1 2017-10-12 Bin Cheng <bin.cheng@arm.com>
3 * tree-loop-distribution.c (struct builtin_info): New struct.
4 (struct partition): Refactor fields into struct builtin_info.
5 (partition_free): Free struct builtin_info.
6 (build_size_arg_loc, build_addr_arg_loc): Delete.
7 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
8 information from struct builtin_info.
9 (find_single_drs): New function refactored from classify_partition.
10 Also moved builtin validity checks to this function.
11 (compute_access_range, alloc_builtin): New functions.
12 (classify_builtin_st, classify_builtin_ldst): New functions.
13 (classify_partition): Refactor code into functions find_single_drs,
14 classify_builtin_st and classify_builtin_ldst.
15 (distribute_loop): Don't do runtime alias check when distributing
17 (find_seed_stmts_for_distribution): New function.
18 (pass_loop_distribution::execute): Refactor code finding seed
19 stmts into above function. Support distribution for the innermost
20 two-level loop nest. Adjust dump information.
22 2017-10-12 Bin Cheng <bin.cheng@arm.com>
24 * tree-loop-distribution.c: Adjust the general comment.
25 (NUM_PARTITION_THRESHOLD): New macro.
26 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
27 (classify_partition): Skip builtin pattern of loop nest's inner loop.
28 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
29 in call to build_partition_graph.
30 (finalize_partitions): New parameter. Make loop distribution more
31 conservative by fusing more partitions.
32 (distribute_loop): Don't do runtime alias check in case of loop nest
34 (find_seed_stmts_for_distribution): New function.
35 (prepare_perfect_loop_nest): New function.
36 (pass_loop_distribution::execute): Refactor code finding seed stmts
37 and loop nest into above functions. Support loop nest distribution.
38 Adjust dump information accordingly.
40 2017-10-12 Bin Cheng <bin.cheng@arm.com>
42 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
43 and set PTYPE_SEQUENTIAL for merged partition.
45 2017-10-12 Richard Biener <rguenther@suse.de>
47 PR tree-optimization/69728
49 2017-09-19 Richard Biener <rguenther@suse.de>
51 PR tree-optimization/69728
52 * graphite-sese-to-poly.c (schedule_error): New global.
53 (add_loop_schedule): Handle empty domain by failing the
55 (build_original_schedule): Handle schedule_error.
57 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
58 domain by returning an unchanged schedule.
60 2017-10-12 Jakub Jelinek <jakub@redhat.com>
62 * genrecog.c (validate_pattern): For VEC_SELECT verify that
63 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
65 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
67 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
70 2017-10-12 Jakub Jelinek <jakub@redhat.com>
73 * expr.c (store_field): Don't optimize away bitsize == 0 store
74 from CALL_EXPR with addressable return type.
76 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
78 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
79 * config/rs6000/rs6000.md (sel): Delete mode attribute.
80 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
81 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
82 TARGET_ISEL instead of TARGET_ISEL<sel>.
84 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
86 * config/rs6000/rs6000.c
87 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
89 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
91 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
93 (reg_or_cint_operand): Fix comment.
94 (reg_or_zero_operand): New predicate.
95 * config/rs6000/rs6000-protos.h (output_isel): Delete.
96 * config/rs6000/rs6000.c (output_isel): Delete.
97 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
98 instead of reg_or_cint_operand. Output instruction directly (not via
100 (isel_unsigned_<mode>): Ditto.
101 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
102 gpc_reg_operand. Add an instruction alternative for this. Output
103 instruction directly.
104 (*isel_reversed_unsigned_<mode>): Ditto.
106 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
108 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
109 (TARGET_CANONICALIZE_COMPARISON): Define.
111 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
114 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
115 Check whether the dest is REG before adding REG_EQUIV note.
117 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
120 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
122 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
123 (make_hard_regno_born, make_hard_regno_dead): Update
124 bb_killed_pseudos and bb_gen_pseudos.
126 2017-10-11 Nathan Sidwell <nathan@acm.org>
128 * incpath.h (enum incpath_kind): Name enum, prefix values.
129 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
130 * incpath.c (heads, tails): Use INC_MAX.
131 (add_env_var_paths, add_standard_paths): Use incpath_kind.
132 (merge_include_chains, split_quote_chain,
133 register_include_chains): Update incpath_kind names.
134 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
135 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
137 (add_framework_path, darwin_register_objc_includes): Likewise.
138 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
140 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
142 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
143 Do not use float_operator operator predicate.
144 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
145 * config/i386/predicates.md (float_operator): Remove predicate.
147 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
149 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
150 (*jccxf_i387): Ditto.
151 (*jcc<mode>_i387): Ditto.
152 (*jccu<mode>_i387): Ditto.
153 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
154 (*jcc_*_i387 splitters): Remove.
155 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
156 * config/i386/i386.c (ix86_split_fp_branch): Remove.
157 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
160 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
162 * profile-count.h (slow_safe_scale_64bit): New function.
163 (safe_scale_64bit): New inline.
164 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
165 * profile-count.c: Include wide-int.h
166 (slow_safe_scale_64bit): New.
168 2017-10-11 Nathan Sidwell <nathan@acm.org>
170 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
171 HAS_DECL_ASSEMBLER_NAME_P.
172 * gimple-expr.c (gimple_decl_printable_name: Check
173 HAS_DECL_ASSEMBLER_NAME_P too.
174 * ipa-utils.h (type_in_anonymous_namespace_p): Check
175 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
176 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
177 * passes.c (rest_of_decl_compilation): Check
178 HAS_DECL_ASSEMBLER_NAME_P too.
179 * recog.c (verify_changes): Likewise.
180 * tree-pretty-print.c (dump_decl_name): Likewise.
181 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
183 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
184 (SET_DECL_ASSEMBLER_NAME): Use it.
185 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
186 (COPY_DECL_ASSEMBLER_NAME): Likewise.
187 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
189 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
191 * config.gcc (i386, x86_64): Add extra objects.
192 * i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
193 (ix86_min_insn_size): Declare.
194 (ix86_issue_rate): Declare.
195 (ix86_adjust_cost): Declare.
196 (ia32_multipass_dfa_lookahead): Declare.
197 (ix86_macro_fusion_p): Declare.
198 (ix86_macro_fusion_pair_p): Declare.
199 (ix86_bd_has_dispatch): Declare.
200 (ix86_bd_do_dispatch): Declare.
201 (ix86_core2i7_init_hooks): Declare.
202 (ix86_atom_sched_reorder): Declare.
203 * i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
204 (COSTS_N_BYTES): Move to x86-tune-costs.h.
205 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
206 (rip_relative_addr_p): Rename to ...
207 (ix86_rip_relative_addr_p): ... this one; export.
208 (memory_address_length): Update.
209 (ix86_issue_rate): Move to x86-tune-sched.c.
210 (ix86_flags_dependent): Move to x86-tune-sched.c.
211 (ix86_agi_dependent): Move to x86-tune-sched.c.
212 (exact_dependency_1): Move to x86-tune-sched.c.
213 (exact_store_load_dependency): Move to x86-tune-sched.c.
214 (ix86_adjust_cost): Move to x86-tune-sched.c.
215 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
216 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
217 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
218 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
219 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
220 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
221 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
222 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
223 (min_insn_size): Rename to ...
224 (ix86_min_insn_size): ... this one; export.
225 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
226 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
227 (core2i7_first_cycle_multipass_backtrack): Move to
228 x86-tune-sched-core.c.
229 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
230 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
231 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
232 (ix86_avoid_jump_mispredicts): Update.
233 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
234 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
235 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
236 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
237 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
238 (MAX_INSN): Move to ix86-tune-sched-bd.c.
239 (MAX_IMM): Move to ix86-tune-sched-bd.c.
240 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
241 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
242 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
243 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
244 (MAX_STORE): Move to ix86-tune-sched-bd.c.
245 (BIG): Move to ix86-tune-sched-bd.c.
246 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
247 (enum insn_path): Move to ix86-tune-sched-bd.c.
248 (get_mem_group): Move to ix86-tune-sched-bd.c.
249 (is_cmp): Move to ix86-tune-sched-bd.c.
250 (dispatch_violation): Move to ix86-tune-sched-bd.c.
251 (is_branch): Move to ix86-tune-sched-bd.c.
252 (is_prefetch): Move to ix86-tune-sched-bd.c.
253 (init_window): Move to ix86-tune-sched-bd.c.
254 (allocate_window): Move to ix86-tune-sched-bd.c.
255 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
256 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
257 (process_end_window): Move to ix86-tune-sched-bd.c.
258 (allocate_next_window): Move to ix86-tune-sched-bd.c.
259 (find_constant): Move to ix86-tune-sched-bd.c.
260 (get_num_immediates): Move to ix86-tune-sched-bd.c.
261 (has_immediate): Move to ix86-tune-sched-bd.c.
262 (get_insn_path): Move to ix86-tune-sched-bd.c.
263 (get_insn_group): Move to ix86-tune-sched-bd.c.
264 (count_num_restricted): Move to ix86-tune-sched-bd.c.
265 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
266 (add_insn_window): Move to ix86-tune-sched-bd.c.
267 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
268 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
269 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
270 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
271 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
272 (do_dispatch): Move to ix86-tune-sched-bd.c.
273 (has_dispatch): Move to ix86-tune-sched-bd.c.
274 * i386/t-i386: Add new object files.
275 * i386/x86-tune-costs.h: New file.
276 * i386/x86-tune-sched-atom.c: New file.
277 * i386/x86-tune-sched-bd.c: New file.
278 * i386/x86-tune-sched-core.c: New file.
279 * i386/x86-tune-sched.c: New file.
281 2017-10-11 Liu Hao <lh_mouse@126.com>
283 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
284 the generic version below instead.
285 (should_colorize): Recognize Windows consoles as terminals
287 * pretty-print.c [__MINGW32__] (write_all): New function.
288 [__MINGW32__] (find_esc_head): Likewise.
289 [__MINGW32__] (find_esc_terminator): Likewise.
290 [__MINGW32__] (eat_esc_sequence): Likewise.
291 [__MINGW32__] (mingw_ansi_fputs): New function that handles
293 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
296 2017-10-11 Richard Biener <rguenther@suse.de>
298 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
299 Properly call analyze_scalar_evolution with the loop of the stmt.
301 2017-10-11 Richard Biener <rguenther@suse.de>
303 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
304 * tree-core.h (tree_base): Add chrec_var union member.
305 * tree.h (CHREC_VAR): Remove.
306 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
307 * tree-chrec.h (build_polynomial_chrec): Adjust.
308 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
309 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
311 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
313 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
314 * match.pd: ... here.
315 ((T) X == (T) Y): Relax condition.
317 2017-10-11 Bin Cheng <bin.cheng@arm.com>
319 PR tree-optimization/82472
320 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
322 (break_alias_scc_partitions): Update postorder number.
324 2017-10-11 Martin Liska <mliska@suse.cz>
327 * opts.c (parse_no_sanitize_attribute): Do not use error_value
329 * opts.h (parse_no_sanitize_attribute): Remove last argument.
331 2017-10-11 Martin Liska <mliska@suse.cz>
333 * print-rtl.c (print_insn): Move declaration of idbuf
334 to same scope as name.
336 2017-10-11 Martin Liska <mliska@suse.cz>
341 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
344 2017-10-11 Martin Liska <mliska@suse.cz>
347 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
350 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
352 * config.gcc (default_gnu_indirect_function): Default to yes for
353 arm*-*-linux* with glibc.
355 2017-10-11 Richard Biener <rguenther@suse.de>
357 * tree-scalar-evolution.c (get_scalar_evolution): Handle
358 default-defs and types we do not want to analyze.
359 (interpret_loop_phi): Replace unreachable code with an assert.
360 (compute_scalar_evolution_in_loop): Remove and inline ...
361 (analyze_scalar_evolution_1): ... here, replacing condition with
362 what makes the intent clearer. Remove handling of cases
363 get_scalar_evolution now handles.
365 2017-10-10 Jim Wilson <wilson@tuliptree.org>
367 PR rtl-optimization/81434
368 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
369 comment for main loop. In sched_group_found if, also add checks for
370 pass and min_cost_group.
372 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
374 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
375 (rs6000_insn_cost): New function.
376 * config/rs6000/rs6000.md (cost): New attribute.
378 2017-10-10 Jakub Jelinek <jakub@redhat.com>
379 H.J. Lu <hongjiu.lu@intel.com>
383 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
384 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
385 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
386 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
387 to OPTION_MASK_ISA_AVX512VL - builtins that have both
388 OPTION_MASK_ISA_MMX and some other bit set require both
389 mmx and the ISAs without the mmx bit.
390 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
391 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
392 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
393 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
394 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
395 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
396 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
397 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
398 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
399 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
400 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
401 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
402 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
403 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
404 Add OPTION_MASK_ISA_MMX.
406 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
408 * config.gcc (armv7*-*-freebsd*): New target.
409 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
411 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
413 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
414 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
417 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
419 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
420 a template parameter.
421 (WIDE_INT_REF_FOR): Update accordingly.
422 * tree.h (wi::int_traits <const_tree>): Delete.
423 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
424 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
425 (wi::tree_to_wide_ref): New typedef.
426 (wi::to_wide): New function.
427 * calls.c (get_size_range): Use wi::to_wide when operating on
429 * cgraph.c (cgraph_node::create_thunk): Likewise.
430 * config/i386/i386.c (ix86_data_alignment): Likewise.
431 (ix86_local_alignment): Likewise.
432 * dbxout.c (stabstr_O): Likewise.
433 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
434 * expr.c (const_vector_from_tree): Likewise.
435 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
436 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
437 (fold_negate_expr_1, int_const_binop_1, const_binop)
438 (fold_convert_const_int_from_real, optimize_bit_field_compare)
439 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
440 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
441 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
442 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
443 (fold_not_const, round_up_loc): Likewise.
444 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
445 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
446 (alloca_call_type): Likewise.
447 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
448 * godump.c (go_output_typedef): Likewise.
449 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
450 * internal-fn.c (get_min_precision): Likewise.
451 * ipa-cp.c (ipcp_store_vr_results): Likewise.
452 * ipa-polymorphic-call.c
453 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
454 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
455 (ipa_modify_call_arguments): Likewise.
456 * match.pd: Likewise.
457 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
458 * print-tree.c (print_node_brief, print_node): Likewise.
459 * stmt.c (expand_case): Likewise.
460 * stor-layout.c (layout_type): Likewise.
461 * tree-affine.c (tree_to_aff_combination): Likewise.
462 * tree-cfg.c (group_case_labels_stmt): Likewise.
463 * tree-data-ref.c (dr_analyze_indices): Likewise.
464 (prune_runtime_alias_test_list): Likewise.
465 * tree-dump.c (dequeue_and_dump): Likewise.
466 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
467 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
468 * tree-pretty-print.c (dump_generic_node): Likewise.
469 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
470 (simple_iv_with_niters): Likewise.
471 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
472 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
473 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
474 * tree-ssa-loop-niter.c (split_to_var_and_offset)
475 (refine_value_range_using_guard, number_of_iterations_ne_max)
476 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
477 (get_cst_init_from_scev, record_nonwrapping_iv)
478 (scev_var_range_cant_overflow): Likewise.
479 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
480 * tree-ssa-pre.c (compute_avail): Likewise.
481 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
482 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
483 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
484 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
485 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
486 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
487 (dump_case_nodes, try_switch_expansion): Likewise.
488 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
489 (vect_do_peeling): Likewise.
490 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
491 * tree-vect-stmts.c (vectorizable_load): Likewise.
492 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
493 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
494 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
495 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
496 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
497 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
498 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
499 (evrp_dom_walker::before_dom_children): Likewise.
500 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
501 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
502 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
503 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
504 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
505 (get_range_pos_neg): Likewise.
506 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
507 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
508 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
509 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
510 * config/avr/avr.c (avr_fold_builtin): Likewise.
511 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
512 * config/msp430/msp430.c (msp430_attr): Likewise.
513 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
514 * config/powerpcspe/powerpcspe-c.c
515 (altivec_resolve_overloaded_builtin): Likewise.
516 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
517 (rs6000_expand_ternop_builtin): Likewise.
518 * config/rs6000/rs6000-c.c
519 (altivec_resolve_overloaded_builtin): Likewise.
520 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
521 (rs6000_expand_ternop_builtin): Likewise.
522 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
524 2017-10-10 Bin Cheng <bin.cheng@arm.com>
526 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
527 when copying loop nest with only one inner loop.
529 2017-10-10 Richard Biener <rguenther@suse.de>
531 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
532 blocks if SCEV is active.
533 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
535 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
536 (scev_initialize): Assert we are not yet initialized.
538 2017-10-10 Bin Cheng <bin.cheng@arm.com>
540 * tree-loop-distribution.c (generate_loops_for_partition): Remove
541 inner loop's exit stmt by making it always exit the loop, otherwise
542 we would generate an infinite empty loop.
544 2017-10-10 Bin Cheng <bin.cheng@arm.com>
546 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
547 renaming variables in new preheader if it's deleted.
549 2017-10-10 Bin Cheng <bin.cheng@arm.com>
551 * tree-loop-distribution.c (struct partition): Remove unused field
552 loops of the structure.
553 (partition_alloc, partition_free): Ditto.
554 (build_rdg_partition_for_vertex): Ditto.
556 2017-10-09 Jeff Law <law@redhat.com>
558 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
559 return type to match prototype and documentation.
561 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
563 * config/rs6000/rs6000.c (processor_costs): Move to ...
564 * config/rs6000/rs6000.h: ... here.
565 (rs6000_cost): Declare.
567 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
569 * except.c (setjmp_fn): New global variable.
570 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
571 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
572 if DONT_USE_BUILTIN_SETJMP is defined.
574 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
576 * target.def (insn_cost): New hook.
577 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
578 * doc/tm.texi: Regenerate.
579 * rtlanal.c (insn_cost): Use the new hook.
581 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
583 * combine.c (combine_validate_cost): Compute the new insn_cost,
584 not just pattern_cost.
585 (try_combine): Adjust comment.
587 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
589 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
591 * combine.c (uid_insn_cost): Adjust comment.
592 (combine_validate_cost): Adjust comment. Use pattern_cost instead
594 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
595 * dse.c (find_shift_sequence): Ditto.
596 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
597 (bb_valid_for_noce_process_p): Use pattern_cost.
598 * rtl.h (insn_rtx_cost): Delete.
599 (pattern_cost): New prototype.
600 (insn_cost): New prototype.
601 * rtlanal.c (insn_rtx_cost): Rename to...
602 (pattern_cost): ... this.
605 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
607 * config/i386/i386.md (*jcc_2): Remove insn pattern.
608 (*jcc<mode>_0_r_i387): Ditto.
609 (*jccxf_r_i387): Ditto.
610 (*jcc<mode>_r_i387): Ditto.
611 (*jccu<mode>_r_i387): Ditto.
612 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
613 (*jcc): Rename from *jcc_1.
615 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
617 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
618 deferred rescans after the lvx/stvx recombination pre-pass.
620 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
622 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
623 memory operation instruction support.
624 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
625 (rs6000-ibm-aix[789]*): Likewise.
626 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
627 Document new functions.
629 2017-10-09 Richard Biener <rguenther@suse.de>
631 PR tree-optimization/82397
632 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
633 equality only for semantically equal trees.
635 2017-10-09 Richard Biener <rguenther@suse.de>
637 PR tree-optimization/82449
638 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
639 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
640 allow constant addresses.
641 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
644 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
646 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
649 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
652 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
655 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
658 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
660 2017-10-09 Jakub Jelinek <jakub@redhat.com>
663 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
664 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
666 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
668 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
669 (WI_BINARY_PREDICATE_RESULT): Likewise.
670 (wi::binary_traits::operator_result): New type.
671 (wi::binary_traits::predicate_result): Likewise.
672 (generic_wide_int::operator~, unary generic_wide_int::operator-)
673 (generic_wide_int::operator==, generic_wide_int::operator!=)
674 (generic_wide_int::operator&, generic_wide_int::and_not)
675 (generic_wide_int::operator|, generic_wide_int::or_not)
676 (generic_wide_int::operator^, generic_wide_int::operator+
677 (binary generic_wide_int::operator-, generic_wide_int::operator*):
679 (operator~, unary operator-, operator==, operator!=, operator&)
680 (operator|, operator^, operator+, binary operator-, operator*): New
682 * expr.c (get_inner_reference): Use wi::bit_and_not.
683 * fold-const.c (fold_binary_loc): Likewise.
684 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
685 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
686 (bit_value_binop): Likewise.
687 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
688 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
689 (extract_range_from_binary_expr_1): Likewise.
690 (masked_increment): Likewise.
691 (simplify_bit_ops_using_ranges): Likewise.
693 2017-10-09 Martin Jambor <mjambor@suse.cz>
696 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
697 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
698 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
700 (hsa_fixup_mov_insn_type): New function.
701 (hsa_op_with_type::get_in_type): Use it.
702 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
703 immediates in an assert.
704 (hsa_op_with_type::extend_int_to_32bit): New method.
705 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
706 types. Convert to dest type if necessary.
707 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
708 (reg_for_gimple_ssa): Pass false as min32int to
709 hsa_type_for_scalar_tree_type.
710 (gen_hsa_addr): Fixup type when creating addresable temporary.
711 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
712 (gen_hsa_unary_operation): Extend operands and convert to dest type if
713 necessary. Call hsa_fixup_mov_insn_type.
714 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
715 extend operands and convert to dest type if necessary.
716 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
717 to dest type if necessary.
718 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
719 if conversion nt necessary and size matches.
720 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
721 to dest type if necessary.
722 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
723 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
725 (gen_hsa_clrsb): Likewise.
726 (gen_hsa_ffs): Likewise.
727 (gen_hsa_divmod): Extend operands and convert to dest type if
729 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
731 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
733 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
734 Remove empty default arguments. Use a brace block as output
736 (conditional return): Ditto.
738 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
739 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
741 (group_ending_nop): Ditto.
743 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
744 (splitters for those): Ditto.
746 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
748 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
749 a conditional jump (and the compare for it) so that pc_rtx is the
751 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
752 for the deleted and renamed ctr<mode>_internal[234] patterns.
753 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
754 Delete second conditional return pattern.
755 (ctr<mode>_internal2): Delete this second bdnz pattern.
756 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
757 (ctr<mode>_internal4): Delete this second bdz pattern.
759 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
761 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
762 (always_initialized_rtx_for_ssa_name_p): New predicate.
763 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
764 (finish_out_of_ssa): Free new field of SA.
765 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
766 * tree-ssa-coalesce.c: Include tree-ssa.h.
767 (get_parm_default_def_partitions): Remove extern keyword.
768 (get_undefined_value_partitions): New function.
769 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
770 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
772 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
774 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
776 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
778 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
780 * i386.c (ix86_expand_set_or_movmem): Disable 512bit loops for targets
783 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
785 * i386.c (has_dispatch): Disable for Ryzen.
787 2017-10-08 Olivier Hainque <hainque@adacore.com>
789 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
790 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
792 (thumb_set_return_address): Likewise.
794 2017-10-08 Olivier Hainque <hainque@adacore.com>
796 * common/config/arm/arm-common.c (arm_except_unwind_info):
797 Handle DWARF2_UNWIND_INFO.
799 2017-10-07 Michael Collison <michael.collison@arm.com>
801 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
804 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
806 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
807 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
808 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
809 defined, force the creation of a new block for a dispatch label.
811 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
813 * invoke.texi (Wsuggest-attribute=cold): Document.
814 * common.opt (Wsuggest-attribute=cold): New
815 * ipa-pure-const.c (warn_function_cold): New function.
816 * predict.c (compute_function_frequency): Use it.
817 * predict.h (warn_function_cold): Declare.
819 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
821 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
824 2017-10-06 Martin Liska <mliska@suse.cz>
826 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
827 keyword for member functions.
828 (struct sanopt_tree_couple): New struct.
829 (struct sanopt_tree_couple_hash): New function.
830 (struct sanopt_ctx): Add new hash_map.
831 (has_dominating_ubsan_ptr_check): New function.
832 (record_ubsan_ptr_check_stmt): Likewise.
833 (maybe_optimize_ubsan_ptr_ifn): Likewise.
834 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
835 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
837 2017-10-06 Sudakshina Das <sudi.das@arm.com>
840 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
841 aarch64_simd_valid_immediate on CONST_VECTORs.
842 (aarch64_reg_or_bic_imm): Likewise.
844 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
846 PR rtl-optimization/82396
847 * haifa-sched.c (ready_sort_real): Disable qsort checking.
849 2017-10-06 Sebastian Pop <sebpop@gmail.com>
851 * graphite-dependences.c (scop_get_reads): Move code to...
852 (scop_get_must_writes): Move code to...
853 (scop_get_may_writes): Move code to...
854 (scop_get_reads_and_writes): ... here.
855 (scop_get_dependences): Call scop_get_reads_and_writes.
857 2017-10-06 Jakub Jelinek <jakub@redhat.com>
859 PR tree-optimization/82434
860 * fold-const.h (can_native_encode_type_p,
861 can_native_encode_string_p): Remove.
862 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
863 don't encode anything, just return what would be otherwise returned.
864 (native_encode_fixed, native_encode_complex, native_encode_vector):
866 (native_encode_string): Likewise. Inline by hand
867 can_native_encode_string_p.
868 (can_native_encode_type_p): Remove.
869 (can_native_encode_string_p): Remove.
870 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
871 STRING_CSTs using can_native_encode_string_p, test all
872 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
873 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
874 argument from native_encode_expr.
875 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
876 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
879 2017-10-06 Richard Biener <rguenther@suse.de>
881 PR tree-optimization/82397
882 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
883 operand_equal_p but rely on data_ref_compare_tree for detecting
885 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
886 to match up with dr_group_sort_cmp.
888 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
891 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
893 * config/s390/s390-builtin-types.def: Regenerate.
895 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
898 * config/s390/s390-builtin-types.def: Regenerate.
899 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
900 Change flag from B_VXE to B_VX.
901 (s390_vec_min_dbl): Remove B_VXE flag.
903 2017-10-06 Richard Biener <rguenther@suse.de>
905 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
906 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
907 translate_isl_ast_to_gimple::is_valid_rename,
908 translate_isl_ast_to_gimple::get_rename,
909 translate_isl_ast_to_gimple::get_def_bb_for_const,
910 translate_isl_ast_to_gimple::get_new_name,
911 translate_isl_ast_to_gimple::collect_all_ssa_names,
912 translate_isl_ast_to_gimple::copy_loop_phi_args,
913 translate_isl_ast_to_gimple::collect_all_ssa_names,
914 translate_isl_ast_to_gimple::copy_loop_phi_args,
915 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
916 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
917 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
918 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
919 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
920 translate_isl_ast_to_gimple::copy_cond_phi_args,
921 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
922 translate_isl_ast_to_gimple::edge_for_new_close_phis,
923 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
924 translate_isl_ast_to_gimple::rename_uses,
925 translate_isl_ast_to_gimple::rename_all_uses): Remove.
