short)
19372 vector unsigned int vec_popcnt (vector signed int)
19373 vector unsigned int vec_popcnt (vector unsigned int)
19374 vector unsigned long long vec_popcnt (vector signed long long)
19375 vector unsigned long long vec_popcnt (vector unsigned long long)
19376 vector signed long long vec_slo (vector signed long long,
19377 vector signed char)
19378 vector signed long long vec_slo (vector signed long long,
19379 vector unsigned char)
19380 vector unsigned long long vec_slo (vector unsigned long long,
19381 vector signed char)
19382 vector unsigned long long vec_slo (vector unsigned long long,
19383 vector unsigned char)
19384 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
19385 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
19386 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
19387 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
19388 * doc/extend.texi: Update the built-in documentation file for the
19389 new built-in functions.
19391 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
19393 * attribs.h (sorted_attr_string): Move machine independent
19394 functions for target clone support from the i386 port to common
19395 code. Rename ix86_function_versions to common_function_versions.
19396 Rename make_name to make_unique_name.
19397 (common_function_versions): Likewise.
19398 (make_unique_name): Likewise.
19399 (make_dispatcher_decl): Likewise.
19400 (is_function_default_version): Likewise.
19401 * attribs.c (attr_strcmp): Likewise.
19402 (sorted_attr_string): Likewise.
19403 (common_function_versions): Likewise.
19404 (make_unique_name): Likewise.
19405 (make_dispatcher_decl): Likewise.
19406 (is_function_default_version): Likewise.
19407 * config/i386/i386.c (attr_strcmp): Likewise.
19408 (sorted_attr_string): Likewise.
19409 (ix86_function_versions): Likewise.
19410 (make_name): Likewise.
19411 (make_dispatcher_decl): Likewise.
19412 (is_function_default_version): Likewise.
19413 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
19415 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19418 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
19419 Account for direct move costs for vec_construct of integer
19422 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
19425 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
19426 (UNSPEC_STX_ATOMIC): Ditto.
19427 (loaddi_via_sse): New insn.
19428 (storedi_via_sse): Ditto.
19429 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
19430 Update corresponding peephole2 patterns.
19431 (atomic_storedi_fpu): Ditto.
19433 2017-05-11 Julia Koval <julia.koval@intel.com>
19435 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
19436 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
19438 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
19439 (__builtin_ia32_rsqrt14ss_mask): New builtins.
19440 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
19442 2017-05-11 Nathan Sidwell <nathan@acm.org>
19444 * graphite-poly.c: Include dumpfile.h.
19446 * dumpfle.h (dump_function): Declare here ...
19447 * tree-dump.h (dump_function): ... not here.
19448 * dumpfile.c: #include tree-cfg.h.
19449 (dump_function): Move here from ...
19450 * tree-dump.c (dump_function): ... here.
19451 * gimplify.c: #include splay-tree.h, not tree-dump.h.
19452 * graphite-poly.c: Don't include tree-dump.h.
19453 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
19454 * print-tree.c: Likewise.
19455 * stor-layout.c: Likewise.
19456 * tree-nested.c: Likewise.
19458 * dumpfile.c (dump_start): Use TDF_FLAGS.
19459 (dump_enable_all): Fix TDF_KIND check thinko.
19461 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
19463 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19464 array entries to represent two legal parameterizations of the
19465 overloaded __builtin_cmpb function, as represented by the
19466 P6_OV_BUILTIN_CMPB constant.
19467 (altivec_resolve_overloaded_builtin): Add special case handling
19468 for the __builtin_cmpb function, as represented by the
19469 P6_OV_BUILTIN_CMPB constant.
19470 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
19471 (BU_P6_64BIT_2): New macro.
19472 (BU_P6_OVERLOAD_2): New macro
19473 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
19474 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
19475 (CMPB): Add overload support to represent both 32-bit and 64-bit
19476 compare-bytes function.
19477 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
19478 support for TARGET_CMPB.
19479 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
19480 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19481 documentation of the __builtin_cmpb overloaded built-in function.
19483 2017-05-11 Richard Biener <rguenther@suse.de>
19485 PR tree-optimization/80705
19486 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
19487 bases are not vectorizable.
19489 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19491 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
19492 when counting register pressure.
19494 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19496 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
19497 (iv_ca_more_deps): Renamed to ...
19498 (iv_ca_compare_deps): ... this.
19499 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
19501 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19503 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
19505 (determine_group_iv_costs): ... here.
19506 (find_inv_vars_cb): Record inv var if it's not recorded before.
19508 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19510 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
19511 (get_shiftadd_cost): Ditto.
19513 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19515 * tree-ssa-address.c: Include header file.
19516 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
19518 (add_to_parts): Refactor.
19519 (addr_to_parts): New parameter. Update use of move_hint_to_base.
19520 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
19523 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19525 PR tree-optimization/53090
19526 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
19528 (extract_cond_operands): Detect condition with IV on both sides
19529 and return COMP_IV_EXPR_2.
19530 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
19531 (rewrite_use_compare): Simplify by removing call to function
19532 extract_cond_operands.
19534 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19536 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
19537 (extract_cond_operands): Detect condition comparing against non-
19538 invariant bound and return appropriate enum value.
19539 (find_interesting_uses_cond): Update use of extract_cond_operands.
19540 Handle its return value accordingly.
19541 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
19543 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19545 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
19546 nonlinear iv_use computation in loop invariant sensitive way.
19548 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19550 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
19551 (find_iv_candidates): Call relate_compare_use_with_all_cands.
19553 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19555 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
19556 (dump_cand): Support iv_cand.inv_exprs.
19557 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
19559 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
19562 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19564 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
19566 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
19567 as local function. Include necessary header files.
19568 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
19570 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19572 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
19574 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19576 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
19577 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
19578 RSHIFT_EXPR and BIT_NOT_EXPR.
19580 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19582 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
19583 (adjust_setup_cost): New parameter supporting round up adjustment.
19584 (struct address_cost_data): Delete.
19585 (force_expr_to_var_cost): Don't bound cost with spill_cost.
19586 (split_address_cost, ptr_difference_cost): Delete.
19587 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
19588 (struct ainc_cost_data): New struct.
19589 (get_address_cost_ainc): New function.
19590 (get_address_cost, get_computation_cost): Reimplement.
19591 (determine_group_iv_cost_address): Record inv_expr for all uses of
19593 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
19594 (iv_ca_has_deps): Reimplemented to ...
19595 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
19597 (iv_ca_extend): Call iv_ca_more_deps.
19599 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19601 * tree-ssa-address.c (struct mem_address): Move to header file.
19602 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
19603 * tree-ssa-address.h (struct mem_address): Move from C file.
19604 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
19606 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19608 * tree-affine.h (aff_combination_type): New interface.
19609 (aff_combination_zero_p): Remove static.
19610 (aff_combination_const_p): New interface.
19611 (aff_combination_singleton_var_p): New interfaces.
19613 2017-05-11 Richard Biener <rguenther@suse.de>
19615 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
19616 Skip unreachable blocks and destinations.
19617 (eliminate): Move stmt removal and fixup ...
19618 (fini_eliminate): ... here. Skip inserted exprs.
19619 (pass_pre::execute): Move fini_pre after fini_eliminate.
19620 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
19621 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
19622 PRE to get rid of dead code that has invalid SSA form and
19623 split critical edges again.
19625 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19627 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19629 2017-05-11 Richard Biener <rguenther@suse.de>
19631 * passes.c (execute_function_todo): Verify loops if they are
19632 said to be up-to-date.
19633 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
19634 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
19636 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
19639 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
19640 handle calling assemble_external ourself.
19643 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
19644 modes with zero size. Enhance comment.
19646 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19648 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
19649 built-ins for vec_xl and vec_xst with short and char pointer
19652 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
19654 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
19655 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
19656 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
19657 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
19658 (_mm_maskz_min_round_ss): New intrinsics.
19659 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
19660 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
19661 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
19662 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
19663 (__builtin_ia32_minss_mask_round): New builtins.
19664 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19665 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
19666 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
19668 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
19669 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
19671 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
19674 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
19676 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
19677 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
19678 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
19679 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
19680 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
19681 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
19682 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
19683 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19684 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
19685 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
19686 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
19687 (__builtin_ia32_mulss_mask_round): New builtins.
19688 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19689 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
19690 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
19692 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
19693 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
19695 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
19698 2017-05-10 Julia Koval <julia.koval@intel.com>
19700 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
19701 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
19702 (_mm256_setr_m128i): New intrinsics.
19704 2017-05-10 Julia Koval <julia.koval@intel.com>
19706 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
19707 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
19708 (_mm_maskz_rcp14_ss): New intrinsics.
19709 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
19710 (__builtin_ia32_rcp14ss_mask): New builtins.
19711 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
19713 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
19715 PR tree-optimization/51513
19716 * tree-cfg.c (gimple_seq_unreachable_p): New function.
19717 (assert_unreachable_fallthru_edge_p): Use it.
19718 (group_case_labels_stmt): Likewise.
19719 * tree-cfg.h: Prototype it.
19720 * stmt.c: Include cfghooks.h and tree-cfg.h.
19721 (emit_case_dispatch_table) <gap_label>: New local variable.
19722 Use it to fill dispatch table gaps.
19723 Test for default_label before updating probabilities.
19724 (expand_case) <default_label>: Remove unneeded initialization.
19725 Test for unreachable default case statement and remove its edge.
19726 Set default_label accordingly.
19727 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
19729 2017-05-10 Carl Love <cel@us.ibm.com>
19731 * config/rs6000/rs6000-c: Add support for built-in functions
19732 vector signed char vec_neg (vector signed char)
19733 vector signed short int vec_neg (vector short int)
19734 vector signed int vec_neg (vector signed int)
19735 vector signed long long vec_neg (vector signed long long)
19736 vector float vec_neg (vector float)
19737 vector double vec_neg (vector double)
19738 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
19740 * config/rs6000/altivec.h: Add define for vec_neg
19741 * doc/extend.texi: Update the built-in documentation for the
19742 new built-in functions.
19744 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19746 PR tree-optimization/77644
19747 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
19749 2017-05-10 Nathan Sidwell <nathan@acm.org>
19751 * dumpfile.h (TDI_lang_all): New.
19752 (TDF_KIND): New. Renumber others
19753 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
19755 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
19757 (get_dump_file_name): Adjust suffix generation.
19758 (dump_enable_all): Use TDF_KIND.
19759 * doc/invoke.texi (-fdump-lang-all): Document.
19761 * dumpfile.h: Tabify.
19763 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
19766 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
19767 Move member access before delete.
19769 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
19771 * tree-inline.c (expand_call_inline): Split block at stmt
19774 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
19777 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
19778 are now unused after splitting mov{sf,sd}_hardfloat.
19779 (f32_lr2): Likewise.
19780 (f32_lm): Likewise.
19781 (f32_lm2): Likewise.
19782 (f32_li): Likewise.
19783 (f32_li2): Likewise.
19784 (f32_lv): Likewise.
19785 (f32_sr): Likewise.
19786 (f32_sr2): Likewise.
19787 (f32_sm): Likewise.
19788 (f32_sm2): Likewise.
19789 (f32_si): Likewise.
19790 (f32_si2): Likewise.
19791 (f32_sv): Likewise.
19792 (f32_dm): Likewise.
19793 (f32_vsx): Likewise.
19794 (f32_av): Likewise.
19795 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
19796 For movsf, order stores so the VSX stores occur before the GPR
19797 store which encourages the register allocator to use a traditional
19798 FPR instead of a GPR. For movsd, order the stores so that the GPR
19799 store comes before the VSX stores to allow the power6 to work.
19800 This is due to the power6 not having a 32-bit integer store
19801 instruction from a FPR.
19802 (movsf_hardfloat): Likewise.
19803 (movsd_hardfloat): Likewise.
19805 2017-05-09 Martin Sebor <msebor@redhat.com>
19807 PR translation/80280
19808 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
19811 PR translation/80280
19812 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
19813 data member added in r247778.
19814 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
19816 2017-05-09 Nathan Sidwell <nathan@acm.org>
19818 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
19820 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
19823 2017-05-09 Marek Polacek <polacek@redhat.com>
19825 * doc/invoke.texi: Fix typo.
19827 2017-05-09 Richard Biener <rguenther@suse.de>
19829 * tree-vrp.c (vrp_val_is_max): Adjust comment.
19830 (vrp_val_is_min): Likewise.
19831 (set_value_range_to_value): Likewise.
19832 (set_value_range_to_nonnegative): Likewise.
19833 (gimple_assign_nonzero_p): Likewise.
19834 (gimple_stmt_nonzero_p): Likewise.
19835 (vrp_int_const_binop): Likewise. Remove unreachable case.
19836 (adjust_range_with_scev): Adjust comments.
19837 (compare_range_with_value): Likewise.
19838 (extract_range_from_phi_node): Likewise.
19839 (test_for_singularity): Likewise.
19841 2017-05-09 Richard Biener <rguenther@suse.de>
19843 * tree-vrp.c (get_single_symbol): Add assert that we don't
19844 get overflowed constants as invariant part.
19845 (compare_values_warnv): Add comment before the TREE_NO_WARNING
19846 checks. Use wi::cmp instead of recursing for integer constants.
19847 (compare_values): Just ignore whether we assumed undefined
19848 overflow instead of failing the compare.
19849 (extract_range_for_var_from_comparison_expr): Add comment before the
19850 TREE_NO_WARNING sets.
19851 (test_for_singularity): Likewise.
19852 (extract_range_from_comparison): Do not disable optimization
19853 when we assumed undefined overflow.
19854 (extract_range_basic): Remove init of unused var.
19856 2017-05-09 Richard Biener <rguenther@suse.de>
19858 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
19859 (extract_range_from_multiplicative_op_1): Adjust.
19860 (extract_range_from_binary_expr_1): Use int_const_binop.
19862 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
19865 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
19866 rs6000_store_data_bypass_p in seven define_bypass directives and
19867 in several comments.
19868 * config/rs6000/rs6000-protos.h: Add prototype for
19869 rs6000_store_data_bypass_p function.
19870 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
19871 function implements slightly different (rs6000-specific) semantics
19872 than store_data_bypass_p, returning false rather than aborting
19873 with assertion error when arguments do not satisfy the
19874 requirements of store data bypass.
19875 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
19876 rs6000_store_data_bypass_p.
19878 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
19880 * config/xtensa/xtensa-protos.h
19881 (xtensa_initial_elimination_offset): New declaration.
19882 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
19883 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
19884 macro definition, add case for FRAME_POINTER_REGNUM when
19885 FRAME_GROWS_DOWNWARD.
19886 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
19887 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
19888 xtensa_initial_elimination_offset.
19890 2017-05-08 Nathan Sidwell <nathan@acm.org>
19892 * doc/invoke.texi: Alphabetize -fdump options.
19894 2017-05-08 Martin Sebor <msebor@redhat.com>
19896 PR translation/80280
19897 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
19899 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
19901 * target.def (compute_frame_layout): New optional target hook.
19902 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
19903 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
19904 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
19906 * reload1.c (verify_initial_elim_offsets): Likewise.
19907 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
19908 (use_simple_return_p): Call arm_compute_frame_layout if needed.
19909 (arm_get_frame_offsets): Split up into this ...
19910 (arm_compute_frame_layout): ... and this function.
19912 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
19914 * config/aarch64/constraints.md (Usa): New constraint.
19915 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
19917 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
19919 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
19920 with_multilib_list after it has been checked.
19922 2017-05-08 Richard Biener <rguenther@suse.de>
19924 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
19925 (bitmap_set_subtract_values): Likewise.
19927 2017-05-08 Richard Biener <rguenther@suse.de>
19929 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
19930 (gimple_assign_nonzero): ... this and remove strict_overflow_p
19932 (gimple_stmt_nonzero_warnv_p): Rename to ...
19933 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
19935 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
19936 (extract_range_basic): Adjust, do not disable propagation on
19937 strict overflow sensitive simplification.
19938 (vrp_visit_cond_stmt): Likewise.
19940 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
19942 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
19943 body size unconditionally.
19945 2017-05-07 Jeff Law <law@redhat.com>
19948 2017-05-06 Jeff Law <law@redhat.com>
19949 PR tree-optimization/78496
19950 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
19953 PR tree-optimization/78496
19954 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
19955 (simplify_stmt_using_ranges): Call it.
19956 (vrp_dom_walker::before_dom_children): Extract equivalences
19957 from an ASSERT_EXPR with an equality comparison against a
19960 2017-05-06 Jeff Law <law@redhat.com>
19962 PR tree-optimization/78496
19963 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
19966 PR tree-optimization/78496
19967 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
19968 (simplify_stmt_using_ranges): Call it.
19969 (vrp_dom_walker::before_dom_children): Extract equivalences
19970 from an ASSERT_EXPR with an equality comparison against a
19973 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
19975 * lra-constraints.c (lra_copy_reg_equiv): New function.
19976 (split_reg): Use it to copy equivalence information from the
19977 original register to the spill register.
19979 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
19981 PR rtl-optimization/75964
19982 * simplify-rtx.c (simplify_const_relational_operation): Remove
19983 invalid handling of comparisons of integer ABS.
19985 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
19987 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
19988 initialize to zero.
19989 (init_regs): Remove declaration.
19990 (function_arg_advance_32): Initialize error_p as boolean variable.
19992 2017-05-05 Nathan Sidwell <nathan@acm.org>
19994 * store-motion.c (remove_reachable_equiv_notes): Reformat long
19995 lines. Use for (;;).
19997 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19999 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
20000 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
20001 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
20002 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
20003 VF=2 that require versioning.
20005 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20007 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
20010 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20012 * diagnostic.h (diagnostic_override_option_index): Convert from
20013 macro to inline function.
20015 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20017 * diagnostic.c (last_module_changed_p): New function.
20018 (set_last_module): New function.
20019 (diagnostic_report_current_module): Convert macro usage to
20020 the above functions.
20021 * diagnostic.h (diagnostic_context::last_module): Strengthen
20022 from const line_map * to const line_map_ordinary *.
20023 (diagnostic_last_module_changed): Delete macro.
20024 (diagnostic_set_last_module): Delete macro.
20026 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20028 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
20029 with diagnostic_report_diagnostic.
20030 (diagnostic_n_impl_richloc): Likewise.
20031 * diagnostic.h (report_diagnostic): Delete macro.
20032 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
20033 with diagnostic_report_diagnostic.
20034 * substring-locations.c (format_warning_va): Likewise.
20036 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20038 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
20039 save/restor of format_spec. Move option-printing code to...
20040 (print_option_information): ...this new function, and
20041 reimplement by simply printing to the pretty_printer,
20042 rather than appending to the format string.
20044 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20046 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
20047 handling logic into...
20048 (update_effective_level_from_pragmas): ...this new function.
20050 2017-05-04 Andrew Waterman <andrew@sifive.com>
20052 * config/riscv/riscv.opt (mstrict-align): New option.
20053 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
20054 (SLOW_UNALIGNED_ACCESS): Define.
20055 (riscv_slow_unaligned_access): Declare.
20056 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
20058 (riscv_slow_unaligned_access): New variable.
20059 (rocket_tune_info): Set slow_unaligned_access to true.
20060 (optimize_size_tune_info): Set slow_unaligned_access to false.
20061 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
20062 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
20063 (riscv_option_override): Set riscv_slow_unaligned_access.
20064 * doc/invoke.texi: Add -mstrict-align to RISC-V.
20066 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
20068 * config/riscv/riscv.md: Unify indentation.
20070 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20075 * config/rs6000/rs6000.md (u code attribute): Add FIX and
20077 (extendsi<mode>2): Add support for doing sign extension via
20078 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
20079 don't have ISA 3.0 instructions.
20080 (extendsi<mode>2 splitter): Likewise.
20081 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
20082 generate the normal insns since SImode can now go in vector
20083 registers. Disallow the special UNSPECs needed for previous
20084 machines to hide SImode being used. Add new insns
20085 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
20086 (fix_trunc<mode>si2_stfiwx): Likewise.
20087 (fix_trunc<mode>si2_internal): Likewise.
20088 (fixuns_trunc<mode>si2): Likewise.
20089 (fixuns_trunc<mode>si2_stfiwx): Likewise.
20090 (fctiw<u>z_<mode>_smallint): Likewise.
20091 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
20092 of floating point to 32-bit integer from doing a direct move to
20093 the GPR registers to do a store.
20094 (fctiwz_<mode>): Break long line.
20096 2017-05-05 Bin Cheng <bin.cheng@arm.com>
20098 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
20099 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
20100 (addr_list, addr_offset_valid_p): New.
20101 (split_address_groups): Check offset validity with above function.
20102 (gt-tree-ssa-loop-ivopts.h): Include header file.
20104 2017-05-05 Nathan Sidwell <nathan@acm.org>
20106 * config.gcc (arm*-*-*): Add missing 'fi'.
20108 2017-05-05 Steve Ellcey <sellcey@cavium.com>
20110 * doc/invoke.texi (-fopt-info): Explicitly say order of options
20111 included in -fopt-info does not matter.
20112 * doc/optinfo.texi (-fopt-info): Fix description of default
20113 behavour. Explicitly say order of options included in -fopt-info
20116 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
20118 * config.gcc: Allow combinations of aprofile and rmprofile values for
20119 --with-multilib-list.
20120 * config/arm/t-multilib: New file.
20121 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
20122 variables. Remove setting of ISA and floating-point ABI in
20123 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
20124 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
20125 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
20126 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
20128 * config/arm/t-rmprofile: Likewise except for the matches changes.
20129 * doc/install.texi (--with-multilib-list): Document the combination of
20130 aprofile and rmprofile values and warn about pitfalls in doing that.
20132 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
20134 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
20135 (movdi_aarch64): Likewise.
20137 2017-05-05 Jakub Jelinek <jakub@redhat.com>
20139 PR tree-optimization/80632
20140 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
20142 (build_arrays): Initialize it for virtual phis.
20143 (fix_phi_nodes): Use it for virtual phis.
20145 PR tree-optimization/80558
20146 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
20147 [x, y] op z into [x op, y op z] for op & or | if conditions
20150 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
20151 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20154 * config/arm/arm.md (use_literal_pool): Remove.
20155 (64-bit immediate split): No longer takes cost into consideration
20156 if arm_disable_literal_pool is enabled.
20157 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
20158 used when arm_disable_literal_pool is enabled.
20159 (arm_max_const_double_inline_cost): Remove use of
20160 arm_disable_literal_pool.
20161 (push_minipool_fix): Add assert.
20162 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
20163 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
20164 (no_literal_pool_sf_immediate): New.
20166 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20168 PR tree-optimization/80613
20169 * tree-ssa-dce.c (propagate_necessity): Remove cases for
20170 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
20172 2017-05-05 Richard Biener <rguenther@suse.de>
20174 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
20176 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
20178 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
20179 of this flag from insn conditions due to removal from r247495.
20181 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
20183 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
20185 (arm_early_store_addr_dep_ptr): Likewise.
20186 * config/arm/aarch-common-protos.h
20187 (arm_early_load_addr_dep_ptr): Add prototype.
20188 (arm_early_store_addr_dep_ptr): Likewise.
20189 * config/arm/cortex-a53.md: Add new bypasses.
20191 2017-05-05 Jakub Jelinek <jakub@redhat.com>
20193 * tree.c (next_type_uid): Change type to unsigned.
20194 (type_hash_canon): Decrement back next_type_uid if
20195 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
20196 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
20199 2017-05-04 Martin Sebor <msebor@redhat.com>
20201 * builtins.c: Fix a trivial typo in a comment.
20203 PR middle-end/79234
20204 * builtins.c (check_sizes): Adjust to handle reading past the end.
20205 Avoid printing excessive upper bound of ranges. Use %E to print
20206 tree nodes instead of converting them to %wu.
20207 (expand_builtin_memchr): New function.
20208 (compute_dest_size): Rename...
20209 (compute_objsize): ...to this.
20210 (expand_builtin_memcpy): Adjust.
20211 (expand_builtin_mempcpy): Adjust.
20212 (expand_builtin_strcat): Adjust.
20213 (expand_builtin_strcpy): Adjust.
20214 (check_strncat_sizes): Adjust.
20215 (expand_builtin_strncat): Adjust.
20216 (expand_builtin_strncpy): Adjust and simplify.
20217 (expand_builtin_memset): Adjust.
20218 (expand_builtin_bzero): Adjust.
20219 (expand_builtin_memcmp): Adjust.
20220 (expand_builtin): Handle memcmp.
20221 (maybe_emit_chk_warning): Check strncat just once.
20223 2017-05-04 Martin Sebor <msebor@redhat.com>
20225 PR preprocessor/79214
20226 PR middle-end/79222
20227 PR middle-end/79223
20228 * builtins.c (check_sizes): Add inlining context and issue
20229 warnings even when -Wno-system-headers is set.
20230 (check_strncat_sizes): Same.
20231 (expand_builtin_strncat): Same.
20232 (expand_builtin_memmove): New function.
20233 (expand_builtin_stpncpy): Same.
20234 (expand_builtin): Handle memmove and stpncpy.
20236 2017-05-04 Bin Cheng <bin.cheng@arm.com>
20238 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
20239 which is not used any more.
20241 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20243 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
20245 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20247 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
20248 (cortexa53_tunings): Likewise.
20249 (cortexa57_tunings): Likewise.
20250 (cortexa72_tunings): Likewise.
20251 (cortexa73_tunings): Likewise.
20253 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20255 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
20256 Set loop alignment to 8.
20258 2017-05-04 Martin Sebor <msebor@redhat.com>
20260 PR translation/80280
20261 * builtins.c (expand_builtin_object_size): Add missing quoting to
20262 %D and like directives.
20263 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
20264 (hsa_type_for_tree_type): Same.
20265 (verify_function_arguments): Same.
20266 * symtab.c (symbol_table::change_decl_assembler_name): Same.
20267 * varasm.c (get_section): Same.
20270 2017-05-04 Martin Sebor <msebor@redhat.com>
20272 PR translation/80280
20273 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
20275 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20277 * config/aarch64/aarch64.c (generic_addrcost_table):
20278 Change HI/TI mode setting.
20280 2017-05-04 Martin Jambor <mjambor@suse.cz>
20282 PR tree-optimization/80622
20283 * tree-sra.c (comes_initialized_p): New function.
20284 (build_accesses_from_assign): Only set write lazily when
20285 comes_initialized_p is false.
20286 (analyze_access_subtree): Use comes_initialized_p.
20287 (propagate_subaccesses_across_link): Assert !comes_initialized_p
20288 instead of testing for PARM_DECL.
20290 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20292 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
20293 constraint on operand 0 to allow more general addressing modes.
20294 Adjust output template.
20295 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
20297 * config/aarch64/aarch64-protos.h
20298 (aarch64_address_valid_for_prefetch_p): Declare prototype.
20299 * config/aarch64/constraints.md (Dp): New address constraint.
20300 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
20303 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
20305 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
20306 update use of estimate_ipcp_clone_size_and_time.
20307 (estimate_local_effects): Update use of
20308 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
20309 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
20310 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
20311 Return nonspecialized time.
20313 2017-05-04 Richard Biener <rguenther@suse.de>
20315 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
20316 for the last VUSE which def dominates the PHI. Directly call
20318 (get_continuation_for_phi_1): Remove.
20320 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
20322 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
20323 to explain the use of truncating division. Cap the number of
20324 iterations to the maximum given by nb_iterations_upper_bound,
20327 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20329 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
20330 * configure: Regenerate.
20331 * config.in: Regenerate.
20332 * config/i386/driver-mingw32.c: new file.
20333 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
20334 * config.host: Link driver-mingw32.o on MinGW host.
20335 * doc/install.texi: Document new --enable-mingw-wildcard configure
20338 2017-05-04 Marek Polacek <polacek@redhat.com>
20340 PR tree-optimization/80612
20341 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
20343 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20344 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
20346 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
20347 (movt splitter): Likewise.
20348 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
20349 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
20350 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
20351 block for Thumb-1 with MOVT.
20352 (thumb2_legitimate_address_p): Move code block ...
20353 (can_avoid_literal_pool_for_label_p): ... into this new function.
20354 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
20356 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
20357 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
20358 "M-profile targets with the MOVT instruction".
20360 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20362 * config/arm/arm-builtins.c (arm_init_builtins): Rename
20363 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
20364 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
20366 2017-05-04 Martin Liska <mliska@suse.cz>
20368 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
20369 variable cond_code.
20371 2017-05-04 Richard Biener <rguenther@suse.de>
20373 * tree.c (array_at_struct_end_p): Handle arrays at struct
20374 end with flexarrays more conservatively. Refactor and treat
20375 arrays of arrays or aggregates more strict. Fix
20376 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
20377 * tree.c (array_at_struct_end_p): Adjust prototype.
20378 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
20379 * gimple-fold.c (get_range_strlen): Likewise.
20380 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
20382 2017-05-04 Richard Biener <rguenther@suse.de>
20384 PR tree-optimization/31130
20385 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
20387 (supports_overflow_infinity): Likewise.
20388 (is_negative_overflow_infinity): Likewise.
20389 (is_positive_overflow_infinity): Likewise.
20390 (is_overflow_infinity): Likewise.
20391 (stmt_overflow_infinity): Likewise.
20392 (overflow_infinity_range_p): Likewise.
20393 (usable_range_p): Remove as always returning true.
20394 (make_overflow_infinity): Remove.
20395 (negative_overflow_infinity): Likewise.
20396 (positive_overflow_infinity): Likewise.
20397 (avoid_overflow_infinity): Likewise.
20398 (set_value_range): Adjust accordingly.
20399 (set_value_range_to_nonnegative): Likewise, remove now unused
20400 overflow_infinity arg.
20401 (vrp_operand_equal_p): Adjust.
20402 (update_value_range): Likewise.
20403 (range_int_cst_singleton_p): Likewise.
20404 (operand_less_p): Likewise.
20405 (compare_values_warnv): Likewise.
20406 (extract_range_for_var_from_comparison_expr): Likewise.
20407 (vrp_int_const_binop): Likewise.
20408 (zero_nonzero_bits_from_vr): Likewise.
20409 (extract_range_from_multiplicative_op_1): Likewise.
20410 (extract_range_from_binary_expr_1): Likewise.
20411 (extract_range_from_unary_expr): Likewise.
20412 (extract_range_from_comparison): Likewise.
20413 (extract_range_basic): Likewise.
20414 (adjust_range_with_scev): Likewise.
20415 (compare_ranges): Likewise.
20416 (compare_range_with_value): Likewise.
20417 (dump_value_range): Likewise.
20418 (test_for_singularity): Likewise, remove strict_overflow_p parameter
20420 (simplify_cond_using_ranges): Adjust.
20422 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
20424 * brig-builtins.def: Added a builtin for class_f64.
20425 * builtin-types.def: Added a builtin type needed by class_f64.
20427 2017-05-03 Jason Merrill <jason@redhat.com>
20429 * timevar.def: Add TV_CONSTEXPR.
20431 2017-05-03 David Malcolm <dmalcolm@redhat.com>
20433 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
20435 2017-05-03 Martin Jambor <mjambor@suse.cz>
20437 * ipa-prop.c (ipa_update_after_lto_read): Removed.
20438 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
20439 * ipa-cp.c (ipcp_propagate_stage): Do not call
20440 ipa_update_after_lto_read.
20441 * ipa-inline.c (ipa_inline): Likewise.
20443 2017-05-03 Martin Jambor <mjambor@suse.cz>
20445 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
20446 tag. Added a default constructor and a destructor.
20447 (ipa_edge_args_sum_t): New class;
20448 (ipa_edge_args_sum): Declare.
20449 (ipa_edge_args_vector): Remove declaration.
20450 (IPA_EDGE_REF): Use ipa_edge_args_sum.
20451 (ipa_free_edge_args_substructures): Remove declaration.
20452 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
20453 (ipa_edge_args_info_available_for_edge_p): Likewise.
20454 * ipa-prop.c (ipa_edge_args_vector): Removed.
20455 (edge_removal_hook_holder): Likewise.
20456 (edge_duplication_hook_holder): Likewise.
20457 (ipa_edge_args_sum): New variable.
20458 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
20459 ipa_edge_args_vector.
20460 (ipa_free_edge_args_substructures): Likewise.
20461 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
20462 ipa_edge_args_vector.
20463 (ipa_edge_removal_hook): Turned into method
20464 ipa_edge_args_sum_t::remove.
20465 (ipa_edge_duplication_hook): Turned into method
20466 ipa_edge_args_sum_t::duplicate.
20467 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
20468 registering edge hooks.
20469 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
20470 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
20471 ipa_edge_args_sum instead of ipa_edge_args_vector.
20472 * ipa-profile.c (ipa_profile): Likewise.
20474 2017-05-03 Martin Jambor <mjambor@suse.cz>
20476 * symbol-summary.h (function_summary): New method exists.
20477 (function_summary::symtab_removal): Deallocate through release.
20478 (call_summary): New class.
20479 (gt_ggc_mx): New overload.
20480 (gt_pch_nx): Likewise.
20481 (gt_pch_nx): Likewise.
20483 2017-05-03 Jeff Law <law@redhat.com>
20485 PR tree-optimization/78496
20486 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
20487 from simplify_cond_using_ranges. Split off code to walk
20488 backwards through casts into ...
20489 (simplify_cond_using_ranges_2): New function.
20490 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
20491 (execute_vrp): After identifying jump threads, call
20492 simplify_cond_using_ranges_2.
20494 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
20497 * ipa-inline.h (inline_summary): Add ctor.
20498 (create_ggc): Do not use ggc_cleared_alloc.
20500 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
20501 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20503 * gcc.c (handle_braces): Support escaping in switch matching
20505 * doc/invoke.texi (Spec Files): Document it.
20506 Remove superfluous @code markup in items.
20508 2017-05-03 David Malcolm <dmalcolm@redhat.com>
20510 * diagnostic-show-locus.c (struct column_range): New struct.
20511 (get_affected_columns): New function.
20512 (get_printed_columns): New function.
20513 (struct correction): New struct.
20514 (correction::ensure_capacity): New function.
20515 (correction::ensure_terminated): New function.
20516 (struct line_corrections): New struct.
20517 (line_corrections::~line_corrections): New dtor.
20518 (line_corrections::add_hint): New function.
20519 (layout::print_trailing_fixits): Reimplement in terms of the new
20521 (selftest::test_overlapped_fixit_printing): New function.
20522 (selftest::diagnostic_show_locus_c_tests): Call it.
20524 2017-05-03 Nathan Sidwell <nathan@acm.org>
20526 Canonicalize canonical type hashing
20527 * tree.h (type_hash_canon_hash): Declare.
20528 * tree.c (type_hash_list, attribute_hash_list): Move into
20529 type_hash_canon_hash.
20530 (build_type_attribute_qual_variant): Break out hash code calc into
20531 type_hash_canon_hash.
20532 (type_hash_canon_hash): New. Generic type hash computation.
20533 (build_range_type_1, build_array_type_1, build_function_type,
20534 build_method_type_directly, build_offset_type, build_complex_type,
20535 make_vector_type): Call it.
20537 2017-05-03 Richard Biener <rguenther@suse.de>
20539 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
20540 When all DRs have unknown misaligned do not always peel
20541 when there is a store but apply the same costing model as if
20542 there were only loads.
20544 2017-05-03 Richard Biener <rguenther@suse.de>
20547 PR tree-optimization/80492
20548 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
20549 compare_base_decls returning dont-know properly.
20551 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
20553 * config/arm/iterators.md (CCSI): New mode iterator.
20554 (arch): New mode attribute.
20555 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
20556 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
20557 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
20558 code iterator for success result mode.
20559 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
20560 the corresponding new insn generators.
20562 2017-05-03 Bin Cheng <bin.cheng@arm.com>
20565 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20566 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
20568 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
20570 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
20571 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
20572 (DDR_A): Wrap DDR argument in brackets.
20573 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
20574 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
20575 (DDR_REVERSED_P): Likewise.
20577 2017-05-03 Jakub Jelinek <jakub@redhat.com>
20579 PR tree-optimization/79472
20580 * tree-switch-conversion.c (struct switch_conv_info): Add
20581 contiguous_range and default_case_nonstandard fields.
20582 (collect_switch_conv_info): Compute contiguous_range and
20583 default_case_nonstandard fields, don't clear final_bb if
20584 contiguous_range and only the default case doesn't have the required
20586 (check_all_empty_except_final): Set default_case_nonstandard instead
20587 of failing if contiguous_range and the default case doesn't have empty
20589 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
20590 and only the default case doesn't have the required constants. Skip
20592 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
20593 if default_case_nonstandard.
20594 (build_constructors): Build constant 1 just once. Assert that default
20595 values aren't inserted in between cases if contiguous_range. Skip
20597 (build_arrays): Skip virtual phis.
20598 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
20599 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
20600 Handle virtual phis.
20601 (gen_inbound_check): Handle default_case_nonstandard case.
20602 (process_switch): Adjust check_final_bb caller. Call
20603 gather_default_values with the first non-default case instead of
20604 default case if default_case_nonstandard.
20606 2017-05-02 Nathan Sidwell <nathan@acm.org>
20608 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
20609 check. Fix formatting.
20611 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
20613 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
20614 errors when comparing specialized and unspecialized times.