926 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
927 (set_rename_for_each_def): Likewise.
928 (graphite_copy_stmts_from_block): Handle debug stmt resetting
929 here. Handle rewriting SCEV analyzable uses here.
930 (copy_bb_and_scalar_dependences): Generate code for PHI
932 (graphite_regenerate_ast_isl): Adjust.
933 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
934 (add_write, add_read): New functions.
935 (build_cross_bb_scalars_def): Use it and simplify.
936 (build_cross_bb_scalars_use): Likewise.
937 (graphite_find_cross_bb_scalar_vars): Inline into...
938 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
939 simulating out-of-SSA. Compute liveout and add dependencies.
940 (build_scops): Force an empty entry block.
941 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
943 (sese_build_liveouts): Declare.
944 (sese_trivially_empty_bb_p): Likewise.
945 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
946 compute liveout and debug_liveout.
947 (sese_bad_liveouts_use): Remove.
948 (sese_reset_debug_liveouts_bb): Likewise.
949 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
950 (sese_build_liveouts): Build liveout and debug_liveout and store
952 (new_sese_info): Adjust.
953 (free_sese_info): Likewise.
954 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
955 do not build liveout here.
956 (move_sese_in_condition): Adjust region entry.
957 (scev_analyzable_p): Match up with chrec_apply requirements.
958 (sese_trivially_empty_bb_p): New.
959 * tree-into-ssa.c (get_reaching_def): Properly support generating
960 default-defs for incremental rewrite of anonymous names.
962 2017-10-06 Richard Biener <rguenther@suse.de>
964 * graphite-sese-to-poly.c (extract_affine): For casts increasing
965 precision do not perform modulo reduction.
967 2017-10-06 Richard Biener <rguenther@suse.de>
969 PR tree-optimization/82436
970 * tree-vect-slp.c (vect_supported_load_permutation_p): More
971 conservatively choose the vectorization factor when checking
972 whether we can perform the required load permutation.
973 (vect_transform_slp_perm_load): Assert when we may not fail.
975 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
977 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
978 message for incompatible -msdata=* and -mcall-* options.
980 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
982 * i386.c (ia32_multipass_dfa_lookahead): Default to issue rate
983 for post-reload scheduling.
985 2017-10-05 Tamar Christina <tamar.christina@arm.com>
987 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
989 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
991 * i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
992 to improve monte carlo in scimark.
994 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
996 * i386.c (ix86_size_cost, i386_cost, i486_cost, pentium_cost,
997 lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
998 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
999 pentium4_cost, nocona_cost): Set reassociation width to 1.
1000 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
1001 width to 2 for fp operations and 1 otherwise.
1002 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
1004 (atom_cost): Set reassociation width to 2.
1005 (slm_cost, generic_cost): Set fp reassociation width
1006 to 2 and 1 otherwise.
1007 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
1008 (core_cost): Set fp reassociation width to 4 and vector to 2.
1009 (ix86_reassociation_width): Rewrite using cost table; special case
1010 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
1011 and TARGET_AVX128_OPTIMAL.
1012 * i386.h (processor_costs): Add
1013 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
1014 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
1015 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
1016 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
1017 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
1018 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
1020 2017-10-05 Nathan Sidwell <nathan@acm.org>
1022 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
1024 2017-10-05 Tamar Christina <tamar.christina@arm.com>
1026 * config/arm/arm.c (arm_test_fpu_data): New.
1027 (arm_run_selftests): Call arm_test_fpu_data.
1029 2017-10-04 Nathan Sidwell <nathan@acm.org>
1031 * toplev.c (toplev::main): Remove excess parens on pretty_printer
1033 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
1035 2017-10-04 Sudakshina Das <sudi.das@arm.com>
1037 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
1038 check type for aarch64_simd_valid_immediate.
1039 (aarch64_output_simd_mov_immediate): Update prototype.
1040 (aarch64_simd_valid_immediate): Update prototype.
1041 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
1042 support for ORR-immediate.
1043 (and<mode>3): modified pattern to add support for BIC-immediate.
1044 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
1045 now checks for valid immediate for BIC and ORR based on new enum
1047 (aarch64_output_simd_mov_immediate): Function now used to output
1048 BIC/ORR imm as well based on new enum argument.
1049 * config/aarch64/constraints.md (Do): New vector immediate constraint.
1051 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
1052 (aarch64_reg_or_bic_imm): Likewise.
1054 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1056 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
1057 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
1059 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
1063 PR rtl-optimization/82396
1064 * haifa-sched.c (autopref_multipass_init): Simplify
1066 (autopref_rank_data): Simplify sort order.
1067 * sched-int.h (autopref_multipass_data_): Remove
1068 multi_mem_insn_p, min_offset and max_offset.
1070 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1072 * doc/sourcebuild.texi: Document vect_peeling_profitable.
1074 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1076 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
1079 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1081 * doc/sourcebuild.texi: Document vect_long_mult.
1083 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
1085 PR tree-optimization/82413
1086 * fold-const.c (build_range_check): Use widest_int when comparing
1087 the maximum ETYPE value with HIGH.
1089 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
1091 PR rtl-optimization/82396
1092 * haifa-sched.c (autopref_multipass_init): Simplify
1094 (autopref_rank_data): Simplify sort order.
1095 * sched-int.h (autopref_multipass_data_): Remove
1096 multi_mem_insn_p, min_offset and max_offset.
1098 2017-10-04 Jakub Jelinek <jakub@redhat.com>
1100 PR tree-optimization/82381
1101 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
1102 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
1105 PR tree-optimization/82374
1106 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
1107 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
1108 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
1109 current_function_decl to the new decl.
1111 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
1113 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
1114 helper macro for IEEE float128 hardware built-in functions.
1115 (SQRTF128_ODD): Add built-in functions with the round-to-odd
1117 (TRUNCF128_ODD): Likewise.
1118 (ADDF128_ODD): Likewise.
1119 (SUBF128_ODD): Likewise.
1120 (MULF128_ODD): Likewise.
1121 (DIVF128_ODD): Likewise.
1122 (FMAF128_ODD): Likewise.
1123 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
1124 UNSPEC_TRUNC_ROUND_TO_ODD.
1125 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
1126 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
1127 floating point round to odd instructions.
1128 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
1129 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
1130 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
1131 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
1132 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
1133 (trunc<mode>sf2_hw): Change the truncate with round to odd
1134 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
1135 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
1136 to odd hardware instructions.
1137 (sub<mode>3_odd): Likewise.
1138 (mul<mode>3_odd): Likewise.
1139 (div<mode>3_odd): Likewise.
1140 (sqrt<mode>2_odd): Likewise.
1141 (fma<mode>4_odd): Likewise.
1142 (fms<mode>4_odd): Likewise.
1143 (nfma<mode>4_odd): Likewise.
1144 (nfms<mode>4_odd): Likewise.
1145 (trunc<mode>df2_odd): Change the truncate with round to odd
1146 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
1148 * doc/extend.texi (PowerPC built-in functions): Update documentation
1149 for existing IEEE float128-bit built-in functions. Add built-in
1150 functions that generate the IEEE 128-bit floating point round to
1153 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
1155 PR rtl-optimization/77729
1156 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
1157 to (X&(C1&~C2))|C2 transformations.
1159 2017-10-03 Martin Jambor <mjambor@suse.cz>
1161 PR tree-optimization/82363
1162 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
1163 mismatch, mark lacc written regardless of racc.
1165 2017-10-03 Jakub Jelinek <jakub@redhat.com>
1167 PR tree-optimization/82381
1168 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
1169 stmt_to_insert nor wheather SSA_NAMEs are default defs.
1170 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
1171 fallthrough into reassoc_stmt_dominates_stmt_p.
1174 * combine.c (combine_instructions): Don't combine in unreachable
1177 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
1180 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
1181 function to not use the have_cpu variable. Do not set cpu_index,
1182 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
1184 (rs6000_valid_attribute_p): Remove duplicate initializations of
1185 old_optimize and func_optimize.
1186 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
1187 (rs6000_activate_target_options): Make global.
1188 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
1191 2017-10-02 Jakub Jelinek <jakub@redhat.com>
1193 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
1194 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
1195 Set *poffset to 0 and *psize and *pmax_size to -1 if
1196 *poffset + *psize overflows in HOST_WIDE_INT.
1198 PR tree-optimization/82387
1199 PR tree-optimization/82388
1200 PR tree-optimization/82389
1201 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
1202 instead of live_bytes non-NULL.
1204 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
1207 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
1210 2017-10-02 Richard Biener <rguenther@suse.de>
1212 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
1213 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
1214 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
1216 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
1218 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
1219 requested precision matches the type's.
1220 * calls.c (alloc_max_size): Calculate the new candidate size as
1221 a widest_int and use wi::to_widest when comparing it with the
1222 current candidate size.
1223 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
1224 zero rather than integer_zero_node.
1225 * match.pd: Check for a no-op conversion before using wi::add
1226 rather than after. Use tree_to_uhwi when summing small shift
1227 counts into an unsigned int.
1229 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
1230 Alan Hayward <alan.hayward@arm.com>
1231 David Sherwood <david.sherwood@arm.com>
1234 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
1235 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
1236 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
1237 POINTER_AND_FP_REGS.
1239 2017-10-02 Richard Biener <rguenther@suse.de>
1241 PR tree-optimization/82355
1242 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
1243 a mapping for the enclosing loop but avoid generating one for
1245 (copy_bb_and_scalar_dependences): Remove premature codegen
1246 error on PHIs in blocks duplicated into multiple places.
1247 * graphite-scop-detection.c
1248 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
1249 in the region use it as loop and nest to analyze the DR in.
1250 (try_generate_gimple_bb): Likewise.
1251 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
1252 (add_loop_constraints): For blocks in a loop not in the region
1253 create a dimension with a single iteration.
1254 * sese.h (gbb_loop_at_index): Remove assert.
1256 2017-10-01 Kevin Buettner <kevinb@redhat.com>
1258 * omp-expand.c (adjust_context_scope): New function.
1259 (expand_parallel_call): Call adjust_context_scope.
1261 2017-10-01 Jeff Law <law@redhat.com>
1263 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
1264 dom_opt_dom_walker class with direct access to private members.
1265 Add comments. Call test_for_singularity.
1266 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
1267 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
1268 m_dummy_cond anymore.
1269 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
1271 (pass_dominator:execute): Build the dummy_cond here and pass it
1272 to the dom_opt_dom_walker ctor.
1273 (test_for_singularity): New function.
1275 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
1276 Maya Rashish <coypu@sdf.org>
1278 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
1279 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
1280 (alpha*-*-netbsd*) Use nbsd_tm_file.
1281 (arm*-*-netbsdelf*) Likewise.
1282 (i[34567]86-*-netbsdelf*) Likewise.
1283 (x86_64-*-netbsd*) Likewise.
1284 (mips*-*-netbsd*) Likewise.
1285 (powerpc-*-netbsd*) Likewise.
1286 (sh*-*-netbsd*) Likewise.
1287 (sparc-*-netbsdelf*) Likewise.
1288 (sparc64-*-netbsd*) Likewise.
1289 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
1291 (vax-*-netbsdelf*) Likewise.
1292 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
1293 (UINT_FAST8_TYPE) Likewise.
1294 (INT_FAST16_TYPE) Check CHAR_FAST16.
1295 (UINT_FAST16_TYPE) Likewise.
1297 2017-09-30 Jakub Jelinek <jakub@redhat.com>
1300 * config/i386/i386.md
1301 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
1302 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
1303 *divmodsi4_zext_2): New define_insn_and_split.
1304 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
1305 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
1306 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
1307 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
1308 New define_insn_and_split.
1309 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
1310 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
1311 operands[1] having DImode when mode is SImode.
1313 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
1314 always SImode for DIV and MOD in REG_EQUAL notes.
1316 2017-09-29 Yury Gribov <tetra2005@gmail.com>
1319 * match.pd: Fix handling of NaNs in pattern.
1321 2017-09-29 Jeff Law <law@redhat.com>
1323 * sbitmap.c (bitmap_bit_in_range_p): New function.
1324 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
1325 * tree-ssa-dse.c (live_bytes_read): New function.
1326 (dse_classify_store): Ignore reads of dead bytes.
1328 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
1329 typos and whitespace errors.
1330 * config/i386/predicates.md (address_no_seg_operand): Likewise.
1331 * config/s390/s390.c (s390_emit_prologue): Likewise.
1333 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
1336 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
1337 with a symbol for LRA.
1339 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
1341 PR rtl-optimization/82338
1342 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
1344 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
1346 * genmodes.c (calc_wider_mode): Suppress qsort macro.
1347 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
1348 (qsort_chk): Declare.
1349 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
1350 (qsort_chk): New function.
1352 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1354 PR tree-optimization/82337
1355 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
1356 phi definition if the PHI result appears in an abnormal PHI.
1357 (find_basis_for_base_expr): Don't record a basis if the LHS of the
1358 basis appears in an abnormal PHI.
1360 2017-09-29 Richard Biener <rguenther@suse.de>
1362 * graphite-isl-ast-to-gimple.c
1363 (translate_isl_ast_to_gimple::set_codegen_error): New function.
1364 (binary_op_to_tree): Use it.
1365 (get_rename_from_scev): Likewise.
1366 (copy_loop_phi_nodes): Likewise.
1367 (copy_bb_and_scalar_dependences): Likewise.
1368 (translate_pending_phi_nodes): Likewise.
1370 2017-09-29 Jakub Jelinek <jakub@redhat.com>
1373 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
1374 for movabsq $(i32 << shift), r64.
1376 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
1378 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
1379 index when encoding %esp as %rsp to avoid 0x67 prefix.
1381 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
1383 * config/i386/i386.md (*movsf_internal, *movdf_internal):
1384 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
1386 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
1388 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
1389 Extensions with more than 16 double VFP registers.
1390 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
1391 to_clear_mask and all code related to it. Replace the remaining
1392 entry by a sbitmap and adapt code accordingly.
1394 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
1396 * brig-builtins.def: Change pure attributes to const.
1398 2017-09-28 Joseph Myers <joseph@codesourcery.com>
1400 * config.gcc (default_gnu_indirect_function): Default to yes for
1401 sparc*-*-linux* with glibc.
1403 2017-09-28 Joseph Myers <joseph@codesourcery.com>
1405 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
1406 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
1407 when creating .init_array and .fini_array sections with priority
1410 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
1413 * config/aarch64/aarch64.c
1414 (aarch64_builtin_support_vector_misalignment): Always return false
1415 when misalignment is unknown.
1417 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
1419 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
1420 this function to return false if the definition used by the swap
1421 instruction is artificial, or if the memory address from which the
1422 constant value is loaded is not represented by a base address held
1423 in a register or if the base address register is a frame or stack
1424 pointer. Additionally, return false if the base address of the
1425 loaded constant is a SYMBOL_REF but is not considered to be a
1427 (replace_swapped_load_constant): New function.
1428 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
1429 loaded constant vector with a load of a swapped constant vector.
1431 2017-09-27 Carl Love <cel@us.ibm.com>
1433 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
1434 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
1435 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
1438 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
1440 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
1441 first, always call autopref_rank_data otherwise.
1443 2017-09-27 Richard Biener <rguenther@suse.de>
1445 * graphite-scop-detection.c (find_scop_parameters): Move
1446 loop bound handling ...
1447 (gather_bbs::before_dom_children): ... here, avoiding the need
1448 to build scop_info->loop_nest.
1449 (record_loop_in_sese): Remove.
1450 * sese.h (sese_info_t::loop_nest): Remove.
1451 * sese.c (new_sese_info): Do not allocate loop_nest.
1452 (free_sese_info): Do not free loop_nest.
1454 2017-09-27 Jakub Jelinek <jakub@redhat.com>
1457 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
1458 lhs from calls if the lhs has addressable type.
1460 2017-09-27 Richard Biener <rguenther@suse.de>
1462 * graphite.h (scop::max_alias_set): New member.
1463 * graphite-scop-detection.c: Remove references to non-existing
1464 --param in comments.
1465 (build_alias_sets): Record the maximum alias set used for drs.
1466 (build_scops): Support zero as unlimited for
1467 --param graphite-max-arrays-per-scop.
1468 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
1470 (build_poly_sr_1): ... here. Compute alias set based on the
1471 maximum alias set used for drs rather than
1472 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
1474 2017-09-27 Richard Biener <rguenther@suse.de>
1476 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
1477 --param loop-block-tile-size=0 to disable tiling.
1479 2017-09-27 Richard Biener <rguenther@suse.de>
1481 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
1482 (graphite-max-nb-scop-params): Document special value zero.
1483 * domwalk.h (dom_walker::STOP): New symbolical constant.
1484 (dom_walker::dom_walker): Add optional parameter for bb to
1486 (dom_walker::~dom_walker): Declare.
1487 (dom_walker::before_dom_children): Document STOP return value.
1488 (dom_walker::m_user_bb_to_rpo): New member.
1489 (dom_walker::m_bb_to_rpo): Likewise.
1490 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
1491 mapping here if not provided by the user.
1492 (dom_walker::~dom_walker): Free bb to RPO mapping if not
1493 provided by the user.
1494 (dom_walker::STOP): Define.
1495 (dom_walker::walk): Do not compute bb to RPO mapping here.
1496 Support STOP return value from before_dom_children to stop
1498 * graphite-optimize-isl.c (optimize_isl): If the schedule
1499 is the same still generate code if -fgraphite-identity
1500 or -floop-parallelize-all are given.
1501 * graphite-scop-detection.c: Include cfganal.h.
1502 (gather_bbs::gather_bbs): Get and pass through bb to RPO
1504 (gather_bbs::before_dom_children): Return STOP for BBs
1506 (build_scops): Compute bb to RPO mapping and pass it to
1507 the domwalk. Treat --param graphite-max-nb-scop-params=0
1508 as not limiting the number of params.
1509 * graphite.c (graphite_initialize): Remove limit on the
1510 number of basic-blocks in a function.
1511 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
1512 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
1513 default value of 10.
1515 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
1517 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
1518 Adjust code to eliminate needing to do the shift right 32-bits
1519 operation after XSCVDPSPN.
1521 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1523 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
1524 ((X / Y) != 0 -> X >= Y): Likewise.
1526 2017-09-26 Carl Love <cel@us.ibm.com>
1528 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
1529 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
1530 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
1531 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
1532 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
1533 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
1534 definitions and overloading.
1535 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
1536 statement for P9V_BUILTIN_XST_LEN_R.
1537 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
1538 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
1539 define_expand and define_insn for the instructions and builtins.
1540 * doc/extend.texi: Update the built-in documentation file for the new
1542 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
1543 define_insn for the instructions
1545 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
1548 * gcc/config/netbsd-protos.h: New file.
1549 * gcc/config/netbsd.c: New file.
1550 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
1551 * gcc/config/t-netbsd: New file.
1552 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
1553 (tmake_file) Add t-netbsd.
1554 (extra_objs) Add netbsd.o.
1556 2017-09-26 Janus Weil <janus@gcc.gnu.org>
1560 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
1562 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
1564 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
1565 sign extension from a vector register to a GPR by doing a 32-bit
1566 direct move and then an EXTSW.
1567 (extendsi<mode>2 splitter): Likewise.
1568 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
1569 right or vector extract after doing XSCVDPSPN. Use
1570 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
1572 (movdi_from_sf_zero_ext): Likewise.
1573 (reload_gpr_from_vsxsf): Likewise.
1574 (p8_mfvsrd_4_disf): Delete, no longer used.
1575 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
1576 then needing to move the SFmode to a GPR to use the XSCVDPSP
1577 instruction instead of FRSP and XSCVDPSPN.
1578 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
1579 it is adjacent to the other XSCVSPDP insns.
1580 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
1581 SFmode to be in traditional Altivec registers.
1582 (vsx_xscvdpspn): Eliminate useless alternative constraint.
1583 (vsx_xscvspdpn): Likewise.
1584 (vsx_xscvspdpn_scalar): Likewise.
1586 2017-09-26 Martin Jambor <mjambor@suse.cz>
1588 * tree-sra.c (compare_access_positions): Put integral types first,
1589 stabilize sorting of integral types, remove conditions putting
1590 non-full-precision integers last.
1591 (sort_and_splice_var_accesses): Disable scalarization if a
1592 non-integert would be represented by a non-full-precision integer.
1594 2017-09-26 Joseph Myers <joseph@codesourcery.com>
1596 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
1597 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
1598 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
1599 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
1600 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
1601 conditionals inside the function instead of around it. Call
1602 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
1603 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
1605 2017-09-26 Richard Biener <rguenther@suse.de>
1607 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
1609 (scop_detection::build_scop_breadth): ... this. Removed.
1610 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
1611 (scop_detection::harmful_stmt_in_bb): Likewise.
1612 (scop_detection::graphite_can_represent_stmt): Likewise.
1613 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
1614 (scop_detection::can_represent_loop): Remove recursion, fold in ...
1615 (scop_detection::can_represent_loop_1): ... this. Removed.
1616 (scop_detection::harmful_loop_in_region): Simplify after inlining
1617 the above and remove more quadraticness.
1618 (build_scops): Adjust.
1619 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
1622 2017-09-26 Jakub Jelinek <jakub@redhat.com>
1625 * config/i386/i386.c (ix86_print_operand_address_as): Only test
1626 REGNO (base) == SP_REG if base is a REG.
1629 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
1630 if it is different SSA_NAME.
1631 (optimize_range_tests_cmp_bitwise): New function.
1632 (optimize_range_tests): Call it.
1634 2017-09-26 Richard Biener <rguenther@suse.de>
1636 PR tree-optimization/82321
1637 * graphite.c (canonicalize_loop_closed_ssa): Properly check
1638 for the def being inside the loop.
1640 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1642 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
1644 * config/s390/s390-builtins.def: Fix constraint on op4.
1646 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1648 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
1649 independent expanders.
1650 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
1651 ("vec_ordered", "vec_unordered"): New expanders.
1653 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1655 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
1658 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1660 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
1661 vec_unpacks_lo_v16qi.
1662 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
1664 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1666 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
1667 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
1668 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
1670 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1672 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
1674 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
1676 ("*vec_slb<mode>"): New insn pattern.
1677 ("vec_shr_<mode>"): New expander.
1678 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
1679 and force the shift count operand to V16QImode.
1680 ("vec_srb<mode>"): Set shift count mode to V16QI.
1682 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1684 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
1685 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
1686 ("vec_widen_smult_hi_<mode>"): New expander definitions.
1688 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
1691 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
1693 2017-09-26 Richard Biener <rguenther@suse.de>
1695 PR tree-optimization/82320
1696 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
1699 2017-09-25 Jeff Law <law@redhat.com>
1701 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
1702 prototype for new argument.
1703 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
1704 mostly extracted from rs6000_emit_allocate_stack.
1705 (rs6000_emit_probe_stack_range_stack_clash): New function.
1706 (rs6000_emit_allocate_stack): Call
1707 rs6000_emit_probe_stack_range_stack_clash as needed.
1708 (rs6000_emit_probe_stack_range): Add additional argument
1709 to call to gen_probe_stack_range{si,di}.
1710 (output_probe_stack_range): New.
1711 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
1712 (output_probe_stack_range_stack_clash): New.
1713 (rs6000_emit_prologue): Emit notes into dump file as requested.
1714 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
1715 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
1716 Add additional operand and pass it to output_probe_stack_range.
1718 2017-09-25 Bin Cheng <bin.cheng@arm.com>
1720 PR tree-optimization/82163
1721 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
1722 (checking_verify_loop_closed_ssa): New parameter.
1723 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
1724 (check_loop_closed_ssa_stmt): Delete.
1725 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
1726 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
1727 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
1730 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
1732 * brig-builtins.def: Treat HSAIL barrier builtins as
1733 setjmp/longjump style functions.
1735 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
1737 * target.def (constant_alignment): New hook.
1738 * defaults.h (CONSTANT_ALIGNMENT): Delete.
1739 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
1740 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
1741 * doc/tm.texi: Regenerate.
1742 * targhooks.h (default_constant_alignment): Declare.
1743 (constant_alignment_word_strings): Likewise.
1744 * targhooks.c (default_constant_alignment): New function.
1745 (constant_alignment_word_strings): Likewise.
1746 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
1747 instead of CONSTANT_ALIGNMENT.
1748 * varasm.c (align_variable, get_variable_align, build_constant_desc)
1749 (force_const_mem): Likewise.
1750 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
1751 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
1752 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
1753 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1754 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
1756 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
1757 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1758 constant_alignment_word_strings.
1759 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
1760 (CONSTANT_ALIGNMENT): Likewise.
1761 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1762 (arm_constant_alignment): New function.
1763 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
1764 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1765 constant_alignment_word_strings.
1766 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
1767 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1768 constant_alignment_word_strings.
1769 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
1770 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1771 (cris_constant_alignment): New function.
1772 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
1773 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1774 (epiphany_constant_alignment): New function.
1775 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
1776 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1777 constant_alignment_word_strings.
1778 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
1779 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1780 constant_alignment_word_strings.
1781 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
1782 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1783 constant_alignment_word_strings.
1784 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
1785 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
1786 * config/i386/i386.c (ix86_constant_alignment): Make static.
1787 Use the same interface as the target hook.
1788 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1789 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
1790 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1791 constant_alignment_word_strings.
1792 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
1793 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
1794 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1795 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
1796 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1797 constant_alignment_word_strings.
1798 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
1799 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1800 constant_alignment_word_strings.
1801 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
1802 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1803 constant_alignment_word_strings.
1804 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
1805 * config/microblaze/microblaze.c (microblaze_constant_alignment):
1807 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1808 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
1809 * config/mips/mips.c (mips_constant_alignment): New function.
1810 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1811 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
1812 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
1813 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1814 (mmix_constant_alignment): Make static. Use the same interface
1816 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
1817 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1818 constant_alignment_word_strings.
1819 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
1820 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1821 constant_alignment_word_strings.
1822 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
1823 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1824 constant_alignment_word_strings.
1825 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
1826 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1827 (rs6000_constant_alignment): New function.
1828 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
1829 * config/riscv/riscv.c (riscv_constant_alignment): New function.
1830 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1831 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
1832 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1833 (rs6000_constant_alignment): New function.
1834 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
1835 * config/s390/s390.c (s390_constant_alignment): New function.
1836 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1837 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
1838 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1839 constant_alignment_word_strings.
1840 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
1841 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1842 (sparc_constant_alignment): New function.
1843 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
1844 * config/spu/spu.c (spu_constant_alignment): New function.
1845 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1846 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
1847 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1848 constant_alignment_word_strings.
1849 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
1850 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1851 constant_alignment_word_strings.
1852 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
1853 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1854 constant_alignment_word_strings.
1855 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
1856 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1857 (visium_constant_alignment): New function.
1858 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
1859 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1860 (xtensa_constant_alignment): New function.
1861 * system.h (CONSTANT_ALIGNMENT): Poison.
1863 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
1865 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
1866 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
1867 (rs6000_builtin_valid_without_lhs): New helper function.
1868 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1869 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
1871 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
1873 * target.h (vec_perm_indices): Use unsigned short rather than
1875 (auto_vec_perm_indices): Likewise.
1876 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
1877 Use unsigned int rather than unsigned char.
1878 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
1880 2017-09-25 Richard Biener <rguenther@suse.de>
1882 * cfgloop.h (sort_sibling_loops): Declare.
1883 * cfgloop.c (sort_sibling_loops_cmp): New helper.
1884 (sort_sibling_loops): New function sorting the sibling loop list
1886 * graphite.c (graphite_transform_loops): Sort sibling loops.
1888 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
1890 * target.def (vec_perm_const_ok): Change sel parameter to
1892 * optabs-query.c (can_vec_perm_p): Update accordingly.
1893 * doc/tm.texi: Regenerate.
1894 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
1895 auto_vec_perm_indices and remove separate nelt field.
1896 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
1897 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
1898 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
1899 (aarch64_expand_vec_perm_const): Update accordingly.
1900 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
1901 to vec_perm_indices.
1902 * config/arm/arm.c (expand_vec_perm_d): Change perm to
1903 auto_vec_perm_indices and remove separate nelt field.
1904 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
1905 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
1906 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
1908 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
1909 to vec_perm_indices.
1910 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
1911 sel to vec_perm_indices.
1912 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
1913 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
1914 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
1916 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
1919 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
1922 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
1923 on the FUNCTION_DECL function context if it has a DIE that is a
1926 2017-09-25 Richard Biener <rguenther@suse.de>
1928 PR tree-optimization/82285
1929 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
1932 2017-09-25 Tom de Vries <tom@codesourcery.com>
1936 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
1939 2017-09-25 Richard Biener <rguenther@suse.de>
1941 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
1942 ISL errors other than isl_error_quota happen. Dump if the
1943 schedule is the same.
1944 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
1945 errors instead of aborting inside ISL.
1947 2017-09-25 Iain Sandoe <iain@codesourcery.com>
1950 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
1951 of libgcc_eh for m64.
1952 * config/i386/darwin64.h: Likewise.
1954 2017-09-25 Richard Biener <rguenther@suse.de>
1957 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
1958 attribute for incomplete types nor twice for complete ones.
1960 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
1963 * config/i386/i386.c (ix86_print_operand_address_as): Encode
1964 %esp as %rsp to avoid 0x67 prefix if there is no index or base
1967 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
1970 * config/i386/i386.opt (mprefer-avx256): Use
1971 ix86_target_flags variable.
1972 * config/i386/i386.c (ix86_target_string): Move
1973 -mprefer-avx256 to flag2_opts.
1975 2017-09-22 Jakub Jelinek <jakub@redhat.com>
1978 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
1979 and x != -1 | y != -1 into (x & y) != -1.
1981 2017-09-22 Steve Ellcey <sellcey@cavium.com>
1983 * config.gcc: Add new case statement to set
1984 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
1985 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
1986 s390x-*-linux* case statements. Added aarch64 to the list of
1987 supported architectures.
1989 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1991 PR tree-optimization/82289
1992 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
1993 STMT_VINFO_RELEVANT_P.
1995 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1996 Alan Hayward <alan.hayward@arm.com>
1997 David Sherwood <david.sherwood@arm.com>
1999 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
2000 for VR_RANGE only; don't allow VR_ANTI_RANGE.
2001 (extract_range_from_binary_expr_1): Don't call
2002 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
2004 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2005 Alan Hayward <alan.hayward@arm.com>
2006 David Sherwood <david.sherwood@arm.com>
2008 * target.def (preferred_vector_alignment): New hook.
2009 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
2011 * doc/tm.texi: Regenerate.
2012 * targhooks.h (default_preferred_vector_alignment): Declare.
2013 * targhooks.c (default_preferred_vector_alignment): New function.
2014 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
2016 (DR_TARGET_ALIGNMENT): New macro.
2017 (aligned_access_p): Update commentary.
2018 (vect_known_alignment_in_bytes): New function.
2019 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
2021 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
2022 Calculate the misalignment based on the target alignment rather than
2024 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
2025 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
2026 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
2027 the target alignment, rather than masking the element misalignment
2028 with the number of elements in a vector. Also use the target
2029 alignment when calculating the maximum number of peels.
2030 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
2031 instead of TYPE_ALIGN_UNIT.
2032 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
2033 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
2034 (vect_create_addr_base_for_vector_ref): Update call accordingly.
2035 (vect_create_data_ref_ptr): Likewise.
2036 (vect_setup_realignment): Realign by ANDing with
2037 -DR_TARGET_MISALIGNMENT.
2038 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
2039 the number of peels based on DR_TARGET_ALIGNMENT.
2040 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
2041 with the guaranteed alignment boundary when deciding whether
2043 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
2044 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
2045 (ensure_base_align): Remove stmt_info parameter. Get the
2046 target base alignment from DR_TARGET_ALIGNMENT.
2047 (vectorizable_store): Update call accordingly. Interpret
2048 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
2050 (vectorizable_load): Likewise.
2052 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2053 Alan Hayward <alan.hayward@arm.com>
2054 David Sherwood <david.sherwood@arm.com>
2056 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
2057 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
2058 (vect_enhance_data_refs_alignment): Likewise.
2060 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
2062 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
2063 error. Only quit immediately if parsing is complete.
2064 (BEGIN): Initialize fatal_err and parse_done.
2065 (begin fpu, end fpu): Check number of arguments.
2066 (begin arch, end arch): Likewise.
2067 (begin cpu, end cpu): Likewise.
2068 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
2069 (optalias): Likewise.
2071 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
2073 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
2074 * config/arm/arm-isa.h: Delete. Move definitions to ...
2075 * arm-cpus.in: ... here. Use new feature and fgroup values.
2076 * config/arm/arm.c (arm_option_override): Use lower case for feature
2078 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
2079 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
2080 * config/arm/parsecpu.awk (END): Add new command 'isa'.
2082 (print_isa_bits_for): New function.
2083 (gen_isa): New function.
2084 (gen_comm_data): Use print_isa_bits_for.
2085 (define feature): New keyword.
2086 (define fgroup): New keyword.
2087 * config/arm/t-arm (TM_H): Remove.
2088 (GTM_H): Add arm-isa.h.
2089 (arm-isa.h): Add rule to generate file.
2090 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
2091 case for feature bit names.
2093 2017-09-22 Richard Biener <rguenther@suse.de>
2095 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
2097 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
2098 print of no dependency loops ...
2099 * graphite.c (graphite_transform_loops): ... here.
2100 (canonicalize_loop_closed_ssa_form): Work from inner to outer
2102 (same_close_phi_node, remove_duplicate_close_phi,
2103 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
2104 (canonicalize_loop_closed_ssa): ... here and simplify.
2105 * graphite-optimize-isl.c: Include tree-vectorizer.h.
2106 (optimize_isl): Use dump_printf_loc to tell when we stopped
2107 optimizing because of an ISL timeout.
2109 2017-09-22 Richard Biener <rguenther@suse.de>
2111 PR tree-optimization/82291
2112 * tree-if-conv.c (predicate_mem_writes): Make sure to
2113 remove writes in blocks predicated with false.
2115 2017-09-22 Richard Biener <rguenther@suse.de>
2117 * sese.c: Include cfganal.h.
2118 (if_region_set_false_region): Remove.
2119 (create_if_region_on_edge): Likewise.
2120 (move_sese_in_condition): Re-implement without destroying
2123 2017-09-22 Richard Biener <rguenther@suse.de>
2125 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
2126 Verify both BBs contain loop PHI nodes before dispatching to
2128 (graphite_regenerate_ast_isl): Do not recompute dominators,
2129 do not verify three times. Restructure for clarity.
2130 * graphite-scop-detection.c (same_close_phi_node,
2131 remove_duplicate_close_phi, make_close_phi_nodes_unique,
2132 defined_in_loop_p, canonicalize_loop_closed_ssa,
2133 canonicalize_loop_closed_ssa_form): Simplify, remove excess
2134 checking and SSA rewrite, move to ...
2135 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
2136 (graphite_initialize): Do not pass in ctx, do not reset the
2137 SCEV cache, compute only dominators.
2138 (graphite_transform_loops): Allocate ISL ctx after
2139 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
2140 Maintain post-dominators only around build_scops.
2141 * sese.c (if_region_set_false_region): Make static. Free
2142 and recompute dominators.
2143 (move_sese_in_condition): Assert we don't get called with
2144 post-dominators computed.
2145 * sese.h (if_region_set_false_region): Remove.
2147 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
2149 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
2150 mode attribute for TARGET_AVX512VL.
2152 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
2154 * config/i386/i386.opt (mprefer-avx256): New option.
2155 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
2157 (ix86_preferred_simd_mode): Return 256-bit AVX modes
2158 for TARGET_PREFER_AVX256.
2159 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
2161 2017-09-21 Jeff Law <law@redhat.com>
2163 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
2164 Fix dump output if the only stack space is for pushed registers.
2166 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2168 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
2171 2017-09-21 Martin Sebor <msebor@redhat.com>
2174 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
2175 code (in C++) or code that triggers warnings.
2177 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
2179 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
2181 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
2183 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
2184 * sched-rgn.c: Ditto.
2185 * sel-sched-ir.c: Ditto.
2187 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
2189 * toplev.h (set_random_seed): Adjust return type.
2190 * toplev.c (init_local_tick): Move eager initialization of random_seed
2191 to get_random_seed. Adjust comment.
2192 (init_random_seed): Inline to get_random_seed, delete.
2193 (get_random_seed): Initialize random_seed lazily.
2194 (set_random_seed): Do not return previous value.
2195 (print_switch_value): Do not call get_random_seed.
2197 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
2199 * cgraph.c (delete_function_version): New, broken out from...
2200 (cgraph_node::delete_function_version): ...here. Rename to
2201 cgraph_node::delete_function_version_by_decl. Update all uses.
2202 (cgraph_node::remove): Call delete_function_version.
2204 2017-09-21 Jakub Jelinek <jakub@redhat.com>
2207 * tree-inline.c (expand_call_inline): Emit clobber stmts for
2208 VAR_DECLs to which addressable non-volatile parameters are mapped
2209 and for id->retvar after the return value assignment. Clear
2210 id->retval and id->retbnd after inlining.
2212 2017-09-21 Richard Biener <rguenther@suse.de>
2214 PR tree-optimization/82276
2215 PR tree-optimization/82244
2216 * tree-vrp.c (build_assert_expr_for): Set
2217 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
2219 (remove_range_assertions): Revert earlier change.
2221 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
2224 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
2226 2017-09-21 Richard Biener <rguenther@suse.de>
2228 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
2229 Restore valid IL after code generation errors.
2230 * graphite.c (graphite_transform_loops): Diagnose code
2231 generation issues as MSG_MISSED_OPTIMIZATION and continue
2232 with processing SCOPs.
2234 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2235 Alan Hayward <alan.hayward@arm.com>
2236 David Sherwood <david.sherwood@arm.com>
2238 * calls.c (compute_argument_addresses): Use simplify_gen_binary
2239 rather than choosing between plus_constant and gen_rtx_<CODE>.
2240 * expr.c (emit_push_insn): Likewise.
2241 (expand_expr_real_2): Likewise.
2243 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2244 Alan Hayward <alan.hayward@arm.com>
2245 David Sherwood <david.sherwood@arm.com>
2247 * loop-unroll.c (split_iv): Call copy_rtx on the step.
2249 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2250 Alan Hayward <alan.hayward@arm.com>
2251 David Sherwood <david.sherwood@arm.com>
2253 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
2254 calling tree_to_uhwi.
2256 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2257 Alan Hayward <alan.hayward@arm.com>
2258 David Sherwood <david.sherwood@arm.com>
2260 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
2261 INTEGER_CST rather than a negative test for ADDR_EXPR.
2263 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2264 Alan Hayward <alan.hayward@arm.com>
2265 David Sherwood <david.sherwood@arm.com>
2267 * tree-vrp.c (extract_range_from_binary_expr_1): Check
2268 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
2270 2017-09-21 Richard Biener <rguenther@suse.de>
2272 PR tree-optimization/71351
2273 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
2274 graphite_create_new_loop_guard): Remove, fold remaining parts
2276 (translate_isl_ast_node_for): ... here and simplify.
2278 2017-09-21 Jakub Jelinek <jakub@redhat.com>
2281 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
2282 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
2283 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
2284 alternative always use QI mode, for -Os imov (=R,R) alternative
2285 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
2288 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2289 Jeff Law <law@redhat.com>
2291 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
2292 (allocate_stack_space): New function, partially extracted from
2294 (s390_emit_prologue): Track offset to most recent stack probe.
2295 Code to allocate space moved into allocate_stack_space.
2296 Dump actions when no stack is allocated.
2297 (s390_prologue_plus_offset): New function.
2298 (s390_emit_stack_probe): Likewise.
2300 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
2302 * common.opt (Wa, Wl, Wp, g, gz=): Add
2304 (gno-column-info): Remove.
2305 (gcolumn-info): Drop RejectNegative.
2307 (gno-record-gcc-switches): Remove.
2308 (grecord-gcc-switches): Drop RejectNegative.
2309 (gno-split-dwarf): Remove.
2310 (gsplit-dwarf): Drop RejectNegative.
2311 (gno-strict-dwarf): Remove.
2312 (gstrict-dwarf): Drop RejectNegative.
2313 * config/darwin.opt (gfull, gused): Add RejectNegative.
2314 * dwarf2out.c (gen_producer_string): Drop
2315 gno-record-gcc-switches handler.
2316 * optc-gen.awk: Add g to prefixes with negative forms.
2317 * opts-common.c (remapping_prefix_p): New.
2318 (find_opt): Check it.
2319 (generate_canonical_option): Test g prefix.
2320 (option_map): Add -gno- mapping.
2321 (add_misspelling_candidates): Check remapping_prefix_p.
2323 2017-09-20 Jeff Law <law@redhat.com>
2325 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
2326 thinko in stack clash protection support.
2328 * explow.c (compute_stack_clash_protection_loop_data): Use
2329 CONST_INT_P instead of explicit test. Verify object is a
2330 CONST_INT_P before looking at INTVAL.
2331 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
2332 instead of explicit test.
2334 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
2337 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
2338 address instead of to r1 and r11.
2340 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
2342 * config.gcc: Support "knm".
2343 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
2344 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2346 * config/i386/i386.c (m_KNM): Define.
2347 (processor_target_table): Add "knm".
2349 (ix86_option_override_internal): Add "knm".
2350 (ix86_issue_rate): Add PROCESSOR_KNM.
2351 (ix86_adjust_cost): Ditto.
2352 (ia32_multipass_dfa_lookahead): Ditto.
2353 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
2354 (fold_builtin_cpu): Add M_INTEL_KNM.
2355 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
2356 (processor_type): Add PROCESSOR_KNM.
2357 * config/i386/x86-tune.def: Add m_KNM.
2358 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
2360 2017-09-20 Richard Biener <rguenther@suse.de>
2362 PR tree-optimization/80213
2363 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
2364 are allowed in empty BBs as well.
2365 (canonicalize_loop_closed_ssa): Also look for other complex
2367 (scop_detection::get_sese): Include the loop-closed PHI block
2369 (scop_detection::merge_sese): Remove code adding extra blocks.
2370 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
2371 (build_scops): Assert the final returned scop is invalid.
2373 2017-09-20 Richard Biener <rguenther@suse.de>
2375 PR tree-optimization/82264
2376 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
2378 (vn_phi_lookup): Likewise.
2379 (vn_phi_insert): Likewise.
2381 2017-09-20 Jakub Jelinek <jakub@redhat.com>
2383 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
2384 that fits into uhwi or shwi, add DW_AT_const_value regardless
2385 of early_dwarf without going through RTL, using add_AT_unsigned
2388 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
2389 (DEBUG_LTO_ABBREV_SECTION): Likewise.
2390 (DEBUG_LTO_MACINFO_SECTION): Likewise.
2391 (DEBUG_MACRO_SECTION): Likewise.
2392 (DEBUG_LTO_MACRO_SECTION): Likewise.
2393 (DEBUG_STR_DWO_SECTION): Likewise.
2394 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
2395 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
2396 (DEBUG_LTO_DWO_LINE_SECTION): Define.
2397 (DEBUG_LTO_LINE_STR_SECTION): Define.
2398 (init_sections_and_labels): Initialize debug_line_str_section
2399 variable. Initialize debug_loc_section for -gdwarf-5 to
2400 DEBUG_LOCLISTS_SECTION. Formatting fixes.
2402 2017-09-20 Richard Biener <rguenther@suse.de>
2404 * graphite-sese-to-poly.c (extract_affine): Properly handle
2405 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
2407 2017-09-20 Richard Biener <rguenther@suse.de>
2409 PR tree-optimization/81373
2410 * graphite-scop-detection.c (build_cross_bb_scalars_def):
2411 Force SESE live-out defs to be handled even if they are
2414 2017-09-19 Jeff Law <law@redhat.com>
2416 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
2417 nothing for stack adjustments with REG_STACK_CHECK.
2418 * sched-deps.c (parse_add_or_inc): Reject insns with
2419 REG_STACK_CHECK from dependency breaking.
2420 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
2421 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
2422 * reg-notes.def (STACK_CHECK): New note.
2424 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
2425 (ix86_expand_prologue): Dump stack clash info as needed.
2426 Call ix86_adjust_stack_and_probe_stack_clash as needed.
2428 * function.c (dump_stack_clash_frame_info): New function.
2429 * function.h (dump_stack_clash_frame_info): Prototype.
2430 (enum stack_clash_probes): New enum.
2432 * config/alpha/alpha.c (alpha_expand_prologue): Also check
2433 flag_stack_clash_protection.
2434 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
2435 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
2436 (arm_frame_pointer_required): Likewise.
2437 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
2438 (ia64_expand_prologue): Likewise.
2439 * config/mips/mips.c (mips_expand_prologue): Likewise.
2440 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
2441 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
2442 (sparc_flat_expand_prologue): Likewise.
2443 * config/spu/spu.c (spu_expand_prologue): Likewise.
2445 * explow.c: Include "params.h".
2446 (anti_adjust_stack_and_probe_stack_clash): New function.
2447 (get_stack_check_protect): Likewise.
2448 (compute_stack_clash_protection_loop_data): Likewise.
2449 (emit_stack_clash_protection_loop_start): Likewise.
2450 (emit_stack_clash_protection_loop_end): Likewise.
2451 (allocate_dynamic_stack_space): Use get_stack_check_protect.
2452 Use anti_adjust_stack_and_probe_stack_clash.
2453 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
2454 (emit_stack_clash_protection_loop_start): Likewise.
2455 (emit_stack_clash_protection_loop_end): Likewise.
2456 * rtl.h (get_stack_check_protect): Prototype.
2457 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
2458 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
2459 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
2461 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
2463 * doc/tm.texi: Rebuilt.
2464 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
2465 get_stack_check_protect.
2466 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
2467 * config/arm/arm.c (arm_expand_prologue): Likewise.
2468 (arm_frame_pointer_required): Likewise.
2469 * config/i386/i386.c (ix86_expand_prologue): Likewise.
2470 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
2471 * config/mips/mips.c (mips_expand_prologue): Likewise.
2472 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
2473 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
2474 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
2475 (sparc_flat_expand_prologue): Likewise.
2477 * common.opt (-fstack-clash-protection): New option.
2478 * flag-types.h (enum stack_check_type): Note difference between
2479 -fstack-check= and -fstack-clash-protection.
2480 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
2481 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
2482 * toplev.c (process_options): Issue warnings/errors for cases
2483 not handled with -fstack-clash-protection.
2484 * doc/invoke.texi (-fstack-clash-protection): Document new option.
2485 (-fstack-check): Note additional problem with -fstack-check=generic.
2486 Note that -fstack-check is primarily for Ada and refer users
2487 to -fstack-clash-protection for stack-clash-protection.
2488 Document new params for stack clash protection.
2490 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
2492 * config/i386/i386.c (ix86_split_long_move): Do not handle
2493 address used for LEA in a special way.
2495 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
2497 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
2499 2017-09-19 Martin Sebor <msebor@redhat.com>
2502 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
2503 of incompatible types.
2505 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
2507 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
2508 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
2509 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2510 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
2512 2017-09-19 Richard Biener <rguenther@suse.de>
2514 PR tree-optimization/82244
2515 * tree-vrp.c (remove_range_assertions): Do not propagate
2516 a constant to abnormals but replace the assert with a copy.
2518 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
2520 PR rtl-optimization/57878
2521 PR rtl-optimization/68988
2522 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
2523 avoidance test involving non_reload_pseudos. Move frequency test
2524 below the general fragmentation avoidance test.
2526 2017-09-19 Richard Biener <rguenther@suse.de>
2528 PR tree-optimization/69728
2529 * graphite-sese-to-poly.c (schedule_error): New global.
2530 (add_loop_schedule): Handle empty domain by failing the
2532 (build_original_schedule): Handle schedule_error.
2534 2017-09-19 Richard Biener <rguenther@suse.de>
2536 * graphite-scop-detection.c (scop_detection::can_represent_loop):
2537 Do not iterate to sibling loops but only to siblings of inner
2540 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
2543 * config/m68k/m68k.md (moveq feeding equality comparison): Check
2544 that the registers are different.
2546 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
2548 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
2549 to processor_model and "amdfam17h" to arch_names_table.
2550 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
2552 2017-09-18 Jakub Jelinek <jakub@redhat.com>
2555 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
2557 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2558 Alan Hayward <alan.hayward@arm.com>
2559 David Sherwood <david.sherwood@arm.com>
2561 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
2563 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
2565 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
2566 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
2567 vect_schedule_slp_instance.
2568 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
2569 Update call to vect_slp_analyze_node_operations. Simplify return
2571 (vect_slp_analyze_bb_1): Update call accordingly.
2572 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
2573 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
2574 (vect_schedule_slp): Update call accordingly.
2576 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2577 Alan Hayward <alan.hayward@arm.com>
2578 David Sherwood <david.sherwood@arm.com>
2580 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
2581 with types that aren't in fact scalar.
2583 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2585 * tree-vect-slp.c (vect_record_max_nunits): New function,
2587 (vect_build_slp_tree_1): ...here.
2588 (vect_build_slp_tree_2): Call it for phis too.
2590 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2592 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
2593 to vect_get_vec_def_for_operand when getting the mask operand.
2595 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2596 Alan Hayward <alan.hayward@arm.com>
2597 David Sherwood <david.sherwood@arm.com>
2599 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
2602 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2603 Alan Hayward <alan.hayward@arm.com>
2604 David Sherwood <david.sherwood@arm.com>
2606 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
2607 calculation for vector booleans.
2609 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2610 Alan Hayward <alan.hayward@arm.com>
2611 David Sherwood <david.sherwood@arm.com>
2613 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
2615 (vect_transform_stmt): ...here.
2616 (vect_analyze_stmt): Use it instead of calling
2617 vectorizable_live_operation directly.
2619 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
2621 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
2622 non-SIMT targets in acc vector loops.
2624 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
2626 * configure.ac: Add arc and check if assembler supports gdwarf2.
2627 * configure: Regenerate.
2629 2017-09-18 Richard Biener <rguenther@suse.de>
2631 PR tree-optimization/82220
2632 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
2633 epilogue niters from the min_profitable_iters compute.
2635 2017-09-18 Jakub Jelinek <jakub@redhat.com>
2638 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
2639 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
2640 (ix86_init_pic_reg): Revert 2017-09-01 changes.
2642 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
2645 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
2646 switching to a new text section.
2648 2017-09-18 Richard Biener <rguenther@suse.de>
2650 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
2652 (build_alias_set): Reject aliases with no access function.
2654 2017-09-18 Richard Biener <rguenther@suse.de>
2656 PR tree-optimization/79622
2657 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
2659 (build_cross_bb_scalars_use): Likewise.
2661 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
2663 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
2665 2017-09-18 Alan Modra <amodra@gmail.com>
2668 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
2669 stack_pointer_rtx for count 0. Update comments. Break up
2670 large rtl expression.
2672 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
2674 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
2675 Increase to 20 bytes.
2676 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
2677 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
2678 or avx version of the stub.
2680 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
2683 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
2684 compute the minimum stack alignment. Also update preferred stack
2685 boundary for leaf functions.
2687 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
2689 PR tree-optimization/82228
2690 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
2693 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
2695 * common/config/nds32/nds32-common.c
2696 (nds32_option_optimization_table): Refine formatting.
2697 (nds32_option_optimization_table): Use -fsched-pressure and
2698 -fomit-frame-pointer for specific optimization level.
2700 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
2702 * config/nds32/nds32.c: Refine formatting and comments.
2703 * config/nds32/nds32.h: Likewise.
2704 * config/nds32/nds32.md: Likewise.
2705 * config/nds32/nds32-cost.c: Likewise.
2706 * config/nds32/nds32-isr.c: Likewise.
2707 * config/nds32/nds32-md-auxiliary.c: Likewise.
2708 * config/nds32/nds32-multiple.md: Likewise.
2709 * config/nds32/nds32-predicates.c: Likewise.
2711 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
2712 Jakub Jelinek <jakub@redhat.com>
2714 Add support for -std=c++2a.
2715 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
2717 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
2719 2017-09-15 Steve Ellcey <sellcey@cavium.com>
2722 * doc/extend.texi (Common Function Attributes): Add
2723 references to ARM, AArch64, and S/390 specific attributes.
2724 (Function Specific Option Pragmas): Add AArch64 and S/390
2725 to list of back ends that support the target pragma.
2727 2017-09-15 Nathan Sidwell <nathan@acm.org>
2729 * doc/standards.texi: Fix C++17 description. Update URLs for
2732 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
2734 * common.opt (Wcast-align=strict): New warning option.
2735 * doc/invoke.texi: Document -Wcast-align=strict.
2737 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
2739 * cgraph.h (cgraph_thunk_info): Add comments.
2740 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
2741 assert for VIRTUAL_* arguments stricter.
2743 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
2745 PR tree-optimization/71026
2746 * match.pd: Move RDIV patterns from fold-const.c
2747 * fold-const.c (distribute_real_division): Removed.
2748 (fold_binary_loc): Remove calls to distribute_real_divison.
2750 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2752 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
2753 c++1z and gnu++1z as deprecated. Change other references to
2754 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
2755 Change -Wc++1z-compat to -Wc++17-compat.
2756 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
2757 * dwarf2out.c (highest_c_language): Handle C++17.
2758 (gen_compile_unit_die): Likewise.
2760 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2762 PR rtl-optimization/82192
2763 * combine.c (make_extraction): Don't look through non-paradoxical
2764 SUBREGs or TRUNCATE if pos + len is or might be bigger than
2767 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2768 Alan Hayward <alan.hayward@arm.com>
2769 David Sherwood <david.sherwood@arm.com>
2771 * target.def (function_arg_offset): New hook.
2772 * targhooks.h (default_function_arg_offset): Declare.