20616 2017-05-02 David Malcolm <dmalcolm@redhat.com>
20618 * diagnostic-show-locus.c
20619 (layout::should_print_annotation_line_p): Make private.
20620 (layout::print_annotation_line): Make private.
20621 (layout::annotation_line_showed_range_p): Make private.
20622 (layout::show_ruler): Make private.
20623 (layout::print_source_line): Make private. Pass in line and
20624 line_width, rather than calling location_get_source_line. Drop
20626 (layout::print_leading_fixits): New method.
20627 (layout::print_any_fixits): Rename to...
20628 (layout::print_trailing_fixits): ...this, and make private.
20629 Don't print newline fixits.
20630 (diagnostic_show_locus): Move logic for printing one row into...
20631 (layout::print_line): ...this new function. Move the
20632 location_get_source_line call and error-handling from
20633 print_source_line to here. Call print_leading_fixits, and rename
20634 print_any_fixits to print_trailing_fixits.
20635 (selftest::test_fixit_insert_containing_newline): Update now that
20636 newlines are partially supported.
20637 (selftest::test_fixit_insert_containing_newline_2): New test.
20638 (selftest::test_fixit_replace_containing_newline): Update comments.
20639 (selftest::diagnostic_show_locus_c_tests): Call the new test.
20640 * edit-context.c (class added_line): New class.
20641 (class edited_line): Describe newline handling in comment.
20642 (edited_line::actually_edited_p): New method.
20643 (edited_line::print_content): Delete redundant decl.
20644 (edited_line::m_predecessors): New field.
20645 (edited_file::print_content): Call edited_line::print_content.
20646 (edited_file::print_diff): Update to support newlines.
20647 (edited_file::print_diff_hunk): Likewise.
20648 (edited_file::print_run_of_changed_lines): New function.
20649 (edited_file::print_diff_line): Convert to...
20650 (print_diff_line): ...this.
20651 (edited_file::get_effective_line_count): New function.
20652 (edited_line::edited_line): Initialize new field m_predecessors.
20653 (edited_line::~edited_line): Clean up m_predecessors.
20654 (edited_line::apply_fixit): Handle newlines.
20655 (edited_line::get_effective_line_count): New function.
20656 (edited_line::print_content): New function.
20657 (edited_line::print_diff_lines): New function.
20658 (selftest::test_applying_fixits_insert_containing_newline): New
20660 (selftest::test_applying_fixits_replace_containing_newline): New
20662 (selftest::insert_line): New function.
20663 (selftest::test_applying_fixits_multiple_lines): Add example of
20665 (selftest::edit_context_c_tests): Call the new tests.
20667 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20669 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
20670 parameter cand. Update dump information.
20671 (get_computation_cost): Update uses.
20673 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20675 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
20676 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
20677 (get_computation_at, rewrite_use_address): Update use of
20678 get_computation_aff.
20680 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20682 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
20683 (get_computation): Delete.
20684 (get_computation_cost): Implement like get_computation_cost_at.
20685 Use get_computation_at.
20686 (get_computation_cost_at): Delete.
20687 (rewrite_use_nonlinear_expr): Use get_computation_at.
20688 (rewrite_use_compare, remove_unused_ivs): Ditto.
20690 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20692 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
20694 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20696 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
20697 (ivopts_global_cost_for_size): Rename parameter and update uses.
20698 (iv_ca_recount_cost): Update uses.
20699 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
20700 candidates seperately in n_invs and n_cands.
20701 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
20703 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20705 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
20706 (find_inv_vars_cb): New.
20707 (find_depends): Renamed to ...
20708 (find_inv_vars): ... this.
20709 (add_candidate_1, force_var_cost): Call find_inv_vars.
20710 (split_address_cost, determine_group_iv_cost_cond): Ditto.
20712 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20714 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
20715 inv_vars. Add inv_exprs.
20716 (struct iv_cand): Rename depends_on to inv_vars.
20717 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
20718 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
20719 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
20720 (dump_cand): Dump inv_vars.
20721 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
20722 (record_invariant, find_depends, add_candidate_1): Ditto.
20723 (set_group_iv_cost, force_var_cost): Ditto.
20724 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
20725 (get_computation_cost_at, get_computation_cost): Ditto.
20726 (determine_group_iv_cost_generic): Ditto.
20727 (determine_group_iv_cost_address): Ditto.
20728 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
20729 (determine_group_iv_costs): Ditto.
20730 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
20731 (iv_ca_set_remove_invariants): Renamed to ...
20732 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
20733 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
20734 (iv_ca_set_add_invariants): Renamed to ...
20735 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
20736 (iv_ca_set_cp): Use iv_ca_set_add_invs.
20737 (iv_ca_has_deps): Support inv_vars and inv_exprs.
20738 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
20739 (create_new_ivs): Remove useless dump.
20741 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20743 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
20745 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
20746 (iv_ca_set_no_cp, create_new_iv): Ditto.
20748 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20750 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
20752 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
20754 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
20755 function tree_check2.
20757 2017-05-02 Martin Liska <mliska@suse.cz>
20759 * doc/gcov.texi: Add missing preposition.
20760 * gcov.c (function_info::function_info): Properly fill up
20761 all member variables.
20763 2017-05-02 Tamar Christina <tamar.christina@arm.com>
20765 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
20767 2017-05-02 Tamar Christina <tamar.christina@arm.com>
20769 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
20771 2017-05-02 Martin Liska <mliska@suse.cz>
20774 * lto-streamer-in.c (lto_read_tree_1): Remove
20775 LTO_STREAMER_DEBUG.
20776 * lto-streamer.c (struct tree_hash_entry): Likewise.
20777 (struct tree_entry_hasher): Likewise.
20778 (tree_entry_hasher::hash): Likewise.
20779 (tree_entry_hasher::equal): Likewise.
20780 (lto_streamer_init): Likewise.
20781 (lto_orig_address_map): Likewise.
20782 (lto_orig_address_get): Likewise.
20783 (lto_orig_address_remove): Likewise.
20784 * lto-streamer.h: Likewise.
20785 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
20786 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
20788 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
20790 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
20791 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
20792 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
20793 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
20794 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
20795 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
20796 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
20797 (mm_maskz_sub_ss): New intrinsics.
20798 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20799 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
20800 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
20801 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
20802 (__builtin_ia32_subss_mask_round): New builtins.
20803 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20804 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20805 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
20807 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
20808 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
20810 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
20813 2017-05-02 Martin Jambor <mjambor@suse.cz>
20815 PR tree-optimization/78687
20816 * tree-sra.c (access): New field parent.
20817 (process_subtree_disqualification): New function.
20818 (disqualify_candidate): Call it.
20819 (build_accesses_from_assign): Reset write flag if creating an
20821 (build_access_subtree): Fill in parent field and also prpagate
20822 down grp_write flag.
20823 (create_artificial_child_access): New parameter set_grp_write, set
20824 grp_write to its value.
20825 (propagate_subaccesses_across_link): Also propagate grp_write flag
20827 (propagate_all_subaccesses): Push the closest parent back to work
20828 queue if add_access_to_work_queue returned true.
20830 2017-05-02 Richard Biener <rguenther@suse.de>
20832 * common.opt (fstrict-overflow): Alias negative to fwrapv.
20833 * doc/invoke.texi (fstrict-overflow): Remove all traces of
20834 -fstrict-overflow documentation.
20835 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
20836 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
20837 flag_strict_overflow.
20838 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
20839 * lto-opts.c (lto_write_options): Do not stream it.
20840 * lto-wrapper.c (merge_and_complain): Do not handle it.
20841 * opts.c (default_options_table): Do not set -fstrict-overflow.
20842 (finish_options): Likewise do not clear it when sanitizing.
20843 * simplify-rtx.c (simplify_const_relational_operation): Do not
20844 test flag_strict_overflow.
20846 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
20848 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
20849 using enabled attribute.
20850 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
20851 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
20852 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
20853 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
20854 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
20855 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
20856 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
20857 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
20858 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
20859 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
20861 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
20863 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
20865 2017-05-02 Richard Biener <rguenther@suse.de>
20867 PR tree-optimization/80591
20869 2017-04-10 Richard Biener <rguenther@suse.de>
20871 * tree-ssa-structalias.c (find_func_aliases): Properly handle
20874 2017-05-02 Richard Biener <rguenther@suse.de>
20876 PR tree-optimization/80549
20877 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
20878 (cleanup_tree_cfg_noloop): Create forwarders to known loop
20879 headers if they do not have a preheader.
20881 2017-05-02 Martin Liska <mliska@suse.cz>
20884 * common.opt: Fix typo.
20885 * doc/invoke.texi: Likewise.
20887 2017-05-01 Jan Beulich <jbeulich@suse.com>
20889 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
20890 swapping, add (x,x,m,x,n) alternative.
20892 2017-05-01 Nathan Sidwell <nathan@acm.org>
20894 * calls.c (combine_pending_stack_adjustment_and_call): Remove
20895 unnecessary unadjusted_alignment check.
20897 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
20900 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
20902 * gimplify.c (gimplify_cilk_detach): New function.
20903 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
20904 * tree-core.h: Document EXPR_CILK_SPAWN.
20905 * tree.h (EXPR_CILK_SPAWN): Define.
20907 2017-05-01 David Malcolm <dmalcolm@redhat.com>
20909 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
20910 to use new fixit_hint representation, using the "replace" logic.
20911 (get_line_span_for_fixit_hint): Likewise.
20912 (layout::print_any_fixits): Likewise.
20913 (selftest::test_one_liner_many_fixits): Rename to...
20914 (selftest::test_one_liner_many_fixits_1): ...this, and update
20915 comment and expected output to reflect that the multiple fix-it
20916 hints are now consolidated into one insertion.
20917 (selftest::test_one_liner_many_fixits_2): New test.
20918 (selftest::test_diagnostic_show_locus_one_liner): Update for
20920 (selftest::test_fixit_consolidation): Update for fix-it API
20922 * diagnostic.c (print_parseable_fixits): Likewise.
20923 * edit-context.c (edited_line::m_line_events): Convert from
20924 auto_vec <line_event *> to auto_vec <line_event>.
20925 (class line_event): Convert from abstract base class to a concrete
20926 class, taking over the role of replace_event.
20927 (class insert_event): Delete.
20928 (class replace_event): Rename to class line_event. Convert to
20930 (edit_context::add_fixits): Reimplement.
20931 (edit_context::apply_insert): Delete.
20932 (edit_context::apply_replace): Rename to...
20933 (edit_context::apply_fixit): ...this. Convert to half-open range.
20934 (edited_file::apply_insert): Delete.
20935 (edited_file::apply_replace): Rename to...
20936 (edited_file::apply_fixit): ...this.
20937 (edited_line::~edited_line): Drop deletion of events.
20938 (edited_line::apply_insert): Delete.
20939 (edited_line::apply_replace): Rename to...
20940 (edited_line::apply_fixit): ...this. Convert to half-open range.
20941 Update for change to type of m_line_events.
20942 * edit-context.h (edit_context::apply_insert): Delete.
20943 (edit_context::apply_replace): Rename to...
20944 (edit_context::apply_fixit): ...this.
20946 2017-05-01 Martin Sebor <msebor@redhat.com>
20948 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
20951 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
20954 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
20955 __get_cpuid_max returns 0.
20956 (__get_cpuid_count): Ditto.
20958 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
20960 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
20961 replacement expression is another instance of one of its arguments.
20963 2017-05-01 Jakub Jelinek <jakub@redhat.com>
20966 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
20967 check for stack push/pop autoinc.
20968 * config/i386/i386.c (ix86_agi_dependent): Return false
20969 if the only reason why modified_in_p returned true is that
20970 addr is SP based and set_insn is a push or pop.
20972 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
20974 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
20977 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
20980 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
20981 (account_size_time): Use two predicates - exec_pred and
20983 (evaluate_conditions_for_known_args): Compute both clause and
20985 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
20986 (inline_summary_t::duplicate): Update.
20987 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
20989 (compute_inline_parameters): Likewise.
20990 (estimate_edge_size_and_time): Update caluclation of time.
20991 (estimate_node_size_and_time): Compute both time and nonspecialized
20993 (estimate_ipcp_clone_size_and_time): Update.
20994 (inline_merge_summary): Update.
20995 (do_estimate_edge_time): Update.
20996 (do_estimate_edge_size): Update.
20997 (do_estimate_edge_hints): Update.
20998 (inline_read_section, inline_write_summary): Stream both new predicates.
20999 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
21001 (compute_inlined_call_time): Cleanup.
21002 (big_speedup_p): Update.
21003 (edge_badness): Update.
21004 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
21005 (size_time_entry): Replace predicate by exec_predicate and
21006 nonconst_predicate.
21007 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
21008 (estimate_edge_time): Return also nonspec_time.
21009 (reset_edge_growth_cache): Update.
21011 2017-04-29 Jakub Jelinek <jakub@redhat.com>
21013 PR rtl-optimization/80491
21014 * ifcvt.c (noce_process_if_block): When looking for x setter
21015 with missing else_bb, don't check only the insn right before
21016 cond_earliest, but look for the last insn that x is modified in
21017 within the same bb.
21019 PR rtl-optimization/80491
21020 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
21022 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
21024 PR tree-optimization/80487
21025 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
21027 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21029 PR tree-optimization/79697
21030 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
21031 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
21032 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
21034 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
21035 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
21037 2017-04-28 Martin Sebor <msebor@redhat.com>
21039 PR tree-optimization/80523
21040 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
21041 (init_target_to_host_charmap, target_to_host, target_strtol10): New
21043 (maybe_warn, format_directive, parse_directive): Use new functions.
21044 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
21046 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
21048 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
21050 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
21052 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
21053 target_header_dir): Set correctly.
21054 * configure: Regenerated.
21055 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21056 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21057 instead of SYSTEM_HEADER_DIR.
21059 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
21061 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
21062 (estimate_local_effects): Likewise.
21063 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
21064 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
21065 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
21066 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
21067 do_estimate_edge_time, estimate_edge_time): Likewise.
21068 * ipa-inline-analysis.c (estimate_node_size_and_time,
21069 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
21070 (estimate_time_after_inlining): Remove.
21072 2017-04-28 Martin Liska <mliska@suse.cz>
21074 * doc/gcov.texi: Enhance documentation of gcov.
21076 2017-04-28 Martin Liska <mliska@suse.cz>
21078 * doc/gcov.texi: Sort options in alphabetic order.
21079 * doc/gcov-dump.texi: Likewise.
21080 * doc/gcov-tool.texi: Likewise.
21081 * gcov.c (print_usage): Likewise.
21082 * gcov-dump.c (print_usage): Likewise.
21083 * gcov-tool.c (print_merge_usage_message): Likewise.
21084 (print_rewrite_usage_message): Likewise.
21085 (print_overlap_usage_message): Likewise.
21087 2017-04-28 Martin Liska <mliska@suse.cz>
21089 PR gcov-profile/53915
21090 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
21092 2017-04-28 Martin Liska <mliska@suse.cz>
21094 PR gcov-profile/79891
21095 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
21096 is marked by compiler as living on a line.
21097 (get_cycles_count): Remove usage of the union.
21098 (output_intermediate_file): Likewise.
21099 (find_source): Fix GNU coding style.
21100 (accumulate_line_counts): Remove old non-all block mode.
21101 (output_lines): Remove usage of the union.
21102 * profile.c (output_location): Include all BBs, even if
21103 belonging to a same line (and file) as a previous BB.
21105 2017-04-28 Martin Liska <mliska@suse.cz>
21107 * gcov.c (process_args): Handle new argument 'w'.
21108 (read_graph_file): Assign ID to BBs.
21109 (output_branch_count): Display BB # if verbose flag is set.
21110 (output_lines): Likewise for arcs.
21111 (print_usage): Add '--verbose' option help.
21112 * doc/gcov.texi: Document --verbose (-w) option.
21114 2017-04-28 Martin Liska <mliska@suse.cz>
21116 * gcov.c (struct block_location_info): New struct.
21117 (process_file): Fill up the new structure.
21118 (read_graph_file): Replace usage of encoding by the newly added
21120 (add_line_counts): Likewise.
21121 (accumulate_line_counts): Remove usage of the union.
21122 (function_info::function_info): New function.
21123 (function_info::~function_info): Likewise.
21124 (process_file): Call delete instead of release_function.
21125 (release_function): Release the function.
21126 (release_structures): Call delete instead of release_function.
21127 (solve_flow_graph): Replace usage of num_blocks.
21128 (find_exception_blocks): Likewise.
21129 (output_lines): Fix GNU coding style.
21131 2017-04-28 Martin Liska <mliska@suse.cz>
21134 * coverage.c (coverage_remove_note_file): New function.
21135 * coverage.h: Declare the function.
21136 * toplev.c (finalize): Clean if an error has been seen.
21138 2017-04-28 Martin Liska <mliska@suse.cz>
21140 PR gcov-profile/80031
21141 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
21142 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
21143 * gcov.c (read_graph_file): Read just number of blocks.
21144 * profile.c (branch_prob): Do not stream 0 flags per a basic
21147 2017-04-28 Martin Liska <mliska@suse.cz>
21149 * gcov-dump.c (tag_*): Add new argument to declarations.
21150 (dump_gcov_file): Likewise.
21151 (tag_blocks): Add and use new argument depth.
21152 (tag_arcs): Likewise.
21153 (tag_lines): Likewise.
21154 (tag_counters): Likewise.
21155 (tag_summary): Likewise.
21156 (dump_working_sets): Use depth to do a proper indentation.
21158 2017-04-28 Jakub Jelinek <jakub@redhat.com>
21161 * cgraph.h (symtab_node::debug_symtab): No longer inline.
21162 * symtab.c (symtab_node::debug_symtab): Move definition here.
21164 2017-04-28 Richard Biener <rguenther@suse.de>
21166 * lto-streamer.h (LTO_major_version): Bump to 7.
21168 2017-04-28 Richard Biener <rguenther@suse.de>
21170 * tree-vrp.c (assert_info): New struct.
21171 (add_assert_info): New helper.
21172 (register_edge_assert_for_2): Refactor to add asserts to a vector
21174 (register_edge_assert_for_1): Likewise.
21175 (register_edge_assert_for): Likewise.
21176 (finish_register_edge_assert_for): New helper actually registering
21177 asserts where live on edge.
21178 (find_conditional_asserts): Adjust.
21179 (find_switch_asserts): Likewise.
21180 (evrp_dom_walker::try_find_new_range): Generalize.
21181 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
21183 2017-04-27 Marek Polacek <polacek@redhat.com>
21186 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
21187 arg10 and arg11 to itype.
21189 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
21191 * doc/extend.texi (Object Size Checking): Improve grammar.
21193 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
21196 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
21197 that the logic for permitting reciprocal estimates matches that
21200 2017-04-27 Jakub Jelinek <jakub@redhat.com>
21203 * tree.c (type_cache_hasher::equal): Only compare
21204 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
21205 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
21206 non-aggregate element types.
21207 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
21208 about the flag on ARRAY_TYPEs in the comment, formatting fix.
21210 2017-04-27 Richard Biener <rguenther@suse.de>
21212 PR middle-end/80533
21213 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
21214 stripping ARRAY_REFs from MEM_EXPR make sure we're not
21215 keeping a reference to a trailing array.
21217 2017-04-27 Richard Biener <rguenther@suse.de>
21219 PR middle-end/80539
21220 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
21221 being in loop-closed SSA form conservatively.
21222 (chrec_fold_multiply_poly_poly): Likewise.
21224 2017-04-27 Tamar Christina <tamar.christina@arm.com>
21226 PR middle-end/79665
21227 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
21228 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
21230 2017-04-27 Jakub Jelinek <jakub@redhat.com>
21233 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
21234 (aarch64_function_arg_alignment): Return unsigned int again, but still
21235 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
21236 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
21237 Don't emit -Wpsabi note.
21238 (aarch64_function_arg_boundary): Likewise.
21239 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
21242 2017-04-26 Nathan Sidwell <nathan@acm.org>
21244 * tree.h (crc32_unsigned_n): Declare.
21245 (crc32_unsigned, crc32_unsigned): Make inline.
21246 * tree.c (crc32_unsigned_bits): Replace with ...
21247 (crc32_unsigned_n): ... this.
21248 (crc32_unsigned, crc32_byte): Remove.
21249 (crc32_string): Remove unnecessary braces.
21251 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
21253 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
21254 * ipa-inline-analysis.c (MAX_TIME): Remove.
21255 (account_size_time): Use sreal for time.
21256 (dump_inline_summary): Update.
21257 (estimate_function_body_sizes): Update.
21258 (estimate_edge_size_and_time): Update.
21259 (estimate_calls_size_and_time): Update.
21260 (estimate_node_size_and_time): Update.
21261 (inline_merge_summary): Update.
21262 (inline_update_overall_summary): Update.
21263 (estimate_time_after_inlining): Update.
21264 (inline_read_section): Update.
21265 (inline_write_summary): Update.
21266 * ipa-inline.c (compute_uninlined_call_time): Update.
21267 (compute_inlined_call_time): Update.
21268 (recursive_inlining): Update.
21269 (inline_small_functions): Update.
21270 (dump_overall_stats): Update.
21271 * ipa-inline.h: Include sreal.h.
21272 (size_time_entry): Turn time to sreal.
21273 (inline_summary): Turn self_time nad time to sreal.
21275 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
21277 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
21279 (sreal::stream_out, sreal::stream_in): New.
21280 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
21282 2017-04-25 Jakub Jelinek <jakub@redhat.com>
21284 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
21287 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
21290 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
21292 (dimode_scalar_chain::compute_convert_gain): Ditto.
21293 (dimode_scalar_chain::make_vector_copies): Ditto.
21294 (dimode_scalar_chain::convert_reg): Ditto.
21295 (dimode_scalar_chain::convert_insn): Ditto.
21296 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
21297 (VI248_AVX512BW_1): New mode iterator.
21298 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
21299 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
21302 2017-04-25 Martin Sebor <msebor@redhat.com>
21304 PR tree-optimization/80497
21305 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
21306 constants are representable in HOST_WIDE_INT.
21307 (parse_directive): Ditto.
21309 2017-04-25 Martin Sebor <msebor@redhat.com>
21312 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
21313 (new_zero_array): Adjust signature.
21314 (dom_info::dom_init): Used unsigned rather that size_t.
21315 (dom_info::dom_info): Same.
21317 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21318 Jakub Jelinek <jakub@redhat.com>
21321 * config/arm/arm.c: Include gimple.h.
21322 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
21323 returns negative, increment ncrn only if it returned positive.
21324 (arm_needs_doubleword_align): Return int instead of bool,
21325 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
21326 members, but if there is any such non-FIELD_DECL
21327 > PARM_BOUNDARY aligned decl, return -1 instead of false.
21328 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
21329 returns negative, increment nregs only if it returned positive.
21330 (arm_setup_incoming_varargs): Likewise.
21331 (arm_function_arg_boundary): Emit -Wpsabi note if
21332 arm_needs_doubleword_align returns negative, return
21333 DOUBLEWORD_ALIGNMENT only if it returned positive.
21335 2017-04-25 Marek Polacek <polacek@redhat.com>
21338 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
21339 first argument to type.
21341 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
21344 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
21345 type checks to test for compatibility instead of equality.
21347 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21348 Jakub Jelinek <jakub@redhat.com>
21351 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
21353 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
21354 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
21355 the alignment computation, but return their maximum in warn_alignment.
21356 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
21357 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
21359 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
21360 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
21363 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21365 * config/arc/simdext.md (dmpyh): Fix typo.
21367 2017-04-25 Richard Biener <rguenther@suse.de>
21369 PR tree-optimization/80492
21370 * alias.c (compare_base_decls): Handle registers with asm
21371 specification conservatively.
21372 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
21373 compare_base_decls returning dont-know properly.
21375 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21377 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
21378 (legitimate_offset_address_p): New function.
21379 (arc_legitimate_address_p): Use above function.
21381 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21383 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
21385 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21387 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
21388 ACCH registers whenever they are available.
21390 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21392 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
21393 double regs fix when not used.
21395 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21397 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
21399 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
21400 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
21402 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21404 * config/arc/arc.c (arc_output_addsi): Check for h-register class
21405 when emitting short ADD instructions.
21407 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21409 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
21411 (cmpsi_cc_c_insn): Likewise.
21412 (cbranchsi4_scratch): Compute proper instruction length using
21413 compact_hreg_operand.
21414 * config/arc/predicates.md (compact_hreg_operand): New predicate.
21416 2017-04-25 Richard Biener <rguenther@suse.de>
21418 PR middle-end/80509
21419 * passes.c (pass_manager::pass_manager): Initialize
21420 m_name_to_pass_map.
21422 2017-04-25 Richard Biener <rguenther@suse.de>
21424 PR tree-optimization/79201
21425 * tree-ssa-sink.c (statement_sink_location): Handle calls.
21427 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21430 * config/s390/vector.md: Split MEM->GPR vector moves for
21431 non-s_operand addresses.
21433 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21436 * config/s390/predicates.md (reload_const_wide_int_operand): New
21438 * config/s390/s390.md ("movti"): Remove d/P alternative.
21439 ("movti_bigconst"): New pattern definition.
21441 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
21444 * s390-protos.h (s390_expand_cs_hqi): Removed.
21445 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
21446 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
21447 modes as well as CCZ1mode and CCZmode.
21448 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
21449 signature of s390_emit_compare_and_swap.
21450 (s390_expand_cs_hqi): Likewise, make static.
21451 (s390_expand_cs_tdsi): Generate an explicit compare before trying
21452 compare-and-swap, in some cases.
21453 (s390_expand_cs): Wrapper function.
21454 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
21456 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
21457 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
21458 patterns for small and large integers. Forbid symref memory operands.
21459 Move expander to s390.c. Require cc register.
21460 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
21461 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
21462 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
21463 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
21464 symref memory operands. Remove CC mode and call s390_match_ccmode
21466 ("atomic_exchange<mode>"): Allow and implement all integer modes.
21468 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
21470 * config/s390/s390.md (define_peephole2): New peephole to help
21471 combining the load-and-test pattern with volatile memory.
21473 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
21475 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
21476 with CCZmode for TARGET_Z196.
21478 2017-04-25 Jakub Jelinek <jakub@redhat.com>
21480 PR rtl-optimization/80501
21481 * combine.c (make_compound_operation_int): Set subreg_code to SET
21482 even for AND with mask of the sign bit of mode.
21484 PR rtl-optimization/80500
21485 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
21486 sum's initial value.
21488 2017-04-25 Julian Brown <julian@codesourcery.com>
21489 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21491 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
21493 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
21495 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
21497 2017-04-25 Julian Brown <julian@codesourcery.com>
21498 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21500 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
21501 (thunderx2t99_sha): New Reservation.
21503 2017-04-25 Julian Brown <julian@codesourcery.com>
21504 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21506 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
21507 type for 1-element load.
21509 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
21511 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
21513 2017-04-24 Martin Jambor <mjambor@suse.cz>
21515 PR tree-optimization/80293
21516 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
21517 char arrays not totally scalarizable if it is false.
21518 (analyze_all_variable_accesses): Pass correct value in the new
21519 parameter. Add a statistics counter.
21521 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
21523 PR middle-end/79931
21524 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
21526 2017-04-24 Richard Biener <rguenther@suse.de>
21528 PR tree-optimization/80494
21529 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
21530 out for complex types.
21532 2017-04-24 Richard Biener <rguenther@suse.de>
21534 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
21535 * tree-ssa-sccvn.c (print_scc): Print SCC size.
21536 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
21537 (DFS): Adjust and never fail.
21538 (sccvn_dom_walker::fail): Remove.
21539 (sccvn_dom_walker::before_dom_children): Adjust.
21540 (run_scc_vn): Likewise and never fail.
21541 * tree-ssa-pre.c (pass_pre::execute): Adjust.
21542 (pass_fre::execute): Likewise.
21544 2017-04-24 Richard Biener <rguenther@suse.de>
21546 PR tree-optimization/79725
21547 * tree-ssa-sink.c (statement_sink_location): Return whether
21548 failure reason was zero uses. Move that check later.
21549 (sink_code_in_bb): Deal with zero uses by removing the stmt
21552 2017-04-24 Richard Biener <rguenther@suse.de>
21555 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
21556 pointer-based references.
21558 2017-04-24 Richard Biener <rguenther@suse.de>
21561 * pass_manager.h (pass_manager::operator new): Remove.
21562 (pass_manager::operator delete): Likewise.
21563 * passes.c (pass_manager::operator new): Remove.
21564 (pass_manager::operator delete): Likewise.
21565 (pass_manager::pass_manager): Zero individual pass members.
21567 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
21570 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
21571 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
21572 Check "XEXP (src, 1)" operand here.
21573 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
21574 Check "XEXP (src, 1)" operand here.
21575 (dimode_scalar_chain::make_vector_copies): Detect count register
21576 of a shift instruction. Zero extend count register from QImode
21577 to DImode to satisfy vector shift pattern count operand predicate.
21578 Substitute vector shift count operand with a DImode copy.
21579 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
21582 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
21584 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
21585 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
21586 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
21587 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
21588 (UNSPEC_NOREX_MEM): Remove definition.
21590 2017-04-21 Richard Biener <rguenther@suse.de>
21592 PR tree-optimization/79547
21593 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
21594 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
21595 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
21596 without any constraints.
21598 2017-04-21 Richard Biener <rguenther@suse.de>
21600 PR tree-optimization/78847
21601 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
21603 2017-04-21 Richard Biener <rguenther@suse.de>
21605 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
21606 (build_distinct_type_copy): Likewise.
21607 (build_variant_type_copy): Likewise.
21608 * tree.c (build_qualified_type): Pass down mem-stat info.
21609 (build_distinct_type_copy): Likewise.
21610 (build_variant_type_copy): Likewise.
21612 2017-04-21 Richard Biener <rguenther@suse.de>
21614 PR tree-optimization/80237
21615 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
21617 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
21618 for a simplified result.
21620 2016-04-21 Richard Biener <rguenther@suse.de>
21622 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
21623 sth as strict as a simple_iv but a chrec without symbols and an
21624 operand defined in the loop we are peeling (and not some subloop).
21625 (propagate_constants_for_unrolling): Propagate all constants.
21627 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
21630 * config/i386/i386.c (print_reg): Remove assert for disalowed
21631 regno values, call output_operand_lossage instead.
21633 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
21636 * config/i386/constraints.md (Yc): New register constraint.
21637 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
21638 Use Yc constraint for alternative 2 of operand 0. Remove
21639 preferred_for_speed attribute.
21641 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
21643 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
21644 lastprivate clauses in SIMT case.
21646 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
21648 * doc/invoke.texi (-Wextra-semi): Document new warning option.
21650 2017-04-20 Richard Biener <rguenther@suse.de>
21652 PR tree-optimization/57796
21653 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
21654 as N scalar stores.
21655 (vect_model_load_cost): Cost gathers as N scalar loads.
21657 2017-04-20 Richard Biener <rguenther@suse.de>
21659 * ggc-page.c (ggc_allocated_p): Rename to ...
21660 (safe_lookup_page_table_entry): ... this and return the lookup
21662 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
21664 2017-04-20 Richard Biener <rguenther@suse.de>
21666 PR tree-optimization/80453
21667 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
21668 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
21669 from the conditions.
21670 (vn_phi_eq): Pass them down.
21671 (vn_phi_lookup): Record them.
21672 (vn_phi_insert): Likewise.
21674 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
21676 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
21677 uninitialized variable warning to avoid buffer overrun.
21679 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
21682 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
21683 is suppressed for '{ 0 }' in C.
21685 2017-04-20 Jakub Jelinek <jakub@redhat.com>
21687 * BASE-VER: Set to 8.0.0.
21689 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
21691 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
21692 priority .init_array and .fini_array section with SECTION_NOTYPE
21695 2017-04-20 Jakub Jelinek <jakub@redhat.com>
21697 PR middle-end/80423
21698 * tree.h (build_array_type): Add typeless_storage default argument.
21699 * tree.c (type_cache_hasher::equal): Also compare
21700 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
21701 (build_array_type): Add typeless_storage argument, set
21702 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
21704 (build_nonshared_array_type): Adjust build_array_type_1 caller.
21705 (build_array_type): Likewise. Add typeless_storage argument.
21707 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
21708 Jakub Jelinek <jakub@redhat.com>
21710 PR tree-optimization/80426
21711 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
21712 operation on symbolic operands, also compute the overflow for the
21713 invariant part when the operation degenerates into a negation.
21715 2017-04-19 Jakub Jelinek <jakub@redhat.com>
21718 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
21719 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
21722 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
21724 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
21727 * config/avr/avr.c (tree.h): Include it.
21728 (cgraph.h): Include it.
21729 (avr_encode_section_info): Don't warn for uninitialized progmem
21730 variable if it's just an alias.
21732 2017-04-19 Richard Biener <rguenther@suse.de>
21735 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
21736 when needed by AutoPGO.
21738 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
21741 * doc/lto.texi: Remove an extra 'that'.
21743 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
21745 PR rtl-optimization/80429
21746 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
21747 are only used in debug insns.
21749 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
21750 Vladimir Makarov <vmakarov@redhat.com>
21752 * config/sparc/predicates.md (input_operand): Add comment. Return
21753 true for any memory operand when LRA is in progress.
21754 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
21756 2017-04-18 Jeff Law <law@redhat.com>
21759 * mips.md ({return,simple_return}_internal): Do not overwrite
21762 2017-04-18 Jakub Jelinek <jakub@redhat.com>
21764 PR tree-optimization/80443
21765 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
21766 instead of adding 1, subtract -1 and similarly instead of subtracting
21769 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
21771 PR rtl-optimization/80357
21772 * haifa-sched.c (tmp_bitmap): New variable.
21773 (model_recompute): Handle duplicate use records.
21774 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
21775 (free_global_sched_pressure_data): Free it.
21777 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
21780 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
21781 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21782 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21783 instead of SYSTEM_HEADER_DIR.
21785 2017-04-18 Jeff Law <law@redhat.com>
21787 PR middle-end/80422
21788 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
21789 predecessors after walking up the insn chain.
21791 2017-04-18 Jakub Jelinek <jakub@redhat.com>
21794 * dwarf2out.c (modified_type_die): Try harder not to emit internal
21795 sizetype type into debug info.
21797 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
21800 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
21801 unneeded test for TARGET_UPPER_REGS_SF.
21802 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
21804 2017-04-18 Jakub Jelinek <jakub@redhat.com>
21807 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
21808 instead of gsi_after_labels.
21810 2017-04-18 Jeff Law <law@redhat.com>
21812 * regcprop.c (maybe_mode_change): Avoid creating copies of the
21816 2017-04-13 Jeff Law <law@redhat.com>
21817 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
21818 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
21820 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
21823 * config/avr/avr.c (intl.h): Include it.
21824 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
21826 2017-04-18 Martin Liska <mliska@suse.cz>
21828 PR gcov-profile/78783
21829 * gcov-tool.c (gcov_output_files): Validate that destination
21830 file is either removed by the tool or by a user.
21832 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
21833 Guy Benyei <guybe@mellanox.com>
21835 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
21836 block, and do not negate it, the stored id is already negative.
21838 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
21840 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
21842 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
21845 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
21846 masks of options that should be turned off if the VSX vector
21847 options are turned off.
21848 (OTHER_P8_VECTOR_MASKS): Likewise.
21849 (OTHER_VSX_VECTOR_MASKS): Likewise.
21850 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
21851 rs6000_disable_incompatible_switches to validate no type switches
21853 (rs6000_incompatible_switch): New function to disallow turning on
21854 other vector options if -mno-vsx, -mno-power8-vector, or
21855 -mno-power9-vector are specified.
21857 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21859 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
21861 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21863 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
21864 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
21865 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
21866 (ARG_POINTER_CFA_OFFSET): Likewise.
21868 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21870 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
21871 conditions to take advantage of various optimizations.
21873 2017-04-13 Jeff Law <law@redhat.com>
21875 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
21876 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
21877 (zero_extendsidi2_dext): Likewise.
21879 2017-04-13 Jakub Jelinek <jakub@redhat.com>
21882 * fold-const.c (fold_ternary_loc): Revert
21883 use op0 instead of fold_convert_loc (loc, type, arg0) part of
21886 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
21888 PR rtl-optimization/80343
21889 * lra-remat.c (update_scratch_ops): Assign original hard reg to
21890 new scratch pseudo.
21892 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
21895 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
21896 to ubsan_encode_value.
21898 2017-04-13 Jeff Law <law@redhat.com>
21900 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
21901 appearing in DEBUG_INSNs.
21903 2017-04-13 Martin Liska <mliska@suse.cz>
21905 PR gcov-profile/80413
21906 * gcov-io.c (gcov_write_string): Copy to buffer just when
21907 allocated size is greater than zero.
21909 2017-04-13 Jakub Jelinek <jakub@redhat.com>
21912 * dwarf2out.c (decls_for_scope): Ignore declarations of
21913 current_function_decl in BLOCK_NONLOCALIZED_VARS.
21915 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
21918 * ipa-visibility.c (non_local_p): Fix typos.
21919 (localize_node): When localizing symbol in same comdat group,
21920 dissolve the group only when we know external symbols are going
21922 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
21924 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21926 PR tree-optimization/79390
21927 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
21928 order does not result in usable sequence, retry with reversed operand
21934 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
21935 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
21936 op0 instead of fold_convert_loc (loc, type, arg0).