2773 * targhooks.c (default_function_arg_offset): New function.
2774 * function.c (locate_and_pad_parm): Use
2775 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
2776 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
2777 (TARGET_FUNCTION_ARG_OFFSET): ...this.
2778 * doc/tm.texi: Regenerate.
2779 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
2780 * config/spu/spu.c (spu_function_arg_offset): New function.
2781 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
2782 * system.h (FUNCTION_ARG_OFFSET): Poison.
2784 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2785 Alan Hayard <alan.hayward@arm.com>
2786 David Sherwood <david.sherwood@arm.com>
2788 * target.def (truly_noop_truncation): New hook.
2789 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
2790 than TRULY_NOOP_TRUNCATION.
2791 * hooks.h (hook_bool_uint_uint_true): Declare.
2792 * hooks.c (hook_bool_uint_uint_true): New function.
2793 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
2794 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
2795 * doc/tm.texi: Regenerate.
2796 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
2797 rather than TRULY_NOOP_TRUNCATION in comments.
2798 (simplify_comparison): Likewise.
2799 (record_truncated_value): Likewise.
2800 * expmed.c (extract_bit_field_1): Likewise.
2801 (extract_split_bit_field): Likewise.
2802 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
2803 instead of TRULY_NOOP_TRUNCATION.
2804 * function.c (assign_parm_setup_block): Likewise.
2805 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
2806 * rtlhooks.c: Include target.h.
2807 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
2808 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
2809 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
2810 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
2811 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
2812 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
2813 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
2814 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
2815 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
2816 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
2817 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
2818 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
2819 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
2820 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
2821 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
2822 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
2823 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
2824 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
2825 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
2826 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
2827 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
2828 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
2829 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
2830 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
2831 * config/mips/mips.c (mips_truly_noop_truncation): New function.
2832 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2833 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2834 rather than TRULY_NOOP_TRUNCATION in comments.
2835 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
2836 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
2837 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
2838 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
2839 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
2840 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
2841 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
2842 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
2843 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
2844 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
2845 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
2846 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2847 rather than TRULY_NOOP_TRUNCATION in comments.
2848 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
2849 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
2850 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
2851 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
2852 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
2853 TRULY_NOOP_TRUNCATION condition.
2854 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
2855 (TRULY_NOOP_TRUNCATION): Delete.
2856 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
2857 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
2858 * config/spu/spu.c (spu_truly_noop_truncation): New function.
2859 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2860 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
2861 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
2862 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
2863 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2864 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2865 rather than TRULY_NOOP_TRUNCATION in comments.
2866 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
2867 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
2868 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
2869 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
2870 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
2871 * system.h (TRULY_NOOP_TRUNCATION): Poison.
2873 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
2876 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
2877 (*cmp_ior): Likewise.
2878 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
2879 (*ior_scc_scc_cmp): Likewise.
2880 (*and_scc_scc): Likewise.
2881 (*and_scc_scc_cmp): Likewise.
2883 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2884 Alan Hayard <alan.hayward@arm.com>
2885 David Sherwood <david.sherwood@arm.com>
2887 * target.def (can_change_mode_class): New hook.
2888 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
2889 (hard_regno_nregs): Likewise.
2890 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
2891 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
2892 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
2893 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
2894 (LOAD_EXTEND_OP): Update accordingly.
2895 * doc/tm.texi: Regenerate.
2896 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
2897 CANNOT_CHANGE_MODE_CLASS.
2898 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
2899 (REG_CAN_CHANGE_MODE_P): ...this new macro.
2900 * combine.c (simplify_set): Update accordingly.
2901 * emit-rtl.c (validate_subreg): Likewise.
2902 * recog.c (general_operand): Likewise.
2903 * regcprop.c (mode_change_ok): Likewise.
2904 * reload1.c (choose_reload_regs): Likewise.
2905 (inherit_piecemeal_p): Likewise.
2906 * rtlanal.c (simplify_subreg_regno): Likewise.
2907 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
2908 instead of CANNOT_CHANGE_MODE_CLASS.
2909 (reload_cse_simplify_operands): Likewise.
2910 * reload.c (push_reload): Use targetm.can_change_mode_class
2911 instead of CANNOT_CHANGE_MODE_CLASS.
2912 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
2913 REG_CANNOT_CHANGE_MODE_P.
2914 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2915 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
2916 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2917 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2918 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2919 (arm_can_change_mode_class): New function.
2920 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
2921 than CANNOT_CHANGE_MODE_CLASS in comments.
2922 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2923 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
2924 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
2925 (ix86_can_change_mode_class): ...this new function, inverting the
2926 sense of the return value.
2927 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2928 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2929 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2930 (ia64_can_change_mode_class): New function.
2931 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2932 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
2933 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
2934 (m32c_can_change_mode_class): ...this new function, inverting the
2935 sense of the return value.
2936 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2937 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2938 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
2939 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
2940 (mips_can_change_mode_class): ...this new function, inverting the
2941 sense of the return value.
2942 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2943 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2944 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2945 (msp430_can_change_mode_class): New function.
2946 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2947 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
2948 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2949 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2950 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2951 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
2952 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2953 (pa_cannot_change_mode_class): Replace with...
2954 (pa_can_change_mode_class): ...this new function, inverting the
2955 sense of the return value.
2956 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
2957 than CANNOT_CHANGE_MODE_CLASS in comments.
2958 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2959 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
2960 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2961 (pdp11_cannot_change_mode_class): Replace with...
2962 (pdp11_can_change_mode_class): ...this new function, inverting the
2963 sense of the return value.
2964 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2965 * config/powerpcspe/powerpcspe-protos.h
2966 (rs6000_cannot_change_mode_class_ptr): Delete.
2967 * config/powerpcspe/powerpcspe.c
2968 (rs6000_cannot_change_mode_class_ptr): Delete.
2969 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2970 (rs6000_option_override_internal): Assign to
2971 targetm.can_change_mode_class instead of
2972 rs6000_cannot_change_mode_class_ptr.
2973 (rs6000_cannot_change_mode_class): Replace with...
2974 (rs6000_can_change_mode_class): ...this new function, inverting the
2975 sense of the return value.
2976 (rs6000_debug_cannot_change_mode_class): Replace with...
2977 (rs6000_debug_can_change_mode_class): ...this new function.
2978 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2979 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
2980 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2981 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2982 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
2984 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
2985 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2986 (rs6000_option_override_internal): Assign to
2987 targetm.can_change_mode_class instead of
2988 rs6000_cannot_change_mode_class_ptr.
2989 (rs6000_cannot_change_mode_class): Replace with...
2990 (rs6000_can_change_mode_class): ...this new function, inverting the
2991 sense of the return value.
2992 (rs6000_debug_cannot_change_mode_class): Replace with...
2993 (rs6000_debug_can_change_mode_class): ...this new function.
2994 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2995 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
2996 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
2997 (s390_can_change_mode_class): ...this new function, inverting the
2998 sense of the return value.
2999 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3000 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3001 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
3002 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3003 (sh_cannot_change_mode_class): Replace with...
3004 (sh_can_change_mode_class): ...this new function, inverting the
3005 sense of the return value.
3006 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3007 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3008 (sparc_can_change_mode_class): New function.
3009 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3010 * config/spu/spu.c (spu_can_change_mode_class): New function.
3011 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3012 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3013 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3014 (visium_can_change_mode_class): New function.
3015 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
3017 2017-09-15 Richard Biener <rguenther@suse.de>
3019 PR tree-optimization/82217
3020 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
3021 but not undefined case.
3023 2017-09-15 Jakub Jelinek <jakub@redhat.com>
3026 * postreload.c (reload_cse_simplify_operands): Skip
3027 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
3029 2017-09-15 Richard Biener <rguenther@suse.de>
3031 PR tree-optimization/68823
3032 * graphite-scop-detection.c (build_alias_set): If we have a
3033 possible dependence check whether we can handle them by just
3034 looking at the DRs DR_ACCESS_FNs.
3035 (build_scops): If build_alias_set fails, fail the SCOP.
3037 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
3039 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
3040 to support float128 built-in functions that require the ISA 3.0
3042 (BU_FLOAT128_3_HW): Likewise.
3043 (SQRTF128): Add support for the IEEE 128-bit square root and fma
3045 (FMAF128): Likewise.
3047 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
3048 support for built-in functions that need the ISA 3.0 IEEE 128-bit
3049 floating point instructions.
3050 (rs6000_invalid_builtin): Likewise.
3051 (rs6000_builtin_mask_names): Likewise.
3052 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
3053 (RS6000_BTM_FLOAT128_HW): Likewise.
3054 (RS6000_BTM_COMMON): Likewise.
3055 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
3057 * doc/extend.texi (RS/6000 built-in functions): Document the
3058 IEEE 128-bit floating point square root and fused multiply-add
3061 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
3063 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
3064 reg (r2) isn't in the set of registers defined in the prologue.
3066 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3067 Alan Hayward <alan.hayward@arm.com>
3068 David Sherwood <david.sherwood@arm.com>
3070 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
3071 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
3072 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
3073 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
3074 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
3076 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
3077 max_vectorization_factor.
3078 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
3080 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3081 Alan Hayward <alan.hayward@arm.com>
3082 David Sherwood <david.sherwood@arm.com>
3084 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
3085 (vect_worthwhile_without_simd_p): Declare.
3086 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
3087 (vectorizable_reduction): Use it.
3088 * tree-vect-stmts.c (vectorizable_shift): Likewise.
3089 (vectorizable_operation): Likewise.
3091 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3092 Alan Hayward <alan.hayward@arm.com>
3093 David Sherwood <david.sherwood@arm.com>
3095 * tree-vectorizer.h (vect_get_num_copies): New function.
3096 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
3097 * tree-vect-loop.c (vectorizable_reduction): Likewise.
3098 (vectorizable_induction): Likewise.
3099 (vectorizable_live_operation): Likewise.
3100 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
3101 (vectorizable_bswap): Likewise.
3102 (vectorizable_call): Likewise.
3103 (vectorizable_conversion): Likewise.
3104 (vectorizable_assignment): Likewise.
3105 (vectorizable_shift): Likewise.
3106 (vectorizable_operation): Likewise.
3107 (vectorizable_store): Likewise.
3108 (vectorizable_load): Likewise.
3109 (vectorizable_condition): Likewise.
3110 (vectorizable_comparison): Likewise.
3111 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
3113 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3114 Alan Hayward <alan.hayward@arm.com>
3115 David Sherwood <david.sherwood@arm.com>
3117 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
3118 of vect_init_vector.
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 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
3125 an inline wrapper that provides a location.
3126 (gimple_build_vector): Likewise.
3127 * gimple-fold.c (gimple_build_vector_from_val): New function.
3128 (gimple_build_vector): Likewise.
3129 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
3130 functions to build the initial value. Always return a gimple value.
3131 (get_initial_defs_for_reduction): Likewise. Only compute
3133 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
3134 vect_init_vector on the results from get_initial_def(s)_for_reduction.
3135 (vectorizable_induction): Use gimple_build_vector rather than
3138 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3139 Alan Hayward <alan.hayward@arm.com>
3140 David Sherwood <david.sherwood@arm.com>
3142 * target.h (vec_perm_indices): New typedef.
3143 (auto_vec_perm_indices): Likewise.
3144 * optabs-query.h: Include target.h
3145 (can_vec_perm_p): Take a vec_perm_indices *.
3146 * optabs-query.c (can_vec_perm_p): Likewise.
3147 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
3148 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
3149 * tree-vect-generic.c (lower_vec_perm): Likewise.
3150 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
3151 (vect_grouped_load_supported): Likewise.
3152 (vect_shift_permute_load_chain): Likewise.
3153 (vect_permute_store_chain): Use auto_vec_perm_indices.
3154 (vect_permute_load_chain): Likewise.
3155 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
3156 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
3157 Update uses of can_vec_perm_p.
3158 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
3159 mode with a number of elements. Take a vec_perm_indices *.
3160 (vect_create_epilog_for_reduction): Update accordingly.
3161 Use auto_vec_perm_indices.
3162 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
3163 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
3164 (vect_transform_slp_perm_load): Likewise.
3165 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
3166 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
3167 (vect_gen_perm_mask_checked): Likewise.
3168 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
3169 (vect_gen_perm_mask_checked): Likewise.
3170 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
3171 (vectorizable_store): Likewise.
3172 (vectorizable_load): Likewise.
3173 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
3174 (vectorizable_bswap): Likewise.
3176 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3177 Alan Hayward <alan.hayward@arm.com>
3178 David Sherwood <david.sherwood@arm.com>
3180 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
3181 * tree.c (build_vector): Likewise.
3182 (build_vector_from_ctor): Update accordingly.
3183 (build_vector_from_val): Likewise.
3184 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
3185 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
3186 * tree-vect-generic.c (add_rshift): Likewise.
3187 (expand_vector_divmod): Likewise.
3188 (optimize_vector_constructor): Likewise.
3189 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
3190 (vect_transform_slp_perm_load): Likewise.
3191 (vect_schedule_slp_instance): Likewise.
3192 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
3193 (vectorizable_call): Likewise.
3194 (vect_gen_perm_mask_any): Likewise. Add elements in order.
3195 * expmed.c (make_tree): Likewise.
3196 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
3197 a vector passed to build_vector.
3198 (fold_convert_const): Likewise.
3199 (exact_inverse): Likewise.
3200 (fold_ternary_loc): Likewise.
3201 (fold_relational_const): Likewise.
3202 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
3203 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
3204 (const_unop): Likewise. Store the reduction accumulator in a
3205 variable rather than an array.
3206 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
3207 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
3208 the new vector, rather than constructing it after the input arrays.
3209 (native_interpret_vector): Use auto_vec<tree> when building
3210 a vector passed to build_vector. Add elements in order.
3211 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
3212 auto_vec<tree> when building a vector passed to build_vector.
3213 (vect_create_epilog_for_reduction): Likewise.
3214 (vectorizable_induction): Likewise.
3215 (get_initial_def_for_reduction): Likewise. Fix indentation of
3217 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
3219 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
3220 passed to build_vector.
3222 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3223 Alan Hayward <alan.hayward@arm.com>
3224 David Sherwood <david.sherwood@arm.com>
3226 * tree-core.h (tree_base::u): Add an "nelts" field.
3227 (tree_vector): Use VECTOR_CST_NELTS as the length.
3228 * tree.c (tree_size): Likewise.
3229 (make_vector): Initialize VECTOR_CST_NELTS.
3230 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
3231 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
3232 TYPE_VECTOR_SUBPARTS.
3233 * expr.c (const_vector_mask_from_tree): Consistently use "units"
3234 as the number of units, setting it from VECTOR_CST_NELTS.
3235 (const_vector_from_tree): Likewise.
3236 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
3237 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
3238 (fold_negate_expr_1): Likewise.
3239 (fold_convert_const): Likewise.
3240 (const_binop): Likewise. Differentiate the number of output and
3242 (const_unop): Likewise.
3243 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
3244 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
3245 in cases that did the opposite.
3247 2017-09-14 Richard Biener <rguenther@suse.de>
3249 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
3252 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
3254 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
3256 2017-09-14 Jakub Jelinek <jakub@redhat.com>
3259 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
3260 if and where to split a bb, except for splitting before debug insn
3261 sequences followed by non-label real insn. Delete debug insns
3262 in between basic blocks.
3264 * combine.c (make_compound_operation_int): Formatting fixes.
3266 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
3267 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3268 * config/netbsd.h (LINK_EH_SPEC): Likewise.
3269 * config/sol2.h (LINK_EH_SPEC): Likewise.
3270 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3271 * config/s390/linux.h (LINK_SPEC): Likewise.
3272 * config/freebsd.h (LINK_EH_SPEC): Likewise.
3273 * config/openbsd.h (LINK_EH_SPEC): Likewise.
3274 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3275 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
3276 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
3277 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3278 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
3279 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
3281 2017-09-13 Jakub Jelinek <jakub@redhat.com>
3283 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
3285 (ENDFILE_LINUX_SPEC): Likewise.
3286 (LINK_EH_SPEC): Likewise.
3287 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
3288 (LINK_OS_LINUX_SPEC32): Likewise.
3289 (LINK_OS_LINUX_SPEC64): Likewise.
3290 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
3291 (LINK_OS_LINUX_SPEC): Likewise.
3293 2017-09-13 Martin Liska <mliska@suse.cz>
3296 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
3297 CASE_HIGH is NULL_TREE.
3299 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
3300 Alan Hayward <alan.hayward@arm.com>
3301 David Sherwood <david.sherwood@arm.com>
3303 * target.def (secondary_memory_needed): New hook.
3304 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
3305 instead of SECONDARY_MEMORY_NEEDED.
3306 (secondary_memory_needed_mode): Likewise.
3307 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
3308 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
3309 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
3310 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
3311 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
3312 * doc/tm.texi: Regenerate.
3313 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
3314 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
3315 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3316 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
3317 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
3318 * config/i386/i386.c (inline_secondary_memory_needed): Put the
3319 mode argument first and change the reg_class arguments to reg_class_t.
3320 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
3321 Make static. Update the call to inline_secondary_memory_needed.
3322 (ix86_register_move_cost): Update the call to
3323 inline_secondary_memory_needed.
3324 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3325 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
3327 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
3328 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
3330 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
3331 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
3332 * config/mips/mips.c (mips_secondary_memory_needed): Make static
3333 and match hook interface. Add comment from mips.h.
3334 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3335 * config/mmix/mmix.md (truncdfsf2): Refer to
3336 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
3338 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
3339 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
3340 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3341 (pa_secondary_memory_needed): New function.
3342 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
3343 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
3344 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3345 (pdp11_secondary_memory_needed): Make static and match hook interface.
3346 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
3347 * config/powerpcspe/powerpcspe-protos.h
3348 (rs6000_secondary_memory_needed_ptr): Delete.
3349 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
3351 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3352 (rs6000_option_override_internal): Assign to
3353 targetm.secondary_memory_needed rather than
3354 rs6000_secondary_memory_needed_ptr.
3355 (rs6000_secondary_memory_needed): Match hook interface.
3356 (rs6000_debug_secondary_memory_needed): Likewise.
3357 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
3358 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
3359 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
3360 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3361 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
3362 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
3364 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
3365 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3366 (rs6000_option_override_internal): Assign to
3367 targetm.secondary_memory_needed rather than
3368 rs6000_secondary_memory_needed_ptr.
3369 (rs6000_secondary_memory_needed): Match hook interface.
3370 (rs6000_debug_secondary_memory_needed): Likewise.
3371 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
3372 * config/s390/s390.c (s390_secondary_memory_needed): New function.
3373 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3374 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
3375 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3376 (sparc_secondary_memory_needed): New function.
3377 * lra-constraints.c (check_and_process_move): Refer to
3378 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
3380 (curr_insn_transform): Likewise.
3381 (process_alt_operands): Use targetm.secondary_memory_needed
3382 instead of TARGET_SECONDARY_MEMORY_NEEDED.
3383 (check_secondary_memory_needed_p): Likewise.
3384 (choose_split_class): Likewise.
3385 * reload.c: Unconditionally include code that was previously
3386 conditional on SECONDARY_MEMORY_NEEDED.
3387 (push_secondary_reload): Use targetm.secondary_memory_needed
3388 instead of TARGET_SECONDARY_MEMORY_NEEDED.
3389 (push_reload): Likewise.
3390 * reload1.c: Unconditionally include code that was previously
3391 conditional on SECONDARY_MEMORY_NEEDED.
3392 (choose_reload_regs): Use targetm.secondary_memory_needed
3393 instead of TARGET_SECONDARY_MEMORY_NEEDED.
3394 (gen_reload): Likewise.
3395 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
3397 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
3398 Alan Hayward <alan.hayward@arm.com>
3399 David Sherwood <david.sherwood@arm.com>
3401 * target.def (secondary_memory_needed_mode): New hook:
3402 * targhooks.c (default_secondary_memory_needed_mode): Declare.
3403 * targhooks.h (default_secondary_memory_needed_mode): New function.
3404 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
3405 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
3406 * doc/tm.texi: Regenerate.
3407 * lra-constraints.c (check_and_process_move): Use
3408 targetm.secondary_memory_needed_mode instead of
3409 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
3410 (curr_insn_transform): Likewise.
3411 * reload.c (get_secondary_mem): Likewise.
3412 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3413 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
3415 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3416 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3417 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
3418 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3419 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
3421 * config/powerpcspe/powerpcspe-protos.h
3422 (rs6000_secondary_memory_needed_mode): Delete.
3423 * config/powerpcspe/powerpcspe.c
3424 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3425 (rs6000_secondary_memory_needed_mode): Make static.
3426 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3427 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
3429 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
3431 (rs6000_secondary_memory_needed_mode): Make static.
3432 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3433 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
3434 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3435 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3436 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
3438 (sparc_secondary_memory_needed_mode): New function.
3439 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
3441 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
3443 * config/aarch64/constraints.md (Umq): New constraint.
3444 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
3446 (mov<mode>): Update condition.
3448 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3450 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
3451 when bitposition is the same.
3453 2017-09-13 Richard Biener <rguenther@suse.de>
3455 * dwarf2out.c (output_die_symbol): Remove.
3456 (output_die): Do not output a DIEs symbol.
3458 2017-09-13 Richard Biener <rguenther@suse.de>
3461 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
3462 default-def to avoid breaking iterator update with the weird
3463 interaction with cgraph_update_edges_for_call_stmt_node.
3465 2017-09-13 Richard Biener <rguenther@suse.de>
3467 * tree-cfg.c (verify_gimple_assign_binary): Add verification
3468 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
3469 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
3470 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
3472 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
3474 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
3475 Disable pc relative literal load irrespective of
3476 TARGET_FIX_ERR_A53_84341 for default.
3478 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
3480 * config/sparc/sparc.c (output_return): Output the source location of
3481 the insn in the delay slot, if any.
3482 (output_sibcall): Likewise.
3484 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
3487 * common.opt (-static-pie): New alias.
3488 (shared): Negate static-pie.
3489 (-no-pie): Update help text.
3491 (static-pie): New option.
3492 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
3493 -static-pie support.
3494 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
3495 (LINK_EH_SPEC): Likewise.
3496 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3497 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
3498 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
3499 * gcc.c (LINK_COMMAND_SPEC): Likewise.
3500 (init_gcc_specs): Likewise.
3501 (init_spec): Likewise.
3502 (display_help): Update help message for -pie.
3503 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
3506 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
3508 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
3509 (movdi_aarch64): Likewise.
3510 (movti_aarch64): Likewise.
3512 2017-09-12 Simon Wright <simon@pushface.org>
3515 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
3516 calculation of the minor version, always output as 0.
3518 2017-09-12 Jakub Jelinek <jakub@redhat.com>
3521 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
3522 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
3523 on it early, rather than manual conversion late. For
3524 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
3525 instead of performing manual conversion.
3527 2017-09-12 Carl Love <cel@us.ibm.com>
3529 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
3530 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
3532 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
3533 VMULOSW): Add definitions.
3534 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
3535 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
3536 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
3537 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
3538 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
3540 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
3542 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
3544 (movti_aarch64): Likewise.
3545 (movdf_aarch64): Likewise.
3546 (movtf_aarch64): Likewise.
3547 (load_pairdi): Likewise.
3548 (store_pairdi): Likewise.
3549 (load_pairdf): Likewise.
3550 (store_pairdf): Likewise.
3551 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
3552 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
3553 (ldr_got_small_<mode>): Likewise.
3554 (ldr_got_small_28k_<mode>): Likewise.
3555 (ldr_got_tiny): Likewise.
3556 * config/aarch64/iterators.md (ldst_sz): New.
3557 (ldpstp_sz): Likewise.
3558 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
3560 (thunderx_load): Split load_8 to load_16.
3561 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
3563 (thunderx2t99_storepair_basic): Split store_8 to store_16.
3564 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
3565 (xgene1_store_pair): Split store_8 to store_16.
3566 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
3567 (falkor_st_0_st_sd): Split store_8 to store_16.
3569 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
3571 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
3572 and store1/2/3/4 to store_4/8/12/16.
3573 * config/aarch64/aarch64.md: Update for rename.
3574 * config/arm/arm.md: Likewise.: Likewise.
3575 * config/arm/arm.c: Likewise.
3576 * config/arm/thumb1.md: Likewise.
3577 * config/arm/thumb2.md: Likewise.
3578 * config/arm/vfp.md: Likewise.
3579 * config/arm/arm-generic.md: Likewise.
3580 * config/arm/arm1020e.md: Likewise.
3581 * config/arm/arm1026ejs.md: Likewise.
3582 * config/arm/arm1136jfs.md: Likewise.
3583 * config/arm/arm926ejs.md: Likewise.
3584 * config/arm/cortex-a15.md: Likewise.
3585 * config/arm/cortex-a17.md: Likewise.
3586 * config/arm/cortex-a5.md: Likewise.
3587 * config/arm/cortex-a53.md: Likewise.
3588 * config/arm/cortex-a57.md: Likewise.
3589 * config/arm/cortex-a7.md: Likewise.
3590 * config/arm/cortex-a8.md: Likewise.
3591 * config/arm/cortex-a9.md: Likewise.
3592 * config/arm/cortex-m4.md: Likewise.
3593 * config/arm/cortex-m7.md: Likewise.
3594 * config/arm/cortex-r4.md: Likewise.
3595 * config/arm/exynos-m1.md: Likewise.
3596 * config/arm/fa526.md: Likewise.
3597 * config/arm/fa606te.md: Likewise.
3598 * config/arm/fa626te.md: Likewise.
3599 * config/arm/fa726te.md: Likewise.
3600 * config/arm/fmp626.md: Likewise.
3601 * config/arm/iwmmxt.md: Likewise.
3602 * config/arm/ldmstm.md: Likewise.
3603 * config/arm/marvell-pj4.md: Likewise.
3604 * config/arm/xgene1.md: Likewise.
3605 * config/aarch64/thunderx.md: Likewise.
3606 * config/aarch64/thunderx2t99.md: Likewise.
3607 * config/aarch64/falkor.md: Likewise.
3609 2017-09-12 Martin Liska <mliska@suse.cz>
3611 * attribs.c (private_lookup_attribute): New function.
3612 * attribs.h (private_lookup_attribute): Declared here.
3613 (lookup_attribute): Called from this place.
3615 2017-09-12 Richard Biener <rguenther@suse.de>
3617 PR tree-optimization/82157
3618 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
3619 stmts with side-effects.
3621 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3622 Alan Hayward <alan.hayward@arm.com>
3623 David Sherwood <david.sherwood@arm.com>
3625 * target.def (hard_regno_nregs): New hook.
3626 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
3627 * targhooks.h (default_hard_regno_nregs): Declare.
3628 * targhooks.c (default_hard_regno_nregs): New function.
3629 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
3630 (TARGET_HARD_REGNO_NREGS): ...this hook.
3631 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
3632 (CLASS_MAX_NREGS): Likewise.
3633 * doc/tm.texi: Regenerate.