21938 2017-04-12 Jeff Law <law@redhat.com>
21940 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
21941 has a delay slot in the generated code.
21943 * config/cris/cris.md (cris_preferred_reload_class): Return
21944 GENNONACR_REGS rather than GENERAL_REGS.
21946 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21949 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
21950 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
21951 signedness of the result type.
21953 2017-04-12 Richard Biener <rguenther@suse.de>
21954 Jeff Law <law@redhat.com>
21956 PR tree-optimization/80359
21957 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
21958 trim stores to TARGET_MEM_REFs.
21960 2017-04-12 Richard Biener <rguenther@suse.de>
21962 PR tree-optimization/79390
21963 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
21964 threading case even more.
21966 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
21969 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
21970 for quad_address_p for TImode, instead of just not indexed_address.
21972 2017-04-12 Richard Biener <rguenther@suse.de>
21973 Bernd Edlinger <bernd.edlinger@hotmail.de>
21975 PR middle-end/79671
21976 * alias.c (component_uses_parent_alias_set_from): Handle
21977 TYPE_TYPELESS_STORAGE.
21978 (get_alias_set): Likewise.
21979 * tree-core.h (tree_type_common): Add typeless_storage flag.
21980 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
21981 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
21982 for types containing members with TYPE_TYPELESS_STORAGE.
21983 (place_field): Likewise.
21984 (layout_type): Likewise for ARRAY_TYPE.
21985 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
21986 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
21987 TYPE_TYPELESS_STORAGE.
21988 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
21990 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21993 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
21994 first argument to type.
21996 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22000 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
22001 CONST0_RTX (mode) rather than const0_rtx where appropriate.
22002 (rs6000_expand_binop_builtin): Likewise.
22003 (rs6000_expand_ternop_builtin): Likewise; also add missing
22004 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
22005 vshasigma built-ins.
22006 * doc/extend.texi: Document that vec_xxpermdi's third argument
22007 must be a constant.
22009 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
22011 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
22012 Use shift_const cost parameter when calculating gain of STV shifts.
22014 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
22016 PR rtl-optimization/70478
22017 * lra-constraints.c (process_alt_operands): Check memory for
22018 disfavoring memory insn operand.
22020 2017-04-11 Jakub Jelinek <jakub@redhat.com>
22022 PR middle-end/80100
22023 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
22024 left shift in unsigned HOST_WIDE_INT type.
22026 PR rtl-optimization/80385
22027 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
22028 (not (neg X)) into (plus X -1) for complex or non-integral modes.
22031 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
22032 if they have any depend clauses.
22034 2017-04-11 Martin Liska <mliska@suse.cz>
22037 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
22038 * ipa-split.c (split_function): Create a local comdat symbol
22039 if caller is in a comdat group.
22041 2017-04-11 Martin Liska <mliska@suse.cz>
22044 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
22047 2017-04-11 Martin Sebor <msebor@redhat.com>
22049 PR middle-end/80364
22050 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
22051 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
22053 (directive::set_width, directive::set_precision, format_character):
22055 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
22058 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
22061 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
22062 conflict, set target->arch_name instead of target->cpu_name.
22064 2017-04-11 Richard Biener <rguenther@suse.de>
22066 PR tree-optimization/80374
22067 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
22068 build_zero_cst, remove fold_convertible_p check again.
22070 2017-04-11 Martin Liska <mliska@suse.cz>
22073 * ubsan.c (instrument_object_size): Do not instrument register
22076 2017-04-11 Jakub Jelinek <jakub@redhat.com>
22079 * config/i386/i386-builtin-types.def
22080 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
22081 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
22082 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
22083 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
22084 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
22085 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
22086 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
22087 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
22088 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
22089 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
22090 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
22091 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
22092 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
22093 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
22094 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
22095 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
22096 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
22097 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
22098 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
22099 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
22100 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
22101 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
22102 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
22103 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
22104 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
22105 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
22106 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
22107 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
22108 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
22109 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
22110 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
22111 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
22112 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
22113 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
22114 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
22115 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
22116 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
22117 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
22118 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
22119 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
22120 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
22121 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
22122 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
22123 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
22124 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
22125 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
22127 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
22128 flag to second_arg_count, handle 4 argument function type _COUNT
22129 aliases, handle second_arg_count on second argument rather than last.
22131 2017-04-10 Jeff Law <law@redhat.com>
22133 PR tree-optimization/80374
22134 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
22135 record anything if we can not convert integer_zero_node to the
22138 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
22141 * config/rs6000/rs6000.c (rs6000_option_override_internal):
22142 Enhance special handling given to the TARGET_P9_MINMAX option in
22143 relation to certain other options.
22145 2017-04-10 Bin Cheng <bin.cheng@arm.com>
22147 PR tree-optimization/80153
22148 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
22149 remove POINTER_PLUS_EXPR's base part directly, rather than through
22152 2017-04-10 Richard Biener <rguenther@suse.de>
22153 Bin Cheng <bin.cheng@arm.com>
22155 PR tree-optimization/80153
22156 * tree-affine.c (aff_combination_to_tree): Get base pointer from
22157 the first element of pointer type aff_tree. Build result expr in
22159 (add_elt_to_tree): Convert to type unconditionally. Remove other
22160 fold_convert calls.
22161 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
22162 (rewrite_use_nonlinear_expr): Check invariant using iv information.
22164 2017-04-10 Richard Biener <rguenther@suse.de>
22166 * tree-ssa-structalias.c (find_func_aliases): Properly handle
22169 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
22171 PR rtl-optimization/70478
22172 * lra-constraints.c (curr_small_class_check): New.
22173 (update_and_check_small_class_inputs): New.
22174 (process_alt_operands): Update curr_small_class_check. Disfavor
22175 alternative insn memory operands. Check available regs for small
22178 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
22181 * config/mips/mips.opt (-mvirt): Update description.
22182 * doc/invoke.texi (-mvirt): Likewise.
22184 2017-04-10 Richard Biener <rguenther@suse.de>
22186 PR middle-end/80362
22187 * fold-const.c (fold_binary_loc): Look at unstripped ops when
22188 looking for NEGATE_EXPR in -A / -B to A / B folding.
22190 2017-04-10 Martin Liska <mliska@suse.cz>
22192 PR gcov-profile/80224
22193 * gcov.c (print_usage): Fix usage string.
22194 (get_gcov_intermediate_filename): Remove.
22195 (output_gcov_file): Use both for normal and intermediate format.
22196 (generate_results): Do not initialize special file for
22197 intermediate format.
22199 2017-04-10 Richard Biener <rguenther@suse.de>
22201 PR tree-optimization/80304
22202 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
22205 2017-04-10 Nathan Sidwell <nathan@acm.org>
22208 * config/rs6000/rs6000.c (rs6000_vector_type): New.
22209 (rs6000_init_builtins): Use it.
22211 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22213 * config/arm/arm.md (<mrc>): Add mode to SET source.
22214 (<mrrc>): Likewise.
22216 2017-04-10 Richard Biener <rguenther@suse.de>
22218 PR middle-end/80344
22219 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
22221 2017-04-10 Jakub Jelinek <jakub@redhat.com>
22224 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
22225 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
22226 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
22227 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
22228 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
22229 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
22230 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
22231 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
22232 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
22233 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
22234 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
22235 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
22236 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
22237 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
22238 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
22239 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
22240 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
22241 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
22242 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
22243 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
22244 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
22245 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
22246 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
22248 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
22250 PR rtl-optimization/70478
22251 * lra-constraints.c: Reverse the last patch.
22253 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
22255 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
22256 Add comment for WCHAR_T.
22258 2017-04-08 Martin Liska <mliska@suse.cz>
22261 2017-04-07 Martin Liska <mliska@suse.cz>
22264 * ipa-split.c (split_function): Add function part to a same comdat
22267 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22270 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
22272 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
22274 * rs6000/rs6000.c (vec_load_pendulum): Rename...
22275 (vec_pairing): ...to this.
22276 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
22277 (rs6000_sched_init): Adjust for name change.
22278 (struct rs6000_sched_context): Likewise.
22279 (rs6000_init_sched_context): Likewise.
22280 (rs6000_set_sched_context): Likewise.
22282 2017-04-07 Jakub Jelinek <jakub@redhat.com>
22288 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
22290 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
22291 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
22292 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
22294 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
22296 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
22298 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
22300 PR rtl-optimization/70703
22301 * ira-color.c (update_conflict_hard_regno_costs): Use
22302 int64_t instead of HOST_WIDE_INT.
22304 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
22306 PR rtl-optimization/70478
22307 * lra-constraints.c (process_alt_operands): Disfavor alternative
22308 insn memory operands.
22310 2017-04-07 Jeff Law <law@redhat.com>
22312 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
22313 CALL and NOTE_INSN_CALL_ARG_LOCATION.
22315 2017-04-07 Martin Liska <mliska@suse.cz>
22318 * config/aarch64/aarch64.c (aarch64_process_target_attr):
22319 Show error message instead of an ICE.
22321 2017-04-07 Martin Liska <mliska@suse.cz>
22324 * ipa-split.c (split_function): Add function part to a same comdat
22327 2017-04-07 Richard Biener <rguenther@suse.de>
22329 PR middle-end/80341
22330 * tree.c (get_unwidened): Also handle ! for_type case for
22332 * convert.c (do_narrow): Split out from ...
22333 (convert_to_integer_1): ... here. Do not pass final truncation
22334 type to get_unwidened for TRUNC_DIV_EXPR.
22336 2017-04-07 Richard Biener <rguenther@suse.de>
22338 * tree-affine.c (wide_int_ext_for_comb): Take type rather
22340 (aff_combination_const): Adjust.
22341 (aff_combination_scale): Likewise.
22342 (aff_combination_add_elt): Likewise.
22343 (aff_combination_add_cst): Likewise.
22344 (aff_combination_convert): Likewise.
22345 (add_elt_to_tree): Likewise. Remove unused argument.
22346 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
22348 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
22350 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
22352 * config/arm/arm.c (arm_default_short_enums): Use
22353 ARM_DEFAULT_SHORT_ENUMS.
22354 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
22356 2017-04-06 Jakub Jelinek <jakub@redhat.com>
22359 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
22360 members with redundant out-of-class redeclaration.
22362 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
22365 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
22366 * config/i386/i386.md (*zero_extendsidi2):
22367 Add (?*x,*x) and (?*v,*v) alternatives.
22369 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
22372 * config/i386/i386.c (ix86_expand_builtin)
22373 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
22374 mode from insn data. Convert operands to insn operand mode.
22375 Copy operands that don't satisfy insn predicate to a register.
22377 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
22379 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
22382 2017-04-06 Richard Biener <rguenther@suse.de>
22384 PR tree-optimization/80334
22385 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
22386 preserve alignment of accesses.
22388 2017-04-06 Richard Biener <rguenther@suse.de>
22390 PR tree-optimization/80262
22391 * tree-sra.c (build_ref_for_offset): Preserve address-space
22393 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
22394 Drop useless address-space information on MEM_REF offsets.
22396 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
22398 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
22400 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
22402 PR rtl-optimization/70703
22403 * ira-color.c (update_conflict_hard_regno_costs): Use
22404 HOST_WIDE_INT instead of long.
22406 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
22409 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
22410 not defined for x86_64 target. Add -mmmx target option when __SSE2__
22412 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
22413 for x86_64 target. Handle -m3dnowa option.
22415 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
22417 PR rtl-optimization/70703
22418 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
22419 (update_conflict_hard_regno_costs): Use long instead of unsigned
22420 arithmetic for cost calculation.
22422 2017-04-05 Jakub Jelinek <jakub@redhat.com>
22423 Bernd Edlinger <bernd.edlinger@hotmail.de>
22426 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
22429 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
22432 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
22433 ptr_mode with Pmode throughout.
22434 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
22435 into probe_stack_range and use DImode.
22437 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
22440 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
22441 call_eh_return is true.
22443 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22445 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
22446 Initialize last_match_fntype_index.
22448 2017-04-05 Jakub Jelinek <jakub@redhat.com>
22451 * tree-nvr.c: Include internal-fn.h.
22452 (pass_return_slot::execute): Ignore internal calls without
22455 2017-04-04 Jakub Jelinek <jakub@redhat.com>
22456 Richard Biener <rguenther@suse.de>
22459 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
22460 captures used multiple times, except for the last use.
22461 * generic-match-head.c: Include gimplify.h.
22463 2017-04-04 Jakub Jelinek <jakub@redhat.com>
22465 PR tree-optimization/79390
22466 * target.h (struct noce_if_info): Declare.
22467 * targhooks.h (default_noce_conversion_profitable_p): Declare.
22468 * target.def (noce_conversion_profitable_p): New target hook.
22469 * ifcvt.h (struct noce_if_info): New type, moved from ...
22470 * ifcvt.c (struct noce_if_info): ... here.
22471 (noce_conversion_profitable_p): Renamed to ...
22472 (default_noce_conversion_profitable_p): ... this. No longer
22474 (noce_try_store_flag_constants, noce_try_addcc,
22475 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
22476 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
22477 instead of noce_conversion_profitable_p.
22478 * config/i386/i386.c: Include ifcvt.h.
22479 (ix86_option_override_internal): Don't override
22480 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
22481 (ix86_noce_conversion_profitable_p): New function.
22482 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
22483 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
22484 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
22485 * doc/tm.texi: Regenerated.
22487 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22489 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
22492 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22495 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
22496 instructions for small multiply cores.
22498 2017-04-04 Jeff Law <law@redhat.com>
22500 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
22502 (mips_expand_vec_perm_const): Initialize elements in orig_perm
22503 that are not set by the loop over the elements.
22505 2017-04-04 Jakub Jelinek <jakub@redhat.com>
22508 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
22509 int mode, convert_modes it to mode as unsigned, otherwise use
22510 lowpart_subreg to mode rather than SImode.
22511 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
22512 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
22513 Use DImode instead of SImode for the shift count operand.
22514 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
22517 2017-04-04 Richard Biener <rguenther@suse.de>
22519 PR middle-end/80281
22520 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
22521 arithmetic done for the negate or the plus. Simplify.
22522 (A - (-B) -> A + B): Likewise.
22523 * fold-const.c (split_tree): Make sure to not negate pointers.
22525 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
22527 PR rtl-optimization/60818
22528 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
22529 a compare of comparisons with the thing compared if this results
22530 in a different machine mode.
22532 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
22534 * alias.c (base_alias_check): Fix typo in comment.
22535 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
22536 * cgraphunit.c (symbol_table::compile): Likewise.
22537 * collect2.c (maybe_run_lto_and_relink): Likewise.
22538 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
22539 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
22540 * config/avr/avr.c (avr_map_op_t): Likewise.
22541 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
22542 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
22543 * config/epiphany/epiphany.md (movcc): Likewise.
22544 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
22545 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
22547 * config/mips/mips.c (mips_save_restore_reg): Likewise.
22548 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
22549 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
22550 * config/sh/sh.c (sh_rtx_costs): Likewise.
22551 * fold-const.c (fold_truth_andor): Likewise.
22552 * genautomata.c (collapse_flag): Likewise.
22553 * gengtype.h (struct type::u::s): Likewise.
22554 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
22555 * input.c (FORMAT_AMOUNT): Likewise.
22556 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
22557 (known_aggs_to_agg_replacement_list): Likewise.
22558 * ipa-inline-analysis.c: Likewise.
22559 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
22560 * ipa-polymorphic-call.c
22561 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
22562 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
22563 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
22565 * modulo-sched.c (apply_reg_moves): Likewise.
22566 * omp-expand.c (build_omp_regions_1): Likewise.
22567 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
22568 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
22569 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
22570 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
22571 * value-prof.c: Likewise.
22572 * var-tracking.c (val_reset): Likewise.
22574 2017-04-03 Richard Biener <rguenther@suse.de>
22576 PR tree-optimization/80275
22577 * fold-const.c (split_address_to_core_and_offset): Handle
22580 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
22582 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
22583 descriptors is at least equal to that of functions.
22585 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
22587 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
22589 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
22592 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
22593 (mov<IMOD4:mode>): New expander.
22594 (*mov<IMOD4:mode>_internal): New insn and split pattern.
22596 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
22598 PR rtl-optimization/79405
22599 * fwprop.c (propagations_left): New variable.
22600 (forward_propagate_into): Decrement it.
22601 (fwprop_init): Initialize it.
22602 (fw_prop): If the variable has reached zero, stop propagating.
22603 (fwprop_addr): Ditto.
22605 2017-03-31 Jakub Jelinek <jakub@redhat.com>
22608 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
22609 a FUNCTION_DECL, pass it as decl instead of origin to
22612 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
22614 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
22617 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
22620 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
22621 TARGET_VSX_SMALL_INTEGER.
22623 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22625 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
22626 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
22628 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
22630 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
22631 extraction from odd-numbered MSA register.
22633 2017-03-31 Jakub Jelinek <jakub@redhat.com>
22635 PR middle-end/80173
22636 * expmed.c (store_bit_field_1): Don't attempt to create
22637 a word subreg out of hard registers wider than word if they
22638 have HARD_REGNO_NREGS of 1 for their mode.
22640 PR middle-end/80163
22641 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
22642 conversions to integer types wider than word and pointer.
22645 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
22646 (rtx_equal_for_cselib_p): Pass 0 to it.
22647 * cselib.c (cselib_hasher::equal): Likewise.
22648 (rtx_equal_for_cselib_1): Add depth argument. If depth
22649 is 128, don't look up VALUE locs and punt. Increment
22650 depth in recursive calls when walking VALUE locs.
22652 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
22654 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
22655 (make_gcov_file_name): Use the canonical path name for generating
22657 (read_line): Fix handling of files with ascii null bytes.
22659 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
22661 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
22662 to initialise a vector register instead
22663 of using a const_int.
22665 2017-03-30 Jakub Jelinek <jakub@redhat.com>
22667 PR translation/80189
22668 * gimplify.c (omp_default_clause): Use %qs instead of %s in
22669 diagnostic messages.
22671 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
22674 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
22675 (dfp_diex_<mode>): Update mode of operand 1.
22676 * doc/extend.texi (dxex, dxexq): Document change to return type.
22677 (diex, diexq): Document change to argument type.
22679 2017-03-30 Martin Jambor <mjambor@suse.cz>
22682 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
22683 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
22684 it reflects the signature changes performed at the callee side.
22685 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
22686 to cgraph_build_function_type_skip_args.
22687 (build_function_decl_skip_args): Adjust call to the above function.
22689 2017-03-30 Jakub Jelinek <jakub@redhat.com>
22692 * config/i386/sse.md
22693 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
22694 register as dest whenever it is a MEM not rtx_equal_p to the
22695 corresponding dup operand, and when forcing into reg move the
22696 reg into the memory afterwards.
22697 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
22698 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
22699 for the force_reg mode.
22700 (avx512vl_vextractf128<mode>): Use register as dest either
22701 always when a MEM, or when it is a MEM not rtx_equal_p to the
22702 corresponding dup operand, or even not when it is a CONST_VECTOR
22703 depending on the mode and lo vs. hi.
22704 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
22706 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
22707 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
22708 Likewise. Require that operands[2] is even.
22709 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
22710 Remove extraneous parens. Require that operands[2] is a multiple
22712 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
22713 operands[0] is a MEM if <mask_applied>, the predicates/constraints
22714 disallow memory then.
22716 2017-03-30 Richard Biener <rguenther@suse.de>
22718 PR tree-optimization/77498
22719 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
22720 to non-constants over backedges.
22722 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
22724 PR rtl-optimization/80233
22725 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
22726 as last_combined_insn. Do not test for BARRIER_P separately.
22728 2017-03-29 Andreas Schwab <schwab@suse.de>
22731 * calls.c (prepare_call_address): Convert funexp to Pmode before
22732 copying to temp reg.
22734 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22736 PR tree-optimization/80158
22737 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
22738 Handle possible future case of more than one alternate
22740 (replace_rhs_if_not_dup): Likewise.
22741 (replace_one_candidate): Likewise.
22743 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
22745 PR rtl-optimization/80193
22746 * ira.c (ira): Do not check allocation for LRA.
22748 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
22750 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
22751 (nvptx_output_simt_exit): Declare.
22752 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
22753 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
22754 (init_softstack_frame): Move initialization of crtl->is_leaf to...
22755 (nvptx_declare_function_name): ...here. Emit declaration of local
22756 memory space buffer for omp_simt_enter insn.
22757 (nvptx_output_unisimt_switch): New.
22758 (nvptx_output_softstack_switch): New.
22759 (nvptx_output_simt_enter): New.
22760 (nvptx_output_simt_exit): New.
22761 * config/nvptx/nvptx.h (struct machine_function): New fields
22762 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
22763 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
22764 (UNSPECV_SIMT_EXIT): Ditto.
22765 (omp_simt_enter_insn): New insn.
22766 (omp_simt_enter): New expansion.
22767 (omp_simt_exit): New insn.
22768 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
22770 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
22771 (expand_GOMP_SIMT_ENTER_ALLOC): New.
22772 (expand_GOMP_SIMT_EXIT): New.
22773 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
22774 (GOMP_SIMT_ENTER_ALLOC): Ditto.
22775 (GOMP_SIMT_EXIT): Ditto.
22776 * target-insns.def (omp_simt_enter): New insn.
22777 (omp_simt_exit): Ditto.
22778 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
22780 (lower_rec_simd_input_clauses): Implement SIMT privatization.
22781 (lower_rec_input_clauses): Likewise.
22782 (lower_lastprivate_clauses): Handle SIMT privatization.
22784 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
22785 (ompdevlow_adjust_simt_enter): New.
22786 (find_simtpriv_var_op): New.
22787 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
22788 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
22790 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
22791 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
22792 (copy_decl_for_dup_finish): Ditto.
22794 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
22796 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
22799 * config/i386/i386.c (ix86_option_override_internal): Always
22800 allow -mpreferred-stack-boundary=3 for 64-bit targets.
22802 2017-03-28 Bin Cheng <bin.cheng@arm.com>
22804 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
22806 2017-03-28 Bin Cheng <bin.cheng@arm.com>
22808 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
22809 mark new edge's irreducible flag accordign to it.
22810 (vect_do_peeling): Check loop preheader edge's irreducible flag
22811 and pass it to function slpeel_add_loop_guard.
22813 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
22815 PR tree-optimization/80218
22816 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
22817 Update block frequencies and counts.
22819 2017-03-28 Richard Biener <rguenther@suse.de>
22821 PR tree-optimization/78644
22822 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
22823 of a simplification result we may not use it at all.
22825 2017-03-28 Richard Biener <rguenther@suse.de>
22828 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
22829 without arguments, generate default definition of a SSA name.
22831 2017-03-28 Richard Biener <rguenther@suse.de>
22833 PR middle-end/80222
22834 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
22835 TYPE_REF_CAN_ALIAS_ALL references.
22836 * fold-const.c (fold_indirect_ref_1): Likewise.
22838 2017-03-28 Martin Liska <mliska@suse.cz>
22841 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
22842 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
22844 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
22845 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
22847 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
22848 (EXTRA_SPECS): Define.
22849 (SUBTARGET_EXTRA_SPECS): Likewise.
22850 (SUBTARGET_CPP_SPEC): Likewise.
22851 * config/arc/elf.h (EXTRA_SPECS): Renamed to
22852 SUBTARGET_EXTRA_SPECS.
22853 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
22855 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
22857 * config/arc/simdext.md (vst64_insn): Update pattern.
22858 (vld32wh_insn): Likewise.
22859 (vld32wl_insn): Likewise.
22860 (vld64_insn): Likewise.
22861 (vld32_insn): Likewise.
22863 2017-03-28 Marek Polacek <polacek@redhat.com>
22866 * fold-const.c (fold_comparison): Use protected_set_expr_location
22867 instead of SET_EXPR_LOCATION.
22869 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
22871 * tree.c (add_expr): Avoid name lookup warning.
22873 2017-03-27 Jeff Law <law@redhat.com>
22875 PR tree-optimization/80216
22876 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
22877 function name. Limit recursion depth.
22878 (record_temporary_equivalences): Corresponding changes.
22880 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
22882 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
22885 2017-03-27 Jakub Jelinek <jakub@redhat.com>
22888 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
22890 * cfgcleanup.c (reg_note_cfa_p): New array.
22891 (insns_have_identical_cfa_notes): New function.
22892 (old_insns_match_p): Don't cross-jump in between /f
22893 and non-/f instructions. If both i1 and i2 are frame related,
22894 verify all CFA notes, their order and content.
22896 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
22899 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
22900 HImode and SImode with zero extend to DImode to one insn.
22901 (bswap<mode>2_extenddi): Likewise.
22902 (bswapsi2_extenddi): Likewise.
22903 (bswaphi2_extendsi): Likewise.
22904 (bswaphi2): Combine bswap HImode and SImode into one insn.
22905 Separate memory insns from swapping register.
22906 (bswapsi2): Likewise.
22907 (bswap<mode>2): Likewise.
22908 (bswaphi2_internal): Delete, no longer used.
22909 (bswapsi2_internal): Likewise.
22910 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
22911 store, and gpr<-gpr swap insns.
22912 (bswap<mode>2_store): Likewise.
22913 (bswaphi2_reg): Register only splitter, combine with the splitter.
22914 (bswaphi2 splitter): Likewise.
22915 (bswapsi2_reg): Likewise.
22916 (bswapsi2 splitter): Likewise.
22917 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
22918 the insns into load, store, and register/register insns.
22919 (bswapdi2_ldbrx): Likewise.
22920 (bswapdi2_load): Likewise.
22921 (bswapdi2_store): Likewise.
22922 (bswapdi2_reg): Likewise.
22924 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
22926 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
22927 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
22929 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
22932 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
22934 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
22935 special handling for target option conflicts between dform
22936 options (-mpower9-dform, -mpower9-dform-vector,
22937 -mpower9-dform-scalar) and -mno-direct-move.
22939 2017-03-27 Richard Biener <rguenther@suse.de>
22941 PR tree-optimization/80181
22942 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
22944 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22946 * config/arc/predicates.md (move_double_src_operand): Replace the
22947 call to move_double_src_operand with a call to address_operand.
22949 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22951 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
22952 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
22953 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
22955 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22957 * config/arc/predicates.md (long_immediate_loadstore_operand):
22958 Consider scaled addresses cases.
22960 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22962 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
22963 restored when in interrupt.
22964 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
22965 doesn't have delay slot.
22967 2017-03-27 Richard Biener <rguenther@suse.de>
22970 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
22971 inlined thunk clones.
22973 2017-03-27 Jakub Jelinek <jakub@redhat.com>
22976 * asan.c (instrument_derefs): Copy over last operand from
22977 original COMPONENT_REF to the new COMPONENT_REF with
22978 DECL_BIT_FIELD_REPRESENTATIVE.
22979 * ubsan.c (instrument_object_size): Likewise.
22981 2017-03-27 Richard Biener <rguenther@suse.de>
22983 PR tree-optimization/80170
22984 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
22985 sure DR/SCEV didnt fold in constants we do not see when looking
22986 at the reference base alignment.
22988 2017-03-27 Richard Biener <rguenther@suse.de>
22990 PR middle-end/80171
22991 * gimple-fold.c (fold_ctor_reference): Properly guard against
22992 NULL return value from canonicalize_constructor_val.
22994 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
22997 * config/i386/i386.c (ix86_expand_builtin)
22998 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
22999 flags reg setting and flags reg using instructions.
23000 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
23001 clobbering instructions to zero extend op2.
23003 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
23005 * doc/install.texi (Configuration) <--with-aix-soname>:
23006 Update link to AIX ld.
23008 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
23010 PR rtl-optimization/80160
23011 PR rtl-optimization/80159
23012 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
23013 reg_alternate_class into account.
23015 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
23018 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
23019 to consider in curr_insn_transform.
23021 2017-03-24 Jakub Jelinek <jakub@redhat.com>
23023 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
23024 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
23025 and emit_mode_inner.
23027 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23029 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
23030 argument to the overloaded builtin variants. Use the new flag to
23031 deprecate certain builtin variants.
23032 * config/s390/s390-builtin-types.def: Add new builtin types.
23033 * config/s390/s390-builtins.h: Support new flags field for
23034 overloaded builtins.
23035 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
23036 (s390_macro_to_expand): Enable vector float data type.
23037 (s390_cpu_cpp_builtins_internal): Indicate support of the new
23038 builtins by incrementing the __VEC__ version number.
23039 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
23041 (s390_resolve_overloaded_builtin): Emit error messages depending
23042 on the builtin flags.
23043 * config/s390/s390.c (s390_expand_builtin): Support additional
23044 flags argument. Change error message to match the messages
23045 emitted in s390-c.c.
23046 * config/s390/s390.md: New UNSPEC_* constants.
23047 (op_type): Add new instruction types.
23048 * config/s390/vecintrin.h: Add new builtins and test data class
23050 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
23051 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
23052 (VEC_INEXACT, VEC_NOINEXACT): New constants.
23053 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
23054 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
23055 ("vec_mergel<mode>"): V_HW -> VEC_HW.
23057 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
23058 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
23059 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
23060 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
23062 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
23063 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
23064 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
23065 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
23067 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
23068 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
23069 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
23070 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
23071 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
23072 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
23073 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
23075 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
23076 ("vec_scatter_element<V_HW_4:mode>_DI")
23077 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
23078 ("vec_fpint<mode>", "vflls")
23079 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
23080 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
23081 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
23082 ("*vec_cmphe<mode>_cc"): ... these.
23084 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
23085 mode constant instead of magic value.
23087 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23089 * config/s390/s390.c (s390_expand_vec_compare): Support other
23090 vector floating point modes than just V2DF.
23091 (s390_expand_vcond): Likewise.
23092 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
23093 (s390_cannot_change_mode_class): Prevent mode changes between TF
23094 and V1TF in vector registers.
23095 * config/s390/s390.md (DF, SF): New mode attributes.
23096 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
23097 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
23098 SFmode support for VRs.
23099 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
23101 (VFT, VF_HW): New mode iterators.
23102 (vw, sdx): New mode attributes.
23103 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
23104 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
23105 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
23106 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
23107 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
23108 also the new vector floating point modes. Renaming to ...
23110 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
23111 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
23112 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
23113 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
23114 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
23115 ("vec_unordered<mode>"): ... these.
23117 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
23118 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
23119 ("*vec_extendv2df"): New insn definitions.
23121 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23123 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
23124 ("mulditi3_2", "*muldi3_sign"): New patterns.
23125 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
23126 rename the pattern definition.
23128 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23130 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
23132 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
23134 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23136 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
23137 instruction if possible.
23138 * config/s390/vector.md (vec_halfnumelts): New mode
23140 ("*vec_vllezlf<mode>"): New pattern.
23142 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23144 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
23145 ("popcountv4si2", "popcountv2di2"): Rename to ...
23146 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
23147 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
23149 ("popcount<mode>2_vxe"): New pattern.
23151 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23153 * common/config/s390/s390-common.c (processor_flags_table): Add
23155 * config.gcc: Add arch12.
23156 * config/s390/driver-native.c (s390_host_detect_local_cpu):
23157 Default to arch12 for unknown CPU model numbers.
23158 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
23159 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
23160 PROCESSOR_max sanity check.
23161 * config/s390/s390-opts.h (enum processor_type): Add
23163 * config/s390/s390.c (processor_table): Add arch12.
23164 (s390_expand_builtin): Add check for B_VXE flag.
23165 (s390_issue_rate): Add PROCESSOR_ARCH12.
23166 (s390_get_sched_attrmask): Likewise.
23167 (s390_get_unit_mask): Likewise.
23168 (s390_sched_score): Enable z13 scheduling for arch12.
23169 (s390_sched_reorder): Likewise.
23170 (s390_sched_variable_issue): Likewise.
23171 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
23173 (s390_tune_attr): Use z13 scheduling also for arch12.
23174 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
23175 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
23176 (TARGET_VXE_P): New macros.
23177 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
23178 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
23179 * config/s390/s390.opt: Add arch12 as processor_type.
23181 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23183 * config/s390/s390.md
23184 ("fixuns_truncdddi2", "fixuns_trunctddi2")
23185 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
23186 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
23188 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
23189 Rename expanders to ...
23191 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
23192 ("fixuns_truncdddi2_emu"): ... these.
23194 ("fixuns_trunc<mode>si2_emu"): New expander.
23196 ("*fixuns_truncdfdi2_z13"): Rename to ...
23197 ("*fixuns_truncdfdi2_vx"): ... this.
23199 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23201 * config/s390/2964.md: Remove the single element vector compare
23202 instructions which are no longer used.
23203 * config/s390/s390.c (s390_select_ccmode): Remove handling of
23205 (s390_canonicalize_comparison): Remove handling of DFmode
23207 (s390_expand_vec_compare_scalar): Remove function.
23208 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
23209 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
23211 ("*cmp<mode>_ccs"): Add wfcdb instruction.
23213 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23215 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
23217 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
23218 will anyway by matched by mov<mode>_64dfp.
23220 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23222 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
23223 vlef/vstef. Add missing operand to vleif.
23225 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23227 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
23228 pair for all vector types with 64 bit elements.
23229 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
23230 * config/s390/vector.md (V_HW_64): ... here.
23231 (V_128_NOSINGLE): New mode iterator.
23232 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
23233 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
23234 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
23235 ("*vec_load_pairv2di"): Change to ...
23236 ("*vec_load_pair<mode>"): ... this one.
23238 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23240 * config/s390/constraints.md: Add comments.
23241 (jKK): Reject element sizes > 8 bytes.
23242 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
23244 * config/s390/s390.md: Add the s_operand checks formerly in
23245 s390_split_ok_p to various splitters where they are still
23247 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
23248 for 128 bit vectors. Plus two splitters.
23250 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23252 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
23255 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23258 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
23259 error if the boundary argument is not constant.
23261 2017-03-24 Jakub Jelinek <jakub@redhat.com>
23263 PR rtl-optimization/80112
23264 * loop-doloop.c (doloop_condition_get): Don't check condition
23265 if cmp isn't SET with IF_THEN_ELSE src.
23267 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23269 PR tree-optimization/80158
23270 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
23271 replacing a candidate statement, also replace it for the
23272 candidate's alternate interpretation.
23273 (replace_rhs_if_not_dup): Likewise.
23274 (replace_one_candidate): Likewise.
23276 2017-03-24 Richard Biener <rguenther@suse.de>
23278 PR tree-optimization/80167
23279 * graphite-isl-ast-to-gimple.c
23280 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
23282 (translate_isl_ast_to_gimple::get_rename): Likewise.
23284 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
23286 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
23287 handling of certain combinations of target options, including the
23288 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
23289 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
23291 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23294 * config/arm/arm.md (*load_multiple): Add reload_completed to
23295 matching condition.
23297 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23298 Richard Biener <rguenth@suse.de>
23300 PR tree-optimization/79908
23301 PR tree-optimization/80136
23302 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
23303 been cast away, gimplify_and_add suffices.
23305 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
23307 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
23309 2017-03-23 Richard Biener <rguenther@suse.de>
23311 PR tree-optimization/80032
23312 * gimplify.c (gimple_push_cleanup): Forced unconditional
23313 cleanups still have to go to the conditional_cleanups
23316 2017-03-22 Jakub Jelinek <jakub@redhat.com>
23318 PR tree-optimization/80072
23319 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
23321 (next_operand_entry_id): Change type to unsigned int.
23322 (sort_by_operand_rank): Make sure to return the right return value
23323 even if unsigned fields are bigger than INT_MAX.
23324 (struct oecount): Change cnt and id type to unsigned int.
23325 (oecount_hasher::equal): Formatting fix.
23326 (oecount_cmp): Make sure to return the right return value
23327 even if unsigned fields are bigger than INT_MAX.
23328 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
23331 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
23332 TREE_READONLY on result if writing it more than once.
23335 * doc/invoke.texi (-fsanitize=thread): Document that with
23336 -fnon-call-exceptions atomics are not able to throw
23340 * tsan.c: Include tree-eh.h.
23341 (instrument_builtin_call): Call maybe_clean_eh_stmt or
23342 maybe_clean_or_replace_eh_stmt where needed.
23343 (instrument_memory_accesses): Add cfg_changed argument.
23344 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
23345 if it returned true.
23346 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
23348 PR rtl-optimization/63191
23349 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
23350 wrapper function, moved the whole old content into ...
23351 (ix86_delegitimize_address_1): ... this. New inline function.
23352 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
23353 true as last argument instead of ix86_delegitimize_address.
23355 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
23357 * config/aarch64/aarch64.c (generic_branch_cost): Copy
23358 cortexa57_branch_cost.
23360 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
23362 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
23364 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23367 * doc/md.texi (Constraints): Document wA constraint.
23368 * config/rs6000/constraints.md (wA): New.
23369 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
23370 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
23371 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
23372 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
23374 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
23377 * gimplify.c (is_oacc_declared): New function.
23378 (oacc_default_clause): Use it to set default flags for acc declared
23379 variables inside parallel regions.
23380 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
23381 declared variables.
23382 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
23383 declare attribute to any decl as necessary.
23385 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
23388 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
23389 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
23390 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
23391 (arm_arch_lpae): This.
23392 * config/arm/arm.c (arm_arch7ve): Rename into ...
23393 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
23394 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
23397 2017-03-22 Martin Liska <mliska@suse.cz>
23400 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
23401 error message instead of an ICE.
23403 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23405 * doc/extend.texi (6.11 Additional Floating Types): Revise.
23407 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
23409 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
23411 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
23414 2017-03-21 Martin Sebor <msebor@redhat.com>
23416 * doc/extend.texi: Use "cannot" instead of "can't."
23417 * doc/hostconfig.texi: Same.
23418 * doc/install.texi: Same.
23419 * doc/invoke.texi: Same.
23420 * doc/loop.texi: Same.
23421 * doc/md.texi: Same.
23422 * doc/objc.texi: Same.
23423 * doc/rtl.texi: Same.
23424 * doc/tm.texi: Same.
23425 * doc/tm.texi.in: Same.
23426 * doc/trouble.texi: Same.
23428 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
23431 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
23432 (collect_checksum_attributes): Set it.
23433 (die_checksum_ordered): Use it.
23435 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23437 PR tree-optimization/79908
23438 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
23439 change: For a VA_ARG whose LHS has been cast away, use
23440 force_gimple_operand to construct the side effects.
23442 2017-03-21 David Malcolm <dmalcolm@redhat.com>
23444 PR translation/80001
23445 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
23446 more amenable to translation.
23447 (oacc_loop_auto_partitions): Likewise.
23449 2017-03-21 Marek Polacek <polacek@redhat.com>
23450 Martin Sebor <msebor@redhat.com>
23452 PR tree-optimization/80109
23453 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
23454 on INTEGRAL_TYPE_P.
23456 2017-03-21 Jakub Jelinek <jakub@redhat.com>
23457 Segher Boessenkool <segher@kernel.crashing.org>
23460 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
23461 check reg_used_between_p between insn and one of succ or succ2
23462 depending on if succ is artificial insn not inserted into insn
23465 2017-03-21 Martin Liska <mliska@suse.cz>
23467 PR gcov-profile/80081
23468 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
23469 * doc/gcc.texi: Include gcov-dump stuff.
23470 * doc/gcov-dump.texi: New file.
23472 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
23474 PR rtl-optimization/79150
23475 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
23476 conditional jump, if the jump is the last insn of the loop.
23478 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23479 Richard Biener <rguenth@suse.de>
23481 PR tree-optimization/79908
23482 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
23483 been cast away, use force_gimple_operand to construct the side
23486 2017-03-21 Martin Liska <mliska@suse.cz>
23488 PR libfortran/79956
23489 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
23492 2017-03-21 Brad Spengler <spender@grsecurity.net>
23495 * plugin.c (htab_hash_plugin): New function.
23496 (add_new_plugin): Use it and adjust.
23497 (parse_plugin_arg_opt): Adjust.
23498 (init_one_plugin): Likewise.
23500 2017-03-21 Richard Biener <rguenther@suse.de>
23502 PR tree-optimization/80032
23503 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
23504 if set force the cleanup to happen unconditionally.
23505 (gimplify_target_expr): Push inserted clobbers with force_uncond
23506 to avoid them being removed by control-dependent DCE.
23508 2017-03-21 Richard Biener <rguenther@suse.de>
23510 PR tree-optimization/80122
23511 * tree-inline.c (copy_bb): Do not expans va-arg packs or
23512 va_arg_pack_len when the inlined call stmt requires pack
23514 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
23516 2017-03-21 Jakub Jelinek <jakub@redhat.com>
23519 * tsan.c (instrument_builtin_call): If the memory model argument
23520 is not a constant, assume it is valid.
23523 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
23526 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
23528 PR rtl-optimization/79910
23529 * combine.c (can_combine_p): Do not allow combining an I0 or I1
23530 if its dest is used by an insn before I2 (other than the combined
23531 insns themselves, which are properly handled already).
23533 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
23536 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
23538 * combine.c (record_used_regs): New static function.
23539 (try_combine): Handle situations where there is an additional
23540 instruction between I2 and I3 which needs to have a LOG_LINK
23544 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
23546 * combine.c (try_combine): Delete redundant i1 test. Call
23547 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
23549 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23552 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
23553 alternatives 13/14.
23555 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23557 PR tree-optimization/80054
23558 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
23559 the optimization if a PHI or any of its arguments is not dominated
23560 by the candidate's basis. Use gphi* rather than gimple* as
23562 (replace_profitable_candidates): Clean up a gimple* variable that
23563 should be a gphi* variable.
23565 2017-03-20 Martin Sebor <msebor@redhat.com>
23568 * doc/extend.texi (attribute constructor): Document present limitation.
23570 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
23573 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
23574 __POWER9_VECTOR__ #ifdef control, change template definition to
23575 use Power9-specific built-in function.
23576 (vec_any_eq): Likewise.
23577 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
23578 to control outcomes from this test.
23579 (vector_ae_<mode>p): For VEC_F modes, likewise.
23581 2017-03-20 Ian Lance Taylor <iant@google.com>
23583 * config/i386/i386.c (ix86_function_regparm): Save an extra
23584 register for -fsplit-stack with DECL_STATIC_CHAIN.
23586 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
23589 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
23590 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
23592 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
23594 * config/riscv/riscv.c (riscv_print_operand): Use "fence
23596 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
23599 2017-03-20 Marek Polacek <polacek@redhat.com>
23602 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
23604 2017-03-20 Richard Biener <rguenther@suse.de>
23606 PR tree-optimization/80113
23607 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
23608 allocate extra SSA name for PHI def.
23609 (add_close_phis_to_outer_loops): Likewise.
23610 (add_close_phis_to_merge_points): Likewise.
23611 (copy_loop_close_phi_args): Likewise.
23612 (copy_cond_phi_nodes): Likewise.
23614 2017-03-20 Martin Liska <mliska@suse.cz>
23616 PR middle-end/79753
23617 * tree-chkp.c (chkp_build_returned_bound): Do not build
23618 returned bounds for a LHS that's not a BOUNDED_P type.
23620 2017-03-20 Martin Liska <mliska@suse.cz>
23624 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
23625 COMPLEX_CST and VECTOR_CST.
23627 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23630 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
23631 target operand. A new splitter adds the clobber statement in case
23632 the target operand is dead anyway.
23634 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
23636 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
23637 to age-old versions of binutils and glibc.
23639 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
23641 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
23643 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
23645 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
23647 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
23649 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
23650 requirement for binutils 2.13.
23652 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
23654 * combine.c (try_combine): Delete redundant i1 test. Call
23655 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
23657 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
23659 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
23660 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
23662 <riscv64-*-elf>: Re-arrange section
23663 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
23664 <riscv32-*-linux>: Likewise.
23665 <riscv64-*-elf>: Likewise
23666 <riscv64-*-linux>: Likewise.
23668 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
23671 * aarch64.opt(verbose-cost-dump): Fix typo.
23673 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
23676 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
23677 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
23679 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
23681 * reload.c (find_reloads): When reloading a nonoffsettable address,
23682 use RELOAD_OTHER for it and its address reloads.
23684 PR rtl-optimization/79910
23685 * combine.c (record_used_regs): New static function.
23686 (try_combine): Handle situations where there is an additional
23687 instruction between I2 and I3 which needs to have a LOG_LINK
23690 2017-03-17 Jeff Law <law@redhat.com>
23692 PR tree-optimization/71437
23693 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
23694 conditional in the hash table first.
23695 (vrp_dom_walker::before_dom_children): Extract condition from
23696 ASSERT_EXPR. Record condition, its inverion and any implied
23697 conditions as well.
23699 2017-03-17 Marek Polacek <polacek@redhat.com>
23700 Markus Trippelsdorf <markus@trippelsdorf.de>
23702 PR tree-optimization/80079
23703 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
23706 2017-03-17 Richard Biener <rguenther@suse.de>
23708 PR middle-end/80075
23709 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
23710 Properly verify the LHS before the RHS possibly claims to be
23712 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
23715 2017-03-17 Martin Jambor <mjambor@suse.cz>
23717 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
23718 (List of -O2 options): Likewise.
23719 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
23722 2017-03-17 Tom de Vries <tom@codesourcery.com>
23724 * gcov-dump.c (print_usage): Print bug_report_url.
23726 2017-03-17 Richard Biener <rguenther@suse.de>
23728 PR middle-end/80050
23729 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
23730 (parser::peek): Likewise.
23732 2017-03-17 Richard Biener <rguenther@suse.de>
23734 PR tree-optimization/80048
23735 * sese.c (free_sese_info): Properly release rename_map and
23736 copied_bb_map elements.
23738 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
23740 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
23741 Add linked-list forward and backlinks. Insert on
23742 construction, remove on destruction.
23743 (class pass_store_merging): Add m_stores_head field.
23744 (pass_store_merging::terminate_and_process_all_chains):
23745 Iterate over m_stores_head list.
23746 (pass_store_merging::terminate_all_aliasing_chains):
23748 (pass_store_merging::execute): Check for debug stmts first.
23749 Push new chains onto the m_stores_head stack.
23751 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
23754 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
23755 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
23756 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
23758 2017-03-16 Jeff Law <law@redhat.com>
23760 PR tree-optimization/71437
23761 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
23762 member function. Implementation moved into after_dom_children
23763 member function and into the threader's thread_outgoing_edges
23765 (dom_opt_dom_walker::after_dom_children): Simplify by moving
23766 some code into new thread_outgoing_edges.
23767 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
23768 definition. Simplify marker handling (do it here). Assume we always
23769 have the available expression and the const/copies tables.
23770 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
23772 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
23773 * tree-vrp.c (equiv_stack): No longer file scoped.
23774 (vrp_dom_walker): New class.
23775 (vrp_dom_walker::before_dom_children): New member function.
23776 (vrp_dom_walker::after_dom_children): Likewise.
23777 (identify_jump_threads): Setup domwalker. Use it rather than
23778 walking edges in a random order by hand. Simplify setup/finalization.
23779 (finalize_jump_threads): Remove.
23780 (vrp_finalize): Do not call identify_jump_threads here.
23781 (execute_vrp): Do it here instead and call thread_through_all_blocks
23784 PR tree-optimization/71437
23785 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
23787 (simplify_stmt_for_jump_threading): Add basic_block argument. All
23789 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
23790 (dom_opt_dom_walker::thread_across_edge): Remove
23791 handle_dominating_asserts argument. All callers changed.
23792 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
23793 changes. Remove calls to lhs_of_dominating_assert. Other
23794 uses of handle_dominating_asserts turn into unconditional code
23795 (simplify_control_stmt_condition_1): Likewise.
23796 (simplify_control_stmt_condition): Likewise.
23797 (thread_through_normal_block, thread_across_edge): Likewise.
23798 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
23799 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
23800 object if it is not an SSA_NAME.
23801 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
23802 before calling into the VRP specific simplifiers.
23803 (identify_jump_threads): Remove handle_dominating_asserts
23806 2017-03-16 Jakub Jelinek <jakub@redhat.com>
23809 * tree-diagnostic.c (default_tree_printer): No longer static.
23810 * tree-diagnostic.h (default_tree_printer): New prototype.
23812 2017-03-16 Tamar Christina <tamar.christina@arm.com>
23814 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
23815 Change ins into fmov.
23817 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23819 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
23820 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
23821 Use h_con constraint for operand 1.
23822 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
23823 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
23825 2017-03-15 Jeff Law <law@redhat.com>
23827 PR tree-optimization/71437
23828 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
23829 (record_temporary_equivalences): Use it.
23831 PR tree-optimization/71437
23832 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
23833 tree-ssa-scopedtables.
23834 (lookup_avail_expr, build_and_record_new_cond): Likewise.
23835 (record_conditions, record_cond, vuse_eq): Likewise.
23836 (record_edge_info): Adjust to API tweak of record_conditions.
23837 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
23838 (record_temporary_equivalences, optimize_stmt): Likewise.
23839 (eliminate_redundant_computations): Likewise.
23840 (record_equivalences_from_stmt): Likewise.
23841 * tree-ssa-scopedtables.c: Include options.h and params.h.
23842 (vuse_eq): New function, moved from tree-ssa-dom.c
23843 (build_and_record_new_cond): Likewise.
23844 (record_conditions): Likewise. Accept vector of conditions rather
23845 than edge_equivalence structure for first argument.
23846 for the first argument.
23847 (avail_exprs_stack::lookup_avail_expr): New member function, moved
23848 from tree-ssa-dom.c.
23849 (avail_exprs_stack::record_cond): Likewise.
23850 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
23851 from tree-ssa-dom.c.
23852 (avail_exprs_stack): Add new member functions lookup_avail_expr
23854 (record_conditions): Declare.
23856 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
23859 * lra-constraints.c (process_alt_operands): Increase reject for
23860 reloading an input/output operand.
23862 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
23865 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
23866 insns to convert from signed/unsigned char/short to IEEE 128-bit
23868 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
23870 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
23873 * config/i386/i386.c (ix86_vector_duplicate_value): Create
23874 subreg of inner mode for values already in registers.
23876 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
23878 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
23879 iteration reg is used after the loop.
23881 2017-03-14 Martin Sebor <msebor@redhat.com>
23883 PR tree-optimization/79800
23884 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
23885 precision in negative-positive range.
23886 (format_floating): Call non-const overload with adjusted precision.
23888 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
23891 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
23894 2017-03-14 Martin Sebor <msebor@redhat.com>
23896 PR middle-end/80020
23897 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
23898 * builtins.def (aligned_alloc): Use it.
23901 * Makefile.in (GTFILES): Add calls.c.
23902 * calls.c: Include "gt-calls.h".
23904 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
23906 PR rtl-optimization/79728
23907 * regs.h (struct target_regs): New field
23908 x_contains_allocatable_regs_of_mode.
23909 (contains_allocatable_regs_of_mode): New macro.
23910 * reginfo.c (init_reg_sets_1): Initialize it, and change
23911 contains_reg_of_mode so it includes global regs as well.
23912 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
23913 rather than contains_regs_of_mode.
23915 2017-03-14 Martin Liska <mliska@suse.cz>
23917 * doc/invoke.texi: Document options that can't be combined with
23918 -fcheck-pointer-bounds.
23920 2017-03-14 Martin Liska <mliska@suse.cz>
23922 PR middle-end/79831
23923 * doc/invoke.texi (-Wchkp): Document the option.
23925 2017-03-14 Martin Liska <mliska@suse.cz>
23927 * Makefile.in: Install gcov-dump.
23929 2017-03-14 Martin Liska <mliska@suse.cz>
23931 * multiple_target.c (expand_target_clones): Bail out for
23932 an invalid attribute.
23934 2017-03-14 Richard Biener <rguenther@suse.de>
23936 * alias.c (struct alias_set_entry): Pack properly.
23937 * cfgloop.h (struct loop): Likewise.
23938 * cse.c (struct set): Likewise.
23939 * ipa-utils.c (struct searchc_env): Likewise.
23940 * loop-invariant.c (struct invariant): Likewise.
23941 * lra-remat.c (struct cand): Likewise.
23942 * recog.c (struct change_t): Likewise.
23943 * rtl.h (struct address_info): Likewise.
23944 * symbol-summary.h (function_summary): Likewise.
23945 * tree-loop-distribution.c (struct partition): Likewise.
23946 * tree-object-size.c (struct object_size_info): Likewise.
23947 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
23948 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
23949 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
23950 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
23951 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
23952 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
23953 (struct _stmt_vec_info): Likewise.
23955 2017-03-14 Martin Liska <mliska@suse.cz>
23958 * multiple_target.c (create_dispatcher_calls): Check that
23959 a target can create a function dispatcher.
23961 2017-03-14 Martin Liska <mliska@suse.cz>
23964 * multiple_target.c (expand_target_clones): Drop local.local
23965 flag for default implementation.
23967 2017-03-14 Richard Biener <rguenther@suse.de>
23969 PR tree-optimization/80030
23970 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
23972 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
23974 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
23975 gcc_fallthrough() instead of __attribute__((fallthrough));
23977 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
23979 * doc/gcc.texi: Remove "up" link to (DIR).
23980 * doc/gccint.texi: Ditto.
23982 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
23984 * doc/install.texi (Specific) <avr>: Remove reference to
23987 2017-03-13 Jeff Law <law@redhat.com>
23989 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
23990 attribute rather than comments.
23992 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
23993 match_scratch operand is highest.
23995 2017-03-13 Martin Liska <mliska@suse.cz>
23997 PR middle-end/78339
23998 * ipa-pure-const.c (warn_function_noreturn): If the declarations
23999 is a CHKP clone, use original declaration.
24001 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24003 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
24004 (arc_conditional_register_usage): Use a different allocation order
24005 when optimizing for size.
24006 * common/config/arc/arc-common.c (arc_option_optimization_table):
24007 Section anchors default on when optimizing for size.
24009 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24011 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
24013 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24015 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
24016 * config/arc/arc.md (cpu_facility): Add cd variant.
24017 (*movqi_insn): Add code density variant.
24018 (*movhi_insn): Likewise.
24019 (*movqi_insn): Likewise.
24020 (*addsi3_mixed): Likewise.
24021 (subsi3_insn): Likewise.
24023 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24025 * config/arc/arc.md (movsi_cond_exec): Update constraint.
24027 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24029 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
24030 expressions with MINUS and UNARY ops.
24032 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24035 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
24037 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
24038 between vec_select and vector argument.
24039 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
24040 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
24041 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
24042 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
24043 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
24044 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
24046 2017-03-13 Richard Biener <rguenther@suse.de>
24049 * params.def (vect-max-peeling-for-alignment): Fix typo.
24051 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
24053 * doc/install.texi (Specific) <mips-*-*>: Remove description of
24054 issue that only occurred with binutils below 2.18.
24056 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
24058 * doc/install.texi (Specific) <cris-axis-elf>: No longer
24059 refer to binutils 2.11/2.12 minimum.
24061 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
24063 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
24064 ftp.kernel.org and simplify binutils requirement.
24066 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
24068 * doc/invoke.texi (Warning Options): Fix spelling of link-time
24070 (Optimize Options): Ditto. Also remove redundancy.
24072 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24074 PR translation/79848
24075 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
24077 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
24078 to G_ to avoid double translation.
24080 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24082 PR translation/79923
24083 * auto-profile.c (get_combined_location): Convert leading
24084 character of diagnostics to lower case and remove trailing period.
24085 (read_profile): Likewise for various diagnostics.
24086 * config/arm/arm.c (arm_option_override): Remove trailing period
24087 from various diagnostics.
24088 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
24089 (msp430_expand_delay_cycles): Likewise.
24091 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24094 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
24095 full command-line argument, rather than just "str".
24096 (aarch64_validate_march): Likewise.
24097 (aarch64_validate_mtune): Likewise.
24099 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
24101 PR rtl-optimization/78911
24102 * lra-assigns.c (must_not_spill_p): New function.
24103 (spill_for): Use it.
24105 2017-03-10 Jakub Jelinek <jakub@redhat.com>
24107 PR tree-optimization/79981
24108 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
24109 ATOMIC_COMPARE_EXCHANGE ifn result.
24110 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
24111 IFN_ATOMIC_COMPARE_EXCHANGE.
24113 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24116 * opts.c (parse_sanitizer_options): Add missing question mark to
24117 "did you mean" message.
24119 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24121 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
24123 (VMULEUH_UNS): Likewise.
24124 (VMULOUB_UNS): Likewise.
24125 (VMULOUH_UNS): Likewise.
24126 * config/rs6000/rs6000.c (builtin_function_type): Remove
24127 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
24129 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24132 * read-rtl-function.c (function_reader::read_rtx_operand): Update
24133 x with result of extra_parsing_for_operand_code_0.
24134 (function_reader::extra_parsing_for_operand_code_0): Convert
24135 return type from void to rtx, returning x. When reading
24136 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
24137 larger size containing struct block_symbol.
24139 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
24141 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
24142 -mfloat128-hardware without -m64.
24144 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
24147 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
24148 entries to the case statement that marks unsigned arguments to
24149 overloaded functions.
24151 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
24153 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
24154 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
24156 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
24159 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
24160 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
24162 2017-03-10 Martin Liska <mliska@suse.cz>
24166 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
24167 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
24170 2017-03-10 Olivier Hainque <hainque@adacore.com>
24172 * tree-switch-conversion (array_value_type): Start by resetting
24173 candidate type to it's main variant.
24175 2017-03-10 Jakub Jelinek <jakub@redhat.com>
24177 PR rtl-optimization/79909
24178 * combine.c (try_combine): Use simplify_replace_rtx on individual
24179 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
24180 of the whole CALL_INSN_FUNCTION_USAGE.
24182 PR tree-optimization/79972
24183 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
24184 get_range_info on SSA_NAMEs. Formatting fixes.
24186 2017-03-10 Richard Biener <rguenther@suse.de>
24187 Jakub Jelinek <jakub@redhat.com>
24189 PR tree-optimization/77975
24190 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
24191 edge to be constant.
24192 (get_val_for): For constant x return it. Formatting fix.
24193 (loop_niter_by_eval): Avoid pointless looping if the next iteration
24194 would use the same bases as the current one.
24196 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24198 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
24199 instead of vec_select for V1TImode.
24200 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
24202 (VSX_LE_128): Add V1TI to this mode iterator.
24203 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
24204 (*vsx_le_perm_store_<mode>): Likewise.
24205 (pre-reload splitter for VSX stores): Likewise.
24206 (post-reload splitter for VSX stores): Likewise.
24207 (*vsx_xxpermdi2_le_<mode>): Likewise.
24208 (*vsx_lxvd2x2_le_<mode>): Likewise.
24209 (*vsx_stxvd2x2_le_<mode>): Likewise.
24211 2017-03-09 Michael Eager <eager@eagercon.com>
24213 Correct failures with --enable-checking=yes,rtl.
24215 * config/microblaze/microblaze.c (microblaze_expand_shift):
24216 Replace GET_CODE test with CONST_INT_P and INTVAL test with
24217 test for const0_rtx.
24218 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
24219 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
24221 2017-03-09 Richard Biener <rguenther@suse.de>
24223 PR tree-optimization/79977
24224 * graphite-scop-detection.c (scop_detection::merge_sese):
24225 Handle the case of extra exits to blocks dominating the entry.
24227 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
24229 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
24232 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
24234 PR rtl-optimization/79949
24235 * lra-constraints.c (process_alt_operands): Check memory when
24236 trying to predict a cycle. Print about the overall increase.
24238 2017-03-09 Richard Biener <rguenther@suse.de>
24240 PR middle-end/79971
24241 * gimple-expr.c (useless_type_conversion_p): Preserve
24242 TYPE_SATURATING for fixed-point types.
24244 2017-03-09 Richard Biener <rguenther@suse.de>
24247 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
24248 alignment of BLKmode params.
24250 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24253 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
24254 (VALL_NO_V2Q): Likewise.
24256 * config/aarch64/aarch64-simd.md
24257 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
24259 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
24260 VALL_NO_V2Q mode iterator.
24261 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
24263 2017-03-09 Martin Liska <mliska@suse.cz>
24265 PR tree-optimization/79631
24266 * tree-chkp-opt.c (chkp_is_constant_addr): Call
24267 tree_int_cst_sign_bit just for INTEGER constants.
24269 2017-03-09 Martin Liska <mliska@suse.cz>
24273 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
24276 2017-03-09 Marek Polacek <polacek@redhat.com>
24279 * tree.c (inchash::add_expr): Handle TREE_VEC.
24281 2017-03-09 Martin Liska <mliska@suse.cz>
24284 (chkp_narrow_size_and_offset): New function.
24285 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
24286 (void chkp_parse_bit_field_ref): New function.
24287 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
24288 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
24290 2017-03-09 Martin Liska <mliska@suse.cz>
24293 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
24294 (chkp_find_bounds_1): Remove gcc_unreachable.
24296 2017-03-09 Jakub Jelinek <jakub@redhat.com>
24299 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
24300 BUILT_IN_SYNC*, determine the access type from the size suffix and
24301 always build a MEM_REF with that type. Handle forgotten
24302 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
24305 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
24306 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
24307 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
24308 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
24309 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
24310 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
24311 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
24312 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
24313 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
24314 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
24315 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
24316 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
24317 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
24318 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
24319 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
24320 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
24321 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
24322 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
24323 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
24324 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
24325 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
24326 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
24327 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
24328 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
24329 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
24330 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
24331 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
24332 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
24333 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
24334 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
24335 definitions outside of __OPTIMIZE__ guarded section.
24338 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
24339 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
24340 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
24341 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
24344 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24346 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
24347 ("vfenez<mode>"): Add missing constraints.
24349 2017-03-08 Martin Sebor <msebor@redhat.com>
24352 * config/nds32/nds32.c (nds32_option_override):
24353 Fix misspelled diagnostic.
24355 2017-03-08 Jakub Jelinek <jakub@redhat.com>
24358 * gimplify.c (gimplify_omp_for): Replace index var in outer
24359 taskloop statement with an artificial variable and add
24360 OMP_CLAUSE_PRIVATE clause for it.
24362 2017-03-08 Richard Biener <rguenther@suse.de>
24364 PR tree-optimization/79955
24365 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
24366 for accesses that are completely outside of the variable.
24368 2017-03-08 Andrew Haley <aph@redhat.com>
24370 PR tree-optimization/79943
24371 * tree-ssa-loop-split.c (compute_new_first_bound): When
24372 calculating the new upper bound, (END-BEG) should be added, not
24375 2017-03-08 Jakub Jelinek <jakub@redhat.com>
24377 * config/avr/avr.md (setmemhi): Make sure match_dup
24378 operand number comes before match_scratch.
24380 2017-03-08 Richard Biener <rguenther@suse.de>
24382 PR tree-optimization/79920
24383 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
24384 with ncopies == 1 to ...
24385 (vect_transform_slp_perm_load): ... here. Properly compute
24386 all element loads by iterating VF times over the group. Do
24387 not handle ncopies (computed in a broken way) in
24388 vect_create_mask_and_perm.
24390 2017-03-08 Jakub Jelinek <jakub@redhat.com>
24393 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
24394 is a uniform vector, use uniform_vector_p return value instead of
24395 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
24397 2017-03-07 Marek Polacek <polacek@redhat.com>
24399 PR middle-end/79809
24400 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
24401 (alloca_call_type): Likewise.
24403 2017-03-07 Martin Liska <mliska@suse.cz>
24405 * gcov.c (process_args): Put comment to correct location.
24407 2017-03-07 Martin Liska <mliska@suse.cz>
24409 PR middle-end/68270
24410 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
24411 Use array_at_struct_end_p instead of DECL_CHAIN (field).
24412 (chkp_narrow_bounds_for_field): Likewise.
24413 (chkp_parse_array_and_component_ref): Pass one more argument to
24416 2017-03-07 Richard Biener <rguenther@suse.de>
24418 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
24421 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
24423 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
24424 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
24426 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24429 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
24430 to end of description.
24431 (PARAM_MAX_STORES_TO_MERGE): Likewise.
24433 2017-03-07 Jakub Jelinek <jakub@redhat.com>
24435 PR rtl-optimization/79901
24436 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
24438 (*avx512f_<code><mode>3<mask_name>): ... this.
24439 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
24440 iterator instead of VI8_AVX2_AVX512BW.
24442 PR rtl-optimization/79901
24443 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
24444 min/max expander, expand it using expand_vec_cond_expr.
24447 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
24450 2017-03-06 Jakub Jelinek <jakub@redhat.com>
24453 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
24454 to void * for PCH reasons.
24455 * dwarf2out.c (output_loc_operands, output_die): Cast
24456 v.val_vec.array to unsigned char *.
24458 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
24461 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
24464 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
24466 PR rtl-optimization/79571
24467 * lra-constraints.c (process_alt_operands): Calculate static
24468 reject and subtract it from overall when only addresses will be
24471 2017-03-06 Julia Koval <julia.koval@intel.com>
24474 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
24475 incoming stack boundary to 128 for 64-bit targets.
24477 2017-03-06 Richard Biener <rguenther@suse.de>
24479 PR tree-optimization/79894
24480 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
24481 to NULL after folding it.
24483 2017-03-06 Richard Biener <rguenther@suse.de>
24485 PR tree-optimization/79824
24486 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
24487 check disabling peeling for gaps.
24489 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
24491 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
24492 attributes): Document gettimeofday.
24494 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
24496 * config/s390/s390.c (s390_option_override_internal): Set
24497 PARAM_MIN_VECT_LOOP_BOUND
24499 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
24501 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
24502 * config/s390/s390.md: Likewise.
24504 2017-03-06 Jakub Jelinek <jakub@redhat.com>
24507 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
24508 (<avx2_avx512>_perm<mode>): Rename to ...
24509 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
24511 (<avx512>_perm<mode>_mask): Rename to ...
24512 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
24514 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
24515 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
24516 instead of VI8F_256_512.
24517 (avx512f_perm<mode>): New define_expand.
24518 (avx512f_perm<mode>_mask): Likewise.
24519 (avx512f_perm<mode>_1<mask_name>): New define_insn.
24520 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
24522 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
24524 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
24525 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
24527 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
24529 2017-03-06 Martin Liska <mliska@suse.cz>
24532 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
24533 when having a SSA NAME w/o VAR_DECL assigned to it.
24535 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
24537 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
24538 msa_dpsub_<su>_d): Fix MODE for vec_select.
24540 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
24542 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
24544 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
24546 2017-03-06 Richard Biener <rguenther@suse.de>
24548 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
24549 * plugin.c (register_plugin_info): Likewise.
24550 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
24552 2017-03-05 Jakub Jelinek <jakub@redhat.com>
24554 * config/i386/sse.md (sse_storehps, sse_storelps,
24555 avx_<castmode><avxsizesuffix>_<castmode>,
24556 avx512f_<castmode><avxsizesuffix>_<castmode>,
24557 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
24558 in condition that at least one operand is not a MEM.
24560 2017-03-03 Jakub Jelinek <jakub@redhat.com>
24562 PR middle-end/79805
24563 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
24564 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
24566 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
24567 gimple_call_nothrow_p flag based on whether original builtin can throw.
24568 If it can, emit following stmts on the fallthrough edge.
24569 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
24570 don't create new bb if inserting just debug stmts on the edge, try to
24571 insert them on the fallthru bb or just reset debug stmts.
24573 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
24576 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
24577 restore recog_data (including the operand rtxes inside it) around
24578 the call to get_insn_template.
24580 2017-03-03 Martin Sebor <msebor@redhat.com>
24582 PR tree-optimization/79699
24583 * context.c (context::~context): Free MPFR caches to avoid
24584 a memory leak on program exit.
24586 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24588 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
24589 Use wide_int::ulow () instead of .elt (0).
24591 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
24593 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
24594 (*pushxf): Limit oF constraint to 32bit targets and add oC
24595 constraint for 64bit targets.
24596 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
24597 (*pushdf): Change rmF constraint to rmC.
24599 2017-03-03 Martin Liska <mliska@suse.cz>
24601 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
24602 Remove unused variable.
24604 2017-03-03 Jakub Jelinek <jakub@redhat.com>
24607 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
24608 is a memory operand, increase num_memory.
24609 (ix86_expand_args_builtin): Likewise.
24611 2017-03-03 Jan Hubicka <jh@suse.cz>
24614 * ipa-devirt.c (maybe_record_node): Properly handle
24615 __cxa_pure_virtual visibility.
24617 2017-03-03 Martin Liska <mliska@suse.cz>
24619 PR tree-optimization/79803
24620 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
24622 (pass_loop_prefetch::execute): Disabled optimization if an
24623 assumption about L1 cache size is not met.
24625 2017-03-03 Martin Liska <mliska@suse.cz>
24627 PR rtl-optimization/79574
24628 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
24629 (hash_scan_set): Likewise.
24630 (dump_hash_table): Likewise.
24631 (hoist_code): Likewise.
24633 2017-03-03 Richard Biener <rguenther@suse.de>
24635 * fixed-value.c (fixed_from_string): Restore use of elt (1)
24636 in place of uhigh ().
24637 (fixed_convert_from_real): Likewise.
24639 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
24642 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
24644 2017-03-03 Richard Biener <rguenther@suse.de>
24646 PR middle-end/79818
24647 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
24648 TYPE_OVERFLOW_UNDEFINED check.
24650 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24652 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
24654 (vector_ae_<mode>_p): Likewise.
24655 (vector_nez_<mode>_p): Likewise.
24656 (vector_ne_v2di_p): Likewise.
24657 (vector_ae_v2di_p): Likewise.
24658 (vector_ne_<mode>_p): Likewise.
24659 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
24661 (vsx_tsqrt<mode>2_fe): Likewise.
24663 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
24666 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
24668 2017-03-02 Jakub Jelinek <jakub@redhat.com>
24670 PR rtl-optimization/79780
24671 * cprop.c (one_cprop_pass): When second and further conditional trap
24672 in a single basic block is turned into an unconditional trap, turn it
24673 into a deleted note to avoid RTL verification failures.
24675 2017-03-02 Richard Biener <rguenther@suse.de>
24677 * fold-const.c (const_binop): Use ulow () instead of elt (0).
24679 2017-03-02 Richard Biener <rguenther@suse.de>
24681 PR tree-optimization/79345
24683 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
24684 param and abort the walk, returning -1 if it is hit.
24685 (walk_aliased_vdefs): Take a limit param and pass it on.
24686 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
24687 defaulting to 0 and return a signed int.
24688 * tree-ssa-uninit.c (struct check_defs_data): New struct.
24689 (check_defs): New helper.
24690 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
24691 about uninitialized memory.
24692 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
24693 bogus uninitialized warning.
24694 (fixed_convert_from_real): Likewise.
24696 2017-03-02 Bin Cheng <bin.cheng@arm.com>
24698 PR tree-optimization/66768
24699 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
24700 iv_use if base object can't be determined.
24702 2017-03-02 Jakub Jelinek <jakub@redhat.com>
24704 PR tree-optimization/79345
24705 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
24706 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
24707 (get_pattern_stats): Initialize it.
24708 * genemit.c (gen_expand): Verify match_scratch numbers come after
24709 match_operand/match_dup numbers.
24710 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
24711 match_scratch numbers.
24712 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
24714 * config/s390/s390.md (trunctdsd2): Likewise.
24716 2017-03-02 Richard Biener <rguenther@suse.de>
24718 * wide-int.h (wide_int_storage::operator=): Implement in terms
24721 2017-03-02 Richard Biener <rguenther@suse.de>
24723 PR tree-optimization/79777
24724 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
24725 the to insert expression to sth existing.
24727 2017-03-01 Martin Sebor <msebor@redhat.com>
24729 PR middle-end/79692
24730 * gimple-ssa-sprintf.c
24731 (directive::known_width_and_precision): New function.
24732 (format_integer): Use it.
24733 (get_mpfr_format_length): Consider the full range of precision
24734 when computing %g output with the # flag. Set the likely byte
24735 count to 3 rather than 1 when precision is indeterminate.
24736 (format_floating): Correct the lower bound of precision.
24738 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24740 * doc/invoke.texi: Document default code model for 64-bit Linux.
24742 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24745 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
24746 udiv rather than div since input pattern is unsigned.
24748 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
24750 * config/i386/i386.c (print_reg): Warn for values of
24751 unsupported size in integer register.
24753 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
24756 * config/rs6000/predicates.md (current_file_function_operand): Do
24757 not allow self calls to be local if the function is replaceable.
24759 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
24762 * config/rs6000/altivec.h (vec_ctz and others): Change the
24763 preprocessor macro that controls conditional compilation from
24764 _ARCH_PWR9 to __POWER9_VECTOR__.
24765 (vec_all_ne): Change parameterization of __altivec_scalar_pred
24766 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
24767 control (instead of _ARCH_PWR9 control) so that template
24768 definition uses power9-specific function.
24769 (vec_any_eq): Likewise.
24770 (vec_all_ne): Change macro definition to use a power9-specific
24771 expansion under #ifdef __POWER9_VECTOR__ control (instead of
24772 _ARCH_PWR9 control).
24773 (vec_any_eq) Likewise.
24774 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
24775 expansion for CMPNEF to remove support for xvcmpnesp instruction.
24776 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
24777 support for xvcmpnedp instruction.
24778 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
24779 macro expansion so that Power9 implementation of vec_all_ne does
24780 not use the AltiVec predicate framework.
24781 (VCMPNEH_P): Likewise.
24782 (VCMPNEW_P): Likewise.
24783 (VCMPNED_P): Likewise.
24784 (VCMPNEFP_P): Likewise.
24785 (VCMPNEDP_P): Likewise.
24786 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
24787 implementation of vec_any_eq to not use AltiVec predicate
24789 (VCMPAEH_P): Likewise.
24790 (VCMPAEW_P): Likewise.
24791 (VCMPAED_P): Likewise.
24792 (VCMPAEFP_P): Likewise.
24793 (VCMPAEDP_P): Likewise.
24794 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
24795 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
24796 not use the AltiVec predicate framework.
24797 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
24798 of vec_any_eq to not use AltiVec predicate framework.