3634 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
3635 instead of HARD_REGNO_NREGS.
3636 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
3637 HARD_REGNO_NREGS in the comment.
3638 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
3639 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
3640 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
3641 Return an unsigned int.
3642 (TARGET_HARD_REGNO_NREGS): Redefine.
3643 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
3644 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
3645 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
3646 (arc_hard_regno_nregs): New function.
3647 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
3648 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
3649 (arm_hard_regno_nregs): New function.
3650 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
3651 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
3652 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
3653 (TARGET_HARD_REGNO_NREGS): Redefine.
3654 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
3655 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
3656 (HARD_REGNO_NREGS): Delete.
3657 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
3658 (cr16_hard_regno_nregs): New function.
3659 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
3660 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
3661 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
3662 (cris_hard_regno_nregs): New function.
3663 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
3664 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
3665 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
3666 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
3667 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
3668 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
3669 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
3670 (frv_hard_regno_nregs): Make static. Take and return an
3672 (frv_class_max_nregs): Remove outdated copy of documentation.
3673 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
3674 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
3675 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
3676 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
3677 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
3678 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
3679 (TARGET_HARD_REGNO_NREGS): Redefine.
3680 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
3681 (CLASS_MAX_NREGS): Update comment.
3682 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
3683 (ia64_hard_regno_nregs): New function.
3684 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
3685 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
3686 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
3687 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
3688 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
3690 (m32c_hard_regno_nregs): Likewise. Make static.
3691 (TARGET_HARD_REGNO_NREGS): Redefine.
3692 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
3693 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
3694 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
3695 (m68k_hard_regno_nregs): New function.
3696 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
3697 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
3698 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
3699 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
3700 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
3701 Take and return an unsigned int.
3702 (TARGET_HARD_REGNO_NREGS): Redefine.
3703 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
3704 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
3705 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
3706 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
3707 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
3708 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
3709 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
3710 (msp430_hard_regno_nregs): Make static. Take and return an
3712 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
3713 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
3714 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
3715 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
3716 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
3717 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
3718 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
3719 (TARGET_HARD_REGNO_NREGS): Redefine.
3720 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
3721 (PA_HARD_REGNO_NREGS): ...this.
3722 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
3723 (PA_HARD_REGNO_NREGS): ...this.
3724 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
3725 (pa_hard_regno_nregs): New function.
3726 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
3727 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
3728 (pdp11_hard_regno_nregs): New function.
3729 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
3730 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
3731 (rs6000_hard_regno_nregs_hook): New function.
3732 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
3733 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
3734 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
3735 Take and return an unsigned int. Move earlier in file.
3736 (TARGET_HARD_REGNO_NREGS): Redefine.
3737 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
3738 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
3739 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
3740 (rl78_hard_regno_nregs): Make static. Take and return an
3742 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
3743 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
3744 (rs6000_hard_regno_nregs_hook): New function.
3745 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
3746 * config/rx/rx.c (rx_hard_regno_nregs): New function.
3747 (TARGET_HARD_REGNO_NREGS): Redefine.
3748 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
3749 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
3750 instead of HARD_REGNO_NREGS.
3751 (s390_hard_regno_nregs): New function.
3752 (s390_hard_regno_mode_ok): Add comment from s390.h.
3753 (TARGET_HARD_REGNO_NREGS): Redefine.
3754 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
3755 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
3756 (sh_hard_regno_nregs): New function.
3757 (sh_pass_in_reg_p): Use it.
3758 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
3759 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
3760 (sparc_hard_regno_nregs): New function.
3761 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
3762 * config/spu/spu.c (spu_hard_regno_nregs): New function.
3763 (spu_function_arg_advance): Use it, supplying a valid register number.
3764 (TARGET_HARD_REGNO_NREGS): Redefine.
3765 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
3766 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
3767 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
3768 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
3769 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
3770 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
3771 (CLASS_MAX_NREGS): Remove copy of old documentation.
3772 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
3773 (visium_hard_regno_nregs): New function.
3774 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
3775 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
3776 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
3777 xtensa_hard_regno_nregs): New function.
3778 * system.h (HARD_REGNO_NREGS): Poison.
3780 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3782 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
3783 hard_regno_nregs instead of HARD_REGNO_NREGS.
3784 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
3785 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
3786 (c6x_expand_epilogue): Likewise.
3787 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
3788 (frv_read_iacc_argument): Likewise.
3789 * config/sh/sh.c: Include regs.h.
3790 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
3791 (regs_used): Likewise.
3792 (output_stack_adjust): Likewise.
3793 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
3794 * expmed.c: Include regs.h.
3795 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
3796 * ree.c: Include regs.h.
3797 (combine_reaching_defs): Use hard_regno_nregs instead of
3799 (add_removable_extension): Likewise.
3801 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3803 * regs.h (hard_regno_nregs): Turn into a function.
3804 (end_hard_regno): Update accordingly.
3805 * caller-save.c (setup_save_areas): Likewise.
3806 (save_call_clobbered_regs): Likewise.
3807 (replace_reg_with_saved_mem): Likewise.
3808 (insert_restore): Likewise.
3809 (insert_save): Likewise.
3810 * combine.c (can_change_dest_mode): Likewise.
3811 (move_deaths): Likewise.
3812 (distribute_notes): Likewise.
3813 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
3814 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
3815 (rs6000_split_multireg_move): Likewise.
3816 (rs6000_register_move_cost): Likewise.
3817 (rs6000_memory_move_cost): Likewise.
3818 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
3819 (rs6000_split_multireg_move): Likewise.
3820 (rs6000_register_move_cost): Likewise.
3821 (rs6000_memory_move_cost): Likewise.
3822 * cselib.c (cselib_reset_table): Likewise.
3823 (cselib_lookup_1): Likewise.
3824 * emit-rtl.c (set_mode_and_regno): Likewise.
3825 * function.c (aggregate_value_p): Likewise.
3826 * ira-color.c (setup_profitable_hard_regs): Likewise.
3827 (check_hard_reg_p): Likewise.
3828 (calculate_saved_nregs): Likewise.
3829 (assign_hard_reg): Likewise.
3830 (improve_allocation): Likewise.
3831 (calculate_spill_cost): Likewise.
3832 * ira-emit.c (modify_move_list): Likewise.
3833 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
3834 (ira_hard_reg_in_set_p): Likewise.
3835 * ira.c (setup_reg_mode_hard_regset): Likewise.
3836 (clarify_prohibited_class_mode_regs): Likewise.
3837 (check_allocation): Likewise.
3838 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3839 (lra_setup_reg_renumber): Likewise.
3840 (setup_try_hard_regno_pseudos): Likewise.
3841 (spill_for): Likewise.
3842 (assign_hard_regno): Likewise.
3843 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
3844 * lra-constraints.c (in_class_p): Likewise.
3845 (lra_constraint_offset): Likewise.
3846 (simplify_operand_subreg): Likewise.
3847 (lra_constraints): Likewise.
3848 (split_reg): Likewise.
3849 (split_if_necessary): Likewise.
3850 (invariant_p): Likewise.
3851 (inherit_in_ebb): Likewise.
3852 * lra-lives.c (process_bb_lives): Likewise.
3853 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
3854 (get_hard_regs): Likewise.
3855 (do_remat): Likewise.
3856 * lra-spills.c (assign_spill_hard_regs): Likewise.
3857 * mode-switching.c (create_pre_exit): Likewise.
3858 * postreload.c (reload_combine_recognize_pattern): Likewise.
3859 * recog.c (peep2_find_free_register): Likewise.
3860 * regcprop.c (kill_value_regno): Likewise.
3861 (set_value_regno): Likewise.
3862 (copy_value): Likewise.
3863 (maybe_mode_change): Likewise.
3864 (find_oldest_value_reg): Likewise.
3865 (copyprop_hardreg_forward_1): Likewise.
3866 * regrename.c (check_new_reg_p): Likewise.
3867 (regrename_do_replace): Likewise.
3868 * reload.c (push_reload): Likewise.
3869 (combine_reloads): Likewise.
3870 (find_dummy_reload): Likewise.
3871 (operands_match_p): Likewise.
3872 (find_reloads): Likewise.
3873 (find_equiv_reg): Likewise.
3874 (reload_adjust_reg_for_mode): Likewise.
3875 * reload1.c (count_pseudo): Likewise.
3876 (count_spilled_pseudo): Likewise.
3877 (find_reg): Likewise.
3878 (clear_reload_reg_in_use): Likewise.
3879 (free_for_value_p): Likewise.
3880 (allocate_reload_reg): Likewise.
3881 (choose_reload_regs): Likewise.
3882 (reload_adjust_reg_for_temp): Likewise.
3883 (emit_reload_insns): Likewise.
3884 (delete_output_reload): Likewise.
3885 * rtlanal.c (subreg_get_info): Likewise.
3886 * sched-deps.c (sched_analyze_reg): Likewise.
3887 * sel-sched.c (init_regs_for_mode): Likewise.
3888 (mark_unavailable_hard_regs): Likewise.
3889 (choose_best_reg_1): Likewise.
3890 (verify_target_availability): Likewise.
3891 * valtrack.c (dead_debug_insert_temp): Likewise.
3892 * var-tracking.c (track_loc_p): Likewise.
3893 (emit_note_insn_var_location): Likewise.
3894 * varasm.c (make_decl_rtl): Likewise.
3895 * reginfo.c (choose_hard_reg_mode): Likewise.
3896 (init_reg_modes_target): Refer directly to
3897 this_target_regs->x_hard_regno_nregs.
3899 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3901 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
3902 instead of hard_regno_nregs.
3904 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3906 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
3907 end_hard_regno instead of hard_regno_nregs.
3908 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
3909 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
3910 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
3911 * ira-color.c (improve_allocation): Likewise.
3912 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3913 * lra-lives.c (mark_regno_live): Likewise.
3914 (mark_regno_dead): Likewise.
3915 * lra-remat.c (operand_to_remat): Likewise.
3916 * lra.c (collect_non_operand_hard_regs): Likewise.
3917 * postreload.c (reload_combine_note_store): Likewise.
3918 (move2add_valid_value_p): Likewise.
3919 * reload.c (regno_clobbered_p): Likewise.
3921 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3923 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
3925 * config/v850/v850.c (v850_reorg): Likewise.
3926 * reload.c (refers_to_regno_for_reload_p): Likewise.
3927 (find_equiv_reg): Likewise.
3928 * reload1.c (reload_reg_reaches_end_p): Likewise.
3930 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3932 * caller-save.c (add_used_regs): Use REG_NREGS instead of
3934 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
3935 * config/arm/arm.c (output_move_neon): Likewise.
3936 (arm_attr_length_move_neon): Likewise.
3937 (neon_split_vcombine): Likewise.
3938 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
3939 (c6x_mark_reg_written): Likewise.
3940 (c6x_dwarf_register_span): Likewise.
3941 * config/i386/i386.c (ix86_save_reg): Likewise.
3942 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
3943 (rws_access_reg): Likewise.
3944 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
3945 * mode-switching.c (create_pre_exit): Likewise.
3946 * ree.c (combine_reaching_defs): Likewise.
3947 (add_removable_extension): Likewise.
3948 * regcprop.c (find_oldest_value_reg): Likewise.
3949 (copyprop_hardreg_forward_1): Likewise.
3950 * reload.c (reload_inner_reg_of_subreg): Likewise.
3951 (push_reload): Likewise.
3952 (combine_reloads): Likewise.
3953 (find_dummy_reload): Likewise.
3954 (reload_adjust_reg_for_mode): Likewise.
3955 * reload1.c (find_reload_regs): Likewise.
3956 (forget_old_reloads_1): Likewise.
3957 (reload_reg_free_for_value_p): Likewise.
3958 (reload_adjust_reg_for_temp): Likewise.
3959 (emit_reload_insns): Likewise.
3960 (delete_output_reload): Likewise.
3961 * sel-sched.c (choose_best_reg_1): Likewise.
3962 (choose_best_pseudo_reg): Likewise.
3964 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3965 Alan Hayward <alan.hayward@arm.com>
3966 David Sherwood <david.sherwood@arm.com>
3968 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
3969 * target.def (slow_unaligned_access): New hook.
3970 * targhooks.h (default_slow_unaligned_access): Declare.
3971 * targhooks.c (default_slow_unaligned_access): New function.
3972 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
3973 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
3974 * doc/tm.texi: Regenerate.
3975 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
3976 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
3977 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
3979 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
3980 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
3982 (rs6000_slow_unaligned_access): New function.
3983 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
3984 (expand_block_compare): Likewise.
3985 (expand_strn_compare): Likewise.
3986 (rs6000_rtx_costs): Likewise.
3987 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
3988 (riscv_slow_unaligned_access): Likewise.
3989 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
3990 (riscv_slow_unaligned_access_p): ...this and make static.
3991 (riscv_option_override): Update accordingly.
3992 (riscv_slow_unaligned_access): New function.
3993 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
3994 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
3995 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
3996 (rs6000_slow_unaligned_access): New function.
3997 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
3998 (rs6000_rtx_costs): Likewise.
3999 * config/rs6000/rs6000-string.c (expand_block_compare)
4000 (expand_strn_compare): Use targetm.slow_unaligned_access instead
4001 of SLOW_UNALIGNED_ACCESS.
4002 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
4003 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
4004 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
4005 of SLOW_UNALIGNED_ACCESS.
4006 * expmed.c (simple_mem_bitfield_p): Likewise.
4007 * expr.c (alignment_for_piecewise_move): Likewise.
4008 (emit_group_load_1): Likewise.
4009 (emit_group_store): Likewise.
4010 (copy_blkmode_from_reg): Likewise.
4011 (emit_push_insn): Likewise.
4012 (expand_assignment): Likewise.
4013 (store_field): Likewise.
4014 (expand_expr_real_1): Likewise.
4015 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
4016 * lra-constraints.c (simplify_operand_subreg): Likewise.
4017 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
4018 * gimple-ssa-store-merging.c: Likewise in block comment at start
4020 * tree-ssa-strlen.c: Include target.h.
4021 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
4022 of SLOW_UNALIGNED_ACCESS.
4023 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
4025 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4027 PR rtl-optimization/82185
4028 * expmed.c (emit_store_flag_int): Only test tem if it has been
4031 2017-09-12 Richard Biener <rguenther@suse.de>
4034 * match.pd ((FTYPE) N CMP CST): Fix typo.
4036 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
4038 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
4040 (mips_near_type_p): Add 'short_call' attribute as a synonym
4042 * doc/extend.texi (short_call): Document new function attribute.
4044 2017-09-12 Jakub Jelinek <jakub@redhat.com>
4047 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
4048 assertion check that in the condition.
4049 (get_atomic_generic_size): Likewise. Before testing if parameter
4050 has pointer type, if it has array type, call for C++
4051 default_conversion to perform array-to-pointer conversion.
4053 2017-09-12 Richard Biener <rguenther@suse.de>
4055 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
4056 for operations we cannot scalarize.
4058 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
4060 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
4061 vectors heap vectors. Clean up comments.
4062 Make visited_bbs a reference.
4063 (profitable_jump_thread_path): Make GC
4064 vectors heap vectors. Clean up comments.
4066 (convert_and_register_jump_thread_path): Make GC vectors heap
4068 (check_subpath_and_update_thread_path): Same. Clean up comments.
4069 Make visited_bbs a reference.
4070 (handle_phi): Abstract common code to to
4071 register_jump_thread_path_if_profitable.
4072 Rename VAR_BB to DEF_BB.
4074 Make GC vectors heap vectors.
4075 Make visited_bbs a reference.
4076 (handle_assignment): Same.
4077 (register_jump_thread_path_if_profitable): New.
4078 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
4080 Make GC vectors heap vectors. Clean up comments.
4081 Make visited_bbs a reference.
4082 (find_jump_threads_backwards): Make visited_bbs live in the stack.
4083 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
4086 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
4089 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
4090 words of E_DImode object are reachable by xtensa_uimm8x4 access.
4092 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
4094 Revert r251800 and r251799.
4096 2017-09-11 Martin Jambor <mjambor@suse.cz>
4099 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
4100 arguments in advance.
4101 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
4102 use it to find predecessor edges.
4103 (naive_outof_ssa): Collect vector of predecessors.
4105 2017-09-08 Jason Merrill <jason@redhat.com>
4107 PR c++/70029 - ICE with ref-qualifier and -flto
4108 * langhooks.h (struct lang_hooks_for_types): Add
4109 copy_lang_qualifiers.
4110 * attribs.c (build_type_attribute_qual_variant): Use it.
4111 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
4113 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
4114 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
4116 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
4119 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
4120 (*mulsi3_sp64): New instruction.
4121 (mulsi3): New expander.
4123 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
4125 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
4127 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4129 * sancov.c: Include memmodel.h.
4131 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
4134 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
4137 2017-09-07 Carl Love <cel@us.ibm.com>
4139 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
4140 the sldi instruction.
4142 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
4144 * sancov.c: Include tm_p.h.
4146 2017-09-07 Jakub Jelinek <jakub@redhat.com>
4149 * output.h (switch_to_other_text_partition): New declaration.
4150 * varasm.c (switch_to_other_text_partition): New function.
4151 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
4152 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
4153 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
4154 to the other text partition before emitting LCL label and switch back
4155 after emitting the word after it.
4157 2017-09-07 Richard Biener <rguenther@suse.de>
4159 * passes.def (pass_split_crit_edges): Remove instance before PRE.
4160 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
4161 critical edges here, after loop init.
4162 (pass_data_pre): Remove PROP_no_crit_edges flags.
4163 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
4164 for valueization of call args to avoid leaking VN_TOP.
4165 (visit_use): Assert we do not visit default defs.
4166 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
4167 Use error_mark_node to more easily detect leaking VN_TOP.
4168 All default-defs are varying, not VN_TOP. Mark them visited.
4169 (run_scc_vn): Make code match comment.
4171 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
4173 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
4174 OPTION_MASK_FLOAT128_KEYWORD.
4175 (POWERPC_MASKS): Likewise.
4176 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
4177 support for the -mfloat128-type option, and make -mfloat128
4178 default on PowerPC Linux systems. Define or undefine
4179 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
4180 Define __float128 to be __ieee128 if IEEE 128-bit support is
4181 enabled, or undefine it.
4182 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
4183 Delete defining __FLOAT128_TYPE__.
4184 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
4185 -mfloat128-type option and make -mfloat128 default on PowerPC
4187 (TARGET_FLOAT128_TYPE): Likewise.
4188 (-mfloat128-type): Likewise.
4189 * config/rs6000/rs6000.c (rs6000_option_override_internal):
4190 Delete the -mfloat128-type option and make -mfloat128 default on
4191 PowerPC Linux systems. Always use __ieee128 to be the keyword for
4192 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
4193 128-bit floating point is enabled. Change tests from using
4194 -mfloat128-type to -mfloat128.
4195 (rs6000_mangle_type): Use the correct mangling for the __float128
4196 type even if normal long double is restricted to 64-bits.
4197 (floatn_mode): Enable the _Float128 type by default on VSX Linux
4199 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
4200 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
4201 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
4203 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
4204 documentation for -mfloat128.
4206 2017-09-06 Olivier Hainque <hainque@adacore.com>
4208 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
4210 2017-09-06 Wish Wu <wishwu007@gmail.com>
4211 Jakub Jelinek <jakub@redhat.com>
4213 * asan.c (initialize_sanitizer_builtins): Add
4214 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
4215 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
4216 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
4217 BT_FN_VOID_UINT64_PTR variables.
4218 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
4219 (BT_FN_VOID_UINT16_UINT16): Likewise.
4220 (BT_FN_VOID_UINT32_UINT32): Likewise.
4221 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
4222 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
4223 (BT_FN_VOID_UINT64_PTR): Likewise.
4224 * common.opt (flag_sanitize_coverage): New variable.
4225 (fsanitize-coverage=trace-pc): Remove.
4226 (fsanitize-coverage=): Add.
4227 * flag-types.h (enum sanitize_coverage_code): New enum.
4228 * fold-const.c (fold_range_test): Disable non-short-circuit
4229 optimization if flag_sanitize_coverage.
4230 (fold_truth_andor): Likewise.
4231 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
4232 * opts.c (COVERAGE_SANITIZER_OPT): Define.
4233 (coverage_sanitizer_opts): New array.
4234 (get_closest_sanitizer_option): Add OPTS argument, handle also
4235 OPT_fsanitize_coverage_.
4236 (parse_sanitizer_options): Adjusted to also handle
4237 OPT_fsanitize_coverage_.
4238 (common_handle_option): Add OPT_fsanitize_coverage_.
4239 * sancov.c (instrument_comparison, instrument_switch): New function.
4240 (sancov_pass): Add trace-cmp support.
4241 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
4242 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
4243 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
4244 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
4245 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
4246 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
4247 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
4248 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
4249 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
4250 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
4252 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
4254 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
4255 error. Only quit immediately if parsing is complete.
4256 (BEGIN): Initialize fatal_err and parse_done.
4257 (begin fpu, end fpu): Check number of arguments.
4258 (begin arch, end arch): Likewise.
4259 (begin cpu, end cpu): Likewise.
4260 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
4261 (optalias): Likewise.
4263 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
4265 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
4266 * config/arm/arm-isa.h: Delete. Move definitions to ...
4267 * arm-cpus.in: ... here. Use new feature and fgroup values.
4268 * config/arm/arm.c (arm_option_override): Use lower case for feature
4270 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
4271 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
4272 * config/arm/parsecpu.awk (END): Add new command 'isa'.
4274 (print_isa_bits_for): New function.
4275 (gen_isa): New function.
4276 (gen_comm_data): Use print_isa_bits_for.
4277 (define feature): New keyword.
4278 (define fgroup): New keyword.
4279 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
4280 (arm-isa.h): Add rule to generate file.
4281 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
4282 case for feature bit names.
4284 2017-09-06 Richard Biener <rguenther@suse.de>
4286 * tree-ssa-pre.c (NECESSARY): Remove.
4287 (create_expression_by_pieces): Do not touch pass-local flags.
4288 (insert_into_preds_of_block): Likewise.
4289 (do_pre_regular_insertion): Likewise.
4290 (eliminate_insert): Likewise.
4291 (eliminate_dom_walker::before_dom_children): Likewise.
4292 (fini_eliminate): Do not look at inserted_exprs.
4293 (mark_operand_necessary): Remove.
4294 (remove_dead_inserted_code): Replace with simple work-list
4295 algorithm based on inserted_exprs and SSA uses.
4296 (pass_pre::execute): Re-order fini_eliminate and
4297 remove_dead_inserted_code.
4299 2017-09-06 Olivier Hainque <hainque@adacore.com>
4301 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
4302 for VxWorks 7. Adjust surrounding comments.
4304 2017-09-06 Richard Biener <rguenther@suse.de>
4306 * gimple-ssa-strength-reduction.c
4307 (find_candidates_dom_walker::before_dom_children): Also allow
4310 2017-09-06 Richard Biener <rguenther@suse.de>
4312 PR tree-optimization/82108
4313 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
4314 for gap in the non-permutation SLP case.
4316 2017-09-06 Martin Jambor <mjambor@suse.cz>
4318 PR tree-optimization/82078
4319 * tree-sra.c (sort_and_splice_var_accesses): Move call to
4320 add_access_to_work_queue...
4321 (build_accesses_from_assign): ...here.
4322 (propagate_all_subaccesses): Make sure racc is the group
4323 representative, if there is one.
4325 2017-09-06 Jakub Jelinek <jakub@redhat.com>
4328 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
4331 2017-09-06 Richard Biener <rguenther@suse.de>
4333 * gimple-ssa-strength-reduction.c
4334 (find_candidates_dom_walker::before_doom_children): Use a
4335 type and not a mode check.
4337 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
4340 * config/arm/predicates.md (arm_general_adddi_operand): Create new
4342 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
4344 2017-09-05 Jeff Law <law@redhat.com>
4346 PR tree-optimization/64910
4347 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
4348 cases where we have 3 or more operands.
4350 2017-09-05 Jakub Jelinek <jakub@redhat.com>
4353 * omp-low.c (lower_omp_for): Recompute tree invariant if
4354 gimple_omp_for_initial/final is ADDR_EXPR.
4357 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
4358 into gimple val before gimplification fo the COND_EXPR.
4360 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
4362 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
4363 REGION_COPY argument.
4364 (thread_through_all_blocks): Remove unused argument to
4365 duplicate_thread_path.
4367 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4368 Alan Hayward <alan.hayward@arm.com>
4369 David Sherwood <david.sherwood@arm.com>
4371 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
4372 Take a scalar_mode rather than a machine_mode.
4373 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
4374 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
4375 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
4376 (aarch64_gen_adjusted_ldpstp): Likewise.
4377 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
4379 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4380 Alan Hayward <alan.hayward@arm.com>
4381 David Sherwood <david.sherwood@arm.com>
4383 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
4384 Take a scalar_int_mode instead of a machine_mode.
4385 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
4386 (aarch64_output_scalar_simd_mov_immediate): Likewise.
4387 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
4388 (aarch64_simd_attr_length_rglist): Delete.
4389 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
4390 a scalar_int_mode instead of a machine_mode.
4391 (aarch64_add_offset): Likewise.
4392 (aarch64_internal_mov_immediate): Likewise
4393 (aarch64_add_constant_internal): Likewise.
4394 (aarch64_add_constant): Likewise.
4395 (aarch64_movw_imm): Likewise.
4396 (aarch64_rtx_arith_op_extract_p): Likewise.
4397 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
4398 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
4399 Remove assert that the mode isn't a vector.
4400 (aarch64_output_scalar_simd_mov_immediate): Likewise.
4401 (aarch64_expand_mov_immediate): Update calls after above changes.
4402 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
4403 (aarch64_and_bitmask_imm): Check for scalar integer modes.
4404 (aarch64_move_imm): Likewise.
4405 (aarch64_can_const_movi_rtx_p): Likewise.
4406 (aarch64_strip_extend): Likewise.
4407 (aarch64_extr_rtx_p): Likewise.
4408 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
4409 a CONST_INT when the mode parameter is VOIDmode.
4410 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
4412 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4414 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
4415 * stor-layout.c (bitwise_mode_for_mode): Likewise.
4416 (bitwise_type_for_mode): Update accordingly.
4418 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4420 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
4421 * stor-layout.c (mode_for_size_tree): Likewise.
4422 (mode_for_array): Update accordingly.
4423 (layout_decl): Likewise.
4424 (compute_record_mode): Likewise. Only set the mode once.
4426 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4428 * target.def (get_mask_mode): Change return type to opt_mode.
4430 * doc/tm.texi: Regenerate.
4431 * targhooks.h (default_get_mask_mode): Return an opt_mode.
4432 * targhooks.c (default_get_mask_mode): Likewise.
4433 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
4434 * optabs-query.c (can_vec_mask_load_store_p): Update use of
4435 targetm.get_mask_mode.
4436 * tree.c (build_truth_vector_type): Likewise.
4438 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4440 * machmode.h (mode_for_vector): Return an opt_mode.