24799 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
24800 support for predefined __POWER9_VECTOR__ macro to indicate that
24801 Power9 instruction selection is enabled.
24802 (altivec_overloaded_builtins): Remove extraneous
24803 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
24804 function argument types RS6000_BTI_bool_V16QI and
24805 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
24806 entry for overloaded function argument types RS6000_BTI_bool_V4SI
24807 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
24808 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
24809 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
24810 Power9 for implementations of vec_cmpne. Change the signature for
24811 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
24812 (representing vec_all_ne) to remove the previously described first
24813 argument of type RS6000_BTI_INTSI, as this was an artifact of
24814 reliance on the AltiVec predicate framework, which is no longer
24815 used in the implementation of these functions. Add
24816 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
24817 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
24818 since, unlike the AltiVec predicate framework implementation, we
24819 do not share function descriptors between vec_alle and vec_anyeq.
24820 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
24821 set of modes that receive special treatment even when
24822 TARGET_P9_VECTOR is true. The special treatment emits code that
24823 does not depend on Power9 instructions.
24824 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
24825 define_expand to not rely on AltiVec predicate framework.
24826 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
24828 (vector_ne_v2di_p): Change this define_expand to not rely on
24829 AltiVec predicate framework.
24830 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
24832 (vector_ne_<mode>_p): Change this define_expand to not rely on
24833 AltiVec predicate framework.
24834 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
24836 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
24837 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
24838 define_insn pattern.
24839 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
24840 define_insn pattern because the xvcmpne<VSs>. instruction is not
24842 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
24843 instruction is not supported.
24845 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24847 * config/nvptx/nvptx.c: Include intl.h.
24849 2017-03-01 Martin Jambor <mjambor@suse.cz>
24852 * ipa-prop.h (ipa_bits): Removed field known.
24853 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
24854 to pointers. Adjusted their comments to warn about their sharing.
24855 (ipcp_transformation_summary): Change bits to a vector of pointers.
24856 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
24857 (ipa_get_ipa_bits_for_value): Declare.
24858 * tree-vrp.h (value_range): Mark as GTY((for_user)).
24859 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
24860 (ipa_bits_hash_table): Likewise.
24861 (ipa_vr_ggc_hash_traits): Likewise.
24862 (ipa_vr_hash_table): Likewise.
24863 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
24864 being pointers and vr_known being removed.
24865 (ipa_set_jf_unknown): Likewise.
24866 (ipa_get_ipa_bits_for_value): New function.
24867 (ipa_set_jfunc_bits): Likewise.
24868 (ipa_get_value_range): New overloaded functions.
24869 (ipa_set_jfunc_vr): Likewise.
24870 (ipa_compute_jump_functions_for_edge): Use the above functions to
24871 construct bits and vr parts of jump functions.
24872 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
24873 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
24875 (ipcp_grow_transformations_if_necessary): Also allocate
24876 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
24878 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
24879 them. Fix too long lines.
24880 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
24881 vr_known being removed.
24882 (ipa_read_jump_function): Use new setter functions to construct bits
24883 and vr parts of jump functions or set them to NULL.
24884 (write_ipcp_transformation_info): Adjust for bits being pointers.
24885 (read_ipcp_transformation_info): Likewise.
24886 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
24888 Include gt-ipa-prop.h.
24889 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
24891 (ipcp_store_bits_results): Likewise.
24892 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
24893 Do not write to existing jump functions but use a temporary instead.
24895 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24898 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
24899 attempt to use its first operand as BIT_FIELD_REF base.
24901 2017-03-01 Richard Biener <rguenther@suse.de>
24903 PR middle-end/79721
24904 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
24905 interpolating formula in wrapping arithmetic.
24906 (chrec_apply): Convert chrec_evaluate return value to wanted type.
24908 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24910 PR tree-optimization/79734
24911 * tree-vect-generic.c (expand_vector_condition): Optimize
24912 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
24913 Handle VEC_COND_EXPR where comparison has different inner width from
24914 type's inner width.
24916 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
24918 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
24919 markup, and similar issues. Remove @opindex entries for things
24920 that aren't options. Add missing -mmpy-option entries.
24922 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24924 PR tree-optimization/79737
24925 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
24926 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
24927 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
24928 instead of byte_size. Formatting fix.
24929 (shift_bytes_in_array_right): Formatting fix.
24931 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
24934 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
24935 condition on optimize for the leaf function test.
24937 2017-02-28 Martin Liska <mliska@suse.cz>
24940 * read-rtl-function.c (function_reader::handle_unknown_directive):
24941 Bail out when one uses -flto.
24943 2017-02-28 Martin Liska <mliska@suse.cz>
24945 * common.opt: Replace space with tabular for options of <number>
24947 * config/i386/i386.opt: Show <number> value for
24948 -mlarge-data-threshold.
24949 * opts.c (print_filtered_help): Do not display number in hexadecimal
24952 2017-02-28 Martin Liska <mliska@suse.cz>
24954 * common.opt: Fix --help=option -Q for options which are of
24957 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
24959 * config/i386/i386.c (print_reg): Error out for values
24960 of 8-bit size in invalid integer register.
24962 2017-02-28 Martin Sebor <msebor@redhat.com>
24964 PR tree-optimization/79691
24965 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
24967 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24970 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
24971 gcc_unreachable with output_operand_lossage.
24973 2017-02-28 Richard Biener <rguenther@suse.de>
24975 PR tree-optimization/79740
24976 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
24978 (visit_nary_op): Insert the nary into the hashtable if we
24979 pattern-matched sth.
24980 * tree-ssa-pre.c (eliminate_insert): Robustify.
24982 2017-02-28 Richard Biener <rguenther@suse.de>
24984 PR middle-end/79731
24985 * fold-const.c (decode_field_reference): Reject out-of-bound
24988 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24990 * config/i386/i386.c: Include intl.h.
24991 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
24992 instead of just cond ? "..." : "...".
24993 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
24994 * coverage.c (read_counts_file): Likewise.
24995 * omp-offload.c: Include intl.h.
24996 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
24997 of just cond ? "..." : "...".
24998 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
24999 of just cond ? "..." : "...".
25001 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
25004 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
25006 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
25007 'tune for' CPU name.
25008 * config/arm/arm-cpu-data.h: Regenerated.
25010 2017-02-28 Richard Biener <rguenther@suse.de>
25012 PR tree-optimization/79732
25013 * tree-inline.c (expand_call_inline): Do not shadow var.
25015 2017-02-28 Richard Biener <rguenther@suse.de>
25017 PR tree-optimization/79723
25018 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
25019 address-space properly.
25021 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
25023 * doc/optinfo.texi (Optimization groups): Fix option used for
25025 * doc/invoke.texi (-fopt-info): Document "omp".
25026 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
25027 (OPTGROUP_ALL): Add OPTGROUP_OMP.
25028 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
25029 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
25030 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
25032 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
25034 * dumpfile.c (optgroup_options): Instead of "openmp", associate
25035 OPTGROUP_OMP with "omp".
25037 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
25040 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
25041 for arithmetic shift of unsigned V2DI.
25043 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
25045 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
25046 arc/linux.h headers.
25047 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
25048 (LINK_SPEC): Likewise.
25049 (ARC_TLS_EXTRA_START_SPEC): Likewise.
25050 (EXTRA_SPECS): Likewise.
25051 (STARTFILE_SPEC): Likewise.
25052 (ENDFILE_SPEC): Likewise.
25053 (LIB_SPEC): Likewise.
25054 (TARGET_SDATA_DEFAULT): Likewise.
25055 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
25056 (MULTILIB_DEFAULTS): Likewise.
25057 (DWARF2_UNWIND_INFO): Likewise.
25058 * config/arc/big.h: New file.
25059 * config/arc/elf.h: Likewise.
25060 * config/arc/linux.h: Likewise.
25061 * config/arc/t-uClibc: Remove.
25063 2017-02-27 Bin Cheng <bin.cheng@arm.com>
25065 PR tree-optimization/77536
25066 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
25067 (tree_transform_and_unroll_loop): Use above function to compute the
25068 estimated niter of unrolled loop and use it when scaling profile.
25069 Also use count info rather than frequency if it's non-zero.
25070 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
25071 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
25072 (vect_transform_loop): Call above function.
25074 2017-02-27 Richard Biener <rguenther@suse.de>
25076 PR tree-optimization/45397
25077 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
25078 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
25079 (visit_nary_op): Add pattern matching for CSEing sign-changed
25080 or truncated operations with wider ones.
25082 2017-02-27 Richard Biener <rguenther@suse.de>
25084 PR tree-optimization/79690
25085 * tree-vect-stmts.c (vectorizable_store): Use vector type
25086 built from the DR with address-space.
25088 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
25090 * doc/invoke.texi (Optimize Options): Refine the description
25091 of asan-use-after-return.
25093 2017-02-25 Alan Modra <amodra@gmail.com>
25095 PR rtl-optimization/79584
25096 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
25097 base, not ad->base_term, the reg within base. Remove assertion
25098 that ad->base == ad->base_term. Replace gen_int_mode using
25099 bogus mode with const0_rtx.
25101 2017-02-25 Jakub Jelinek <jakub@redhat.com>
25103 PR middle-end/79396
25104 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
25105 FMA_EXPR like tcc_binary or tcc_unary.
25107 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
25110 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
25112 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
25113 (output_loc_operands): Handle DW_OP_call_ref and
25114 DW_OP_GNU_variable_value.
25115 (struct variable_value_struct): New type.
25116 (struct variable_value_hasher): Likewise.
25117 (variable_value_hash): New variable.
25118 (string_types): Remove.
25119 (copy_loc_descr): New function.
25120 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
25121 (prepend_loc_descr_to_each): New function.
25122 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
25123 instead of add_loc_descr_to_each if the first argument is single
25124 location list and the second has multiple.
25125 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
25126 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
25127 when looking for variable value which doesn't have other location info.
25128 (loc_list_from_tree): Formatting fix.
25129 (gen_array_type_die): Simplify DW_AT_string_length handling.
25130 (adjust_string_types): Remove.
25131 (gen_subprogram_die): Don't call adjust_string_types nor test/set
25132 string_types. Call resolve_variable_values.
25133 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
25134 (resolve_addr_in_expr): Likewise. Add A argument.
25135 (copy_deref_exprloc): Remove deref argument. Adjust for the
25136 original expression being DW_OP_GNU_variable_value with optionally
25137 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
25138 optionally after it.
25139 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
25140 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
25141 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
25142 (variable_value_hasher::hash, variable_value_hasher::equal): New
25144 (resolve_variable_value_in_expr, resolve_variable_value,
25145 resolve_variable_values, note_variable_value_in_expr,
25146 note_variable_value): New functions.
25147 (dwarf2out_early_finish): Call note_variable_value on all toplevel
25150 2017-02-24 Jakub Jelinek <jakub@redhat.com>
25153 * opts.h (handle_generated_option): Add GENERATED_P argument.
25154 * opts-common.c (handle_option): Adjust function comment.
25155 (handle_generated_option): Add GENERATED_P argument, pass it to
25157 (control_warning_option): Pass false to handle_generated_option
25159 * opts.c (maybe_default_option): Pass true to handle_generated_option
25161 * optc-gen.awk: Likewise.
25163 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25165 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
25166 a REG, look at the REG it is a SUBREG of.
25167 (splitter for cmpeqsi_t): Ditto.
25169 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25171 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
25172 the special USEs with the pattern of the insn, not the insn itself.
25174 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
25177 * doc/invoke.texi: Document -mload-store-pairs.
25179 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25180 Sandra Loosemore <sandra@codesourcery.com>
25182 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
25183 argument isn't a CONST_INT.
25184 (nios2_alternate_compare_const): Assert op is a CONST_INT.
25185 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
25186 (nios2_validate_compare): Bypass alternate compare logic if *op2
25187 is not a CONST_INT.
25188 (ldstwm_operation_p): Return false if first_base is not a REG or
25189 if first_offset is not a CONST_INT.
25191 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25193 * config/cris/cris.md: Use correct operand in a define_peephole2.
25195 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25197 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
25199 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25201 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
25202 this_insn if it is an INSN or JUMP_INSN.
25203 (force_offsettable): Look at base, not at addr.
25204 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
25205 on things that aren't necessarily CONST_INTs.
25207 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
25209 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
25210 -mfpmath=sse is the default also for x86-32 targets with SSE2
25211 instruction set when @option{-ffast-math} is enabled
25213 2017-02-24 Jeff Law <law@redhat.com>
25215 PR rtl-optimizatoin/79286
25216 * ira.c (update_equiv_regs): Drop may_trap_p exception to
25219 2017-02-24 Richard Biener <rguenther@suse.de>
25221 PR tree-optimization/79389
25222 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
25225 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
25227 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
25228 function comment to reflect reality.
25229 (loop_exits_before_overflow): Fix typo in function description.
25231 2017-02-24 Richard Biener <rguenther@suse.de>
25233 PR tree-optimization/79389
25234 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
25235 properly that a threading opportunity exists. Detect conditional
25236 copy/constant propagation opportunities.
25238 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
25240 * config/visium/visium.md (type): Add trap.
25241 (b): New mode attribute.
25242 (*btst): Rename into...
25243 (*btst<mode>): ...this and adjust.
25244 (*cbranchsi4_btst_insn): Rename into...
25245 (*cbranch<mode>4_btst_insn): ...this and adjust.
25246 (trap): New define_insn.
25248 2017-02-23 Jakub Jelinek <jakub@redhat.com>
25250 PR tree-optimization/79389
25251 * ifcvt.c (struct noce_if_info): Add rev_cond field.
25252 (noce_reversed_cond_code): New function.
25253 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
25254 reversed_comparison_code. Formatting fix.
25255 (noce_try_store_flag): Test rev_cond != NULL in addition to
25256 reversed_comparison_code.
25257 (noce_try_store_flag_constants): Likewise.
25258 (noce_try_store_flag_mask): Likewise.
25259 (noce_try_addcc): Use rev_cond if non-NULL instead of
25260 reversed_comparison_code.
25261 (noce_try_cmove_arith): Likewise. Formatting fixes.
25262 (noce_try_minmax, noce_try_abs): Clear rev_cond.
25263 (noce_find_if_block): Initialize rev_cond.
25264 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
25265 instead of false as last argument never attempt to reverse it
25268 2017-02-23 Bin Cheng <bin.cheng@arm.com>
25270 PR tree-optimization/79663
25271 * tree-predcom.c (combine_chains): Process refs in reverse order
25272 only for ZERO length chains, and add explaining comment.
25274 2017-02-23 Jeff Law <law@redhat.com>
25276 PR tree-optimization/79578
25277 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
25278 in call to operand_equal_p.
25280 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
25283 * config/i386/cpuid.h: Fix another undefined behavior.
25285 2017-02-23 Richard Biener <rguenther@suse.de>
25287 PR tree-optimization/79683
25288 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
25289 vector types for data-refs.
25291 2017-02-23 Martin Liska <mliska@suse.cz>
25293 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
25295 2017-02-23 Jakub Jelinek <jakub@redhat.com>
25297 PR middle-end/79665
25298 * internal-fn.c (get_range_pos_neg): Moved to ...
25299 * tree.c (get_range_pos_neg): ... here. No longer static.
25300 * tree.h (get_range_pos_neg): New prototype.
25301 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
25302 are known to be in between 0 and signed maximum inclusive, try to
25303 expand both unsigned and signed divmod and use the cheaper one from
25306 2017-02-22 Jeff Law <law@redhat.com>
25308 PR tree-optimization/79578
25309 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
25310 to compare base operands.
25312 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
25315 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
25316 gpc_reg_operand instead of fpr_reg_operand.
25318 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
25320 * config/mips/mips.c (mips_return_in_memory): Force FP
25321 vector types to be returned in memory for o32 ABI.
25323 2017-02-22 Jakub Jelinek <jakub@redhat.com>
25325 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
25326 instead of DW_TAG_member for static data member declarations and don't
25327 set no_linkage_name for static inline data members.
25328 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
25331 2017-02-22 Martin Liska <mliska@suse.cz>
25333 * doc/invoke.texi: Replace inequality signs with square brackets
25336 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25338 PR tree-optimization/68644
25339 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
25341 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
25344 * lra-constraints.c (simplify_operand_subreg): Handle
25345 WORD_REGISTER_OPERATIONS targets.
25347 2017-02-22 Jakub Jelinek <jakub@redhat.com>
25350 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
25351 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
25352 elimination by swapping fld*.
25354 2017-02-22 Richard Biener <rguenther@suse.de>
25356 PR tree-optimization/79673
25357 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
25358 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
25359 irrelevant address-space qualifiers and avoiding a
25360 ADDR_SPACE_CONVERT_EXPR from fold_convert.
25362 2017-02-22 Richard Biener <rguenther@suse.de>
25364 PR tree-optimization/79666
25365 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
25366 to not symbolically negate if that may introduce undefined
25369 2017-02-22 Martin Liska <mliska@suse.cz>
25372 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
25373 * data-streamer-out.c (streamer_write_gcov_count_stream):
25375 * value-prof.c (stream_out_histogram_value): Make assert more
25376 precise based on type of counter.
25378 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
25381 * config/i386/i386.md (standard_x87sse_constant_load splitter):
25382 Use nonimmediate_operand instead of memory_operand for operand 1.
25383 (float-extend standard_x87sse_constant_load splitter): Ditto.
25385 2017-02-21 Jeff Law <law@redhat.com>
25387 PR tree-optimization/79621
25388 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
25389 blocks with edges to themselves.
25391 2017-02-21 Jakub Jelinek <jakub@redhat.com>
25394 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
25395 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
25396 Use gimple_call_builtin_p.
25399 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
25400 on temporarily removed DEBUG_INSNs.
25402 PR tree-optimization/79649
25403 * tree-loop-distribution.c (classify_partition): Give up on
25404 non-generic address space loads/stores.
25406 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
25408 * doc/loop.texi (Loop manipulation): Remove nonexistent
25409 tree_ssa_loop_version from the documentation.
25410 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
25412 2017-02-21 Jakub Jelinek <jakub@redhat.com>
25415 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
25416 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
25417 * config/rs6000/rs6000.c: Include except.h.
25418 (rs6000_expand_split_stack_prologue): Call
25419 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
25421 2017-02-21 Martin Jambor <mjambor@suse.cz>
25424 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
25425 have been analyzed.
25427 2017-02-21 Martin Jambor <mjambor@suse.cz>
25429 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
25430 for backward compatibility only.
25431 * doc/invoke.texi (Option Summary): Remove all references to
25432 -fipa-cp-alignment.
25434 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
25438 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25440 * lra-constraints.c (curr_insn_transform): Handle
25441 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
25443 2017-02-21 Martin Liska <mliska@suse.cz>
25445 * config/i386/i386.opt: Replace -masm-dialect with -masm.
25447 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
25449 PR translation/79638
25450 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
25452 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
25455 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
25456 (arm_function_ok_for_sibcall): Return false for an indirect call by
25457 descriptor if all the argument registers are used.
25458 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
25459 alignment of the function.
25461 2017-02-21 Jakub Jelinek <jakub@redhat.com>
25463 PR tree-optimization/61441
25464 * simplify-rtx.c (simplify_const_unary_operation): For
25465 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
25466 the sNaN unmodified.
25468 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
25470 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25471 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25472 instead of SYSTEM_HEADER_DIR.
25474 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
25475 Martin Liška <mliska@suse.cz>
25477 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
25478 Fix typos and grammar, use active voice, and clarify.
25480 2017-02-20 Marek Polacek <polacek@redhat.com>
25482 PR middle-end/79537
25483 * gimplify.c (gimplify_expr): Handle unused *&&L;.
25486 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
25488 2017-02-20 Jakub Jelinek <jakub@redhat.com>
25491 * config/i386/i386.c (ix86_expand_builtin): Handle
25492 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
25493 ix86_builtins_isa[fcode].isa as a requirement of those
25494 flags and any other flag in the bitmask.
25495 (ix86_init_mmx_sse_builtins): Use 0 instead of
25496 ~OPTION_MASK_ISA_64BIT as mask.
25497 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
25498 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
25499 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
25500 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
25502 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25505 * lra-constraints.c (split_reg): Check requested split mode
25506 is supported by the register.
25508 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25510 * lra-constraints.c (simplify_operand_subreg): Remove early
25513 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25516 * lra-constraints.c (curr_insn_transform): Tighten condition
25517 for converting SUBREG reloads from OP_OUT to OP_INOUT.
25519 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25522 * lra-constraints.c (curr_insn_transform): Handle
25523 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
25525 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
25528 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
25530 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
25532 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
25535 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
25538 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25540 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
25543 2017-02-18 Jakub Jelinek <jakub@redhat.com>
25545 * final.c (last_columnnum, override_columnnum): New variables.
25546 (final_start_function): Set last_columnnum, pass it to begin_prologue
25547 hook and pass 0 to dwarf2out_begin_prologue.
25548 (final_scan_insn): Update override_columnnum. Pass last_columnnum
25549 to source_line debug hook.
25550 (notice_source_line): Compute last_columnnum and for debug_column_info
25551 return true on column changes.
25552 * debug.h (struct gcc_debug_hooks): Add column argument to
25553 source_line and begin_prologue hooks.
25554 (debug_nothing_int_charstar_int_bool): Remove prototype.
25555 (debug_nothing_int_int_charstar,
25556 debug_nothing_int_int_charstar_int_bool): New prototypes.
25557 (dwarf2out_begin_prologue): Add column argument.
25558 * debug.c (do_nothing_debug_hooks): Adjust source_line and
25559 begin_prologue hooks.
25560 (debug_nothing_int_charstar_int_bool): Remove.
25561 (debug_nothing_int_int_charstar,
25562 debug_nothing_int_int_charstar_int_bool): New functions.
25563 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
25564 through to dwarf2out_source_line.
25565 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
25566 (dwarf2out_source_line): Add column argument, emit it if requested.
25567 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
25569 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
25570 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
25571 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
25572 through to dwarf2out_begin_prologue.
25573 (vmsdbgout_source_line): Add column argument, pass it through to
25574 dwarf2out_source_line.
25575 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
25576 dbxout_source_line caller.
25577 (dbxout_source_line): Add column argument.
25579 * common.opt (gno-column-info, gcolumn-info): New options.
25580 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
25581 (check_die): Also test for multiple DW_AT_decl_column attributes.
25582 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
25583 DW_AT_decl_column if requested.
25584 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
25586 (gen_variable_die): Likewise.
25587 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
25588 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
25591 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
25592 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
25593 (ix86_handle_option): Handle OPT_m3dnowa.
25594 * doc/invoke.texi (-m3dnowa): Document.
25595 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
25596 -m3dnowa instead of -m3dnow -march=athlon.
25599 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
25600 instead of gcc_assert for K, r and R code checks. Formatting fixes.
25602 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25605 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
25606 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
25607 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
25608 generator for vsx_xxpermdi_<mode>_be.
25609 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
25610 force big-endian semantics.
25611 (vsx_xxpermdi_<mode>_be): New define_expand with same
25612 implementation as previous version of vsx_xxpermdi_<mode>.
25614 2017-02-17 Jakub Jelinek <jakub@redhat.com>
25616 PR tree-optimization/79327
25617 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
25618 variable, its initialization and use.
25620 2017-02-17 Julia Koval <julia.koval@intel.com>
25622 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
25623 (OPTION_MASK_ISA_PKU_UNSET): New.
25624 (ix86_handle_option): Handle -mrdpid.
25625 * config/i386/cpuid.h (bit_RDPID): New.
25626 * config/i386/driver-i386.c (host_detect_local_cpu):
25627 Detect RDPID feature.
25628 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
25629 * config/i386/i386-c.c (ix86_target_macros_internal):
25631 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
25632 (ix86_valid_target_attribute_inner_p): Add "rdpid".
25633 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
25634 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
25635 * config/i386/i386.md (define_insn "rdpid"): New.
25636 * config/i386/i386.opt Add -mrdpid.
25637 * config/i386/immintrin.h (_rdpid_u32): New.
25639 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
25641 PR rtl-optimization/79541
25642 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
25643 instead of transforming it into USE.
25645 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
25647 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
25648 If HONOR_SNANS (SFmode) force the input to a register.
25649 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
25650 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
25651 an frsp or similar insn.
25653 2017-02-17 Martin Liska <mliska@suse.cz>
25655 PR rtl-optimization/79577
25656 * params.def (selsched-max-sched-times): Increase minimum to 1.
25658 2017-02-17 Martin Liska <mliska@suse.cz>
25660 PR rtl-optimization/79574
25661 * gcse.c (want_to_gcse_p): Prevent integer overflow.
25663 2017-02-17 Martin Liska <mliska@suse.cz>
25665 PR tree-optimization/79529
25666 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
25667 ssa_defined_default_def_p to handle cases which are implicitly
25669 * tree-ssa.c (ssa_defined_default_def_p): New function.
25670 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
25671 which are implicitly defined.
25672 * tree-ssa.h (ssa_defined_default_def_p): Declare.
25674 2017-02-17 Richard Biener <rguenther@suse.de>
25676 PR middle-end/79576
25677 * params.def (max-ssa-name-query-depth): Limit to 10.
25679 2017-02-17 Richard Biener <rguenther@suse.de>
25681 PR tree-optimization/79552
25682 * tree-ssa-structalias.c (visit_loadstore): Properly verify
25685 2017-02-17 Richard Biener <rguenther@suse.de>
25688 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
25690 2017-02-17 Marek Polacek <polacek@redhat.com>
25692 PR middle-end/79536
25693 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
25694 (fold_negate_expr): New wrapper.
25696 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
25698 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
25699 Correct terminology and de-emphasize pre-standard behavior.
25701 2017-02-16 Alan Modra <amodra@gmail.com>
25703 PR rtl-optimization/79286
25704 * ira.c (def_dominates_uses): New function.
25705 (update_equiv_regs): Don't create an equivalence for insns that
25706 may trap where the register def does not dominate the use.
25708 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
25710 PR rtl-optimization/78127
25711 * lra.c (lra): Call lra_eliminate before finish the loop after
25714 2017-02-16 Richard Biener <rguenther@suse.de>
25716 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
25718 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
25719 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
25720 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
25721 (isl_val_int_from_wi): New function.
25722 (extract_affine_gmp): Rename to ...
25723 (extract_affine_wi): ... this, take a widest_int.
25724 (extract_affine_int): Just wrap extract_affine_wi.
25725 (add_param_constraints): Use isl_val_int_from_wi.
25726 (add_loop_constraints): Likewise, and extract_affine_wi.
25728 2017-02-15 Jeff Law <law@redhat.com>
25730 PR middle-end/79521
25731 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
25732 ira_init_register_move_cost_if_necessary.
25734 2017-02-15 Martin Sebor <msebor@redhat.com>
25736 PR middle-end/32003
25737 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
25738 removed in a prior commit.
25740 2017-02-15 Bin Cheng <bin.cheng@arm.com>
25742 PR tree-optimization/79347
25743 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
25744 counters during peeling.
25746 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
25748 * Makefile.in (site.exp): Remove "set ISLVER".
25750 2017-02-15 Jakub Jelinek <jakub@redhat.com>
25753 * real.c (real_from_integer): Call real_convert even for decimal.
25755 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
25758 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
25760 2017-02-14 Andrew Pinski <apinski@cavium.com>
25762 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
25763 cores and change the partno/implementer to be correct.
25764 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
25765 the 'B" as the implementer.
25766 * config/aarch64/aarch64-tune.md: Regenerate.
25768 2017-02-14 Carl Love <cel@us.ibm.com>
25770 * config/rs6000/rs6000.c: Add case statement entry to make the
25771 xvcvuxdsp built-in argument unsigned.
25772 * config/rs6000/vsx.md: Fix the source and return operand types so they
25773 match the instruction definitions from the ISA document. Fix typo
25774 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
25777 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
25780 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
25781 member early_clobber_alts.
25782 * lra-lives.c (reg_early_clobber_p): New.
25783 (process_bb_lives): Use it.
25784 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
25785 (debug_operand_data): Initialize early_clobber_alts.
25786 (setup_operand_alternative): Set up early_clobber_alts.
25787 (collect_non_operand_hard_regs): Ditto. Pass early clobber
25788 alternatives to new_insn_reg.
25789 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
25791 (lra_update_insn_regno_info): Pass the new arg.
25793 2017-02-14 Jakub Jelinek <jakub@redhat.com>
25795 PR middle-end/79505
25796 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
25797 (new_oacc_loop_raw): Don't clear already cleared fields.
25800 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
25801 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
25802 _mm512_prefetch_i64gather_ps): New inline functions and macros.
25804 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
25807 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
25809 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
25812 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
25813 the extra instruction to the right place to store 128-bit constant
25816 2017-02-14 Martin Sebor <msebor@redhat.com>
25818 PR middle-end/79448
25819 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
25820 warning for strings of unknown length.
25822 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
25824 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
25826 2017-02-14 Jeff Law <law@redhat.com>
25829 * ira-costs.c (scan_one_insn): Initialize register move costs
25830 for pseudos seen in USE/CLOBBER insns.
25832 PR tree-optimization/79095
25833 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
25834 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
25835 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
25836 if the operands are known to be not equal, then the resulting range
25838 (intersect_ranges): If the new range is ~[0,0] and the old range is
25839 wide, then prefer ~[0,0].
25840 * tree-vrp.c (overflow_comparison_p_1): New function.
25841 (overflow_comparison_p): New function.
25842 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
25843 if NAME is used in an overflow test.
25844 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
25845 overflow check that can be expressed as an equality test, then adjust
25846 ops to be that equality test.
25848 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25850 * config/s390/s390-builtin-types.def: Remove flags argument.
25851 * config/s390/s390.c (s390_init_builtins): Likewise.
25853 2017-02-14 Martin Liska <mliska@suse.cz>
25855 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
25856 vector. Fix trailing white spaces.
25858 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
25860 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
25863 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25865 PR rtl-optimization/68664
25866 * config/arm/arm.c (arm_sched_can_speculate_insn):
25867 New function. Declare prototype.
25868 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
25870 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25872 PR rtl-optimization/68664
25873 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
25875 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
25877 2017-02-14 Amit Pawar <amit.pawar@amd.com>
25879 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
25880 max skip bytes for function, loop and jump.
25882 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25884 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
25885 ABS_EXPR for gimple dump.
25887 2017-02-14 Jakub Jelinek <jakub@redhat.com>
25890 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
25892 PR tree-optimization/79408
25893 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
25894 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
25895 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
25896 also if rhs1 is INTEGER_CST.
25898 2017-02-14 Richard Biener <rguenther@suse.de>
25900 PR middle-end/79432
25901 * tree-into-ssa.c (insert_phi_nodes): When the function can
25902 have abnormal edges rewrite SSA names with broken use-def
25903 dominance out of SSA and register them for PHI insertion.
25905 2017-02-13 Martin Sebor <msebor@redhat.com>
25907 PR middle-end/79496
25908 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
25909 clearing info.nowrite flag when snprintf size argument is a range.
25911 2017-02-13 Jakub Jelinek <jakub@redhat.com>
25913 * cprop.c (cprop_jump): Add missing space in string literal.
25914 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
25915 (get_constraint_for_component_ref): Likewise.
25916 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
25917 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
25918 * lra-constraints.c (process_alt_operands): Likewise.
25919 * ipa-inline.c (inline_small_functions): Likewise.
25920 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
25921 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
25922 * trans-mem.c (diagnose_tm_1_op): Likewise.
25923 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
25924 (grid_parallel_clauses_gridifiable): Likewise.
25926 * config/nvptx/mkoffload.c (process): Add space in between
25929 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
25930 "MOD4_SSE_REGS" and "ALL_REGS".
25932 * spellcheck.c (test_data): Add , in between "foo" and "food".
25934 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25937 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
25938 boundary crossing check and subsequent code generation agree.
25940 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25942 * config/aarch64/aarch64.c (has_memory_op): Delete.
25943 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
25946 2017-02-13 Jakub Jelinek <jakub@redhat.com>
25948 PR rtl-optimization/79388
25949 PR rtl-optimization/79450
25950 * combine.c (distribute_notes): When removing TEM_INSN for which
25951 corresponding dest has last value recorded, invalidate that last
25954 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25956 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
25957 of explicit '@'. Add missing assembly comment marker on branch costs
25960 2017-02-13 Nathan Sidwell <nathan@acm.org>
25962 * gengtype-lex.l (<in_struct>): Add '/'.
25964 2017-02-13 Martin Liska <mliska@suse.cz>
25967 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
25969 2017-02-13 Richard Biener <rguenther@suse.de>
25971 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
25973 * configure: Re-generate.
25974 * config.in: Likewise.
25975 * graphite-dependences.c: Simplify as if
25976 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
25977 * graphite-isl-ast-to-gimple.c: Likewise.
25978 * graphite-optimize-isl.c: Likewise.
25979 * graphite-poly.c: Likewise.
25980 * graphite-sese-to-poly.c: Likewise.
25981 * graphite.h: Likewise.
25982 * toplev.c: Include isl/version.h and use isl_version () for
25983 printing the ISL version.
25984 * doc/install.texi: Update ISL requirement.
25986 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
25988 * doc/standards.texi (Standards): Update reference to
25991 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
25993 * doc/extend.texi (Named Address Spaces): sourceware.org now
25995 * doc/install.texi (Binaries): Ditto.
25998 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
26000 * doc/cpp.texi: Replace "stringify"/"stringification" with C
26001 standard terminology "stringize"/"stringizing" throughout.
26002 * doc/cppinternals.texi: Likewise.
26004 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
26006 * doc/extend.texi: Fix some spelling mistakes and typos.
26007 * doc/invoke.texi: Likewise.
26009 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
26012 * params.def (inline-min-speedup) Change from 10 to 8.
26014 2017-02-11 Jakub Jelinek <jakub@redhat.com>
26016 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
26019 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
26022 * ipa-inline-analysis.c (get_minimal_bb): New function.
26023 (record_modified): Use it.
26024 (remap_edge_change_prob): Handle also ancestor functions.
26026 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
26028 * doc/contrib.texi (Contributors): Remove broken link into
26029 the Mauve CVS repository.
26031 2017-02-11 Jakub Jelinek <jakub@redhat.com>
26033 PR middle-end/79454
26034 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
26035 result computation whenever lhs doesn't have vector mode, not
26036 just when it has BLKmode.
26038 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
26040 * doc/makefile.texi (profiledbootstrap): Refer to the
26041 installation instructions only in textual form.
26043 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26046 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
26048 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
26050 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
26051 (Specific): Update mingw-w64 reference.
26053 (Specific): Remove broken link to Renesas RX processor.
26055 2017-02-10 Richard Biener <rguenther@suse.de>
26057 * toplev.c (process_options): Do not mention obsolete graphite
26058 options when printing sorry message about missing graphite support.
26059 Mention -floop-nest-optimize.
26061 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
26063 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
26064 (vtst_p16): Likewise.
26065 (vtstq_p8): Likewise.
26066 (vtstq_p16): Likewise.
26068 (vtstq_p64): Likewise.
26069 * config/arm/arm_neon.h (vgetq_lane_p64): New.
26070 (vset_lane_p64): New.
26071 (vsetq_lane_p64): New.
26073 2017-02-10 Jakub Jelinek <jakub@redhat.com>
26075 PR tree-optimization/79411
26076 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
26077 stmt operands are SSA_NAMEs used in abnormal phis.
26078 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
26081 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
26084 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
26087 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
26089 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
26091 2017-02-09 Jakub Jelinek <jakub@redhat.com>
26093 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
26097 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
26098 not arbitrary TREE_CONSTANT.
26101 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
26102 "omp declare target link" attribute unless is_global_var.
26103 * omp-offload.c (find_link_var_op): Likewise.
26105 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
26106 Chung-Lin Tang <cltang@codesourcery.com>
26108 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
26110 (gimplify_adjust_omp_clauses): Don't delete TILE.
26111 (gimplify_omp_for): Deal with TILE.
26112 * internal-fn.c (expand_GOACC_TILE): New function.
26113 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
26115 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
26116 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
26118 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
26119 avoid DIV for outermost collapse var.
26120 (expand_oacc_for): Insert tile element loop as needed. Adjust.
26121 Remove out of date comments, fix whitespace.
26122 * omp-general.c (omp_extract_for_data): Deal with tiling.
26123 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
26124 adjust OLF_DIM_BASE value.
26125 (struct omp_for_data): Add tiling field.
26126 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
26127 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
26128 for auto loops. Remove default auto determining, moved to
26129 oacc_loop_fixed_partitions.
26130 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
26131 stmts, add e_mask field.
26132 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
26133 (oacc_thread_numbers): Use oacc_dim_call.
26134 (oacc_xform_tile): New.
26135 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
26136 (finish_oacc_loop): Adjust for ifns vector.
26137 (oacc_loop_discover_walk): Append loop abstraction sites to list,
26138 add case for GOACC_TILE fns.
26139 (oacc_loop_xform_loop): Delete.
26140 (oacc_loop_process): Iterate over call list directly, and add
26141 handling for GOACC_TILE fns.
26142 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
26144 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
26145 vector partitioning to outer loops. Assign 2 partitions to loops
26146 when available. Add TILE handling.
26147 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
26148 (execite_oacc_device_lower): Process GOACC_TILE fns,
26149 ignore unknown specs.
26150 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
26151 * tree.c (omp_clause_num_ops): Adjust TILE ops.
26152 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
26154 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
26156 * configure.ac (ACX_BUGURL): Update.
26157 * configure: Regenerate.
26159 2017-02-09 Richard Biener <rguenther@suse.de>
26161 PR tree-optimization/69823
26162 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
26163 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
26165 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
26167 * config/arc/arc-c.def: Add __NPS400__ definition.
26168 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
26169 (TARGET_NPS400): Define.
26171 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
26173 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
26175 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
26176 pointer, arch_info.
26177 (arc_cpu_types): Fill the arch_info field with a pointer into the
26178 arc_arch_types table.
26179 (arc_selected_cpu): Declare.
26180 * config/arc/arc.c (arc_selected_cpu): Make global.
26181 (arc_selected_arch): Delete.
26182 (arc_base_cpu): Delete.
26183 (arc_override_options): Remove references to deleted variables,
26184 update access to arch information.
26185 (ARC_OPT): Update access to arch information.
26186 (ARC_OPTX): Likewise.
26187 * config/arc/arc.h (arc_base_cpu): Remove declaration.
26188 (TARGET_ARC600): Update access to arch information.
26189 (TARGET_ARC601): Likewise.
26190 (TARGET_ARC700): Likewise.
26191 (TARGET_EM): Likewise.
26192 (TARGET_HS): Likewise.
26193 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
26196 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
26199 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
26200 condition/operands for integer GE/LE/GEU/LEU operations.
26202 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
26204 PR translation/79397
26205 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
26208 2017-02-08 Martin Jambor <mjambor@suse.cz>
26211 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
26212 whether allocation happened.
26213 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
26214 nothing was allocated.
26216 2017-02-08 Jakub Jelinek <jakub@redhat.com>
26218 PR tree-optimization/79408
26219 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
26220 constant, but SSA_NAME with a known integer range, use the minimum
26221 of that range instead of op1 to determine if modulo can be replaced
26222 with its first operand.
26224 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26226 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
26228 2017-02-08 Richard Biener <rguenther@suse.de>
26230 PR tree-optimization/71824
26231 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
26232 Check all loops contained in the merged region.
26234 2017-02-07 Andrew Pinski <apinski@cavium.com>
26236 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
26238 2017-02-07 Andrew Pinski <apinski@cavium.com>
26240 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
26241 (thunderxt88): Likewise.
26242 (thunderxt81): Disable LSE and change v8.1 to v8.
26243 (thunderxt83): Likewise.
26245 2017-02-07 Jakub Jelinek <jakub@redhat.com>
26246 Richard Biener <rguenther@suse.de>
26248 PR middle-end/79399
26249 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
26250 type from int to size_t.
26251 * ira-costs.c (struct_costs_size): Change type from int to size_t.
26253 2017-02-07 Jakub Jelinek <jakub@redhat.com>
26255 PR rtl-optimization/79386
26256 * cprop.c (bypass_conditional_jumps): Initialize
26257 bypass_last_basic_block already before splitting bbs after
26258 unconditional traps...
26259 (bypass_conditional_jumps): ... rather than here.
26262 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
26263 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
26264 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
26265 fix -masm=intel patterns.
26267 2017-02-07 Richard Biener <rguenther@suse.de>
26269 PR tree-optimization/79256
26270 PR middle-end/79278
26271 * builtins.c (get_object_alignment_2): Use min_align_of_type
26272 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
26273 and ADJUST_FIELD_ALIGN.
26275 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
26277 * doc/tm.texi: Regenerate.
26278 * stor-layout.c (layout_decl): Adjust.
26279 (update_alignment_for_field): Likewise.
26280 (place_field): Likewise.
26281 (min_align_of_type): Likewise.
26282 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
26283 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
26284 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
26285 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
26286 * config/frv/frv.c (frv_adjust_field_align): Likewise.
26287 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
26288 * config/i386/i386.c (x86_field_alignment): Likewise.
26289 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
26290 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
26291 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
26292 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
26293 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
26294 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
26298 2017-01-30 Richard Biener <rguenther@suse.de>
26300 PR tree-optimization/79256
26301 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
26302 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
26305 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
26307 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
26308 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
26309 builtins to SImode and emit a zero-extend, if necessary.
26311 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
26313 * docs/invoke.texi (RISC-V Options): Alphabetize.
26315 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
26317 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
26320 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
26322 * config/riscv/riscv.c: New file.
26323 * common/config/riscv/riscv-common.c: Likewise.
26324 * config.gcc: Likewise.
26325 * config/riscv/constraints.md: Likewise.
26326 * config/riscv/elf.h: Likewise.
26327 * config/riscv/generic.md: Likewise.
26328 * config/riscv/linux.h: Likewise.
26329 * config/riscv/multilib-generator: Likewise.
26330 * config/riscv/peephole.md: Likewise.
26331 * config/riscv/pic.md: Likewise.
26332 * config/riscv/predicates.md: Likewise.
26333 * config/riscv/riscv-builtins.c: Likewise.
26334 * config/riscv/riscv-c.c: Likewise.
26335 * config/riscv/riscv-ftypes.def: Likewise.
26336 * config/riscv/riscv-modes.def: Likewise.
26337 * config/riscv/riscv-opts.h: Likewise.
26338 * config/riscv/riscv-protos.h: Likewise.
26339 * config/riscv/riscv.h: Likewise.
26340 * config/riscv/riscv.md: Likewise.
26341 * config/riscv/riscv.opt: Likewise.
26342 * config/riscv/sync.md: Likewise.
26343 * config/riscv/t-elf-multilib: Likewise.
26344 * config/riscv/t-linux: Likewise.
26345 * config/riscv/t-linux-multilib: Likewise.
26346 * config/riscv/t-riscv: Likewise.
26347 * configure.ac: Likewise.
26348 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
26349 Waterman as RISC-V maintainers.
26350 * doc/install.texi: Add RISC-V entries.
26351 * doc/invoke.texi: Add RISC-V options section.
26352 * doc/md.texi: Add RISC-V constraints section.
26353 * configure: Regenerated.
26355 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
26358 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
26359 false values to be constant vectors with all 0 or all 1 bits set.
26360 (vcondu<mode><mode>): Likewise.
26361 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
26363 (fpmask_comparison_operator): Update comment.
26364 (vecint_comparison_operator): New predicate.
26365 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
26366 vector conditionals when the true and false values are constant
26367 vectors with all 0 bits or all 1 bits set.
26369 2017-02-06 Martin Sebor <msebor@redhat.com>
26371 PR tree-optimization/79376
26372 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
26374 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
26376 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
26377 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
26378 to simplify split condition.
26380 2017-02-06 Jakub Jelinek <jakub@redhat.com>
26382 * omp-expand.c (oxpand_omp_atomic_fetch_op,
26383 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
26386 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
26388 PR rtl-optimization/68664
26389 * target.def (can_speculate_insn): New hook.
26390 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
26391 * doc/tm.texi: Regenerate.
26392 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
26393 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
26394 (rs6000_sched_can_speculate_insn): New function.
26396 2017-02-06 Jakub Jelinek <jakub@redhat.com>
26398 PR tree-optimization/79284
26399 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
26400 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
26401 vectorizable_mask_load_store, vectorizable_operation,
26402 vect_is_simple_cond, get_same_sized_vectype): Use it instead
26403 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
26404 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
26405 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
26406 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
26407 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
26408 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
26409 is_gimple_assign (stmt). Replace another such test with
26410 is_gimple_assign (stmt).
26412 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
26415 * config/avr/avr.c (rtl-iter.h): Include it.
26416 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
26417 (avr_legitimate_combined_insn): ...and implementation.
26419 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
26421 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
26422 * config/s390/s390.c (s390_const_operand_ok)
26423 (s390_canonicalize_comparison, s390_extract_part)
26424 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
26425 (s390_contiguous_bitmask_p, s390_rtx_costs)
26426 (legitimize_pic_address): Likewise.
26427 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
26428 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
26429 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
26430 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
26431 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
26433 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
26435 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
26436 REGNO($0) == REGNO($1).
26438 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26440 * config/s390/linux.h(SIZE_TYPE): Add comment.
26442 2017-02-06 Julian Brown <julian@codesourcery.com>
26443 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26444 Virendra Pathak <virendra.pathak@broadcom.com>
26446 * config/aarch64/aarch64-cores.def: Change the scheduler
26448 * config/aarch64/aarch64.md: Include thunderx2t99.md.
26449 * config/aarch64/thunderx2t99.md: New file.
26451 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
26453 * doc/standards.texi (Go Language): Update link to language
26456 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
26458 * tree-eh.c (lower_resx): Sanitize profile.
26459 (cleanup_empty_eh_move_lp): Likewise.
26461 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
26464 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
26466 * cfgloopmanip.h (loop_version): Update prototype.
26467 * modulo-sched.c (sms_schedule): Update call of loop_version.
26468 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
26469 * tree-parloops.c (gen_parallel_loop): Likewise.
26470 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
26471 * tree-ssa-loop-split.c (split_loop): Likewise.
26472 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
26473 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
26475 2017-02-05 Martin Liska <mliska@suse.cz>
26478 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
26480 (print_operand_address): Initialize a struct to zero.
26482 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
26484 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
26485 garbage collector only in textual form.
26487 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
26489 * doc/extend.texi (x86 specific memory model extensions for
26490 transactional memory): Simplify a phrase.
26492 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
26495 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
26496 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
26497 (atomic_storedi_1): Likewise.
26499 2017-02-04 Jakub Jelinek <jakub@redhat.com>
26501 PR tree-optimization/79338
26502 * tree-parloops.c (gather_scalar_reductions): Don't call
26503 vect_analyze_loop_form for loop->inner before destroying loop's
26506 2017-02-03 Martin Sebor <msebor@redhat.com>
26508 PR tree-optimization/79327
26509 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
26510 when precision has resulted in leading zeros.
26511 (format_integer): Adjust the likely counter to assume an unknown
26512 argument that may be zero is non-zero.
26514 2017-02-03 Jason Merrill <jason@redhat.com>
26517 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
26518 avoid copying non-taken branch.
26520 2017-02-03 Jakub Jelinek <jakub@redhat.com>
26522 PR tree-optimization/79340
26523 * tree-vect-loop.c (vectorizable_reduction): Release
26524 vec_defs elements after safe_splicing them into other vectors.
26527 PR tree-optimization/79327
26528 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
26529 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
26531 (format_integer): Use wide_int_to_tree instead of build_int_cst
26532 + to_?hwi. If argmin is NULL, just set argmin and argmax to
26533 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
26534 of shortest and longest sequence.
26536 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
26538 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
26539 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
26541 2017-02-03 Walter Lee <walt@tilera.com>
26544 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
26545 after initial stackframe link reg save.
26546 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
26548 2017-02-03 Jakub Jelinek <jakub@redhat.com>
26551 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
26552 wu for stxssp alternative.
26554 2017-02-03 Martin Sebor <msebor@redhat.com>
26556 PR tree-optimization/79352
26557 * gimple-fold.c (get_range_strlen): Add argument.
26558 (get_range_strlen): Change return type to bool.
26559 (get_maxval_strlen): Pass in a dummy argument.
26560 * gimple-fold.h (get_range_strlen): Change return type to bool.
26561 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
26562 * tree.h (array_at_struct_end_p): Add argument.
26563 * tree.c (array_at_struct_end_p): Handle it.
26565 2017-02-03 Martin Liska <mliska@suse.cz>
26568 * multiple_target.c (create_dispatcher_calls): Redirect edge
26569 from a caller of a dispatcher.
26570 (expand_target_clones): Make the clones local.
26571 (ipa_target_clone): Do both target clones and resolvers.
26572 (ipa_dispatcher_calls): Remove the pass.
26573 (pass_dispatcher_calls::gate): Likewise.
26574 (make_pass_dispatcher_calls): Likewise.
26575 * passes.def (pass_target_clone): Put as very first IPA early
26578 2017-02-03 Martin Liska <mliska@suse.cz>
26580 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
26581 in case of a function with ifunc attribute.
26583 2017-02-03 Martin Liska <mliska@suse.cz>
26585 * cgraph.c (cgraph_node::dump): Dump function version info.
26586 * symtab.c (symtab_node::dump_base): Add missing new line.
26588 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
26590 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
26591 (ifcombine_ifandif): Use it.
26593 2017-02-03 Martin Liska <mliska@suse.cz>
26595 * doc/invoke.texi: Document default value for
26596 use-after-scope-direct-emission-threshold.
26598 2017-02-03 Martin Liska <mliska@suse.cz>
26600 PR tree-optimization/79339
26601 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
26602 (format_floating): Likewise.
26604 2017-02-03 Martin Liska <mliska@suse.cz>
26607 * ipa-prop.c (ipa_node_params_t::insert): Remove current
26609 (ipa_node_params_t::remove): Likewise.
26610 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
26611 initialization from removed ipa_node_params_t::insert.
26612 (ipa_node_params::~ipa_node_params): Move from removed
26613 ipa_node_params_t::release.
26614 * symbol-summary.h (symbol_summary::m_released): New member.
26615 Do not release a summary twice. Do not allow to call finalizer
26616 for types of a summary that live in GGC memory.
26618 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26620 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
26623 2017-02-02 Martin Sebor <msebor@redhat.com>
26625 PR middle-end/79275
26626 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
26627 (format_string): Tighten up the range of output for non-constant
26628 strings and correct the expected range for wide non-constant strings.
26630 2017-02-02 Martin Sebor <msebor@redhat.com>
26632 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
26634 PR middle-end/32003
26635 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
26637 (-fdump-tree-@var): Add to index and document how to come up
26638 with pass-specific option and dump file names.
26639 (-fdump-passes): Clarify where to look for output.
26641 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
26643 PR middle-end/77445
26644 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
26645 statistics of the analyzed path; allow threading for speed when
26646 any of BBs along the path are optimized for speed.
26648 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
26650 PR middle-end/78468
26651 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
26652 settings of the virtual registers.
26655 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
26657 * explow.c (get_dynamic_stack_size): Take known alignment of stack
26658 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
26661 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26663 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
26664 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
26666 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26668 * config/s390/s390.md: Add missing comments with the expanded
26670 * config/s390/vector.md: Likewise.
26671 * config/s390/vx-builtins.md: Likewise.
26673 2017-02-02 Jakub Jelinek <jakub@redhat.com>
26676 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
26677 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
26678 conditions on a single line.
26680 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26682 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
26683 __S390_VX__ to __VX__.
26685 2017-02-01 Andrew Pinski <apinski@cavium.com>
26687 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
26688 stmt_info to record_stmt_cost.
26689 (vect_get_known_peeling_cost): Pass stmt_info if known to
26691 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
26692 cpu_vector_cost field into
26693 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
26694 field into vec_int_stmt_cost and vec_fp_stmt_cost.
26695 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
26696 splitting of scalar_stmt_cost and vec_stmt_cost.
26697 (thunderx_vector_cost): Likewise.
26698 (cortexa57_vector_cost): LIkewise.
26699 (exynosm1_vector_cost): Likewise.
26700 (xgene1_vector_cost): Likewise.
26701 (thunderx2t99_vector_cost): Improve after the splitting of the two
26703 (aarch64_builtin_vectorization_cost): Update for the splitting of
26704 scalar_stmt_cost and vec_stmt_cost.
26706 2017-02-01 Torvald Riegel <triegel@redhat.com>
26707 Richard Henderson <rth@redhat.com>
26709 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
26710 conditional on existance of a fast atomic load.
26711 * optabs-query.c (can_atomic_load_p): New function.
26712 * optabs-query.h (can_atomic_load_p): Declare it.
26713 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
26714 no fast atomic load is available for the particular size of access.
26715 (expand_atomic_compare_and_swap): Likewise.
26716 (expand_atomic_load): Likewise.
26717 (expand_atomic_store): Likewise.
26718 (expand_atomic_fetch_op): Likewise.
26719 * testsuite/lib/target-supports.exp
26720 (check_effective_target_sync_int_128): Remove x86 because it provides
26721 no fast atomic load.
26722 (check_effective_target_sync_int_128_runtime): Likewise.
26724 2017-02-01 Richard Biener <rguenther@suse.de>
26726 * graphite.c: Include tree-vectorizer.h for find_loop_location.
26727 (graphite_transform_loops): Provide opt-info for optimized nests.
26728 * tree-parloop.c (parallelize_loops): Provide opt-info for
26729 parallelized loops.
26731 2017-02-01 Richard Biener <rguenther@suse.de>
26733 PR middle-end/79315
26734 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
26735 was not set before.
26737 2017-02-01 Richard Biener <rguenther@suse.de>
26739 PR tree-optimization/71824
26740 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
26741 Verify the loops are valid in the merged SESE region.
26742 (scop_detection::can_represent_loop_1): Check analyzing the
26743 evolution of the number of iterations in the region succeeds.
26745 2017-01-31 Ian Lance Taylor <iant@golang.org>
26747 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
26748 REG_ARGS_SIZE note to 32-bit push insns and call insn.
26750 2017-01-31 David Malcolm <dmalcolm@redhat.com>
26752 PR preprocessor/79210
26753 * input.c (get_substring_ranges_for_loc): Replace line_width
26754 assertion with error-handling.
26756 2017-01-31 Richard Biener <rguenther@suse.de>
26758 PR tree-optimization/77318
26759 * graphite-sese-to-poly.c (extract_affine): Fix assert.
26760 (create_pw_aff_from_tree): Take loop parameter.
26761 (add_condition_to_pbb): Pass loop of the condition to
26762 create_pw_aff_from_tree.
26764 2017-01-31 Jakub Jelinek <jakub@redhat.com>
26766 * config/s390/s390.c (s390_asan_shadow_offset): New function.
26767 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
26769 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
26773 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
26775 (convert_int_to_float128): Likewise.
26776 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
26777 (convert_int_to_float128): Likewise.
26778 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
26779 (UNSPEC_IEEE128_CONVERT): Likewise.
26780 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
26781 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
26782 Use local variables for IBM extended format.
26783 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
26784 (fix_trunc<mode>si2_fprs): Likewise.
26785 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
26786 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
26787 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
26788 to know that we can now have integers of all sizes in vector
26790 (fix<uns>_<mode>di2_hw): Likewise.
26791 (float<uns>_<mode>si2_hw): Likewise.
26792 (fix_<mode>si2_hw): Likewise.
26793 (fixuns_<mode>si2_hw): Likewise.
26794 (float<uns>_<mode>di2_hw): Likewise.
26795 (float_<mode>di2_hw): Likewise.
26796 (float_<mode>si2_hw): Likewise.
26797 (floatuns_<mode>di2_hw): Likewise.
26798 (floatuns_<mode>si2_hw): Likewise.
26799 (xscvqp<su>wz_<mode>): Delete, no longer used.
26800 (xscvqp<su>dz_<mode>): Likewise.
26801 (xscv<su>dqp_<mode>): Likewise.
26802 (ieee128_mfvsrd_64bit): Likewise.
26803 (ieee128_mfvsrd_32bit): Likewise.
26804 (ieee128_mfvsrwz): Likewise.
26805 (ieee128_mtvsrw): Likewise.
26806 (ieee128_mtvsrd_64bit): Likewise.
26807 (ieee128_mtvsrd_32bit): Likewise.
26809 2017-01-31 Martin Liska <mliska@suse.cz>
26812 * ipa-prop.c (ipa_free_all_node_params): Call release method
26813 instead of ~sumbol_summary to not to trigger double times
26816 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
26818 PR tree-optimization/71691
26819 * bitmap.h (class auto_bitmap): New.
26820 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
26821 is_maybe_undefined instead of ssa_undefined_value_p.
26823 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26825 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
26826 __S390_ARCH_LEVEL__ to __ARCH__.
26828 2017-01-31 Jakub Jelinek <jakub@redhat.com>
26830 PR tree-optimization/79267
26831 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
26832 if should_remove_lhs_p is true.
26834 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
26837 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
26838 (add_alignment_attribute): New.
26839 (base_type_die): Add alignment attribute.
26840 (subrange_type_die): Likewise.
26841 (modified_type_die): Likewise.
26842 (gen_array_type_die): Likewise.
26843 (gen_descr_array_type_die: Likewise.
26844 (gen_enumeration_type_die): Likewise.
26845 (gen_subprogram_die): Likewise.
26846 (gen_variable_die): Likewise.
26847 (gen_field_die): Likewise.
26848 (gen_ptr_to_mbr_type_die): Likewise.
26849 (gen_struct_or_union_type_die): Likewise.
26850 (gen_subroutine_type_die): Likewise.
26851 (gen_typedef_die): Likewise.
26852 (base_type_cmp): Compare alignment attribute.
26854 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26857 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
26858 (setb_unsigned) New pattern for setb with CCUNS.
26859 * config/rs6000/rs6000.c (expand_block_compare): Use a different
26860 subfc./subfe sequence to avoid overflow problems. Generate a
26861 shorter sequence with cmpld/setb for power9.
26862 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
26863 for generating subfc. instruction.
26864 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
26865 now uses this instruction.
26867 2017-01-30 Ian Lance Taylor <iant@google.com>
26870 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
26871 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
26873 2017-01-30 Martin Sebor <msebor@redhat.com>
26875 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
26876 Move constant to the right of a relational operator.
26877 (get_mpfr_format_length, format_character, format_string): Ditto.
26878 (should_warn_p, maybe_warn): Same.
26880 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
26882 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
26885 * asan.c (get_translation_unit_decl): Remove function.
26886 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
26888 2017-01-30 Martin Liska <mliska@suse.cz>
26890 PR gcov-profile/79259
26891 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
26892 -fprofile-generate.
26894 2017-01-30 Martin Liska <mliska@suse.cz>
26897 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
26898 Initialize variables with NULL value.
26900 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
26903 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
26905 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
26907 2017-01-30 Richard Biener <rguenther@suse.de>
26909 PR tree-optimization/79276
26910 * tree-vrp.c (process_assert_insertions): Properly adjust common
26911 when removing a duplicate.
26913 * gcc.dg/torture/pr79276.c: New testcase.
26915 2017-01-30 Richard Biener <rguenther@suse.de>
26917 PR tree-optimization/79256
26918 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
26919 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
26921 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
26923 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
26926 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
26927 ("*r<noxa>sbg_<mode>_sll_bitmask")
26928 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
26929 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
26930 Use contiguous_bitmask_nowrap_operand.
26932 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26935 * config/rs6000/altivec.h (vec_xl): Revise #define.
26936 (vec_xst): Likewise.
26938 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
26940 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
26942 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
26944 PR rtl-optimization/79194
26945 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
26946 traps before call to bypass_conditional_jumps.
26948 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
26950 PR tree-optimization/71374
26951 * lra-constraints.c (check_conflict_input_operands): New.
26952 (match_reload): Use it.
26954 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
26957 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
26958 account to calculate conflict_set.
26960 2017-01-27 Bin Cheng <bin.cheng@arm.com>
26962 PR rtl-optimization/78559
26963 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
26964 other_insn in combine.
26966 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26968 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
26969 uint16_type_node for BT_UINT16.
26971 2017-01-27 David Malcolm <dmalcolm@redhat.com>
26973 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
26974 "RTL Tests" to menu.
26975 (GIMPLE Tests): New node.
26976 (RTL Tests): New node.
26978 2017-01-27 Richard Biener <rguenther@suse.de>
26980 PR tree-optimization/79245
26981 * tree-loop-distribution.c (distribute_loop): Apply cost
26982 modeling also to detected patterns.
26984 2017-01-27 Richard Biener <rguenther@suse.de>
26986 PR tree-optimization/71433
26987 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
26988 (compare_assert_loc): New function.
26989 (process_assert_insertions): Sort and optimize assert locations
26990 to remove duplicates and push down identical assertions on
26991 edges to their destination block.
26993 2017-01-27 Richard Biener <rguenther@suse.de>
26995 PR tree-optimization/79244
26996 * tree-vrp.c (remove_range_assertions): Forcefully propagate
26997 out SSA names even if abnormal.
26999 2017-01-27 Jakub Jelinek <jakub@redhat.com>
27001 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
27002 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
27003 instead of MPFR_RNDN.
27005 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
27008 * arm.c (arm_option_override): Don't call build_target_option_node
27009 until after doing all option overrides.
27010 (arm_valid_target_attribute_tree): Likewise.
27012 2017-01-27 Martin Liska <mliska@suse.cz>
27014 * doc/invoke.texi (-fprofile-arcs): Document profiling support
27015 for {cd}tors and C++ {cd}tors.
27017 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
27019 * config/s390/s390.md ("*setmem_long_and")
27020 ("*setmem_long_and_31z"): Use zero_extend instead of and.
27022 2017-01-26 Martin Sebor <msebor@redhat.com>
27024 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
27027 2017-01-26 Martin Sebor <msebor@redhat.com>
27029 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
27030 HAVE_DFmode before using XFmode or DFmode.
27031 (parse_directive): Avoid using the z length modifier to avoid
27032 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
27034 PR middle-end/78703
27035 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
27036 to accept adjustment as an array.
27037 (get_int_range): New function.
27038 (struct directive): Make width and prec arrays.
27039 (directive::set_width, directive::set_precision): Call get_int_range.
27040 (format_integer, format_floating): Handle width and precision ranges.
27041 (format_string, parse_directive): Same.
27043 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27046 * dwarf2out.c (generate_skeleton_bottom_up): For children with
27047 comdat_type_p set, just clone them, but keep the children in the
27051 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
27052 which have direct callers with -fvar-tracking-assignments enabled
27054 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
27055 inside of type units.
27057 2017-01-26 Martin Sebor <msebor@redhat.com>
27059 PR middle-end/78703
27060 * gimple-ssa-sprintf.c (struct result_range): Add likely and
27062 (struct format_result): Replace number_chars, number_chars_min,
27063 and number_chars_max with a single member of struct result_range.
27065 (format_result::operator+=): Adjust.
27066 (struct fmtresult): Remove bounded. Handle likely and unlikely
27068 (fmtresult::adjust_for_width_or_precision): New function.
27069 (fmtresult:type_max_digits): New function.
27070 (bytes_remaining): Handle likely and unlikely counters.
27071 (min_bytes_remaining): Remove.
27072 (format_percent): Simplify.
27073 (format_integer, format_floating): Set likely and unlikely counters.
27074 (get_string_length, format_character, format_string): Same.
27075 (format_plain, should_warn_p): New function.
27076 (maybe_warn): Call should_warn_p. Update diagnostic messages
27077 and handle those for all directives, including plain strings.
27078 (format_directive): Handle likely and unlikely counters.
27079 Remove unnecessary quoting from diagnostics. Add an informational
27081 (add_bytes): Remove.
27082 (pass_sprintf_length::compute_format_length): Simplify.
27083 (try_substitute_return_value): Handle likely and unlikely counters.
27085 2017-01-26 Carl Love <cel@us.ibm.com>
27087 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
27088 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
27090 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
27093 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
27094 endianess for subregs into account.
27095 * lra-constraints.c (lra_constraints): Do risky transformations
27096 always on the first iteration.
27097 * lra-lives.c (check_pseudos_live_through_calls): Add arg
27098 last_call_used_reg_set.
27099 (process_bb_lives): Define and use last_call_used_reg_set.
27100 * lra.c (lra): Always continue after lra_constraints on the first
27103 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
27105 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
27107 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
27109 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27111 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
27112 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
27113 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
27114 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
27115 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
27116 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
27117 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
27118 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
27119 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
27121 2017-01-26 Marek Polacek <polacek@redhat.com>
27124 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
27125 for the third operand.
27127 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27129 PR middle-end/79236
27130 * omp-low.c (struct omp_context): Add simt_stmt field.
27131 (scan_omp_for): Return omp_context *.
27132 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
27133 context to the _simt_ SIMD stmt.
27134 (lower_omp_for): For combined SIMD with sibling _simt_
27135 SIMD, make sure to use the same decls in _looptemp_
27136 clauses as in the sibling.
27138 2017-01-26 David Sherwood <david.sherwood@arm.com>
27140 PR middle-end/79212
27141 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
27144 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27147 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
27148 emit fld b; fld a; if possible.
27150 * brig-builtins.def: Update copyright years.
27151 * config/arm/arm_acle_builtins.def: Update copyright years.
27153 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
27156 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
27157 constraint instead of o for the stxsd instruction.
27159 2017-01-25 Carl Love <cel@us.ibm.com>
27161 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
27162 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
27164 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
27166 * doc/invoke.texi (C++ Dialect Options): Fix typo.
27168 2017-01-25 Richard Biener <rguenther@suse.de>
27170 PR tree-optimization/69264
27171 * target.def (vector_alignment_reachable): Improve documentation.
27172 * doc/tm.texi: Regenerate.
27173 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
27175 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
27176 earlier changes with respect to TYPE_USER_ALIGN.
27177 (vector_alignment_reachable_p): Likewise. Improve dumping.
27179 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27182 * config/arm/arm.md (xordi3): Force constant operand into a register
27185 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27187 * doc/invoke.texi (-fstore-merging): Correct default optimization
27188 levels at which it is enabled.
27189 (-O): Move -fstore-merging from list to...
27192 2017-01-25 Richard Biener <rguenther@suse.de>
27195 * omp-expand.c: Include debug.h.
27196 (expand_omp_taskreg): Make sure to generate early debug before
27197 outlining anything from a function.
27198 (expand_omp_target): Likewise.
27199 (grid_expand_target_grid_body): Likewise.
27201 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
27204 * asan.c (get_translation_unit_decl): New function.
27205 (asan_add_global): Extract modules file name from globals
27206 TRANSLATION_UNIT_DECL name.
27208 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
27211 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
27212 for long calls with APCS frame and VFP.
27214 2017-01-24 David Malcolm <dmalcolm@redhat.com>
27216 * cfg.c (original_copy_tables_initialized_p): New function.
27217 * cfg.h (original_copy_tables_initialized_p): New decl.
27218 * cfgrtl.c (relink_block_chain): Guard the call to
27219 free_original_copy_tables with a call to
27220 original_copy_tables_initialized_p.
27221 * cgraph.h (symtab_node::native_rtl_p): New decl.
27222 * cgraphunit.c (symtab_node::native_rtl_p): New function.
27223 (symtab_node::needed_p): Don't assert for early assembly output
27224 for __RTL functions.
27225 (cgraph_node::finalize_function): Set "force_output" for __RTL
27227 (cgraph_node::analyze): Bail out early for __RTL functions.
27228 (analyze_functions): Update assertion to support __RTL functions.
27229 (cgraph_node::expand): Bail out early for __RTL functions.
27230 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
27232 * function.h (struct function): Update comment for field
27234 * gimple-expr.c: Include "tree-pass.h".
27235 (gimple_has_body_p): Return false for __RTL functions.
27236 * Makefile.in (OBJS): Add run-rtl-passes.o.
27237 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
27239 (gcc::pass_manager::get_clean_slate): New accessor.
27240 * passes.c: Include "insn-addr.h".
27241 (should_skip_pass_p): Add logging. Update logic for running
27242 "expand" to be compatible with both __GIMPLE and __RTL. Guard
27243 property-provider override so it is only done for gimple passes.
27245 (skip_pass): New function.
27246 (execute_one_pass): Call skip_pass when skipping passes.
27247 * read-md.c (md_reader::read_char): Support filtering
27248 the input to a subset of line numbers.
27249 (md_reader::md_reader): Initialize fields
27250 m_first_line and m_last_line.
27251 (md_reader::read_file_fragment): New function.
27252 * read-md.h (md_reader::read_file_fragment): New decl.
27253 (md_reader::m_first_line): New field.
27254 (md_reader::m_last_line): New field.
27255 * read-rtl-function.c (function_reader::create_function): Only
27256 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
27257 curr_properties. Set DECL_INITIAL to a dummy block.
27258 (read_rtl_function_body_from_file_range): New function.
27259 * read-rtl-function.h (read_rtl_function_body_from_file_range):
27261 * run-rtl-passes.c: New file.
27262 * run-rtl-passes.h: New file.
27264 2017-01-24 Jeff Law <law@redhat.com>
27266 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
27269 2017-01-24 Bin Cheng <bin.cheng@arm.com>
27271 PR tree-optimization/79159
27272 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
27273 (record_nonwrapping_iv): Improve boundary using above function if no
27274 value range information.
27276 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
27277 Martin Jambor <mjambor@suse.cz>
27279 * brig-builtins.def: New file.
27280 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
27281 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
27282 (DEF_HSAIL_SAT_BUILTIN): Likewise.
27283 (DEF_HSAIL_INTR_BUILTIN): Likewise.
27284 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
27285 * builtin-types.def (BT_INT8): New.
27286 (BT_INT16): Likewise.
27287 (BT_UINT8): Likewise.
27288 (BT_UINT16): Likewise.
27289 (BT_FN_ULONG): Likewise.
27290 (BT_FN_UINT_INT): Likewise.
27291 (BT_FN_UINT_ULONG): Likewise.
27292 (BT_FN_UINT_LONG): Likewise.
27293 (BT_FN_UINT_PTR): Likewise.
27294 (BT_FN_ULONG_PTR): Likewise.
27295 (BT_FN_INT8_FLOAT): Likewise.
27296 (BT_FN_INT16_FLOAT): Likewise.
27297 (BT_FN_UINT32_FLOAT): Likewise.
27298 (BT_FN_UINT16_FLOAT): Likewise.
27299 (BT_FN_UINT8_FLOAT): Likewise.
27300 (BT_FN_UINT64_FLOAT): Likewise.
27301 (BT_FN_UINT16_UINT32): Likewise.
27302 (BT_FN_UINT32_UINT16): Likewise.
27303 (BT_FN_UINT16_UINT16_UINT16): Likewise.
27304 (BT_FN_INT_PTR_INT): Likewise.
27305 (BT_FN_UINT_PTR_UINT): Likewise.
27306 (BT_FN_LONG_PTR_LONG): Likewise.
27307 (BT_FN_ULONG_PTR_ULONG): Likewise.
27308 (BT_FN_VOID_UINT64_UINT64): Likewise.
27309 (BT_FN_UINT8_UINT8_UINT8): Likewise.
27310 (BT_FN_INT8_INT8_INT8): Likewise.
27311 (BT_FN_INT16_INT16_INT16): Likewise.
27312 (BT_FN_INT_INT_INT): Likewise.
27313 (BT_FN_UINT_FLOAT_UINT): Likewise.
27314 (BT_FN_FLOAT_UINT_UINT): Likewise.
27315 (BT_FN_ULONG_UINT_UINT): Likewise.
27316 (BT_FN_ULONG_UINT_PTR): Likewise.
27317 (BT_FN_ULONG_ULONG_ULONG): Likewise.
27318 (BT_FN_UINT_UINT_UINT): Likewise.
27319 (BT_FN_VOID_UINT_PTR): Likewise.
27320 (BT_FN_UINT_UINT_PTR: Likewise.
27321 (BT_FN_UINT32_UINT64_PTR): Likewise.
27322 (BT_FN_INT_INT_UINT_UINT): Likewise.
27323 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
27324 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
27325 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
27326 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
27327 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
27328 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
27329 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
27330 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
27331 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
27332 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
27333 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
27334 * doc/frontends.texi: List BRIG FE.
27335 * doc/install.texi (Testing): Add BRIG tesring requirements.
27336 * doc/invoke.texi (Overall Options): Mention BRIG.
27337 * doc/standards.texi (Standards): Doucment BRIG HSA version.
27339 2017-01-24 Richard Biener <rguenther@suse.de>
27341 PR translation/79208
27342 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
27344 2017-01-24 Martin Jambor <mjambor@suse.cz>
27347 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
27348 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
27349 and known_contexts.
27351 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
27353 PR middle-end/79123
27354 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
27355 casts from signed to unsigned really don't have a range.
27357 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
27359 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
27360 GMP_RNDx for compatiblity.
27362 2017-01-24 Martin Liska <mliska@suse.cz>
27365 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
27366 that would prevent us to call alloca with -1 as argument.
27368 2017-01-24 Jakub Jelinek <jakub@redhat.com>
27370 * dwarf2out.c (output_compilation_unit_header, output_file_names):
27371 Avoid -Wformat-security warning.
27373 2017-01-23 Andrew Pinski <apinski@cavium.com>
27375 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
27378 2017-01-23 Martin Sebor <msebor@redhat.com>
27380 PR middle-end/78703
27381 * gimple-ssa-sprintf.c (warn_level): New global.
27382 (format_integer): Use it here and throughout the rest of the file.
27383 Use the same switch to compute sign as base.
27384 (maybe_warn): New function.
27385 (format_directive): Factor out warnings into maybe_warn.
27386 Add debugging output. Use warn_level.
27387 (add_bytes): Use warn_level.
27388 (pass_sprintf_length::compute_format_length): Add debugging output.
27389 (try_substitute_return_value): Same.
27390 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
27392 PR middle-end/78703
27393 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
27394 (struct fmtresult, format_integer, format_floating): Adjust.
27395 (fmtresult::fmtresult): Set max correctly in two argument ctor.
27396 (get_string_length, format_string,format_directive): Same.
27397 (pass_sprintf_length::compute_format_length): Same.
27398 (try_substitute_return_value): Simplify slightly.
27400 PR middle-end/78703
27401 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
27402 (fmtresult::operator+=): Outlined.
27403 (struct fmtresult): Add ctors.
27404 (struct conversion_spec): Rename...
27405 (struct directive): ...to this. Add and remove data members.
27406 (directive::set_width, directive::set_precision): New functions.