4441 * stor-layout.c (mode_for_vector): Likewise.
4442 (mode_for_int_vector): Update accordingly.
4443 (layout_type): Likewise.
4444 * config/i386/i386.c (emit_memmov): Likewise.
4445 (ix86_expand_set_or_movmem): Likewise.
4446 (ix86_expand_vector_init): Likewise.
4447 (ix86_get_mask_mode): Likewise.
4448 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
4450 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
4451 * expmed.c (extract_bit_field_1): Likewise.
4452 * expr.c (expand_expr_real_2): Likewise.
4453 * optabs-query.c (can_vec_perm_p): Likewise.
4454 (can_vec_mask_load_store_p): Likewise.
4455 * optabs.c (expand_vec_perm): Likewise.
4456 * targhooks.c (default_get_mask_mode): Likewise.
4457 * tree-vect-stmts.c (vectorizable_store): Likewise.
4458 (vectorizable_load): Likewise.
4459 (get_vectype_for_scalar_type_and_size): Likewise.
4461 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4463 * machmode.h (mode_for_int_vector): New function.
4464 * stor-layout.c (mode_for_int_vector): Likewise.
4465 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
4466 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
4467 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
4468 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
4469 (s390_expand_vcond): Likewise.
4471 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4473 * machmode.h (opt_machine_mode): New type.
4474 (opt_mode<T>): Allow construction from anything that can be
4476 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
4477 (mode_for_size): Return an opt_machine_mode.
4478 * stor-layout.c (mode_for_size): Likewise.
4479 (mode_for_size_tree): Update call accordingly.
4480 (bitwise_mode_for_mode): Likewise.
4481 (make_fract_type): Likewise.
4482 (make_accum_type): Likewise.
4483 * caller-save.c (replace_reg_with_saved_mem): Update call
4485 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4486 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4487 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4488 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4489 * expmed.c (extract_bit_field_1): Likewise.
4490 * reload.c (get_secondary_mem): Likewise.
4491 * varasm.c (assemble_integer): Likewise.
4492 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
4495 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4497 * machmode.h (decimal_float_mode_for_size): New function.
4498 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
4499 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
4500 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
4501 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
4502 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
4503 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
4505 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4507 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
4508 (get_builtin_sync_mode): Likewise.
4509 (expand_ifn_atomic_compare_exchange): Likewise.
4510 (expand_builtin_atomic_clear): Likewise.
4511 (expand_builtin_atomic_test_and_set): Likewise.
4512 (fold_builtin_atomic_always_lock_free): Likewise.
4513 * calls.c (compute_argument_addresses): Likewise.
4514 (emit_library_call_value_1): Likewise.
4515 (store_one_arg): Likewise.
4516 * combine.c (combine_instructions): Likewise.
4517 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
4518 * config/arm/arm.c (arm_function_value): Likewise.
4519 (aapcs_allocate_return_reg): Likewise.
4520 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
4521 * config/i386/i386.c (construct_container): Likewise.
4522 (ix86_gimplify_va_arg): Likewise.
4523 (ix86_expand_sse_cmp): Likewise.
4524 (emit_memmov): Likewise.
4525 (emit_memset): Likewise.
4526 (expand_small_movmem_or_setmem): Likewise.
4527 (ix86_expand_pextr): Likewise.
4528 (ix86_expand_pinsr): Likewise.
4529 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
4530 * config/microblaze/microblaze.c (microblaze_block_move_straight):
4532 * config/mips/mips.c (mips_function_value_1) Likewise.
4533 (mips_block_move_straight): Likewise.
4534 (mips_expand_ins_as_unaligned_store): Likewise.
4535 * config/powerpcspe/powerpcspe.c
4536 (rs6000_darwin64_record_arg_advance_flush): Likewise.
4537 (rs6000_darwin64_record_arg_flush): Likewise.
4538 * config/rs6000/rs6000.c
4539 (rs6000_darwin64_record_arg_advance_flush): Likewise.
4540 (rs6000_darwin64_record_arg_flush): Likewise.
4541 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
4542 (sparc_function_value_1): Likewise.
4543 * config/spu/spu.c (adjust_operand): Likewise.
4544 (spu_emit_branch_or_set): Likewise.
4545 (arith_immediate_p): Likewise.
4546 * emit-rtl.c (gen_lowpart_common): Likewise.
4547 * expr.c (expand_expr_real_1): Likewise.
4548 * function.c (assign_parm_setup_block): Likewise.
4549 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
4550 * reload1.c (alter_reg): Likewise.
4551 * stor-layout.c (mode_for_vector): Likewise.
4552 (layout_type): Likewise.
4554 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4556 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
4557 (spu_convert_move): Likewise.
4558 * lower-subreg.c (resolve_simple_move): Likewise.
4560 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4563 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
4564 define_insn to a define_expand.
4565 (altivec_vsum2sws_direct): New define_insn.
4566 (altivec_vsumsws): Convert from a define_insn to a define_expand.
4568 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
4570 * config/arm/arm.c (arm_option_params_internal): Improve setting of
4573 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
4578 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
4579 realign stack if stack alignment needed is less than incoming
4582 2017-09-05 Marek Polacek <polacek@redhat.com>
4585 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
4588 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
4590 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
4592 2017-09-05 Richard Biener <rguenther@suse.de>
4594 PR tree-optimization/82084
4595 * fold-const.c (can_native_encode_string_p): Handle wide characters.
4597 2017-09-05 Richard Biener <rguenther@suse.de>
4599 PR tree-optimization/82102
4600 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
4602 2017-09-05 Martin Liska <mliska@suse.cz>
4604 PR tree-optimization/82032
4605 * tree-cfg.c (generate_range_test): New function.
4606 * tree-cfg.h (generate_range_test): Declared here.
4607 * tree-cfgcleanup.c (convert_single_case_switch): New function.
4608 (cleanup_control_expr_graph): Use it.
4609 * tree-switch-conversion.c (try_switch_expansion): Remove
4611 (emit_case_nodes): Use generate_range_test.
4613 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
4616 * config/i386/i386.md (*<btsc><mode>_mask): Add
4617 TARGET_USE_BT to insn constraint.
4618 (*btr<mode>_mask): Ditto.
4620 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
4622 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
4623 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
4625 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
4628 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
4629 TARGET_NEON and TARGET_IWMMXT.
4630 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
4631 TARGET_NEON and TARGET_IWMMXT.
4632 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
4634 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
4636 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
4637 (ix86_rewrite_tls_address): Ditto.
4638 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
4639 (ix86_rewrite_tls_address_1): Ditto.
4640 (ix86_rewrite_tls_address): Ditto.
4641 * config/i386/predicates.md (tls_address_pattern): New predicate.
4642 * config/i386/i386.md (TLS address splitter): New splitter.
4644 2017-09-04 Richard Biener <rguenther@suse.de>
4646 PR tree-optimization/82084
4647 * fold-const.h (can_native_encode_string_p): Declare.
4648 * fold-const.c (can_native_encode_string_p): Factor out from ...
4649 (native_encode_string): ... here.
4650 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
4651 vectorizing stores from constants we later cannot handle.
4653 2017-09-04 Marek Polacek <polacek@redhat.com>
4656 * doc/invoke.texi: Update -Wtautological-compare documentation.
4658 2017-09-04 Jeff Law <law@redhat.com>
4660 PR tree-optimization/64910
4661 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
4662 swap the first and last operand if the last is a constant.
4664 2017-09-04 Marek Polacek <polacek@redhat.com>
4667 * convert.c (do_narrow): When sanitizing signed integer overflows,
4668 bail out for signed types.
4669 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
4671 2017-09-04 Richard Biener <rguenther@suse.de>
4673 PR tree-optimization/82060
4674 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
4675 Move devirtualization after stmt folding and before EH/AB/noreturn
4676 cleanup to get the stmt refs canonicalized. Use a bool instead
4677 of gimple_modified_p since that doesn't work for NOPs. Schedule
4678 NOPs generated by folding for removal.
4680 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4681 Alan Hayward <alan.hayward@arm.com>
4682 David Sherwood <david.sherwood@arm.com>
4684 * coretypes.h (pad_direction): New enum.
4685 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
4686 (FUNCTION_ARG_PADDING): Likewise.
4687 * target.def (function_arg_padding): New hook.
4688 * targhooks.h (default_function_arg_padding): Declare.
4689 * targhooks.c (default_function_arg_padding): New function.
4690 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
4691 (TARGET_FUNCTION_ARG_PADDING): ...this.
4692 * doc/tm.texi: Regenerate.
4693 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
4694 instead of direction.
4695 (compute_argument_addresses): Likewise.
4696 (load_register_parameters): Likewise.
4697 (emit_library_call_value_1): Likewise.
4698 (store_one_arg): Use targetm.calls.function_arg_padding instead
4699 of FUNCTION_ARG_PADDING.
4700 (must_pass_in_stack_var_size_or_pad): Likewise.
4701 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
4702 (emit_group_store): Likewise.
4703 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
4704 instead of FUNCTION_ARG_PADDING.
4705 (emit_push_insn): Likewise, and propagate enum change throughout
4707 * function.h (direction): Delete.
4708 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
4710 * function.c (assign_parm_find_stack_rtl): Likewise.
4711 (assign_parm_setup_block_p): Likewise.
4712 (assign_parm_setup_block): Likewise.
4713 (gimplify_parameters): Likewise.
4714 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
4715 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
4717 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
4718 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
4719 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
4720 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
4721 (aarch64_function_arg_padding): ...this new function.
4722 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
4723 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4724 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
4725 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
4726 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
4727 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4728 (arm_pad_arg_upward): Replace with...
4729 (arm_function_arg_padding): ...this new function.
4730 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
4732 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
4733 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
4734 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4735 (ia64_hpux_function_arg_padding): Replace with...
4736 (ia64_function_arg_padding): ...this new function. Use pad_direction
4737 instead of direction. Check for TARGET_HPUX.
4738 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
4739 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4740 (iq2000_function_arg_padding): New function.
4741 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
4742 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
4743 (mips_function_arg_padding): ...this new function.
4744 (mips_pad_reg_upward): Update accordingly.
4745 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4746 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
4747 targetm.calls.function_arg_padding.
4748 (FUNCTION_ARG_PADDING): Delete.
4749 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
4750 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
4751 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
4752 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
4753 (nios2_block_reg_padding): Return pad_direction instead of direction.
4754 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
4755 instead of direction.
4756 (nios2_function_arg_padding): Likewise. Make static.
4757 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4758 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
4759 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
4760 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
4761 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4762 (pa_function_arg_padding): Make static. Return pad_direction instead
4764 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
4765 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
4766 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
4767 instead of direction. Use targetm.calls.function_arg_padding.
4768 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
4769 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
4770 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
4771 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
4772 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
4774 (function_arg_padding): Rename to...
4775 (rs6000_function_arg_padding): ...this. Make static. Return
4776 pad_direction instead of direction.
4777 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
4778 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
4779 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
4780 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
4781 instead of direction. Use targetm.calls.function_arg_padding.
4782 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
4783 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
4784 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
4785 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
4786 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4787 (function_arg_padding): Rename to...
4788 (rs6000_function_arg_padding): ...this. Make static. Return
4789 pad_direction instead of direction.
4790 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
4791 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
4792 * config/s390/s390.c (s390_function_arg_padding): New function.
4793 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4794 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
4795 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
4796 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4797 (function_arg_padding): Rename to...
4798 (sparc_function_arg_padding): ...this. Make static. Return
4799 pad_direction instead of direction.
4800 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
4801 * config/spu/spu.c (spu_function_arg_padding): New function.
4802 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4803 * system.h (FUNCTION_ARG_PADDING): Poison.
4805 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4806 Alan Hayward <alan.hayward@arm.com>
4807 David Sherwood <david.sherwood@arm.com>
4809 * target.def (modes_tieable_p): New hook.
4810 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
4811 (TARGET_MODES_TIEABLE_P): ...this.
4812 * doc/tm.texi.in: Regenerate.
4813 * hooks.h (hook_bool_mode_mode_true): Declare.
4814 * hooks.c (hook_bool_mode_mode_true): New function.
4815 * combine.c (subst): Use targetm.modes_tieable_p instead of
4817 * dse.c (find_shift_sequence): Likewise.
4818 * expmed.c (extract_low_bits): Likewise.
4819 * lower-subreg.c: Include target.h.
4820 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
4822 * rtlanal.c (rtx_cost): Likewise.
4823 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
4824 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
4825 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
4826 (TARGET_MODES_TIEABLE_P): Redefine.
4827 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
4828 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
4829 (TARGET_MODES_TIEABLE_P): Redefine.
4830 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
4831 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
4832 (arc_modes_tieable_p): New function.
4833 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
4834 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
4835 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
4836 (arm_modes_tieable_p): Make static.
4837 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
4838 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
4839 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
4840 (TARGET_MODES_TIEABLE_P): Redefine.
4841 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
4842 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
4843 (TARGET_MODES_TIEABLE_P): Redefine.
4844 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
4845 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
4846 (cr16_modes_tieable_p): New function.
4847 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
4848 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
4849 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
4850 (TRULY_NOOP_TRUNCATION): Update comment.
4851 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
4852 (TRULY_NOOP_TRUNCATION): Update comment.
4853 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
4854 (frv_modes_tieable_p): New function.
4855 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
4856 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
4857 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
4858 (TARGET_MODES_TIEABLE_P): Redefine.
4859 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
4860 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
4861 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
4862 (TARGET_MODES_TIEABLE_P): Redefine.
4863 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
4864 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
4865 (ia64_modes_tieable_p): New function.
4866 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
4867 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
4868 (iq2000_modes_tieable_p): New function.
4869 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
4870 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
4871 (lm32_modes_tieable_p): New function.
4872 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
4873 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
4874 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
4875 (TARGET_MODES_TIEABLE_P): Redefine.
4876 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
4877 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
4878 (m32r_modes_tieable_p): New function.
4879 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
4880 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
4881 (m68k_modes_tieable_p): New function.
4882 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
4883 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
4884 (mcore_modes_tieable_p): New function.
4885 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
4886 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
4888 (TARGET_MODES_TIEABLE_P): Redefine.
4889 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
4890 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
4891 * config/mips/mips.c (mips_modes_tieable_p): Make static.
4892 (TARGET_MODES_TIEABLE_P): Redefine.
4893 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
4894 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
4895 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
4896 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
4897 (mn10300_modes_tieable_p): ...this and make static.
4898 (TARGET_MODES_TIEABLE_P): Redefine.
4899 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
4900 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
4901 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
4902 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
4903 (msp430_modes_tieable_p): Make static.
4904 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
4905 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
4906 (TARGET_MODES_TIEABLE_P): Redefine.
4907 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
4908 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
4909 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
4910 (TARGET_MODES_TIEABLE_P): Redefine.
4911 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
4912 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
4913 * config/pa/pa.c (pa_modes_tieable_p): Make static.
4914 (TARGET_MODES_TIEABLE_P): Redefine.
4915 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
4916 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
4917 (pdp11_modes_tieable_p): New function.
4918 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
4919 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
4920 (rs6000_modes_tieable_p): New function.
4921 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
4922 * config/powerpcspe/powerpcspe.md: Update comment.
4923 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
4924 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
4925 (TARGET_MODES_TIEABLE_P): Redefine.
4926 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
4927 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
4928 (rl78_modes_tieable_p): New function.
4929 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
4930 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
4931 (rs6000_modes_tieable_p): New function.
4932 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
4933 * config/rs6000/rs6000.md: Update comment.
4934 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
4935 * config/rx/rx.c (rx_modes_tieable_p): New function.
4936 (TARGET_MODES_TIEABLE_P): Redefine.
4937 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
4938 * config/s390/s390.c (s390_modes_tieable_p): New function.
4939 (TARGET_MODES_TIEABLE_P): Redefine.
4940 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
4941 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
4942 (sh_modes_tieable_p): New function.
4943 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
4944 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
4945 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
4946 (sparc_modes_tieable_p): Make static.
4947 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
4948 * config/spu/spu.c (spu_modes_tieable_p): New function.
4949 (TARGET_MODES_TIEABLE_P): Redefine.
4950 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
4951 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
4952 (TARGET_MODES_TIEABLE_P): Redefine.
4953 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
4954 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
4955 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
4956 * config/v850/v850.c (v850_modes_tieable_p): New function.
4957 (TARGET_MODES_TIEABLE_P): Redefine.
4958 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
4959 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
4960 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
4961 (visium_modes_tieable_p): New function.
4962 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
4963 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
4964 (xtensa_modes_tieable_p): New function.
4965 * system.h (MODES_TIEABLE_P): Poison.
4967 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4968 Alan Hayward <alan.hayward@arm.com>
4969 David Sherwood <david.sherwood@arm.com>
4971 * target.def (hard_regno_mode_ok): New hook.
4972 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
4973 (TARGET_HARD_REGNO_MODE_OK): ...this.
4974 * doc/tm.texi.in: Regenerate.
4975 * hooks.h (hook_bool_uint_mode_true): Declare.
4976 * hooks.c (hook_bool_uint_mode_true): New function.
4977 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
4979 * genpreds.c (write_insn_preds_c): Add an include of target.h.
4980 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
4981 instead of HARD_REGNO_MODE_OK.
4982 * caller-save.c: Include target.h.
4983 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
4985 * combine.c (can_combine_p): Likewise.
4986 (combinable_i3pat): Likewise.
4987 (can_change_dest_mode): Likewise.
4988 * expr.c (init_expr_target): Likewise.
4989 (convert_move): Likewise.
4990 (convert_modes): Likewise.
4991 * ira.c (setup_prohibited_class_mode_regs): Likewise.
4992 (setup_prohibited_mode_move_regs): Likewise.
4993 * ira.h (target_ira): Likewise.
4994 * lra-assigns.c (find_hard_regno_for_1): Likewise.
4995 * lra-constraints.c (process_alt_operands): Likewise.
4996 (split_reg): Likewise.
4997 * recog.c (peep2_find_free_register): Likewise.
4998 * ree.c (combine_reaching_defs): Likewise.
4999 * regcprop.c (maybe_mode_change): Likewise.
5000 * reginfo.c (init_reg_sets_1): Likewise.
5001 (choose_hard_reg_mode): Likewise.
5002 (simplifiable_subregs): Likewise.
5003 * regrename.c (check_new_reg_p): Likewise.
5004 * reload.c (find_valid_class): Likewise.
5005 (find_valid_class_1): Likewise.
5006 (reload_inner_reg_of_subreg): Likewise.
5007 (push_reload): Likewise.
5008 (combine_reloads): Likewise.
5009 (find_dummy_reload): Likewise.
5010 (find_reloads): Likewise.
5011 * reload1.c (find_reg): Likewise.
5012 (set_reload_reg): Likewise.
5013 (allocate_reload_reg): Likewise.
5014 (choose_reload_regs): Likewise.
5015 (reload_adjust_reg_for_temp): Likewise.
5016 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
5017 (simplify_subreg_regno): Likewise.
5018 * sel-sched.c (init_regs_for_mode): Likewise.
5019 * varasm.c (make_decl_rtl): Likewise.
5020 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
5021 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
5023 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
5024 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
5025 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5026 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
5027 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
5028 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5029 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
5030 (arc_mode_class): Delete.
5031 (HARD_REGNO_MODE_OK): Delete.
5032 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5033 (arc_hard_regno_mode_ok): Rename old array to...
5034 (arc_hard_regno_mode_ok_modes): ...this.
5035 (arc_conditional_register_usage): Update accordingly.
5036 (arc_mode_class): Make static.
5037 (arc_hard_regno_mode_ok): New function.
5038 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
5039 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
5040 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5041 (arm_hard_regno_mode_ok): Make static.
5042 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
5044 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
5045 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
5046 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
5048 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5049 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
5050 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
5051 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
5052 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
5053 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5054 * config/bfin/predicates.md (valid_reg_operand): Use
5055 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
5056 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
5057 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
5058 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5059 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
5060 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
5061 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5062 (cr16_hard_regno_mode_ok): Make static and return a bool.
5063 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
5064 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5065 (cris_hard_regno_mode_ok): New function.
5066 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
5067 (epiphany_mode_class): Delete.
5068 (HARD_REGNO_MODE_OK): Delete.
5069 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
5070 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5071 (hard_regno_mode_ok): Rename to...
5072 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
5073 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
5074 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
5076 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
5077 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
5078 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5079 (frv_hard_regno_mode_ok): Make static and return a bool.
5080 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
5082 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
5083 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
5084 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
5085 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
5087 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5088 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
5089 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
5090 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
5092 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5093 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
5094 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5095 (ia64_hard_regno_mode_ok): New function.
5096 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
5097 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5098 (iq2000_hard_regno_mode_ok): New function.
5099 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
5100 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5101 (lm32_hard_regno_mode_ok): New function.
5102 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
5103 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
5104 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
5105 instead of HARD_REGNO_MODE_OK.
5106 (m32c_hard_regno_ok): Rename to...
5107 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
5108 (m32c_cannot_change_mode_class): Update accordingly.
5109 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5110 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
5111 (m32r_mode_class): Delete.
5112 (HARD_REGNO_MODE_OK): Delete.
5113 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5114 (m32r_hard_regno_mode_ok): Rename to...
5115 (m32r_hard_regno_modes): ...this.
5116 (m32r_mode_class): Make static.
5117 (m32r_hard_regno_mode_ok): New function.
5118 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
5119 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
5120 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5121 (m68k_hard_regno_mode_ok): Make static.
5122 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
5123 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5124 (mcore_hard_regno_mode_ok): New function.
5125 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
5126 (HARD_REGNO_MODE_OK): Delete.
5127 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
5129 (microblaze_hard_regno_mode_ok_p): ...this and make static.
5130 (microblaze_hard_regno_mode_ok): New function.
5131 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5132 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
5133 (mips_hard_regno_mode_ok): Delete.
5134 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
5135 (mips_hard_regno_mode_ok_p): ...this and make static.
5136 (mips_hard_regno_mode_ok_p): Rename to...
5137 (mips_hard_regno_mode_ok_uncached): ...this.
5138 (mips_hard_regno_mode_ok): New function.
5139 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
5140 of HARD_REGNO_MODE_OK.
5141 (mips_option_override): Update after above name changes.
5142 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5143 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
5144 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
5145 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
5146 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
5147 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5148 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
5149 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
5150 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
5151 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5152 (msp430_hard_regno_mode_ok): Make static and return a bool.
5153 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
5154 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
5155 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
5157 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5158 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
5159 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
5160 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
5161 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
5162 (PA_HARD_REGNO_MODE_OK): ...this
5163 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
5164 (PA_HARD_REGNO_MODE_OK): ...this.
5165 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5166 (pa_hard_regno_mode_ok): New function.
5167 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
5168 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5169 (pdp11_hard_regno_mode_ok): New function.
5170 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
5171 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
5173 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
5175 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5176 (rs6000_hard_regno_mode_ok): Rename to...
5177 (rs6000_hard_regno_mode_ok_uncached): ...this.
5178 (rs6000_init_hard_regno_mode_ok): Update accordingly.
5179 (rs6000_hard_regno_mode_ok): New function.
5180 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
5181 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
5182 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
5183 (riscv_hard_regno_mode_ok): ...this and make static.
5184 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5185 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
5186 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
5187 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5188 (rl78_hard_regno_mode_ok): Make static and return bool.
5189 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
5190 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
5192 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
5193 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5194 (rs6000_hard_regno_mode_ok): Rename to...
5195 (rs6000_hard_regno_mode_ok_uncached): ...this.
5196 (rs6000_init_hard_regno_mode_ok): Update accordingly.
5197 (rs6000_hard_regno_mode_ok): New function.
5198 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
5199 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
5200 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5201 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
5202 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
5203 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
5204 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5205 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
5206 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
5207 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5208 (sh_hard_regno_mode_ok): Make static.
5209 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
5210 instead of HARD_REGNO_MODE_OK.
5211 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
5212 (sparc_mode_class): Delete.
5213 (HARD_REGNO_MODE_OK): Delete.
5214 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5215 (hard_regno_mode_classes): Make static.
5216 (sparc_mode_class): Likewise.
5217 (sparc_hard_regno_mode_ok): New function.
5218 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
5219 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
5220 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
5222 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5223 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
5224 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
5225 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
5226 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
5227 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5228 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
5229 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
5230 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5231 (visium_hard_regno_mode_ok): New function.
5232 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
5233 instead of HARD_REGNO_MODE_OK.
5234 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
5235 (HARD_REGNO_MODE_OK): Delete.
5236 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
5237 (xtensa_hard_regno_mode_ok_p): ...this and make static.
5238 (xtensa_option_override): Update accordingly.
5239 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5240 (xtensa_hard_regno_mode_ok): New function.
5241 * system.h (HARD_REGNO_MODE_OK): Poison.
5243 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5244 Alan Hayward <alan.hayward@arm.com>
5245 David Sherwood <david.sherwood@arm.com>
5247 * target.def (hard_regno_call_part_clobbered): New hook.
5248 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
5249 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
5250 * doc/tm.texi: Regenerate.
5251 * hooks.h (hook_bool_uint_mode_false): Declare.
5252 * hooks.c (hook_bool_uint_mode_false): New function.
5253 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5254 * cselib.c (cselib_process_insn): Use
5255 targetm.hard_regno_call_part_clobbered instead of
5256 HARD_REGNO_CALL_PART_CLOBBERED.
5257 * ira-conflicts.c (ira_build_conflicts): Likewise.
5258 * ira-costs.c (ira_tune_allocno_costs): Likewise.
5259 * lra-constraints.c (need_for_call_save_p): Likewise.
5260 * lra-lives.c: Include target.h.
5261 (check_pseudos_live_through_calls): Use
5262 targetm.hard_regno_call_part_clobbered instead of
5263 HARD_REGNO_CALL_PART_CLOBBERED.
5264 * regcprop.c: Include target.h.
5265 (copyprop_hardreg_forward_1): Use
5266 targetm.hard_regno_call_part_clobbered instead of
5267 HARD_REGNO_CALL_PART_CLOBBERED.
5268 * reginfo.c (choose_hard_reg_mode): Likewise.
5269 * regrename.c (check_new_reg_p): Likewise.
5270 * reload.c (find_equiv_reg): Likewise.
5271 * reload1.c (emit_reload_insns): Likewise.
5272 * sched-deps.c (deps_analyze_insn): Likewise.
5273 * sel-sched.c (init_regs_for_mode): Likewise.
5274 (mark_unavailable_hard_regs): Likewise.
5275 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
5276 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5277 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
5279 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5280 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5281 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
5283 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
5285 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5286 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5287 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
5289 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5290 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5291 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
5293 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5294 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
5296 * config/powerpcspe/powerpcspe.c
5297 (rs6000_hard_regno_call_part_clobbered): New function.
5298 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5299 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5300 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
5302 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5303 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5304 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
5306 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5307 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5308 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
5310 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5311 Alan Hayward <alan.hayward@arm.com>
5312 David Sherwood <david.sherwood@arm.com>
5314 * rtl.h (subreg_memory_offset): Declare.
5315 * emit-rtl.c (subreg_memory_offset): New function.
5316 * expmed.c (store_bit_field_1): Use it.
5317 * expr.c (undefined_operand_subword_p): Likewise.
5318 * simplify-rtx.c (simplify_subreg): Likewise.
5320 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
5322 PR rtl-optimization/57448
5325 * optabs.c (expand_atomic_load): Place compiler memory barriers if
5326 using atomic_load pattern.
5327 (expand_atomic_store): Likewise.
5329 2017-09-04 Jakub Jelinek <jakub@redhat.com>
5332 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
5333 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
5334 handling. Use replace_call_with_value with NULL instead of
5335 gsi_replace, unlink_stmt_vdef and release_defs.
5337 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
5340 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
5342 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5345 * rtl.h (emit_library_call_value_1): Declare.
5346 (emit_library_call): Replace declaration with a series of overloads.
5347 Remove the parameter count argument.
5348 (emit_library_call_value): Likewise.
5349 * calls.c (emit_library_call_value_1): Make global. Replace varargs
5350 with an "rtx_mode_t *".
5351 (emit_library_call_value): Delete.
5352 (emit_library_call): Likewise.
5353 * asan.c (asan_emit_stack_protection): Update calls accordingly.
5354 (asan_emit_allocas_unpoison): Likewise.
5355 * builtins.c (expand_builtin_powi): Likewise.
5356 (expand_asan_emit_allocas_unpoison): Likewise.
5357 * cfgexpand.c (expand_main_function): Likewise.
5358 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
5359 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
5360 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
5361 * config/arm/arm.c (arm_trampoline_init): Likewise.
5362 (arm_call_tls_get_addr): Likewise.
5363 (arm_expand_divmod_libfunc): Likewise.
5364 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
5365 (smulsi3_highpart): Likewise.
5366 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
5367 (c6x_expand_compare): Likewise.
5368 (c6x_expand_movmem): Likewise.
5369 * config/frv/frv.c (frv_trampoline_init): Likewise.
5370 * config/i386/i386.c (ix86_trampoline_init): Likewise.
5371 (ix86_expand_divmod_libfunc): Likewise.
5372 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
5373 (ia64_expand_compare): Likewise.
5374 (ia64_profile_hook): Likewise.
5375 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
5376 (nonlocal_goto): Likewise.
5377 (restore_stack_nonlocal): Likewise.
5378 * config/m32r/m32r.c (block_move_call): Likewise.
5379 (m32r_trampoline_init): Likewise.
5380 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
5381 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
5382 (m68k_call_m68k_read_tp): Likewise.
5383 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
5384 (microblaze_expand_divide): Likewise.
5385 * config/mips/mips.h (mips_args): Likewise.
5386 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
5387 (MIPS_ICACHE_SYNC): Likewise.
5388 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
5389 (nios2_trampoline_init): Likewise.
5390 * config/pa/pa.c (hppa_tls_call): Likewise.
5391 (pa_trampoline_init): Likewise.
5392 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
5393 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
5394 (expand_strn_compare): Likewise.
5395 (rs6000_generate_compare): Likewise.
5396 (rs6000_expand_float128_convert): Likewise.
5397 (output_profile_hook): Likewise.
5398 (rs6000_trampoline_init): Likewise.
5399 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
5400 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
5401 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
5402 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
5403 (rs6000_generate_compare): Likewise.
5404 (rs6000_expand_float128_convert): Likewise.
5405 (output_profile_hook): Likewise.
5406 (rs6000_trampoline_init): Likewise.
5407 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
5408 * config/sh/sh.c (sh_trampoline_init): Likewise.
5409 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
5410 (sparc_emit_float_lib_cmp): Likewise.
5411 (sparc32_initialize_trampoline): Likewise.
5412 (sparc64_initialize_trampoline): Likewise.
5413 (sparc_profile_hook): Likewise.
5414 * config/spu/spu.c (ea_load_store): Likewise.
5415 * config/spu/spu.md (floatunssidf2): Likewise.
5416 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
5417 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
5418 * config/visium/visium.c (expand_block_move_4): Likewise.
5419 (expand_block_move_2): Likewise.
5420 (expand_block_move_1): Likewise.
5421 (expand_block_set_4): Likewise.
5422 (expand_block_set_2): Likewise.
5423 (expand_block_set_1): Likewise.
5424 (visium_trampoline_init): Likewise.
5425 (visium_profile_hook): Likewise.
5426 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
5427 (xtensa_setup_frame_addresses): Likewise.
5428 (xtensa_trampoline_init): Likewise.
5429 * except.c (sjlj_emit_function_enter): Likewise.
5430 (sjlj_emit_function_exit): Likewise.
5431 * explow.c (allocate_dynamic_stack_space): Likewise.
5432 (probe_stack_range): Likewise.
5433 * expr.c (convert_mode_scalar): Likewise.
5434 * optabs.c (expand_binop): Likewise.
5435 (expand_twoval_binop_libfunc): Likewise.
5436 (expand_unop): Likewise.
5437 (prepare_cmp_insn): Likewise.
5438 (prepare_float_lib_cmp): Likewise.
5439 (expand_float): Likewise.
5440 (expand_fix): Likewise.
5441 (expand_fixed_convert): Likewise.
5442 (maybe_emit_sync_lock_test_and_set): Likewise.
5443 (expand_atomic_compare_and_swap): Likewise.
5444 (expand_mem_thread_fence): Likewise.
5445 (expand_atomic_fetch_op): Likewise.
5447 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
5449 * doc/generic.texi (OpenACC): Adjust URL.
5450 * doc/invoke.texi (C Dialect Options): Ditto.
5452 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
5454 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
5455 predicate for operand 1. Add (m,<S>) constraint.
5456 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
5457 Prevent memory operand 1 with register operand 2.
5459 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
5461 PR rtl-optimization/82024
5462 * combine.c (try_combine): If the combination result is a PARALLEL,
5463 and we only need to retain the SET in there that would be placed
5464 at I2, check that we can place that at I3 instead, before doing so.
5466 2017-09-01 Jakub Jelinek <jakub@redhat.com>
5469 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
5471 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
5472 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
5475 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
5476 Jeff Law <law@redhat.com>
5478 * varasm.c (bss_initializer_p): Do not put constants into .bss
5479 (categorize_decl_for_section): Handle bss_initializer_p returning
5480 false when DECL_INITIAL is NULL.
5482 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5485 * config/s390/s390.c (s390_can_inline_p): New function.
5487 2017-09-01 Jeff Law <law@redhat.com>
5489 PR tree-optimization/82052
5490 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
5491 Always initialize the returned slot after a hash table miss
5492 when INSERT is true.
5494 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
5496 * config/s390/s390.md (mem_signal_fence): Remove.
5497 * doc/md.texi (mem_signal_fence): Remove.
5498 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
5500 * target-insns.def (mem_signal_fence): Remove.
5502 2017-09-01 Jakub Jelinek <jakub@redhat.com>
5505 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
5508 * asan.c (create_odr_indicator): Strip name encoding from assembler
5509 name before appending it after __odr_asan_.
5511 2017-09-01 Martin Liska <mliska@suse.cz>
5513 PR tree-optimization/82059
5514 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
5515 frequency only when an edge is redirected.
5517 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5519 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
5520 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
5521 (arc_conditional_register_usage): Remove ARC600 lp_count
5523 (arc_file_start): Emit Tag_ARC_CPU_variation.
5524 (arc_can_use_doloop_p): New conditions to use ZOLs.
5525 (hwloop_fail): New function.
5526 (hwloop_optimize): Likewise.
5527 (hwloop_pattern_reg): Likewise.
5528 (arc_doloop_hooks): New struct, to be used with reorg_loops.
5529 (arc_reorg_loops): New function, calls reorg_loops.
5530 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
5531 (arc600_corereg_hazard): Remove ZOL checking, case handled by
5533 (arc_loop_hazard): Remove function, functionality moved into
5535 (arc_hazard): Remove arc_loop_hazard call.
5536 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
5537 into hwloop_optimize.
5538 (arc_label_align): Remove ZOL handling.
5539 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
5540 * config/arc/arc.md (doloop_begin): Remove pattern.
5541 (doloop_begin_i): Likewise.
5542 (doloop_end_i): Likewise.
5543 (doloop_fallback): Likewise.
5544 (doloop_fallback_m): Likewise.
5545 (doloop_end): Reimplement expand.
5546 (arc_lp): New pattern for LP instruction.
5547 (loop_end): New pattern.
5548 (loop_fail): Likewise.
5549 (decrement_and_branch_until_zero): Likewise.
5550 * config/arc/arc.opt (mlpc-width): New option.
5551 * doc/invoke.texi (mlpc-width): Document option.
5553 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5555 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
5556 (arc_ccfsm_advance): Fix checking for delay slots.
5557 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
5559 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5561 * config/arc/arc.md (movqi_insn): Add stores to save constant long
5563 (movhi_insn): Update store instruction constraint which are saving
5564 6-bit short immediates.
5565 (movsi_insn): Consider also short scaled load operations.
5566 (zero_extendhisi2_i): Use Usd constraint instead of T.
5567 (extendhisi2_i): Add q constraint.
5568 (arc_clzsi2): Add type and length attributes.
5569 (arc_ctzsi2): Likewise.
5570 * config/arc/constraints.md (Usc): Update constraint, the
5571 assembler can parse two relocations for a single instruction.
5573 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5575 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
5576 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
5578 2017-08-31 Olivier Hainque <hainque@adacore.com>
5580 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
5581 match as powerpc-wrs-vxworks*.
5583 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
5585 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
5586 register constraint for by-element operand.
5587 (aarch64_mls_elt_merge<mode>): Likewise.
5589 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
5591 * config/arc/arc.c (arc_can_follow_jump): Check for short
5594 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
5596 * config.gcc: Use g.opt for arc.
5597 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
5598 functionality moved to ...
5599 (legitimate_scaled_address_p): New function, ...here.
5600 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
5601 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
5602 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
5604 (arc_override_options): Handle G option.
5605 (arc_output_pic_addr_const): Correct function definition.
5606 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
5607 (arc_decl_anon_ns_mem_p): Delete.
5608 (arc_in_small_data_p): Overhaul this function to take into
5609 consideration the value given via G option.
5610 (arc_rewrite_small_data_1): Renamed and corrected old
5611 arc_rewrite_small_data function.
5612 (arc_rewrite_small_data): New function.
5613 (small_data_pattern): Don't use pic_offset_table_rtx.
5614 * config/arc/arc.h (CC1_SPEC): Recognize G option.
5615 * config/arc/simdext.md (movmisalignv2hi): Use
5616 prepare_move_operands function.
5618 (movmisalign*): Likewise.
5619 * doc/invoke.texi (ARC options): Document -G option.
5621 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
5623 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
5625 * config/arc/arc.c (arc_print_operand): Output scalled address for
5626 sdata whenever is possible.
5627 (arc_in_small_data_p): Allow sdata for 64bit datum when double
5628 load/stores are available.
5629 (compact_sda_memory_operand): Check for the alignment required by
5630 code density instructions.
5631 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
5633 * config/arc/constraints.md (Usd): Update constraint.
5634 (Us0): New constraint.
5635 (Usc): Update constraint.
5637 2017-08-31 Richard Biener <rguenther@suse.de>
5640 * dwarf2out.c (dwarf2out_early_global_decl): Process each
5643 2017-08-31 Tamar Christina <tamar.christina@arm.com>
5645 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
5646 Resize type_signature.
5648 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
5649 Alan Hayward <alan.hayward@arm.com>
5650 David Sherwood <david.sherwood@arm.com>
5652 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
5653 subregs whose inner modes can be stored in GPRs.
5654 (aarch64_classify_index): Likewise.
5656 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
5657 Alan Hayward <alan.hayward@arm.com>
5658 David Sherwood <david.sherwood@arm.com>
5660 * config/aarch64/iterators.md (V_cmp_result): Rename to...
5661 (V_INT_EQUIV): ...this.
5662 (v_cmp_result): Rename to...
5663 (v_int_equiv): ...this.
5664 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
5665 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
5666 (copysign<mode>3): Likewise.
5667 (aarch64_simd_bsl<mode>_internal): Likewise.
5668 (aarch64_simd_bsl<mode>): Likewise.
5669 (vec_cmp<mode><mode>): Likewise.
5670 (vcond<mode><mode>): Likewise.
5671 (vcond<v_cmp_mixed><mode>): Likewise.
5672 (vcondu<mode><v_cmp_mixed>): Likewise.
5673 (aarch64_cm<optab><mode>): Likewise.
5674 (aarch64_cmtst<mode>): Likewise.
5675 (aarch64_fac<optab><mode>): Likewise.
5676 (vec_perm_const<mode>): Likewise.
5677 (vcond_mask_<mode><v_cmp_result>): Rename to...
5678 (vcond_mask_<mode><v_int_equiv>): ...this.
5679 (vec_cmp<mode><v_cmp_result>): Rename to...
5680 (vec_cmp<mode><v_int_equiv>): ...this.
5682 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
5683 Alan Hayward <alan.hayward@arm.com>
5684 David Sherwood <david.sherwood@arm.com>
5686 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
5688 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
5689 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
5690 (UNSPEC_LD4_DREG): New unspecs.
5691 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
5692 (aarch64_ld2<mode>_dreg_be): Replace with...
5693 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
5695 (aarch64_ld3<mode>_dreg_le)
5696 (aarch64_ld3<mode>_dreg_be): Replace with...
5697 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
5699 (aarch64_ld4<mode>_dreg_le)
5700 (aarch64_ld4<mode>_dreg_be): Replace with...
5701 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
5704 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5706 PR tree-optimization/81987
5707 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
5708 insert an initializer in a location not dominated by the stride
5711 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
5713 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
5714 on the entire header of the finally block in the fallthru case.
5716 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
5718 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
5720 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
5722 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
5723 rs6000_emit_move_from_cr and call renamed function.
5724 (rs6000_emit_prologue): Call renamed functions.
5725 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
5726 movesi_from_cr, remove volatile CRs.
5728 2017-08-30 Jon Beniston <jon@beniston.com>
5729 Richard Biener <rguenther@suse.de>
5731 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
5732 of VECTOR_MODE_P check.
5733 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
5734 element vector types.
5736 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5738 * df.h (df_read_modify_subreg_p): Remove in favor of...
5739 * rtl.h (read_modify_subreg_p): ...this new function. Take a
5740 const_rtx instead of an rtx.
5741 * cprop.c (local_cprop_find_used_regs): Update accordingly.
5742 * df-problems.c (df_word_lr_mark_ref): Likewise.
5743 * ira-lives.c (mark_pseudo_reg_live): Likewise.
5744 (mark_pseudo_reg_dead): Likewise.
5745 (mark_ref_dead): Likewise.
5746 * reginfo.c (init_subregs_of_mode): Likewise.
5747 * sched-deps.c (sched_analyze_1): Likewise.
5748 * df-scan.c (df_def_record_1): Likewise.
5749 (df_uses_record): Likewise.
5750 (df_read_modify_subreg_p): Remove in favor of...
5751 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
5752 const_rtx instead of an rtx.
5754 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5755 Alan Hayward <alan.hayward@arm.com>
5756 David Sherwood <david.sherwood@arm.com>
5758 * rtl.h (partial_subreg_p): New function.
5759 * caller-save.c (save_call_clobbered_regs): Use it.
5760 * calls.c (expand_call): Likewise.
5761 * combine.c (combinable_i3pat): Likewise.
5762 (simplify_set): Likewise.
5763 (make_extraction): Likewise.
5764 (make_compound_operation_int): Likewise.
5765 (gen_lowpart_or_truncate): Likewise.
5766 (force_to_mode): Likewise.
5767 (make_field_assignment): Likewise.
5768 (reg_truncated_to_mode): Likewise.
5769 (record_truncated_value): Likewise.
5770 (move_deaths): Likewise.
5771 * cse.c (record_jump_cond): Likewise.
5772 (cse_insn): Likewise.
5773 * cselib.c (cselib_lookup_1): Likewise.
5774 * expmed.c (extract_bit_field_using_extv): Likewise.
5775 * function.c (assign_parm_setup_reg): Likewise.
5776 * ifcvt.c (noce_convert_multiple_sets): Likewise.
5777 * ira-build.c (create_insn_allocnos): Likewise.
5778 * lra-coalesce.c (merge_pseudos): Likewise.
5779 * lra-constraints.c (match_reload): Likewise.
5780 (simplify_operand_subreg): Likewise.
5781 (curr_insn_transform): Likewise.
5782 * lra-lives.c (process_bb_lives): Likewise.
5783 * lra.c (new_insn_reg): Likewise.
5784 (lra_substitute_pseudo): Likewise.
5785 * regcprop.c (mode_change_ok): Likewise.
5786 (maybe_mode_change): Likewise.
5787 (copyprop_hardreg_forward_1): Likewise.
5788 * reload.c (push_reload): Likewise.
5789 (find_reloads): Likewise.
5790 (find_reloads_subreg_address): Likewise.
5791 * reload1.c (alter_reg): Likewise.
5792 (eliminate_regs_1): Likewise.
5793 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5795 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
5797 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
5798 back to if statements, including unpack.
5800 2017-08-30 Martin Liska <mliska@suse.cz>
5803 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
5805 (func_checker::compare_asm_inputs_outputs): ... this function.
5806 (func_checker::compare_gimple_asm): Use the function to compare
5807 also ASM constrains.
5808 * ipa-icf-gimple.h: Rename the function.
5810 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5811 Alan Hayward <alan.hayward@arm.com>
5812 David Sherwood <david.sherwood@arm.com>
5814 * coretypes.h (complex_mode): New type.
5815 * gdbhooks.py (build_pretty_printer): Handle it.
5816 * machmode.h (complex_mode): New class.
5817 (complex_mode::includes_p): New function.
5818 (is_complex_int_mode): Likewise.
5819 (is_complex_float_mode): Likewise.
5820 * genmodes.c (get_mode_class): Handle complex mode classes.
5821 * function.c (expand_function_end): Use is_complex_int_mode.
5823 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5824 Alan Hayward <alan.hayward@arm.com>
5825 David Sherwood <david.sherwood@arm.com>
5827 * coretypes.h (scalar_mode_pod): New typedef.
5828 * gdbhooks.py (build_pretty_printer): Handle it.
5829 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
5830 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
5831 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
5832 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
5835 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5836 Alan Hayward <alan.hayward@arm.com>
5837 David Sherwood <david.sherwood@arm.com>
5839 * machmode.h (mode_for_vector): Take a scalar_mode instead
5841 * stor-layout.c (mode_for_vector): Likewise.
5842 * explow.c (promote_mode): Use as_a <scalar_mode>.
5843 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
5845 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5846 Alan Hayward <alan.hayward@arm.com>
5847 David Sherwood <david.sherwood@arm.com>
5849 * target.def (preferred_simd_mode): Take a scalar_mode
5850 instead of a machine_mode.
5851 * targhooks.h (default_preferred_simd_mode): Likewise.
5852 * targhooks.c (default_preferred_simd_mode): Likewise.
5853 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
5854 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
5855 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
5856 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
5857 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
5858 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
5859 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
5860 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
5862 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
5863 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
5864 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
5865 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
5866 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
5867 * doc/tm.texi: Regenerate.
5868 * optabs-query.c (can_vec_mask_load_store_p): Return false for
5871 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5872 Alan Hayward <alan.hayward@arm.com>
5873 David Sherwood <david.sherwood@arm.com>
5875 * target.def (scalar_mode_supported_p): Take a scalar_mode
5876 instead of a machine_mode.
5877 * targhooks.h (default_scalar_mode_supported_p): Likewise.
5878 * targhooks.c (default_scalar_mode_supported_p): Likewise.
5879 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
5880 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
5881 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
5882 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
5883 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
5884 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
5885 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
5886 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
5887 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
5888 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
5889 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
5890 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
5892 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
5893 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
5894 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
5895 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
5896 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
5898 * doc/tm.texi: Regenerate.
5900 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5901 Alan Hayward <alan.hayward@arm.com>
5902 David Sherwood <david.sherwood@arm.com>
5904 * coretypes.h (opt_scalar_mode): New typedef.
5905 * gdbhooks.py (build_pretty_printers): Handle it.
5906 * machmode.h (mode_iterator::get_2xwider): Add overload for
5908 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
5910 * expr.c (convert_mode_scalar): Likewise.
5911 * omp-low.c (omp_clause_aligned_alignment): Likewise.
5912 * optabs.c (expand_float): Likewise.
5913 (expand_fix): Likewise.
5914 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5916 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5917 Alan Hayward <alan.hayward@arm.com>
5918 David Sherwood <david.sherwood@arm.com>
5920 * optabs.c (expand_float): Explicitly check for scalars before
5921 using a branching expansion.
5922 (expand_fix): Likewise.
5924 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5925 Alan Hayward <alan.hayward@arm.com>
5926 David Sherwood <david.sherwood@arm.com>
5928 * expr.c (convert_mode): Split scalar handling out into...
5929 (convert_mode_scalar): ...this new function. Treat the modes
5932 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5933 Alan Hayward <alan.hayward@arm.com>
5934 David Sherwood <david.sherwood@arm.com>
5936 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
5938 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
5940 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5941 Alan Hayward <alan.hayward@arm.com>
5942 David Sherwood <david.sherwood@arm.com>
5944 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
5945 rather than a machine_mode.
5946 (fixed_from_string): Likewise.
5947 (fixed_convert): Likewise.
5948 (fixed_convert_from_int): Likewise.
5949 (fixed_convert_from_real): Likewise.
5950 (real_convert_from_fixed): Likewise.
5951 * fixed-value.c (fixed_from_double_int): Likewise.
5952 (fixed_from_string): Likewise.
5953 (fixed_convert): Likewise.
5954 (fixed_convert_from_int): Likewise.
5955 (fixed_convert_from_real): Likewise.
5956 (real_convert_from_fixed): Likewise.
5957 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
5959 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5960 Alan Hayward <alan.hayward@arm.com>
5961 David Sherwood <david.sherwood@arm.com>
5963 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
5964 of separate mode class checks. Do not allow vector modes here.
5965 (immed_wide_int_const): Use as_a <scalar_mode>.
5966 * explow.c (trunc_int_for_mode): Likewise.
5967 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
5968 (wi::shwi): Likewise.
5969 (wi::min_value): Likewise.
5970 (wi::max_value): Likewise.
5971 * dwarf2out.c (loc_descriptor): Likewise.
5972 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
5975 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5976 Alan Hayward <alan.hayward@arm.com>
5977 David Sherwood <david.sherwood@arm.com>
5979 * tree.h (SCALAR_TYPE_MODE): New macro.
5980 * expr.c (expand_expr_addr_expr_1): Use it.
5981 (expand_expr_real_2): Likewise.
5982 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
5983 (fold_convert_const_fixed_from_int): Likewise.
5984 (fold_convert_const_fixed_from_real): Likewise.
5985 (native_encode_fixed): Likewise
5986 (native_encode_complex): Likewise
5987 (native_encode_vector): Likewise.
5988 (native_interpret_fixed): Likewise.
5989 (native_interpret_real): Likewise.
5990 (native_interpret_complex): Likewise.
5991 (native_interpret_vector): Likewise.
5992 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
5993 (simd_clone_adjust_argument_types): Likewise.
5994 (simd_clone_init_simd_arrays): Likewise.
5995 (simd_clone_adjust): Likewise.
5996 * stor-layout.c (layout_type): Likewise.
5997 * tree.c (build_minus_one_cst): Likewise.
5998 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
5999 * tree-inline.c (estimate_move_cost): Likewise.
6000 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
6001 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
6002 (vectorizable_reduction): Likewise.
6003 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
6004 (vect_recog_mixed_size_cond_pattern): Likewise.
6005 (check_bool_pattern): Likewise.
6006 (adjust_bool_pattern): Likewise.
6007 (search_type_for_mask_1): Likewise.
6008 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
6009 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6010 (vectorizable_load): Likewise.
6011 (vectorizable_store): Likewise.
6012 * ubsan.c (ubsan_encode_value): Likewise.
6013 * varasm.c (output_constant): Likewise.
6015 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6016 Alan Hayward <alan.hayward@arm.com>
6017 David Sherwood <david.sherwood@arm.com>
6019 * coretypes.h (scalar_mode): New class.
6020 * machmode.h (scalar_mode): Likewise.
6021 (scalar_mode::includes_p): New function.
6022 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
6023 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
6024 * genmodes.c (get_mode_class): Handle remaining scalar modes.
6025 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
6026 * expmed.c (store_bit_field_1): Likewise.
6027 (extract_bit_field_1): Likewise.
6028 * expr.c (write_complex_part): Likewise.
6029 (read_complex_part): Likewise.
6030 (emit_move_complex_push): Likewise.
6031 (expand_expr_real_2): Likewise.
6032 * function.c (assign_parm_setup_reg): Likewise.
6033 (assign_parms_unsplit_complex): Likewise.
6034 * optabs.c (expand_binop): Likewise.
6035 * rtlanal.c (subreg_get_info): Likewise.
6036 * simplify-rtx.c (simplify_immed_subreg): Likewise.
6037 * varasm.c (output_constant_pool_2): Likewise.
6039 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6040 Alan Hayward <alan.hayward@arm.com>
6041 David Sherwood <david.sherwood@arm.com>
6043 * expmed.c (extract_high_half): Use scalar_int_mode and remove
6045 (expmed_mult_highpart_optab): Likewise.
6046 (expmed_mult_highpart): Likewise.
6048 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6049 Alan Hayward <alan.hayward@arm.com>
6050 David Sherwood <david.sherwood@arm.com>
6052 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
6053 instead of a machine_mode.
6054 (builtin_memset_read_str): Likewise.
6055 * builtins.c (c_readstr): Likewise.
6056 (builtin_memcpy_read_str): Likewise.
6057 (builtin_strncpy_read_str): Likewise.
6058 (builtin_memset_read_str): Likewise.
6059 (builtin_memset_gen_str): Likewise.
6060 (expand_builtin_signbit): Use scalar_int_mode for local variables.
6061 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
6062 instead of a machine_mode.
6063 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
6065 (make_extraction): Likewise.
6066 (try_widen_shift_mode): Take and return scalar_int_modes instead
6068 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
6069 a scalar_int_mode instead of a machine_mode.
6070 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
6071 (avr_addr_space_pointer_mode): Likewise.
6072 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
6073 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
6074 (msp430_unwind_word_mode): Likewise.
6075 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
6076 (spu_addr_space_pointer_mode): Likewise.
6077 (spu_addr_space_address_mode): Likewise.
6078 (spu_libgcc_cmp_return_mode): Likewise.
6079 (spu_libgcc_shift_count_mode): Likewise.
6080 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
6081 (rl78_addr_space_pointer_mode): Likewise.
6082 (fl78_unwind_word_mode): Likewise.
6083 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
6085 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
6086 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
6087 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
6088 (mips_valid_pointer_mode): Likewise.