27407 (format_percent): Use fmtresult ctor.
27408 (get_width_and_precision): Remove.
27409 (format_integer): Make naming changes. Avoid computing width and
27411 (format_floating): Same. Adjust indentation.
27412 (format_character, format_none): New functions.
27413 (format_string): Moved character handling to format_character.
27414 (format_directive): Remove arguments, change return type.
27415 (parse_directive): New function.
27416 (pass_sprintf_length::compute_format_length): Move directive
27417 parsing to parse_directive.
27419 2017-01-23 Jakub Jelinek <jakub@redhat.com>
27421 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
27422 (assign_assembler_name_if_needed): ... this.
27423 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
27424 (assign_assembler_name_if_needed): ... this.
27425 (free_lang_data_in_cgraph): Adjust callers.
27426 * cgraphunit.c (cgraph_node::analyze): Likewise.
27427 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
27430 2017-01-23 Richard Biener <rguenther@suse.de>
27432 PR tree-optimization/79088
27433 PR tree-optimization/79188
27434 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
27435 resetting loop bounds after last path deletion. Reset loop
27436 bounds of the target loop, make code match the comments.
27437 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
27438 Make sure loops need no fixups.
27440 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
27442 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
27443 exponent support with double type for first argument.
27444 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
27445 type returned by __builtin_vec_extract_sig,
27446 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
27447 functions from "vector int" to "vector unsigned int" or from
27448 "vector long long int" to "vector unsigned long long int".
27449 Changed type returned by __builtin_vec_extract_exp,
27450 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
27451 functions from "vector int" to "vector unsigned int" or from
27452 "vector long long int" to "vector unsigned long long int".
27453 Changed return type of __builtin_vec_test_data_class,
27454 __builtin_vec_test_data_class_sp, and
27455 __builtin_vec_test_data_class_dp from "vector int" to
27456 "vector bool int" or from "vector long long int" to "vector bool
27457 long long int" and changed second argument type from "unsigned
27458 int" to "int". Added new overloaded function forms "vector float
27459 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
27460 "vector float __builtin_vec_insert_exp_sp (vector float, vector
27461 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
27462 double, vector unsigned long long int)" and "vector double
27463 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
27464 long int)". Changed return type of
27465 __builtin_scalar_test_data_class and
27466 __builtin_scalar_test_data_class_sp and
27467 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
27468 int" and changed second argument from "unsigned int" to "int".
27469 Changed type returned by __builtin_scalar_test_neg,
27470 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
27471 from "int" to "bool int". Added new overloaded function form
27472 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
27473 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
27474 exponent double-precision with floating point first argument.
27475 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
27476 documentation of scalar_test_data_class, scalar_test_neg,
27477 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
27478 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
27479 vec_test_data_class built-in functions to reflect refinements in
27480 their type signatures.
27482 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
27484 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
27486 (aarch64_elf_asm_destructor): Likewise.
27488 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
27490 PR rtl-optimization/78634
27491 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
27492 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
27493 * ifcvt.c (noce_try_cmove): Add missing cost check.
27495 PR rtl-optimization/71724
27496 * combine.c (if_then_else_cond): Look for situations where it is
27497 beneficial to undo the work of one of the recursive calls.
27499 2017-01-23 Bin Cheng <bin.cheng@arm.com>
27501 PR tree-optimization/70754
27502 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
27503 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
27504 combined stmt before it if not NULL.
27505 (combine_chains): Process refs reversely and compute dominance point
27508 2017-01-23 Martin Liska <mliska@suse.cz>
27510 PR tree-optimization/79196
27511 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
27512 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
27514 (strlen_optimize_stmt): Call the renamed function.
27516 2017-01-23 Michael Matz <matz@suse.de>
27518 PR tree-optimization/78384
27519 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
27521 2017-01-23 Richard Biener <rguenther@suse.de>
27523 PR tree-optimization/79186
27524 * tree-vrp.c (register_new_assert_for): Make sure we've seen
27525 both incoming edges before moving an assert.
27527 2017-01-23 Martin Jambor <mjambor@suse.cz>
27529 * ipa-prop.c (load_from_param_1): Removed.
27530 (load_from_unmodified_param): Bits from load_from_param_1 put back
27532 (load_from_param): Removed.
27533 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
27534 with stmt. Reverted back to use of load_from_unmodified_param.
27536 2017-01-23 Martin Jambor <mjambor@suse.cz>
27539 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
27540 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
27541 field a pointer to garbage collected vector, mark lattices and
27542 ipcp_orig_node with GTY((skip)).
27543 (ipa_get_param_count): Adjust to descriptors being a pointer.
27544 (ipa_get_param): Likewise.
27545 (ipa_get_type): Likewise.
27546 (ipa_get_param_move_cost): Likewise.
27547 (ipa_set_param_used): Likewise.
27548 (ipa_get_controlled_uses): Likewise.
27549 (ipa_set_controlled_uses): Likewise.
27550 (ipa_is_param_used): Likewise.
27551 (ipa_node_params_t): Move into garbage collector. New methods insert
27553 (ipa_node_params_sum): Annotate wth GTY(()).
27554 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
27556 (ipa_load_from_parm_agg): Adjust declaration.
27557 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
27558 * ipa-profile.c (ipa_profile): Likewise.
27559 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
27560 (ipa_populate_param_decls): Make descriptors parameter garbage
27562 (ipa_dump_param): Adjust to descriptors being a pointer.
27563 (ipa_alloc_node_params): Likewise.
27564 (ipa_initialize_node_params): Likewise.
27565 (load_from_param_1): Make descriptors parameter garbage collected.
27566 (load_from_unmodified_param): Likewise.
27567 (load_from_param): Likewise.
27568 (ipa_load_from_parm_agg): Likewise.
27569 (ipa_node_params::~ipa_node_params): Removed.
27570 (ipa_free_all_node_params): Remove call to delete operator.
27571 (ipa_node_params_t::insert): New.
27572 (ipa_node_params_t::remove): Likewise.
27573 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
27574 copy known_csts and known_contexts vectors.
27575 (ipa_read_node_info): Adjust to descriptors being a pointer.
27576 (ipcp_modif_dom_walker): Make m_descriptors field garbage
27578 (ipcp_transform_function): Make descriptors variable garbage
27581 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
27583 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
27584 * config/i386/avx512dqintrin.h: Ditto.
27585 * config/i386/avx512fintrin.h: Ditto.
27586 * config/i386/i386.c: Handle new builtins.
27587 * config/i386/i386-builtin.def: Add new builtins.
27588 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
27589 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
27591 2017-01-23 Jakub Jelinek <jakub@redhat.com>
27592 Martin Liska <mliska@suse.cz>
27594 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
27595 * asan.c (asan_expand_poison_ifn): Support stores and use
27596 appropriate ASAN report function.
27597 * internal-fn.c (expand_ASAN_POISON_USE): New function.
27598 * internal-fn.def (ASAN_POISON_USE): Declare.
27599 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
27600 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
27601 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
27602 ASAN_POISON calls w/o LHS.
27603 * tree-ssa.c (execute_update_addresses_taken): Create clobber
27604 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
27605 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
27606 * gimplify.c (asan_poison_variables): Add attribute
27607 use_after_scope_memory to variables that really needs to live
27609 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
27610 having the attribute.
27612 2017-01-23 Martin Liska <mliska@suse.cz>
27614 * asan.c (create_asan_shadow_var): New function.
27615 (asan_expand_poison_ifn): Likewise.
27616 * asan.h (asan_expand_poison_ifn): New declaration.
27617 * internal-fn.c (expand_ASAN_POISON): Likewise.
27618 * internal-fn.def (ASAN_POISON): New builtin.
27619 * sanopt.c (pass_sanopt::execute): Expand
27620 asan_expand_poison_ifn.
27621 * tree-inline.c (copy_decl_for_dup_finish): Make function
27623 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
27624 * tree-ssa.c (is_asan_mark_p): New function.
27625 (execute_update_addresses_taken): Rewrite local variables
27626 (identified just by use-after-scope as addressable) into SSA.
27628 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
27630 * doc/install.texi (Specific): opensource.apple.com uses https
27631 now. Remove trailing slash.
27633 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
27635 * README.Portability: Remove note on an Irix compatibility issue.
27637 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
27639 * gcov.c (INCLUDE_ALGORITHM): Define.
27640 (INCLUDE_VECTOR): Define.
27641 No longer include <vector> and <algorithm> directly.
27643 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
27645 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
27647 * doc/invoke.texi (Code Gen Options): Ditto.
27649 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
27652 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
27654 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
27656 rtl-optimization/79125
27657 * cprop.c (local_cprop_pass): Handle cases where we make an
27658 unconditional trap.
27660 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
27664 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
27665 read from, for big endian.
27667 2017-01-20 Jiong Wang <jiong.wang@arm.com>
27669 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
27670 register pauth builtins for LP64 only.
27672 2017-01-20 Marek Polacek <polacek@redhat.com>
27675 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
27678 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
27680 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
27682 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
27683 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
27684 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
27686 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27689 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
27690 in big-endian mode when they are not a single duplicated value.
27692 2017-01-20 Richard Biener <rguenther@suse.de>
27694 * BASE-VER: Bump to 7.0.1.
27696 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
27698 * omp-low.c (omplow_simd_context): New struct. Use it...
27699 (lower_rec_simd_input_clauses): ...here and...
27700 (lower_rec_input_clauses): ...here to hold common data. Adjust all
27701 references to idx, lane, max_vf, is_simt.
27703 2017-01-20 Graham Markall <graham.markall@embecosm.com>
27705 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
27708 2017-01-20 Martin Jambor <mjambor@suse.cz>
27710 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
27711 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
27713 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
27714 (GTFILES): Rename hsa.c to hsa-common.c.
27715 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
27716 * hsa-dump.c: Likewise.
27717 * hsa-gen.c: Likewise.
27718 * hsa-regalloc.c: Likewise.
27719 * ipa-hsa.c: Likewise.
27720 * omp-expand.c: Likewise.
27721 * omp-low.c: Likewise.
27722 * toplev.c: Likewise.
27724 2017-01-20 Marek Polacek <polacek@redhat.com>
27727 * doc/invoke.texi: Document -Wduplicated-branches.
27728 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
27729 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
27730 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
27731 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
27732 return 0 only when not OEP_LEXICOGRAPHIC.
27733 (fold_build_cleanup_point_expr): Use the expression
27734 location when building CLEANUP_POINT_EXPR.
27735 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
27736 * tree.c (add_expr): Handle error_mark_node.
27738 2017-01-20 Martin Liska <mliska@suse.cz>
27741 * tree-profile.c (init_ic_make_global_vars): Do not call
27743 (gimple_init_gcov_profiler): Likewise.
27745 2017-01-20 Martin Liska <mliska@suse.cz>
27748 * cgraph.h (maybe_create_reference): Remove argument and
27750 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
27752 * ipa-cp.c (create_specialized_node): Likewise.
27753 * symtab.c (symtab_node::maybe_create_reference): Handle
27754 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
27756 2017-01-20 Martin Liska <mliska@suse.cz>
27758 * read-rtl-function.c (function_reader::create_function): Use
27759 build_decl instread of build_decl_stat.
27761 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
27763 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
27764 * config/i386/avx512dqintrin.h: Ditto.
27765 * config/i386/avx512fintrin.h: Ditto.
27766 * config/i386/i386-builtin-types.def: Add new types.
27767 * config/i386/i386.c: Handle new types.
27768 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
27769 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
27770 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
27771 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
27772 (__builtin_ia32_kshiftridi): New.
27773 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
27775 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
27779 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
27780 define to rs6000_init_stack_protect_guard.
27781 (rs6000_init_stack_protect_guard): New function.
27783 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
27784 Yunqiang Su <yunqiang.su@imgtec.com>
27786 * config.gcc (supported_defaults): Add madd4.
27787 (with_madd4): Add validation.
27788 (all_defaults): Add madd4.
27789 * config/mips/mips.opt (mmadd4): New option.
27790 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
27792 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
27794 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
27795 (ISA_HAS_FUSED_MADD4): Likewise.
27796 * doc/invoke.texi (-mmadd4): Document the new option.
27797 * doc/install.texi (--with-madd4): Document the new option.
27799 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27801 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
27802 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
27803 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
27804 (aarch64_init_pauth_hint_builtins): New.
27805 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
27806 (aarch64_expand_builtin): Expand new builtins.
27808 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27810 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
27811 * combine-stack-adj.c (no_unhandled_cfa): Handle
27812 REG_CFA_TOGGLE_RA_MANGLE.
27813 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
27814 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
27815 info for return address signing.
27816 (aarch64_expand_epilogue): Likewise.
27818 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27820 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
27821 * config/aarch64/aarch64-protos.h
27822 (aarch64_return_address_signing_enabled): New declaration.
27823 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
27825 (aarch64_expand_prologue): Sign return address before it's pushed onto
27827 (aarch64_expand_epilogue): Authenticate return address fetched from
27829 (aarch64_override_options): Sanity check for ILP32 and ISA level.
27830 (aarch64_attributes): New function attributes for "sign-return-address".
27831 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
27832 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
27833 ("*do_return"): Generate combined instructions according to key index.
27834 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
27835 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
27837 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
27838 * config/aarch64/aarch64.opt (msign-return-address=): New.
27839 * doc/extend.texi (AArch64 Function Attributes): Documents
27840 "sign-return-address=".
27841 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
27843 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
27845 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
27846 overall option summary.
27848 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27850 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
27851 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
27852 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
27853 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
27855 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
27857 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
27858 -mpower9-minmax by default for -mcpu=power9.
27859 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
27860 128-bit floating point.
27862 2017-01-20 Alan Modra <amodra@gmail.com>
27864 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
27865 optimizing for size.
27867 2017-01-20 Alan Modra <amodra@gmail.com>
27870 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
27871 for strcmp and strncmp from corresponding builtin decl.
27873 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
27875 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
27876 instead of i386/rtems-64.h.
27877 * config/i386/rtems-64.h: Remove.
27879 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
27883 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
27885 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
27887 2017-01-19 Tamar Christina <tamar.christina@arm.com>
27889 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
27890 Change int to HOST_WIDE_INT.
27891 * config/aarch64/aarch64-protos.h
27892 (aarch64_simd_gen_const_vector_dup): Likewise.
27893 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
27895 2017-01-19 David Malcolm <dmalcolm@redhat.com>
27897 * langhooks-def.h (lhd_type_for_size): New decl.
27898 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
27899 * langhooks.c (lhd_type_for_size): New function, taken from
27902 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
27904 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
27905 define_bypass for CR latency.
27906 (power9-cracked-alu): Update bypass latency and remove power9-branch.
27907 (power9-alu2): Add define_bypass for CR latency.
27909 (power9-mul): Update insn latency.
27910 (power9-mul-compare): Update insn latency, bypass latency and remove
27913 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27915 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
27917 * config/aarch64/aarch64.md
27918 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
27919 aarch64_nopcrelative_literal_loads.
27920 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
27922 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
27924 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
27925 TARGET_LOONGSON_3A.
27926 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
27928 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
27931 * config.gcc (supported_defaults): Add lxc1-sxc1.
27932 (with_lxc1_sxc1): Add validation.
27933 (all_defaults): Add lxc1-sxc1.
27934 * config/mips/mips.opt (mlxc1-sxc1): New option.
27935 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
27937 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
27938 __mips_no_lxc1_sxc1.
27939 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
27940 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
27941 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
27943 2017-01-19 Richard Biener <rguenther@suse.de>
27945 PR tree-optimization/72488
27946 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
27947 sure to restore SSA info.
27948 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
27950 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
27952 PR rtl-optimization/79121
27953 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
27954 of the inner type when shifting an extended value.
27956 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
27959 * symtab.c (symtab_node::equal_address_to): Fix comparing of
27960 interposable aliases.
27962 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
27965 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
27966 Use the evmergelohi instruction.
27967 (mov_si<mode>_e500_subreg4_2_le): Likewise.
27968 (mov_sitf_e500_subreg8_2_be): Likewise.
27969 (mov_sitf_e500_subreg12_2_le): Likewise.
27970 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
27971 (mov_si<mode>_e500_subreg4_2_be): Likewise.
27972 (mov_sitf_e500_subreg8_2_le): Likewise.
27973 (mov_sitf_e500_subreg12_2_be): Likewise.
27975 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27977 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
27978 attribute from vecsimple to vecperm.
27979 (altivec_vbpermq2): Likewise.
27981 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27984 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
27986 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27987 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
27988 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
27989 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
27990 case where N arg is SIZE_MAX.
27991 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
27992 (cmpstrsi): Add pattern.
27994 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27996 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27997 __builtin_vec_revb builtins.
27998 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
27999 built-in functions to support generation of the ISA 3.0 XXBR<x>
28000 vector byte reverse instructions.
28001 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
28002 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
28003 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
28004 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
28005 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
28006 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
28007 (P9V_BUILTIN_VEC_REVB): Likewise.
28008 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
28009 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
28010 (p9_xxbrq_v16qi): Likewise.
28011 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
28012 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
28013 (p9_xxbrh_v8hi): Likewise.
28014 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
28015 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
28016 vec_revb built-in functions.
28018 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
28020 PR rtl-optimization/78952
28021 * config/i386/i386.md (any_extract): New code iterator.
28022 (*insvqi_2): Use any_extract for source operand.
28023 (*insvqi_3): Use any_shiftrt for source operand.
28025 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
28027 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
28029 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
28031 2017-01-18 Matthias Klose <doko@ubuntu.com>
28033 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
28035 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28037 * config/rs6000/altivec.h (vec_bperm): Change #define.
28038 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
28039 (altivec_vbpermq2): New define_insn.
28040 (altivec_vbpermd): Likewise.
28041 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
28042 function interface.
28043 (VBPERMD): Likewise.
28044 (VBPERM): New polymorphic function interface.
28045 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
28046 Add entries for P9V_BUILTIN_VEC_VBPERM.
28047 * doc/extend.texi: Add interfaces for vec_bperm.
28049 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28051 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
28052 first letter of error messages.
28053 (s390_resolve_overloaded_builtin): Likewise.
28054 * config/s390/s390.c (s390_expand_builtin): Likewise.
28055 (s390_invalid_arg_for_unprototyped_fn): Likewise.
28056 (s390_valid_target_attribute_inner_p): Likewise.
28057 * config/s390/s390.md ("tabort"): Likewise.
28059 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
28061 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
28062 (ISA_AVOID_DIV_HILO): New macro.
28063 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
28064 (ISA_HAS_DDIV): Likewise.
28066 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28068 * doc/invoke.texi (fabi-version): Correct number of occurrences.
28070 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28072 * doc/invoke.texi (fabi-version): Spelling fix.
28074 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28077 * doc/invoke.texi (fabi-version): Mention mangling fix for
28080 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28083 * doc/invoke.texi (fabi-version): Document discriminator mangling.
28085 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
28088 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
28089 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
28091 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
28093 (stack_protect_test): Ditto.
28094 * config/rs6000/rs6000.opt (mstack-protector-guard=,
28095 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
28097 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
28098 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
28099 -mstack-protector-guard-offset=.
28100 (RS/6000 and PowerPC Options): Ditto.
28102 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
28104 * config/i386/i386.h (MASK_CLASS_P): New define.
28105 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
28106 there are no registers from different register sets also when
28107 mask registers are used. Update function comment.
28108 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
28109 to (*k/*r) and (*k/*km) alternatives.
28111 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
28113 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
28114 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
28115 (EH_RETURN_HANDLER_RTX): New define.
28116 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
28117 Force frame pointer in EH return functions.
28118 (aarch64_expand_epilogue): Add barrier for eh_return.
28119 (aarch64_final_eh_return_addr): Remove.
28120 (aarch64_eh_return_handler_rtx): New function.
28121 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
28123 (aarch64_eh_return_handler_rtx): New prototype.
28125 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28127 * config/rs6000/altivec.h (vec_rlmi): New #define.
28128 (vec_vrlnm): Likewise.
28129 (vec_rlnm): Likewise.
28130 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
28131 (UNSPEC_VRLNM): Likewise.
28132 (VIlong): New mode iterator.
28133 (altivec_vrl<VI_char>mi): New define_insn.
28134 (altivec_vrl<VI_char>nm): Likewise.
28135 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
28137 (VRLDNM): Likewise.
28138 (RLNM): New polymorphic function entry.
28139 (VRLWMI): New monomorphic function entry.
28140 (VRLDMI): Likewise.
28141 (RLMI): New polymorphic function entry.
28142 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
28143 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
28144 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
28147 2017-01-17 Jakub Jelinek <jakub@redhat.com>
28150 * dwarf2out.c (field_byte_offset): Restore the
28151 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
28152 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
28153 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
28156 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
28159 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
28161 2017-01-17 Jakub Jelinek <jakub@redhat.com>
28164 * dwarf2out.c (add_data_member_location_attribute): For constant
28165 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
28166 instead of DW_AT_data_member_location, DW_AT_bit_offset and
28167 DW_AT_byte_size attributes.
28169 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
28171 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
28172 after forcing to constant memory when the code model is medium.
28174 2017-01-17 Julia Koval <julia.koval@intel.com>
28177 * config/i386/avx512fintrin.h
28178 (_mm512_i32gather_ps): Change __addr type to void const*.
28179 (_mm512_mask_i32gather_ps): Ditto.
28180 (_mm512_i32gather_pd): Ditto.
28181 (_mm512_mask_i32gather_pd): Ditto.
28182 (_mm512_i64gather_ps): Ditto.
28183 (_mm512_mask_i64gather_ps): Ditto.
28184 (_mm512_i64gather_pd): Ditto.
28185 (_mm512_mask_i64gather_pd): Ditto.
28186 (_mm512_i32gather_epi32): Ditto.
28187 (_mm512_mask_i32gather_epi32): Ditto.
28188 (_mm512_i32gather_epi64): Ditto.
28189 (_mm512_mask_i32gather_epi64): Ditto.
28190 (_mm512_i64gather_epi32): Ditto.
28191 (_mm512_mask_i64gather_epi32): Ditto.
28192 (_mm512_i64gather_epi64): Ditto.
28193 (_mm512_mask_i64gather_epi64): Ditto.
28194 (_mm512_i32scatter_ps): Change __addr type to void*.
28195 (_mm512_mask_i32scatter_ps): Ditto.
28196 (_mm512_i32scatter_pd): Ditto.
28197 (_mm512_mask_i32scatter_pd): Ditto.
28198 (_mm512_i64scatter_ps): Ditto.
28199 (_mm512_mask_i64scatter_ps): Ditto.
28200 (_mm512_i64scatter_pd): Ditto.
28201 (_mm512_mask_i64scatter_pd): Ditto.
28202 (_mm512_i32scatter_epi32): Ditto.
28203 (_mm512_mask_i32scatter_epi32): Ditto.
28204 (_mm512_i32scatter_epi64): Ditto.
28205 (_mm512_mask_i32scatter_epi64): Ditto.
28206 (_mm512_i64scatter_epi32): Ditto.
28207 (_mm512_mask_i64scatter_epi32): Ditto.
28208 (_mm512_i64scatter_epi64): Ditto.
28209 (_mm512_mask_i64scatter_epi64): Ditto.
28210 * config/i386/avx512pfintrin.h
28211 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
28212 (_mm512_mask_prefetch_i32gather_ps): Ditto.
28213 (_mm512_mask_prefetch_i64gather_pd): Ditto.
28214 (_mm512_mask_prefetch_i64gather_ps): Ditto.
28215 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
28216 (_mm512_prefetch_i32scatter_ps): Ditto.
28217 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
28218 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
28219 (_mm512_prefetch_i64scatter_pd): Ditto.
28220 (_mm512_prefetch_i64scatter_ps): Ditto.
28221 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
28222 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
28223 * config/i386/avx512vlintrin.h
28224 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
28225 (_mm_mmask_i32gather_ps): Ditto.
28226 (_mm256_mmask_i32gather_pd): Ditto.
28227 (_mm_mmask_i32gather_pd): Ditto.
28228 (_mm256_mmask_i64gather_ps): Ditto.
28229 (_mm_mmask_i64gather_ps): Ditto.
28230 (_mm256_mmask_i64gather_pd): Ditto.
28231 (_mm_mmask_i64gather_pd): Ditto.
28232 (_mm256_mmask_i32gather_epi32): Ditto.
28233 (_mm_mmask_i32gather_epi32): Ditto.
28234 (_mm256_mmask_i32gather_epi64): Ditto.
28235 (_mm_mmask_i32gather_epi64): Ditto.
28236 (_mm256_mmask_i64gather_epi32): Ditto.
28237 (_mm_mmask_i64gather_epi32): Ditto.
28238 (_mm256_mmask_i64gather_epi64): Ditto.
28239 (_mm_mmask_i64gather_epi64): Ditto.
28240 (_mm256_i32scatter_ps): Change __addr type to void*.
28241 (_mm256_mask_i32scatter_ps): Ditto.
28242 (_mm_i32scatter_ps): Ditto.
28243 (_mm_mask_i32scatter_ps): Ditto.
28244 (_mm256_i32scatter_pd): Ditto.
28245 (_mm256_mask_i32scatter_pd): Ditto.
28246 (_mm_i32scatter_pd): Ditto.
28247 (_mm_mask_i32scatter_pd): Ditto.
28248 (_mm256_i64scatter_ps): Ditto.
28249 (_mm256_mask_i64scatter_ps): Ditto.
28250 (_mm_i64scatter_ps): Ditto.
28251 (_mm_mask_i64scatter_ps): Ditto.
28252 (_mm256_i64scatter_pd): Ditto.
28253 (_mm256_mask_i64scatter_pd): Ditto.
28254 (_mm_i64scatter_pd): Ditto.
28255 (_mm_mask_i64scatter_pd): Ditto.
28256 (_mm256_i32scatter_epi32): Ditto.
28257 (_mm256_mask_i32scatter_epi32): Ditto.
28258 (_mm_i32scatter_epi32): Ditto.
28259 (_mm_mask_i32scatter_epi32): Ditto.
28260 (_mm256_i32scatter_epi64): Ditto.
28261 (_mm256_mask_i32scatter_epi64): Ditto.
28262 (_mm_i32scatter_epi64): Ditto.
28263 (_mm_mask_i32scatter_epi64): Ditto.
28264 (_mm256_i64scatter_epi32): Ditto.
28265 (_mm256_mask_i64scatter_epi32): Ditto.
28266 (_mm_i64scatter_epi32): Ditto.
28267 (_mm_mask_i64scatter_epi32): Ditto.
28268 (_mm256_i64scatter_epi64): Ditto.
28269 (_mm256_mask_i64scatter_epi64): Ditto.
28270 (_mm_i64scatter_epi64): Ditto.
28271 (_mm_mask_i64scatter_epi64): Ditto.
28272 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
28273 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
28274 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
28275 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
28276 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
28277 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
28278 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
28279 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
28280 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
28281 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
28282 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
28283 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
28284 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
28285 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
28286 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
28287 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
28288 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
28289 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
28290 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
28291 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
28292 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
28293 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
28294 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
28295 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
28296 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
28297 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
28298 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
28299 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
28300 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
28301 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
28302 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
28303 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
28304 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
28305 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
28306 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
28307 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
28308 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
28309 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
28310 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
28311 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
28312 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
28313 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
28314 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
28315 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
28316 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
28317 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
28318 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
28319 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
28320 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
28321 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
28322 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
28323 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
28324 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
28325 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
28326 definitions accordingly.
28328 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
28329 Kuan-Lin Chen <kuanlinchentw@gmail.com>
28332 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
28335 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
28338 * ira-conflicts.c (ira_build_conflicts): Update total conflict
28339 hard regs for inner regno.
28341 2017-01-17 Martin Liska <mliska@suse.cz>
28344 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
28345 assumption and add comment.
28347 2017-01-17 Nathan Sidwell <nathan@acm.org>
28349 * ipa-visibility.c (localize_node): New function, broken out of ...
28350 (function_and_variable_visibility): ... here. Call it.
28352 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
28354 PR middle-end/77445
28355 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
28356 correctly set frequency of oudgoing edge.
28357 (duplicate_thread_path): Fix profile updating.
28359 2017-01-17 Jakub Jelinek <jakub@redhat.com>
28362 * configure.ac: Add GCC_BASE_VER.
28363 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
28364 version from BASE-VER file.
28365 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
28366 (gcc.o): Depend on $(BASEVER).
28367 * common.opt (dumpfullversion): New option.
28368 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
28369 * doc/invoke.texi: Document -dumpfullversion.
28370 * doc/install.texi: Document --with-gcc-major-version-only.
28371 * configure: Regenerated.
28373 2017-01-17 Richard Biener <rguenther@suse.de>
28375 PR tree-optimization/71433
28376 * tree-vrp.c (register_new_assert_for): Merge same asserts
28377 on all incoming edges.
28378 (process_assert_insertions_for): Handle insertions at the
28381 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
28383 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
28384 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
28386 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
28389 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
28392 2017-01-17 Alan Modra <amodra@gmail.com>
28395 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
28396 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
28397 symbolic stack limit when pic.
28399 2017-01-16 Martin Sebor <msebor@redhat.com>
28401 PR tree-optimization/78608
28402 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
28404 2017-01-16 Jeff Law <law@redhat.com>
28407 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
28408 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
28409 for several include directories that may be relative to sysroot.
28410 * config/i386/x-mingw32 (gplus_includedir): Define.
28411 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
28412 (native_system_includedir): Likewise.
28413 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
28414 override if TARGET_SYSTEM_ROOT is defined.
28415 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
28417 PR tree-optimization/79090
28418 PR tree-optimization/33562
28419 PR tree-optimization/61912
28420 PR tree-optimization/77485
28421 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
28422 and computed trims into the dump file.
28424 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
28426 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
28428 2017-01-16 Jakub Jelinek <jakub@redhat.com>
28431 * gimplify.c (gimplify_init_constructor): If want_value and
28432 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
28436 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
28437 sequence. Formatting fixes.
28438 (doloop_optimize): Formatting fixes.
28441 * gcc.c (debug_level_greater_than_spec_func): New function.
28442 (static_spec_functions): Add debug-level-gt spec function.
28443 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
28445 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
28446 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
28447 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
28448 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
28449 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
28450 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
28452 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
28454 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
28455 QImode fixups to general and mask registers only.
28457 2017-01-16 Carl Love <cel@us.ibm.com>
28459 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
28460 for built-in functions
28461 vector signed char vec_nabs (vector signed char)
28462 vector signed short vec_nabs (vector signed short)
28463 vector signed int vec_nabs (vector signed int)
28464 vector signed long long vec_nabs (vector signed long long)
28465 vector float vec_nabs (vector float)
28466 vector double vec_nabs (vector double)
28467 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
28469 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
28470 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
28471 * doc/extend.texi: Update the documentation file for the new built-in
28474 2017-01-16 Martin Sebor <msebor@redhat.com>
28476 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
28479 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28481 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
28482 UNSPEC_VSX__XXSPLTD to require special splat handling.
28484 2017-01-16 David Malcolm <dmalcolm@redhat.com>
28487 * system.h: Poison strndup.
28489 2017-01-16 Alan Modra <amodra@gmail.com>
28492 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
28495 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
28497 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
28499 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
28501 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
28502 call recog here. Assert that INSN_CODE (insn) is non-negative.
28504 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
28507 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
28509 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
28510 in the currently scheduled RTL fragment.
28512 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
28514 PR rtl-optimization/78751
28515 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
28518 2017-01-14 Jeff Law <law@redhat.com>
28520 PR tree-optimization/79090
28521 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
28522 variable length stores.
28523 (compute_trims): Delete dead assignment to *trim_tail.
28524 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
28527 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
28529 PR rtl-optimization/78626
28530 PR rtl-optimization/78727
28531 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
28532 of a block, and split such blocks after everything else is finished.
28534 2017-01-14 Alan Modra <amodra@gmail.com>
28537 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
28538 target legitimate_combined_insn.
28539 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
28540 (rs6000_legitimate_combined_insn): New function.
28541 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
28543 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
28544 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
28545 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
28547 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
28549 * doc/frontends.texi (G++ and GCC): Remove references to Java.
28551 2017-01-13 Jeff Law <law@redhat.com>
28553 PR tree-optimization/33562
28554 PR tree-optimization/61912
28555 PR tree-optimization/77485
28556 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
28558 (delete_dead_assignment): Likewise.
28559 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
28560 statement to delete_dead_call and delete_dead_assignment.
28562 2017-01-13 David Malcolm <dmalcolm@redhat.com>
28565 * substring-locations.c (format_warning_va): Strengthen case 1 so
28566 that both endpoints of the substring must be within the format
28567 range for just the substring to be printed.
28569 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
28571 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
28572 * config/i386/i386.c (ix86_target_string): Add missing options
28573 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
28574 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
28575 flags_other and ix86_target_other to flags2_other. Display unknown
28577 (ix86_valid_target_attribute_inner_p): Add missing options and
28578 reorder options by implied ISAs, as in ix86_target_string.
28580 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
28582 * hash-table.h (hash_table::too_empty_p): New function.
28583 (hash_table::expand): Use it.
28584 (hash_table::traverse): Likewise.
28585 (hash_table::empty_slot): Use sizeof (value_type) instead of
28586 sizeof (PTR) to convert bytes to elements. Shrink the table
28587 if the current size is excessive for the current number of
28590 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
28592 * ira-costs.c (record_reg_classes): Break from the inner loop
28593 early once alt_fail is known to be true. Update outer loop
28594 handling accordingly.
28596 2017-01-13 Jeff Law <law@redhat.com>
28598 * tree-ssa-dse.c (decrement_count): New function.
28599 (increment_start_addr, maybe_trim_memstar_call): Likewise.
28600 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
28601 when we know the partially dead statement is a mem* function.
28603 PR tree-optimization/61912
28604 PR tree-optimization/77485
28605 * tree-ssa-dse.c: Include expr.h.
28606 (maybe_trim_constructor_store): New function.
28607 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
28609 PR tree-optimization/33562
28610 PR tree-optimization/61912
28611 PR tree-optimization/77485
28612 * doc/invoke.texi: Document new dse-max-object-size param.
28613 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
28614 * tree-ssa-dse.c: Include params.h.
28615 (dse_store_status): New enum.
28616 (initialize_ao_ref_for_dse): New, partially extracted from
28618 (valid_ao_ref_for_dse, normalize_ref): New.
28619 (setup_live_bytes_from_ref, compute_trims): Likewise.
28620 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
28621 (maybe_trim_partially_dead_store): Likewise.
28622 (maybe_trim_complex_store): Likewise.
28623 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
28624 Track what bytes live from the original store. Return tri-state
28625 for dead, partially dead or live.
28626 (dse_dom_walker): Add constructor, destructor and new private members.
28627 (delete_dead_call, delete_dead_assignment): New extracted from
28629 (dse_optimize_stmt): Make a member of dse_dom_walker.
28630 Use initialize_ao_ref_for_dse.
28632 PR tree-optimization/33562
28633 PR tree-optimization/61912
28634 PR tree-optimization/77485
28635 * sbitmap.h (bitmap_count_bits): Prototype.
28636 (bitmap_clear_range, bitmap_set_range): Likewise.
28637 * sbitmap.c (bitmap_clear_range): New function.
28638 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
28640 2017-01-13 Martin Liska <mliska@suse.cz>
28643 * function.c (set_cfun): Add new argument force.
28644 * function.h (set_cfun): Likewise.
28645 * ipa-inline-transform.c (inline_call): Use the function when
28646 strict alising from is dropped for function we inline to.
28648 2017-01-13 Richard Biener <rguenther@suse.de>
28650 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
28651 for dumping GIMPLE INTEGER_CSTs.
28653 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28655 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
28656 to 201112L since C++17.
28658 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
28661 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
28662 if -fsanitize=kernel-address is present.
28664 2017-01-13 Richard Biener <rguenther@suse.de>
28666 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
28667 as _Literal ( type ) number in case usual suffixes do not
28668 preserve all information.
28670 2017-01-13 Richard Biener <rguenther@suse.de>
28672 PR tree-optimization/77283
28673 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
28674 and ssa-iterators.h.
28675 (is_feasible_trace): Implement a cost model based on joiner
28678 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
28681 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
28682 char or short to __float128/_Float128 directly.
28684 2017-01-12 Martin Sebor <msebor@redhat.com>
28686 to -Wformat-overflow.
28687 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
28688 (min_bytes_remaining): Same.
28689 (get_string_length): Same.
28690 (format_string): Same.
28691 (format_directive): Same.
28693 (pass_sprintf_length::handle_gimple_call): Same.
28695 2017-01-12 Jakub Jelinek <jakub@redhat.com>
28697 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
28698 info.nowrite calls with no lhs that can't throw. Return bool
28699 whether gsi_remove has been called or not.
28700 (pass_sprintf_length::handle_gimple_call): Return bool whether
28701 try_substitute_return_value called gsi_remove. Formatting fix.
28702 (pass_sprintf_length::execute): Don't use gsi_remove if
28703 handle_gimple_call returned true.
28706 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
28707 be removed due to side-effects, don't remove following barrier nor
28708 turn the successor edge into fallthru edge.
28710 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28713 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
28714 element-reversing loads and stores as not swappable.
28716 2017-01-12 Nathan Sidwell <nathan@acm.org>
28717 Nicolai Stange <nicstange@gmail.com>
28719 * combine.c (try_combine): Don't ignore result of overlap checking
28720 loop. Combine overlap & asm check into single loop.
28722 2017-01-12 Richard Biener <rguenther@suse.de>
28724 * tree-pretty-print.c (dump_generic_node): Provide -gimple
28725 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
28727 2017-01-12 Richard Biener <rguenther@suse.de>
28729 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
28730 and TS_TARGET_OPTION directly derive from TS_BASE.
28731 * tree-core.h (tree_optimization_option): Derive from tree_base.
28732 (tree_target_option): Likewise.
28734 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
28736 * config/i386/i386.c (memory_address_length): Increase len
28737 only when rip_relative_addr_p returns false.
28739 2017-01-11 Julia Koval <julia.koval@intel.com>
28741 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
28742 (OPTION_MASK_ISA_SGX_SET): New.
28743 (ix86_handle_option): Handle OPT_msgx.
28744 * config.gcc: Added sgxintrin.h.
28745 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
28746 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
28747 * config/i386/i386.c (ix86_target_string): Add -msgx.
28749 (ix86_option_override_internal): Handle new options.
28750 (ix86_valid_target_attribute_inner_p): Add sgx.
28751 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
28752 * config/i386/i386.opt: Add msgx.
28753 * config/i386/sgxintrin.h: New file.
28754 * config/i386/x86intrin.h: Add sgxintrin.h.
28756 2017-01-11 Jakub Jelinek <jakub@redhat.com>
28759 * fold-const.c (maybe_nonzero_address): Return 1 for function
28761 (tree_single_nonzero_warnv_p): Don't handle function local objects
28765 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
28768 2017-01-11 David Malcolm <dmalcolm@redhat.com>
28771 * opts.c: Include "spellcheck.h"
28772 (struct string_fragment): New struct.
28773 (struct edit_distance_traits<const string_fragment &>): New
28775 (get_closest_sanitizer_option): New function.
28776 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
28778 2017-01-11 Jakub Jelinek <jakub@redhat.com>
28780 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
28782 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
28783 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
28784 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
28785 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
28786 for initial die_offset if dwarf_split_debug_info.
28787 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
28788 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
28790 (output_skeleton_debug_sections): Formatting fix. Use
28791 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
28792 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
28794 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
28796 * config/arm/cortex-a53.md: Add bypasses for
28797 cortex_a53_r2f_cvt.
28798 (cortex_a53_r2f): Only use for transfers.
28799 (cortex_a53_f2r): Likewise.
28800 (cortex_a53_r2f_cvt): Add reservation for conversions.
28801 (cortex_a53_f2r_cvt): Likewise.
28803 2017-01-11 Tamar Christina <tamar.christina@arm.com>
28805 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
28806 to all inlined functions, change static to extern.
28808 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
28811 * config/arm/arm.c (legitimize_pic_address): Handle reference to
28813 (arm_assemble_integer): Likewise.
28815 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
28817 * config.gcc: Use new awk script to check CPU, FPU and architecture
28818 parameters for --with-... options.
28819 * config/arm/parsecpu.awk: New file
28820 * config/arm/arm-cpus.in: New file.
28821 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
28823 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
28825 * config/arm/t-arm: Update dependency rules.
28826 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
28827 of processing .def files.
28828 * config/arm/genopt.sh: Deleted.
28829 * config/arm/gentune.sh: Deleted.
28830 * config/arm/arm-cores.def: Deleted.
28831 * config/arm/arm-arches.def: Deleted.
28832 * config/arm/arm-fpus.def: Deleted.
28833 * config/arm/arm-tune.md: Regenerated.
28834 * config/arm/arm-tables.opt: Regenerated.
28835 * config/arm/arm-cpu.h: New generated file.
28836 * config/arm/arm-cpu-data.h: New generated file.
28837 * config/arm/arm-cpu-cdata.h: New generated file.
28839 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
28842 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
28844 (input_varpool_node): Unpack dynamically_initialized bit.
28846 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
28848 PR rtl-optimization/79032
28849 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
28850 the alignment of the adjusted memory reference against that of MODE,
28851 instead of the alignment of the original memory reference.
28853 2017-01-11 Martin Jambor <mjambor@suse.cz>
28855 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
28857 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
28858 decorated functions.
28860 2017-01-11 Richard Biener <rguenther@suse.de>
28862 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
28863 set range/nonnull info for PHI results. Do not set it on
28864 stmts marked for removal.
28866 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
28868 * expr.c (store_field): In the bitfield case, fetch the return value
28869 from the registers before applying a single big-endian adjustment.
28870 Always do a final load for a BLKmode value not larger than a word.
28872 2017-01-10 David Malcolm <dmalcolm@redhat.com>
28875 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
28876 that we correctly handle column numbers greater than
28877 LINE_MAP_MAX_COLUMN_NUMBER.
28879 2017-01-10 Martin Sebor <msebor@redhat.com>
28881 PR middle-end/78245
28882 * gimple-ssa-sprintf.c (get_destination_size): Call
28883 {init,fini}object_sizes.
28884 * tree-object-size.c (addr_object_size): Adjust.
28885 (pass_through_call): Adjust.
28886 (pass_object_sizes::execute): Adjust.
28887 * tree-object-size.h (fini_object_sizes): Declare.
28889 2017-01-10 Martin Sebor <msebor@redhat.com>
28891 PR tree-optimization/78775
28892 * builtins.c (get_size_range): Move...
28893 * calls.c: ...to here.
28894 (alloc_max_size): Accept zero argument.
28895 (operand_signed_p): Remove.
28896 (maybe_warn_alloc_args_overflow): Call get_size_range.
28897 * calls.h (get_size_range): Declare.
28899 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
28901 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
28902 from TI's devices.csv file as of September 2016.
28903 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
28905 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
28907 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
28908 * doc/invoke.texi: Likewise.
28909 * doc/md.texi: Likewise.
28910 * doc/objc.texi: Likewise.
28912 2017-01-10 Joshua Conner <joshconner@google.com>
28914 * config/arm/fuchsia-elf.h: New file.
28915 * config/fuchsia.h: New file.
28916 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
28917 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
28919 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
28921 2016-01-10 Richard Biener <rguenther@suse.de>
28923 PR tree-optimization/79034
28924 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28925 Propagate out degenerate PHIs in the joiner.
28927 2017-01-10 Martin Liska <mliska@suse.cz>
28929 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
28930 (sort_congruence_classes_by_decl_uid): Likewise.
28931 (sort_congruence_class_groups_by_decl_uid): Likewise.
28932 (sem_item_optimizer::merge_classes): Sort class, groups in these
28933 classes and members in the groups by DECL_UID of declarations.
28934 This would make merge operations stable.
28936 2017-01-10 Martin Liska <mliska@suse.cz>
28938 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
28939 usage of m_classes_vec.
28940 (sem_item_optimizer::~sem_item_optimizer): Likewise.
28941 (sem_item_optimizer::get_group_by_hash): Likewise.
28942 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
28943 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
28944 (sem_item_optimizer::verify_classes): Likewise.
28945 (sem_item_optimizer::process_cong_reduction): Likewise.
28946 (sem_item_optimizer::dump_cong_classes): Likewise.
28947 (sem_item_optimizer::merge_classes): Likewise.
28948 * ipa-icf.h (congruence_class_hash): Rename from
28949 congruence_class_group_hash. Remove declaration of m_classes_vec.
28951 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
28953 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
28954 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
28955 * config.gcc: Add avx512vpopcntdqintrin.h.
28956 * config/i386/avx512vpopcntdqintrin.h: New.
28957 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
28958 * config/i386/i386-builtin-types.def: Add new types.
28959 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
28960 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
28961 __builtin_ia32_vpopcountq_v8di_mask): New.
28962 * config/i386/i386-c.c (ix86_target_macros_internal): Define
28963 __AVX512VPOPCNTDQ__.
28964 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
28965 (PTA_AVX512VPOPCNTDQ): Define.
28966 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
28967 TARGET_AVX512VPOPCNTDQ_P): Define.
28968 * config/i386/i386.opt: Add mavx512vpopcntdq.
28969 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
28970 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
28972 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28974 PR middle-end/77484
28975 * predict.def (PRED_CALL): Set to 67.
28977 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
28979 * expr.c (store_field): In the bitfield case, if the value comes from
28980 a function call and is of an aggregate type returned in registers, do
28981 not modify the field mode; extract the value in all cases if the mode
28982 is BLKmode and the size is not larger than a word.
28984 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
28987 * config/i386/cpuid.h: Fix undefined behavior.
28989 2017-01-04 Jeff Law <law@redhat.com>
28991 PR tree-optimization/79007
28992 PR tree-optimization/67955
28993 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
28994 conservative for pt.null when flag_non_call_exceptions is on.
28996 2017-01-09 Jakub Jelinek <jakub@redhat.com>
28998 PR translation/79019
28999 PR translation/79020
29000 * params.def (PARAM_INLINE_MIN_SPEEDUP,
29001 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
29002 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
29004 * config/avr/avr.opt (maccumulate-args): Likewise.
29005 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
29006 * common.opt (freport-bug): Likewise.
29007 * cif-code.def (CIF_FINAL_ERROR): Likewise.
29008 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
29009 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
29010 translatable string.
29011 * config/i386/i386.c (function_value_32): Likewise.
29012 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
29013 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
29015 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
29016 * common/config/msp430/msp430-common.c (msp430_handle_option):
29018 * symtab.c (symtab_node::verify_base): Likewise.
29019 * opts.c (set_debug_level): Likewise.
29020 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
29021 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
29022 missing whitespace to translatable strings.
29023 * config/avr/avr.md (bswapsi2): Fix typo in comment.
29024 * config/sh/superh.h: Likewise.
29025 * config/i386/xopintrin.h: Likewise.
29026 * config/i386/znver1.md: Likewise.
29027 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
29028 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
29029 * double-int.h (struct double_int): Likewise.
29030 * double-int.c (div_and_round_double): Likewise.
29031 * wide-int.cc: Likewise.
29032 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
29033 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
29034 * cfgcleanup.c (crossjumps_occured): Renamed to ...
29035 (crossjumps_occurred): ... this.
29036 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
29039 PR tree-optimization/78899
29040 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
29041 returning bool return struct loop *, NULL for failure and the new
29043 (versionable_outer_loop_p): Don't version outer loop if it has
29044 dont_vectorized bit set.
29045 (tree_if_conversion): When versioning outer loop, ensure
29046 tree_if_conversion is performed also on the inner loop of the
29047 non-vectorizable outer loop copy.
29048 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
29049 LOOP_VECTORIZED in inner loop of the scalar outer loop and
29050 prevent vectorization of it.
29051 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
29052 the outer loop vectorization of the non-scalar version is attempted
29053 before vectorization of the inner loop in scalar version. If
29054 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
29055 vectorization of its inner loop.
29056 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
29057 has 2 inner loops, rename also on edges from bb whose single pred
29058 is outer_loop->header. Fix typo in function comment.
29060 2017-01-09 Martin Sebor <msebor@redhat.com>
29063 * asan.c (asan_emit_stack_protection): Increase local buffer size
29064 to avoid snprintf truncation warning.
29066 2017-01-09 Andrew Pinski <apinski@cavium.com>
29068 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
29069 to reference thunderx2t99 for the tuning structure
29070 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
29072 (thunderx2t99_extra_costs): This.
29073 * config/aarch64/aarch64-tune.md: Regenerate.
29074 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
29075 (vulcan_addrcost_table): This.
29076 (vulcan_regmove_cost): Rename to ...
29077 (thunderx2t99_regmove_cost): This.
29078 (vulcan_vector_cost): Rename to ...
29079 (thunderx2t99_vector_cost): this.
29080 (vulcan_branch_cost): Rename to ...
29081 (thunderx2t99_branch_cost): This.
29082 (vulcan_tunings): Rename to ...
29083 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
29084 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
29086 2017-01-09 Martin Jambor <mjambor@suse.cz>
29090 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
29091 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
29092 (propagate_vr_accross_jump_function): Use the above function for all
29093 value range computations for pass-through jump functions and type
29094 converasion from explicit value range values.
29095 (ipcp_propagate_stage): Do not attempt to deduce types of formal
29096 parameters from TYPE_ARG_TYPES.
29097 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
29098 (ipa_write_node_info): Stream type of the actual argument.
29099 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
29101 2017-01-09 Martin Liska <mliska@suse.cz>
29104 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
29105 (lookup_compiler): Do not show error message with have_E.
29107 2017-01-09 Jakub Jelinek <jakub@redhat.com>
29109 PR tree-optimization/78938
29110 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
29111 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
29112 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
29113 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
29116 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29118 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
29121 2017-01-09 Richard Biener <rguenther@suse.de>
29123 PR tree-optimization/78997
29124 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
29125 name condition properly.
29127 2017-01-09 Richard Biener <rguenther@suse.de>
29130 * dwarf2out.c (is_cxx): New overload with context.
29131 (is_naming_typedef_decl): Use it.
29133 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
29135 * invoke.texi (Option Summary): Correct spacing in option lists
29136 and add line breaks to fix over-long lines.
29138 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
29140 PR middle-end/17660
29142 * extend.texi (Common Variable Attributes): Add xref to GCC
29143 Internals manual to explain mode attribute keywords.
29145 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
29148 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
29149 and Preprocessor Options.
29150 (Options for Linking): Document -pthread here....
29151 (RS/6000 and PowerPC Options): ...not here.
29152 (Solaris 2 Options): ...or here.
29153 * doc/cppopts.texi: Document -pthread.
29155 2017-01-08 Martin Sebor <msebor@redhat.com>
29157 PR middle-end/77708
29158 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
29159 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
29160 New member functions.
29161 (format_directive): Used them.
29163 (pass_sprintf_length::handle_gimple_call): Same.
29164 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
29165 to avoid truncation for any argument.
29166 (extract_affine_mul): Same.
29167 * tree.c (get_file_function_name): Same.
29169 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29171 PR middle-end/77484
29172 * predict.def (PRED_INDIR_CALL): Set to 86.
29174 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
29176 PR preprocessor/54124
29177 * doc/cppopts.texi: Reformat -d subtable to list the full name
29178 of the options. Add cross-reference to the docs for the general
29179 compiler -d options.
29180 * doc/invoke.texi (Developer Options): Add cross-reference to the
29181 preprocessor-specific -d option documentation.
29183 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
29185 PR preprocessor/13498
29186 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
29187 redudant material, and reflect new command-line options.
29188 (System Headers): Likewise.
29190 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
29192 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
29193 -isystem, and -idirafter. Copy-edit.
29194 * doc/cppopts.texi: Copy-edit. Remove contradiction about
29195 default for -ftrack-macro-expansion. Delete obsolete and
29196 badly-formatted implementation details about -fdebug-cpp output.
29197 * doc/cppwarnopts.texi: Copy-edit.
29199 2017-01-07 David Malcolm <dmalcolm@redhat.com>
29202 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
29203 that the transition from a max line width >= 1<<10 to narrower
29204 lines works correctly.
29206 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
29208 * doc/options.texi (PerFunction): New.
29209 * opt-functions.awk (switch_flags): Map both Optimization and
29210 PerFunction to CL_OPTIMIZATION.
29211 * opth-gen.awk: Test for PerFunction flag along with
29213 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
29214 it only when the latter is present. Skip those that don't in
29215 the hash function generator.
29216 * common.opt (fvar-tracking): Mark as PerFunction instead of
29218 (fvar-tracking-assignments): Likewise.
29219 (fvar-tracking-assignments-toggle): Likewise.
29220 (fvar-tracking-uninit): Likewise.
29222 2017-01-07 Jakub Jelinek <jakub@redhat.com>
29224 PR translation/79018
29225 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
29228 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
29231 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
29232 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
29233 TARGET_LONG_JUMP_TABLE_OFFSETS.
29234 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
29235 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
29236 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
29237 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
29238 * config/m68k/m68k.md (tablejump expander): Likewise.
29239 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
29240 TARGET_LONG_JUMP_TABLE_OFFSETS.
29241 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
29242 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
29244 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
29245 David Holsgrove <david.holsgrove@xilinx.com>
29247 * common/config/microblaze/microblaze-common.c
29248 (TARGET_EXCEPT_UNWIND_INFO): Remove.
29249 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
29251 * config/microblaze/microblaze.c (microblaze_must_save_register)
29252 (microblaze_expand_epilogue, microblaze_return_addr): Handle
29254 (microblaze_eh_return): New function.
29255 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
29256 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
29257 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
29258 * config/microblaze/microblaze.md (eh_return): New pattern.
29260 2017-01-06 Jakub Jelinek <jakub@redhat.com>
29262 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
29263 GCC_DIAGNOSTIC_STRINGIFY): Define.
29265 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
29267 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29269 * config/arm/arm.md (<mcrr>): New.
29271 * config/arm/arm.c (arm_arch5te): New.
29272 (arm_option_override): Set arm_arch5te.
29273 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
29275 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
29276 (arm_mcrr_qualifiers): ... this. New.
29277 (MRRC_QUALIFIERS): Define to...
29278 (arm_mrrc_qualifiers): ... this. New.
29279 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
29281 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
29282 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
29283 (MRRCI, mrrc, MRRC): New.
29284 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
29285 VUNSPEC_MRRC2): New.
29287 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29289 * config/arm/arm.md (<mcr>): New.
29291 * config/arm/arm.c (arm_coproc_builtin_available): Add
29292 support for mcr, mrc, mcr2 and mrc2.
29293 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
29294 (arm_mcr_qualifiers): ... this. New.
29295 (MRC_QUALIFIERS): Define to ...
29296 (arm_mrc_qualifiers): ... this. New.
29297 (MCR_QUALIFIERS): Define to ...
29298 (arm_mcr_qualifiers): ... this. New.
29299 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
29301 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
29302 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
29303 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
29304 VUNSPEC_MRC2): New.
29306 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29308 * config/arm/arm.md (*ldc): New.
29312 * config/arm/arm.c (arm_coproc_builtin_available): Add
29313 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
29314 (arm_coproc_ldc_stc_legitimate_address): New.
29315 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
29316 'qualifier_const_pointer'.
29317 (LDC_QUALIFIERS): Define to...
29318 (arm_ldc_qualifiers): ... this. New.
29319 (STC_QUALIFIERS): Define to...
29320 (arm_stc_qualifiers): ... this. New.
29321 * config/arm/arm-protos.h
29322 (arm_coproc_ldc_stc_legitimate_address): New.
29323 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
29324 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
29325 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
29326 stc2, stcl, stc2l): New.
29327 * config/arm/constraints.md (Uz): New.
29328 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
29329 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
29330 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
29331 VUNSPEC_STC2L): New.
29333 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29335 * config/arm/arm.md (<cdp>): New.
29336 * config/arm/arm.c (neon_const_bounds): Rename this ...
29337 (arm_const_bounds): ... this.
29338 (arm_coproc_builtin_available): New.
29339 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
29340 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
29341 (CDP_QUALIFIERS): Define to...
29342 (arm_cdp_qualifiers): ... this. New.
29344 (arm_expand_builtin_args): Add case for 6 arguments.
29345 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
29346 (arm_const_bounds): ... this.
29347 (arm_coproc_builtin_available): New.
29348 * config/arm/arm_acle.h (__arm_cdp): New.
29350 * config/arm/arm_acle_builtins.def (cdp): New.
29352 * config/arm/iterators.md (CDPI,CDP,cdp): New.
29353 * config/arm/neon.md: Rename all 'neon_const_bounds' to
29354 'arm_const_bounds'.
29355 * config/arm/types.md (coproc): New.
29356 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
29357 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
29358 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
29359 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
29361 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29363 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
29364 (UBINOP_QUALIFIERS): New.
29366 (acle_builtin_data): New. Change comment.
29367 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
29368 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
29369 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
29370 arm_acle_builtins.def.
29371 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
29372 (arm_init_acle_builtins): New.
29373 (CRC32_BUILTIN): Remove.
29374 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
29375 crc32cb, crc32ch and crc32cw.
29376 (arm_init_crc32_builtins): Remove.
29377 (arm_init_builtins): Use arm_init_acle_builtins rather
29378 than arm_init_crc32_builtins.
29379 (arm_expand_acle_builtin): New.
29380 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
29381 * config/arm/arm_acle_builtins.def: New.
29383 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29385 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
29386 (arm_builtin_datum): ... this.
29387 (arm_init_neon_builtin): Rename to ...
29388 (arm_init_builtin): ... this. Add a new parameters PREFIX
29389 and USE_SIG_IN_NAME.
29390 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
29391 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
29392 'arm_builtin_datum'.
29393 (arm_init_vfp_builtins): Likewise.
29394 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
29395 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
29396 (arm_expand_neon_args): Rename to ...
29397 (arm_expand_builtin_args): ... this. Rename builtin_arg
29398 enum values and differentiate between ARG_BUILTIN_MEMORY
29399 and ARG_BUILTIN_NEON_MEMORY.
29400 (arm_expand_neon_builtin_1): Rename to ...
29401 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
29402 values, arm_expand_builtin_args and add bool parameter NEON.
29403 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
29404 (arm_expand_vfp_builtin): Likewise.
29405 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
29407 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29409 PR middle-end/77484
29410 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
29411 * predict.c (tree_estimate_probability_bb): Reverse direction of
29412 polymorphic call predictor.
29414 2017-01-06 David Malcolm <dmalcolm@redhat.com>
29416 * passes.c (execute_one_pass): Split out pass-skipping logic into...
29417 (determine_pass_name_match): ...this new function and...
29418 (should_skip_pass_p): ...this new function.
29420 2017-01-06 Nathan Sidwell <nathan@acm.org>
29422 * ipa-visibility.c (function_and_variable_visibility): Reformat
29423 comments and long lines. Remove extrneous if.
29424 * symtab.c (symtab_node::make_decl_local): Fix code format.
29425 (symtab_node::set_section_for_node): Fix comment typo.
29427 2017-01-06 Martin Liska <mliska@suse.cz>
29430 * lra-constraints.c: Rename invariant to lra_invariant.
29431 * predict.c (set_even_probabilities): Initialize e to NULL.
29433 2017-01-05 Martin Sebor <msebor@redhat.com>
29435 PR tree-optimization/78910
29436 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
29437 (format_integer): Correct off-by-one error in the handling
29438 of precision with negative numbers in signed conversions..
29440 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
29442 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
29444 2017-01-05 Jakub Jelinek <jakub@redhat.com>
29446 PR tree-optimization/71016
29447 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
29448 factor_out_conditional_conversion. Formatting fix.
29449 (factor_out_conditional_conversion): Add cond_stmt argument.
29450 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
29451 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
29454 2017-01-05 David Malcolm <dmalcolm@redhat.com>
29456 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
29457 read-rtl-function.o, and selftest-rtl.o.
29458 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
29459 (selftest::aarch64_test_loading_full_dump): New function.
29460 (selftest::aarch64_run_selftests): New function.
29461 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
29462 selftest::aarch64_run_selftests.
29463 * config/i386/i386.c
29464 (selftest::ix86_test_loading_dump_fragment_1): New function.
29465 (selftest::ix86_test_loading_call_insn): New function.
29466 (selftest::ix86_test_loading_full_dump): New function.
29467 (selftest::ix86_test_loading_unspec): New function.
29468 (selftest::ix86_run_selftests): Call the new functions.
29469 * emit-rtl.c (maybe_set_max_label_num): New function.
29470 * emit-rtl.h (maybe_set_max_label_num): New decl.
29471 * function.c (instantiate_decls): Guard call to
29472 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
29473 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
29475 * gensupport.c (gen_reader::gen_reader): Pass "false"
29476 for new "compact" param of rtx_reader.
29477 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
29478 rather than an empty string for NULL strings.
29479 * read-md.c: Potentially include config.h rather than bconfig.h.
29480 Wrap include of errors.h with #ifdef GENERATOR_FILE.
29481 (have_error): New global, copied from errors.c.
29482 (md_reader::read_name): Rename to...
29483 (md_reader::read_name_1): ...this, adding "out_loc" param,
29484 and converting "missing name or number" to returning false, rather
29486 (md_reader::read_name): Reimplement in terms of read_name_1.
29487 (md_reader::read_name_or_nil): New function.
29488 (md_reader::read_string): Handle "(nil)" by returning NULL.
29489 (md_reader::md_reader): Add new param "compact".
29490 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
29491 (md_reader::read_file): New method.
29492 * read-md.h (md_reader::md_reader): Add new param "compact".
29493 (md_reader::read_file): New method.
29494 (md_reader::is_compact): New accessor.
29495 (md_reader::read_name): Convert return type from void to file_location.
29496 (md_reader::read_name_or_nil): New decl.
29497 (md_reader::read_name_1): New decl.
29498 (md_reader::m_compact): New field.
29499 (noop_reader::noop_reader): Pass "false" for new "compact" param
29501 (rtx_reader::rtx_reader): Add new "compact" param.
29502 (rtx_reader::read_rtx_operand): Make virtual and convert return
29503 type from void to rtx.
29504 (rtx_reader::read_until): New decl.
29505 (rtx_reader::handle_any_trailing_information): New virtual function.
29506 (rtx_reader::postprocess): New virtual function.
29507 (rtx_reader::finalize_string): New virtual function.
29508 (rtx_reader::m_in_call_function_usage): New field.
29509 (rtx_reader::m_reuse_rtx_by_id): New field.
29510 * read-rtl-function.c: New file.
29511 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
29512 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
29513 (selftest::verify_three_block_rtl_cfg): New decl.
29514 * read-rtl-function.h: New file.
29515 * read-rtl.c: Potentially include config.h rather than bconfig.h.
29516 For host, include function.h, memmodel.h, and emit-rtl.h.
29517 (one_time_initialization): New function.
29518 (struct compact_insn_name): New struct.
29519 (compact_insn_names): New array.
29520 (find_code): Handle insn codes in compact dumps.
29521 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
29522 (bind_subst_iter_and_attr): Likewise.
29523 (add_condition_to_string): Likewise.
29524 (add_condition_to_rtx): Likewise.
29525 (apply_attribute_uses): Likewise.
29526 (add_current_iterators): Likewise.
29527 (apply_iterators): Likewise.
29528 (initialize_iterators): Guard usage of apply_subst_iterator with
29529 #ifdef GENERATOR_FILE.
29530 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
29531 (md_reader::read_mapping): Likewise.
29532 (add_define_attr_for_define_subst): Likewise.
29533 (add_define_subst_attr): Likewise.
29534 (read_subst_mapping): Likewise.
29535 (check_code_iterator): Likewise.
29536 (rtx_reader::read_rtx): Likewise. Move one-time initialization
29538 (one_time_initialization): New function.
29539 (rtx_reader::read_until): New method.
29540 (read_flags): New function.
29541 (parse_reg_note_name): New function.
29542 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
29543 Handle reuse_rtx ids.
29544 Wrap iterator lookup within #ifdef GENERATOR_FILE.
29545 Add parsing support for RTL dumps, mirroring the special-cases in
29546 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
29547 values, and calling handle_any_trailing_information.
29548 (rtx_reader::read_rtx_operand): Convert return type from void
29549 to rtx, returning return_rtx. Handle case 'e'. Call
29550 finalize_string on XSTR and XTMPL fields.
29551 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
29552 "(nil)" values were omitted. Call the postprocess vfunc on the
29554 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
29555 class ctor. Initialize m_in_call_function_usage. Call
29556 one_time_initialization.
29557 * rtl-tests.c (selftest::test_uncond_jump): Call
29558 set_new_first_and_last_insn.
29559 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
29560 * selftest-rtl.c: New file.
29561 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
29562 (selftest::get_insn_by_uid): New decl.
29563 * selftest-run-tests.c (selftest::run_tests): Call
29564 read_rtl_function_c_tests.
29565 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
29566 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
29569 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
29571 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
29572 operands in a special way. Assert that pos+len <= mode precision.
29574 2017-01-05 Jakub Jelinek <jakub@redhat.com>
29576 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
29577 3 argument Alias with unlimited for the negative form.
29578 (fno-vect-cost-model): Removed.
29580 2017-01-05 Martin Liska <mliska@suse.cz>
29582 * hsa-gen.c (gen_hsa_divmod): New function.
29583 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
29585 2017-01-05 Martin Liska <mliska@suse.cz>
29588 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
29591 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29593 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
29594 small constant length operands.
29596 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29598 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
29599 between loop iterations.
29601 2017-01-05 Martin Liska <mliska@suse.cz>
29604 * gimplify.c (gimplify_decl_expr): Compare to
29605 asan_poisoned_variables instread of checking flags.
29606 (gimplify_target_expr): Likewise.
29607 (gimplify_expr): Likewise.
29608 (gimplify_function_tree): Conditionally initialize
29609 asan_poisoned_variables.
29611 2017-01-04 Jeff Law <law@redhat.com>
29613 PR tree-optimizatin/78812
29614 * rtl.h (contains_mem_rtx_p): Prototype.
29615 * ifcvt.c (containts_mem_rtx_p): Move from here to...
29616 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
29617 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
29618 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
29619 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
29621 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29623 * input.c (assert_char_at_range): Default-initialize actual_range.
29625 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29627 * df-scan.c (df_ref_create_structure): Make regno unsigned,
29628 to match the caller.
29630 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29632 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
29633 insns after final jump in test to emit dummy move.
29635 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29637 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
29638 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
29640 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29642 * multiple_target.c (create_dispatcher_calls): Init e_next.
29643 * tree-ssa-loop-split.c (split_loop): Init border.
29644 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
29647 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
29652 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
29653 (altivec_register_operand): Do not return true if the operand
29654 contains a SUBREG mixing SImode and SFmode.
29655 (vsx_register_operand): Likewise.
29656 (vsx_reg_sfsubreg_ok): New predicate.
29657 (vfloat_operand): Do not return true if the operand contains a
29658 SUBREG mixing SImode and SFmode.
29659 (vint_operand): Likewise.
29660 (vlogical_operand): Likewise.
29661 (gpc_reg_operand): Likewise.
29662 (int_reg_operand): Likewise.
29663 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
29664 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
29665 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
29667 (rs6000_emit_move_si_sf_subreg): New helper function.
29668 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
29669 fixup SUBREGs involving SImode and SFmode.
29670 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
29671 numbers for the new peephole2 optimization.
29672 (peephole2 for SFmode unions): New peephole2 to optimize cases in
29673 the GLIBC math library that do AND/IOR/XOR operations on single
29674 precision floating point.
29675 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
29676 target macros to say whether we need to avoid SUBREGs mixing
29678 (TARGET_ALLOW_SF_SUBREG): Likewise.
29679 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
29680 (UNSPEC_SI_FROM_SF): Likewise.
29681 (iorxor): Change spacing.
29682 (and_ior_xor): New iterator for AND, IOR, and XOR.
29683 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
29684 (movdi_from_sf_zero_ext): Likewise.
29685 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
29686 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
29687 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
29688 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
29689 (fms<mode>4): Likewise.
29690 (fnma<mode>4): Likewise.
29691 (fnms<mode>4): Likewise.
29692 (nfma<mode>4): Likewise.
29693 (nfms<mode>4): Likewise.
29695 2017-01-04 Marek Polacek <polacek@redhat.com>
29698 * doc/invoke.texi: Document -Wpointer-compare.
29700 2017-01-04 Jakub Jelinek <jakub@redhat.com>
29702 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
29705 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
29706 descriptions for -gdwarf-5 and emit them as uleb128 instead of
29709 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
29712 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
29713 documentation of the powerpc_popcntb_ok attribute.
29714 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29715 code to issue warning messages if a requested CPU configuration is
29716 not supported by the binary (assembler and loader) toolchain.
29717 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
29718 made to define a built-in function that has been disabled.
29719 (paired_init_builtins): Add assertion to prevent ICE if attempt is
29720 made to define a built-in function that has been disabled.
29721 (altivec_init_builtins): Add comment explaining why definition
29722 of the DST built-in functions is not preceded by an assertion
29723 check. Add assertions to prevent ICE if attempts are made to
29724 define an altivec predicate or an abs* built-in function that has
29726 (htm_init_builtins): Add comment explaining why definition of the
29727 htm built-in functions is not preceded by an assertion check.
29729 2017-01-04 Jeff Law <law@redhat.com>
29731 PR tree-optimizatin/67955
29732 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
29733 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
29734 the points-to solution does not include pt_null. Use DECL_PT_UID
29737 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
29739 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
29740 Use gen_int_mode instead of gen_lopwart for const_int operands.
29742 2017-01-04 Jakub Jelinek <jakub@redhat.com>
29744 PR tree-optimization/71563
29745 * match.pd: Simplify X << Y into X if Y is known to be 0 or
29746 out of range value - has low bits known to be zero.
29748 2017-01-04 Alan Modra <amodra@gmail.com>
29750 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
29751 * configure: Regenerate.
29752 * config.in: Regenerate.
29754 2017-01-04 Jakub Jelinek <jakub@redhat.com>
29757 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
29758 a substring of the message, but strcmp with the whole message. Ifdef
29759 ENABLE_NLS, translate the message first using dgettext.
29761 2017-01-03 Jeff Law <law@redhat.com>
29763 PR tree-optimizatin/78856
29764 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
29765 (mark_threaded_blocks): Remove code to truncate thread paths that
29766 cross multiple loop headers. Instead invalidate the cached loop
29767 iteration information and handle case of a thread path walking
29768 into an irreducible region.
29770 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
29773 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
29774 assertions. Add support for doing the signbit if the IEEE 128-bit
29775 floating point value is in a GPR.
29776 * config/rs6000/rs6000.md (Fsignbit): Delete.
29777 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
29778 Update the length attribute if the value is in a GPR.
29779 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
29780 the sign or zero extension instruction, since the value is always 0/1.
29781 (signbit<mode>2_dm2): Delete using <Fsignbit>.
29784 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
29785 extracting SImode to a GPR register so that we can generate a
29786 store, limit the vector to be in a traditional Altivec register
29787 for the vextuwrx instruction.
29789 2017-01-03 Ian Lance Taylor <iant@google.com>
29791 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
29793 2017-01-03 Martin Sebor <msebor@redhat.com>
29795 PR tree-optimization/78696
29796 * gimple-ssa-sprintf.c (format_floating): Correct handling of
29797 precision. Use MPFR for %f for greater fidelity. Correct handling
29799 (pass_sprintf_length::compute_format_length): Set width and precision
29800 specified by asrerisk to void_node for vararg functions.
29801 (try_substitute_return_value): Adjust dump output.
29803 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
29805 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
29807 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
29809 * doc/invoke.texi (SPARC options): Document -mlra as the default.
29810 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
29811 -mlra/-mno-lra was passed to the compiler.
29813 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
29815 PR rtl-optimization/65618
29816 * emit-rtl.c (try_split): Move initialization of "before" and
29817 "after" to just before the call to emit_insn_after_setloc.
29819 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
29821 * doc/md.texi (Standard Names): Remove reference to Java frontend.
29823 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
29825 * dwarf2out.c (gen_enumeration_type_die): When
29826 -gno-strict-dwarf, add a DW_AT_encoding attribute.
29828 2017-01-03 Jakub Jelinek <jakub@redhat.com>
29830 PR tree-optimization/78965
29831 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
29832 Change first argument from const call_info & to call_info &. For %n
29833 set info.nowrite to false.
29835 PR middle-end/78901
29836 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
29837 possibly throwing calls.
29839 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
29840 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
29841 and fns handling, rather than in a separate case SSA_NAME.
29843 2017-01-02 Jeff Law <law@redhat.com>
29845 * config/darwin-driver.c (darwin_driver_init): Const-correctness
29846 fixes for first_period and second_period variables.
29848 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
29851 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
29852 (*insvqi_1): New insn pattern.
29853 (*insvqi_1_mem_rex64): Ditto.
29854 (*insvqi_2): Ditto.
29855 (*insvqi_3): Rename from *insvqi.
29857 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
29859 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
29861 * doc/cfg.texi (Edges): Remove reference to Java.
29862 (Maintaining the CFG): Ditto.
29864 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29866 PR middle-end/77674
29867 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
29868 transparent aliases.
29870 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29872 PR middle-end/77484
29873 * predict.def (PRED_CALL): Update hitrate.
29874 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
29875 * predict.c (tree_estimate_probability_bb): Split CALL predictor
29876 into direct/indirect/polymorphic variants.
29878 2017-01-01 Jakub Jelinek <jakub@redhat.com>
29880 Update copyright years.
29882 * gcc.c (process_command): Update copyright notice dates.
29883 * gcov-dump.c (print_version): Ditto.
29884 * gcov.c (print_version): Ditto.
29885 * gcov-tool.c (print_version): Ditto.
29886 * gengtype.c (create_file): Ditto.
29887 * doc/cpp.texi: Bump @copying's copyright year.
29888 * doc/cppinternals.texi: Ditto.
29889 * doc/gcc.texi: Ditto.
29890 * doc/gccint.texi: Ditto.
29891 * doc/gcov.texi: Ditto.
29892 * doc/install.texi: Ditto.
29893 * doc/invoke.texi: Ditto.
29895 Copyright (C) 2017 Free Software Foundation, Inc.
29897 Copying and distribution of this file, with or without modification,
29898 are permitted in any medium without royalty provided the copyright
29899 notice and this notice are preserved.