6089 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
6090 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
6091 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
6093 (ft32_addr_space_address_mode): Likewise.
6094 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
6095 scalar_int_mode instead of a machine_mode.
6096 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
6098 (m32c_addr_space_address_mode): Likewise.
6099 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
6100 (rs6000_eh_return_filter_mode): Likewise.
6101 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
6102 (rs6000_eh_return_filter_mode): Likewise.
6103 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
6104 (s390_libgcc_shift_count_mode): Likewise.
6105 (s390_unwind_word_mode): Likewise.
6106 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
6108 * target.def (mode_rep_extended): Likewise.
6109 (valid_pointer_mode): Likewise.
6110 (addr_space.valid_pointer_mode): Likewise.
6111 (eh_return_filter_mode): Return a scalar_int_mode rather than
6113 (libgcc_cmp_return_mode): Likewise.
6114 (libgcc_shift_count_mode): Likewise.
6115 (unwind_word_mode): Likewise.
6116 (addr_space.pointer_mode): Likewise.
6117 (addr_space.address_mode): Likewise.
6118 * doc/tm.texi: Regenerate.
6119 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
6121 (do_jump): Use scalar_int_mode for local variables.
6122 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
6123 rather than a machine_mode.
6124 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
6125 (scompare_loc_descriptor_wide): Likewise.
6126 (scompare_loc_descriptor_narrow): Likewise.
6127 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
6129 * except.c (sjlj_emit_dispatch_table): Likewise.
6130 (expand_builtin_eh_copy_values): Likewise.
6131 * explow.c (convert_memory_address_addr_space_1): Likewise.
6132 Take a scalar_int_mode rather than a machine_mode.
6133 (convert_memory_address_addr_space): Take a scalar_int_mode rather
6134 than a machine_mode.
6135 (memory_address_addr_space): Use scalar_int_mode for local variables.
6136 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
6137 rather than a machine_mode.
6138 * expmed.c (mask_rtx): Likewise.
6139 (init_expmed_one_conv): Likewise.
6140 (expand_mult_highpart_adjust): Likewise.
6141 (extract_high_half): Likewise.
6142 (expmed_mult_highpart_optab): Likewise.
6143 (expmed_mult_highpart): Likewise.
6144 (expand_smod_pow2): Likewise.
6145 (expand_sdiv_pow2): Likewise.
6146 (emit_store_flag_int): Likewise.
6147 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
6149 (extract_low_bits): Likewise.
6150 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
6152 * expr.c (pieces_addr::adjust): Likewise.
6153 (can_store_by_pieces): Likewise.
6154 (store_by_pieces): Likewise.
6155 (clear_by_pieces_1): Likewise.
6156 (expand_expr_addr_expr_1): Likewise.
6157 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
6158 (expand_expr_real_1): Likewise.
6159 (try_casesi): Likewise.
6160 * final.c (shorten_branches): Likewise.
6161 * fold-const.c (fold_convert_const_int_from_fixed): Change the
6162 type of "mode" to machine_mode.
6163 * internal-fn.c (expand_arith_overflow_result_store): Take a
6164 scalar_int_mode rather than a machine_mode.
6165 (expand_mul_overflow): Use scalar_int_mode for local variables.
6166 * loop-doloop.c (doloop_modify): Likewise.
6167 (doloop_optimize): Likewise.
6168 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
6169 than a machine_mode.
6170 (expand_doubleword_shift_condmove): Likewise.
6171 (expand_doubleword_shift): Likewise.
6172 (expand_doubleword_clz): Likewise.
6173 (expand_doubleword_popcount): Likewise.
6174 (expand_doubleword_parity): Likewise.
6175 (expand_absneg_bit): Use scalar_int_mode for local variables.
6176 (prepare_float_lib_cmp): Likewise.
6177 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
6178 rather than a machine_mode.
6179 (convert_memory_address_addr_space): Likewise.
6180 (get_mode_bounds): Likewise.
6181 (get_address_mode): Return a scalar_int_mode rather than a
6183 * rtlanal.c (get_address_mode): Likewise.
6184 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
6185 than a machine_mode.
6186 * targhooks.c (default_mode_rep_extended): Likewise.
6187 (default_valid_pointer_mode): Likewise.
6188 (default_addr_space_valid_pointer_mode): Likewise.
6189 (default_eh_return_filter_mode): Return a scalar_int_mode rather
6190 than a machine_mode.
6191 (default_libgcc_cmp_return_mode): Likewise.
6192 (default_libgcc_shift_count_mode): Likewise.
6193 (default_unwind_word_mode): Likewise.
6194 (default_addr_space_pointer_mode): Likewise.
6195 (default_addr_space_address_mode): Likewise.
6196 * targhooks.h (default_eh_return_filter_mode): Likewise.
6197 (default_libgcc_cmp_return_mode): Likewise.
6198 (default_libgcc_shift_count_mode): Likewise.
6199 (default_unwind_word_mode): Likewise.
6200 (default_addr_space_pointer_mode): Likewise.
6201 (default_addr_space_address_mode): Likewise.
6202 (default_mode_rep_extended): Take a scalar_int_mode rather than
6204 (default_valid_pointer_mode): Likewise.
6205 (default_addr_space_valid_pointer_mode): Likewise.
6206 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
6208 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
6209 rather than a machine_mode.
6210 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
6211 for local variables.
6212 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6213 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
6214 than a machine_mode.
6216 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6217 Alan Hayward <alan.hayward@arm.com>
6218 David Sherwood <david.sherwood@arm.com>
6220 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
6221 the mode argument to scalar_int_mode.
6222 (do_jump_by_parts_zero_rtx): Likewise.
6223 (do_jump_by_parts_equality_rtx): Likewise.
6224 (do_jump_by_parts_greater): Take a mode argument.
6225 (do_jump_by_parts_equality): Likewise.
6226 (do_jump_1): Update calls accordingly.
6228 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6229 Alan Hayward <alan.hayward@arm.com>
6230 David Sherwood <david.sherwood@arm.com>
6232 * is-a.h (safe_dyn_cast): New function.
6233 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
6234 (jump_table_for_label): Likewise.
6235 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
6236 instead of an rtx_insn *.
6237 (shorten_branches): Use dyn_cast instead of LABEL_P and
6238 JUMP_TABLE_DATA_P. Use jump_table_for_label and
6239 rtx_jump_table_data::get_data_mode.
6240 (final_scan_insn): Likewise.
6242 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6243 Alan Hayward <alan.hayward@arm.com>
6244 David Sherwood <david.sherwood@arm.com>
6246 * combine.c (try_combine): Use is_a <scalar_int_mode> when
6247 trying to combine a full-register integer set with a subreg
6250 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6251 Alan Hayward <alan.hayward@arm.com>
6252 David Sherwood <david.sherwood@arm.com>
6254 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
6255 that is always either address_mode or pointer_mode.
6257 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6258 Alan Hayward <alan.hayward@arm.com>
6259 David Sherwood <david.sherwood@arm.com>
6261 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
6262 when the two are known to be equal.
6264 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6265 Alan Hayward <alan.hayward@arm.com>
6266 David Sherwood <david.sherwood@arm.com>
6268 * simplify-rtx.c (simplify_const_unary_operation): Use
6269 is_a <scalar_int_mode> instead of checking for a nonzero
6270 precision. Forcibly convert op_mode to a scalar_int_mode
6271 in that case. More clearly differentiate the operand and
6272 result modes and use the former when deciding what the value
6273 of a count-bits operation should be. Use is_int_mode instead
6274 of checking for a MODE_INT. Remove redundant check for whether
6275 this mode has a zero precision.
6277 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6278 Alan Hayward <alan.hayward@arm.com>
6279 David Sherwood <david.sherwood@arm.com>
6281 * optabs.c (widen_leading): Change the type of the mode argument
6282 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
6283 (widen_bswap): Likewise.
6284 (expand_parity): Likewise.
6285 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
6286 (expand_ffs): Likewise.
6287 (epand_unop): Check for scalar integer modes before calling the
6290 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6291 Alan Hayward <alan.hayward@arm.com>
6292 David Sherwood <david.sherwood@arm.com>
6294 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
6296 (expand_expr_real_1): Update call accordingly.
6298 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6299 Alan Hayward <alan.hayward@arm.com>
6300 David Sherwood <david.sherwood@arm.com>
6302 * expmed.c (store_bit_field_using_insv): Add op0_mode and
6303 value_mode arguments. Use scalar_int_mode internally.
6304 (store_bit_field_1): Rename the new integer mode from imode
6305 to op0_mode and use it instead of GET_MODE (op0). Update calls
6306 to store_split_bit_field, store_bit_field_using_insv and
6307 store_fixed_bit_field.
6308 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
6309 Use scalar_int_mode internally. Use a bit count rather than a mode
6310 when calculating the largest bit size for get_best_mode.
6311 Update calls to store_split_bit_field and store_fixed_bit_field_1.
6312 (store_fixed_bit_field_1): Add mode and value_mode arguments.
6313 Remove assertion that OP0 has a scalar integer mode.
6314 (store_split_bit_field): Add op0_mode and value_mode arguments.
6315 Update calls to extract_fixed_bit_field.
6316 (extract_bit_field_using_extv): Add an op0_mode argument.
6317 Use scalar_int_mode internally.
6318 (extract_bit_field_1): Rename the new integer mode from imode to
6319 op0_mode and use it instead of GET_MODE (op0). Update calls to
6320 extract_split_bit_field, extract_bit_field_using_extv and
6321 extract_fixed_bit_field.
6322 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
6323 to extract_split_bit_field and extract_fixed_bit_field_1.
6324 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
6325 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
6327 (extract_split_bit_field): Add an op0_mode argument. Update call
6328 to extract_fixed_bit_field.
6330 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6331 Alan Hayward <alan.hayward@arm.com>
6332 David Sherwood <david.sherwood@arm.com>
6334 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
6335 * explow.c (hard_function_value): Likewise.
6336 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
6337 convert_to_mode call outside the loop.
6338 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
6339 for the mode iterator. Require the mode specified by max_pieces
6341 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
6343 (copy_blkmode_to_reg): Likewise.
6344 (set_storage_via_setmem): Likewise.
6345 * optabs.c (prepare_cmp_insn): Likewise.
6346 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6347 * stor-layout.c (finish_bitfield_representative): Likewise.
6349 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6350 Alan Hayward <alan.hayward@arm.com>
6351 David Sherwood <david.sherwood@arm.com>
6353 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
6354 * expr.c (convert_move): Use them.
6355 (convert_modes): Likewise.
6356 (store_expr_with_bounds): Likewise.
6358 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6359 Alan Hayward <alan.hayward@arm.com>
6360 David Sherwood <david.sherwood@arm.com>
6362 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
6363 parameter for the mode of "x". Remove the "known_x", "known_mode"
6364 and "known_ret" arguments. Change the type of the mode argument
6366 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
6367 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
6368 (reg_num_sign_bit_copies_for_combine): Likewise.
6369 * rtlanal.c (nonzero_bits1): Likewise.
6370 (num_sign_bit_copies1): Likewise.
6371 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
6372 (reg_num_sign_bit_copies_general): Likewise.
6373 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
6374 (reg_nonzero_bits_general): Likewise.
6376 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6377 Alan Hayward <alan.hayward@arm.com>
6378 David Sherwood <david.sherwood@arm.com>
6380 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
6381 than in subroutines. Return 1 for non-integer modes.
6382 (cached_num_sign_bit_copies): Change the type of the mode parameter
6384 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
6385 classes. Handle CONST_INT_P first and then check whether X also
6386 has a scalar integer mode. Check the same thing for inner registers
6387 of a SUBREG and for values that are being extended or truncated.
6389 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6390 Alan Hayward <alan.hayward@arm.com>
6391 David Sherwood <david.sherwood@arm.com>
6393 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
6394 in subroutines. Return the mode mask for non-integer modes.
6395 (cached_nonzero_bits): Change the type of the mode parameter
6397 (nonzero_bits1): Likewise. Remove early exit for other mode
6398 classes. Handle CONST_INT_P first and then check whether X
6399 also has a scalar integer mode.
6401 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6402 Alan Hayward <alan.hayward@arm.com>
6403 David Sherwood <david.sherwood@arm.com>
6405 * expr.c (widest_int_mode_for_size): Make the comment match the code.
6406 Return a scalar_int_mode and assert that the size is greater than
6408 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
6409 (op_by_pieces_d::op_by_pieces_d): Likewise.
6410 (op_by_pieces_d::run): Likewise.
6411 (can_store_by_pieces): Likewise.
6413 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6414 Alan Hayward <alan.hayward@arm.com>
6415 David Sherwood <david.sherwood@arm.com>
6417 * combine.c (extract_left_shift): Add a mode argument and update
6419 (make_compound_operation_int): Change the type of the mode parameter
6420 to scalar_int_mode and update the call to extract_left_shift.
6422 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6423 Alan Hayward <alan.hayward@arm.com>
6424 David Sherwood <david.sherwood@arm.com>
6426 * combine.c (simplify_and_const_int): Change the type of the mode
6427 parameter to scalar_int_mode.
6428 (simplify_and_const_int_1): Likewise. Update recursive call.
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 * combine.c (simplify_compare_const): Check that the mode is a
6435 scalar_int_mode (rather than VOIDmode) before testing its
6437 (simplify_comparison): Move COMPARISON_P handling out of the
6438 loop and restrict the latter part of the loop to scalar_int_modes.
6439 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
6440 and when considering SUBREG_REGs. Use is_int_mode instead of
6441 checking GET_MODE_CLASS against MODE_INT.
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 * combine.c (try_widen_shift_mode): Move check for equal modes to...
6448 (simplify_shift_const_1): ...here. Use scalar_int_mode for
6449 shift_unit_mode and for modes involved in scalar shifts.
6451 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6452 Alan Hayward <alan.hayward@arm.com>
6453 David Sherwood <david.sherwood@arm.com>
6455 * combine.c (force_int_to_mode): New function, split out from...
6456 (force_to_mode): ...here. Keep xmode up-to-date and use it
6457 instead of GET_MODE (x).
6459 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6460 Alan Hayward <alan.hayward@arm.com>
6461 David Sherwood <david.sherwood@arm.com>
6463 * optabs-query.h (extraction_insn::struct_mode): Change type to
6464 opt_scalar_int_mode and update comment.
6465 (extraction_insn::field_mode): Change type to scalar_int_mode.
6466 (extraction_insn::pos_mode): Likewise.
6467 * combine.c (make_extraction): Update accordingly.
6468 * optabs-query.c (get_traditional_extraction_insn): Likewise.
6469 (get_optab_extraction_insn): Likewise.
6470 * recog.c (simplify_while_replacing): Likewise.
6471 * expmed.c (narrow_bit_field_mem): Change the type of the mode
6472 parameter to opt_scalar_int_mode.
6474 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6475 Alan Hayward <alan.hayward@arm.com>
6476 David Sherwood <david.sherwood@arm.com>
6478 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
6479 to a scalar_int_mode instead of a machine_mode.
6480 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
6481 (get_best_mode): Return a boolean and use a pointer argument to store
6482 the selected mode. Replace the limit mode parameter with a bit limit.
6483 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
6484 for the values returned by bit_field_mode_iterator::next_mode.
6485 (store_bit_field): Update call to get_best_mode.
6486 (store_fixed_bit_field): Likewise.
6487 (extract_fixed_bit_field): Likewise.
6488 * expr.c (optimize_bitfield_assignment_op): Likewise.
6489 * fold-const.c (optimize_bit_field_compare): Likewise.
6490 (fold_truth_andor_1): Likewise.
6491 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
6492 Update for new type of m_mode.
6493 (get_best_mode): As above.
6495 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6496 Alan Hayward <alan.hayward@arm.com>
6497 David Sherwood <david.sherwood@arm.com>
6499 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
6500 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
6501 (store_bit_field): Check is_a <scalar_int_mode> before calling
6502 strict_volatile_bitfield_p.
6503 (extract_bit_field): Likewise.
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 * target.def (cstore_mode): Return a scalar_int_mode.
6510 * doc/tm.texi: Regenerate.
6511 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
6512 * targhooks.h (default_cstore_mode): Likewise.
6513 * targhooks.c (default_cstore_mode): Likewise, using a forced
6515 * expmed.c (emit_cstore): Expect the target of the cstore to be
6518 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6519 Alan Hayward <alan.hayward@arm.com>
6520 David Sherwood <david.sherwood@arm.com>
6522 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
6524 (niter_desc): Likewise mode.
6525 (iv_analyze): Add a mode parameter.
6527 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
6528 and change its type to scalar_int_mode.
6529 * loop-iv.c: Update commentary at head of file.
6530 (iv_constant): Pass the mode paraeter before the rtx it describes
6531 and change its type to scalar_int_mode. Remove VOIDmode handling.
6532 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
6533 (iv_extend): Likewise.
6534 (shorten_into_mode): Likewise.
6535 (iv_add): Use scalar_int_mode.
6536 (iv_mult): Likewise.
6537 (iv_shift): Likewise.
6538 (canonicalize_iv_subregs): Likewise.
6539 (get_biv_step_1): Pass the outer_mode parameter before the rtx
6540 it describes and change its mode to scalar_int_mode. Also change
6541 the type of the returned inner_mode to scalar_int_mode.
6542 (get_biv_step): Likewise, turning outer_mode from a pointer
6543 into a direct parameter. Update call to get_biv_step_1.
6544 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
6545 iv_constant and get_biv_step.
6546 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
6547 and change its type to scalar_int_mode. Don't initialise iv->mode
6548 to VOIDmode and remove later checks for its still being VOIDmode.
6549 Update calls to iv_analyze_op and iv_analyze_expr. Check
6550 is_a <scalar_int_mode> when changing the mode under consideration.
6551 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
6552 Update call to iv_analyze_expr.
6553 (iv_analyze_op): Add a mode parameter. Reject subregs whose
6554 inner register is not also a scalar_int_mode. Update call to
6556 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
6557 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
6558 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
6559 separate mode class checks. Update calls to iv_analyze. Remove
6560 fix-up of VOIDmodes after iv_analyze_biv.
6561 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
6562 don't have a scalar_int_mode. Update call to biv_p.
6564 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6565 Alan Hayward <alan.hayward@arm.com>
6566 David Sherwood <david.sherwood@arm.com>
6568 * cfgexpand.c (convert_debug_memory_address): Use
6569 as_a <scalar_int_mode>.
6570 * combine.c (expand_compound_operation): Likewise.
6571 (make_extraction): Likewise.
6572 (change_zero_ext): Likewise.
6573 (simplify_comparison): Likewise.
6574 * cse.c (cse_insn): Likewise.
6575 * dwarf2out.c (minmax_loc_descriptor): Likewise.
6576 (mem_loc_descriptor): Likewise.
6577 (loc_descriptor): Likewise.
6578 * expmed.c (init_expmed_one_mode): Likewise.
6579 (synth_mult): Likewise.
6580 (emit_store_flag_1): Likewise.
6581 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
6582 of a comparison with size.
6583 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
6584 (reduce_to_bit_field_precision): Likewise.
6585 * function.c (expand_function_end): Likewise.
6586 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
6587 * loop-doloop.c (doloop_modify): Likewise.
6588 * optabs.c (expand_binop): Likewise.
6589 (expand_unop): Likewise.
6590 (expand_copysign_absneg): Likewise.
6591 (prepare_cmp_insn): Likewise.
6592 (maybe_legitimize_operand): Likewise.
6593 * recog.c (const_scalar_int_operand): Likewise.
6594 * rtlanal.c (get_address_mode): Likewise.
6595 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6596 (simplify_cond_clz_ctz): Likewise.
6597 * tree-nested.c (get_nl_goto_field): Likewise.
6598 * tree.c (build_vector_type_for_mode): Likewise.
6599 * var-tracking.c (use_narrower_mode): Likewise.
6601 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6602 Alan Hayward <alan.hayward@arm.com>
6603 David Sherwood <david.sherwood@arm.com>
6605 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
6606 * builtins.c (expand_builtin_signbit): Use it.
6607 * cfgexpand.c (expand_debug_expr): Likewise.
6608 * dojump.c (do_jump): Likewise.
6609 (do_compare_and_jump): Likewise.
6610 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
6611 * expmed.c (make_tree): Likewise.
6612 * expr.c (expand_expr_real_2): Likewise.
6613 (expand_expr_real_1): Likewise.
6614 (try_casesi): Likewise.
6615 * fold-const-call.c (fold_const_call_ss): Likewise.
6616 * fold-const.c (unextend): Likewise.
6617 (extract_muldiv_1): Likewise.
6618 (fold_single_bit_test): Likewise.
6619 (native_encode_int): Likewise.
6620 (native_encode_string): Likewise.
6621 (native_interpret_int): Likewise.
6622 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
6623 * internal-fn.c (expand_addsub_overflow): Likewise.
6624 (expand_neg_overflow): Likewise.
6625 (expand_mul_overflow): Likewise.
6626 (expand_arith_overflow): Likewise.
6627 * match.pd: Likewise.
6628 * stor-layout.c (layout_type): Likewise.
6629 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
6630 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
6631 * tree-ssanames.c (get_range_info): Likewise.
6632 * tree-switch-conversion.c (array_value_type) Likewise.
6633 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
6634 (vect_recog_divmod_pattern): Likewise.
6635 (vect_recog_mixed_size_cond_pattern): Likewise.
6636 * tree-vrp.c (extract_range_basic): Likewise.
6637 (simplify_float_conversion_using_ranges): Likewise.
6638 * tree.c (int_fits_type_p): Likewise.
6639 * ubsan.c (instrument_bool_enum_load): Likewise.
6640 * varasm.c (mergeable_string_section): Likewise.
6641 (narrowing_initializer_constant_valid_p): Likewise.
6642 (output_constant): Likewise.
6644 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6645 Alan Hayward <alan.hayward@arm.com>
6646 David Sherwood <david.sherwood@arm.com>
6648 * machmode.h (NARROWEST_INT_MODE): New macro.
6649 * expr.c (alignment_for_piecewise_move): Use it instead of
6650 GET_CLASS_NARROWEST_MODE (MODE_INT).
6651 (push_block): Likewise.
6652 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
6654 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6656 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6657 Alan Hayward <alan.hayward@arm.com>
6658 David Sherwood <david.sherwood@arm.com>
6660 * postreload.c (move2add_valid_value_p): Change the type of the
6661 mode parameter to scalar_int_mode.
6662 (move2add_use_add2_insn): Add a mode parameter and use it instead
6664 (move2add_use_add3_insn): Likewise.
6665 (reload_cse_move2add): Update accordingly.
6667 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6668 Alan Hayward <alan.hayward@arm.com>
6669 David Sherwood <david.sherwood@arm.com>
6671 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
6673 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
6674 * optabs.c (expand_unop): Likewise.
6676 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6677 Alan Hayward <alan.hayward@arm.com>
6678 David Sherwood <david.sherwood@arm.com>
6680 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
6681 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
6682 (popcount_loc_descriptor): Likewise.
6683 (bswap_loc_descriptor): Likewise.
6684 (rotate_loc_descriptor): Likewise.
6685 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
6686 calling the functions above.
6688 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6689 Alan Hayward <alan.hayward@arm.com>
6690 David Sherwood <david.sherwood@arm.com>
6692 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
6694 (try_combine): Likewise.
6695 (simplify_if_then_else): Likewise.
6696 * cse.c (cse_insn): Likewise.
6697 * dwarf2out.c (mem_loc_descriptor): Likewise.
6698 * emit-rtl.c (gen_lowpart_common): Likewise.
6699 * simplify-rtx.c (simplify_truncation): Likewise.
6700 (simplify_binary_operation_1): Likewise.
6701 (simplify_const_relational_operation): Likewise.
6702 (simplify_ternary_operation): Likewise.
6703 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
6705 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6706 Alan Hayward <alan.hayward@arm.com>
6707 David Sherwood <david.sherwood@arm.com>
6709 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
6710 * reload.c (push_reload): Likewise.
6711 (find_reloads): Likewise.
6713 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6714 Alan Hayward <alan.hayward@arm.com>
6715 David Sherwood <david.sherwood@arm.com>
6717 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
6718 (make_compound_operation_int): Likewise.
6719 (change_zero_ext): Likewise.
6720 * expr.c (convert_move): Likewise.
6721 (convert_modes): Likewise.
6722 * fwprop.c (forward_propagate_subreg): Likewise.
6723 * loop-iv.c (get_biv_step_1): Likewise.
6724 * optabs.c (widen_operand): Likewise.
6725 * postreload.c (move2add_valid_value_p): Likewise.
6726 * recog.c (simplify_while_replacing): Likewise.
6727 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6728 (simplify_binary_operation_1): Likewise. Remove redundant
6729 mode equality check.
6731 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6732 Alan Hayward <alan.hayward@arm.com>
6733 David Sherwood <david.sherwood@arm.com>
6735 * combine.c (combine_simplify_rtx): Add checks for
6736 is_a <scalar_int_mode>.
6737 (simplify_if_then_else): Likewise.
6738 (make_field_assignment): Likewise.
6739 (simplify_comparison): Likewise.
6740 * ifcvt.c (noce_try_bitop): Likewise.
6741 * loop-invariant.c (canonicalize_address_mult): Likewise.
6742 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6744 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6745 Alan Hayward <alan.hayward@arm.com>
6746 David Sherwood <david.sherwood@arm.com>
6748 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
6749 is_a <scalar_int_mode> instead of != BLKmode.
6751 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6752 Alan Hayward <alan.hayward@arm.com>
6753 David Sherwood <david.sherwood@arm.com>
6755 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
6756 instead of != VOIDmode.
6757 * combine.c (if_then_else_cond): Likewise.
6758 (change_zero_ext): Likewise.
6759 * dwarf2out.c (mem_loc_descriptor): Likewise.
6760 (loc_descriptor): Likewise.
6761 * rtlanal.c (canonicalize_condition): Likewise.
6762 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
6764 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6765 Alan Hayward <alan.hayward@arm.com>
6766 David Sherwood <david.sherwood@arm.com>
6768 * simplify-rtx.c (simplify_binary_operation_1): Use
6769 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
6771 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6772 Alan Hayward <alan.hayward@arm.com>
6773 David Sherwood <david.sherwood@arm.com>
6775 * wide-int.h (int_traits<unsigned char>) New class.
6776 (int_traits<unsigned short>) Likewise.
6777 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
6778 Use GET_MODE_UNIT_PRECISION and remove redundant test for
6780 * combine.c (set_nonzero_bits_and_sign_copies): Use
6781 is_a <scalar_int_mode>.
6782 (find_split_point): Likewise.
6783 (combine_simplify_rtx): Likewise.
6784 (simplify_logical): Likewise.
6785 (expand_compound_operation): Likewise.
6786 (expand_field_assignment): Likewise.
6787 (make_compound_operation): Likewise.
6788 (extended_count): Likewise.
6789 (change_zero_ext): Likewise.
6790 (simplify_comparison): Likewise.
6791 * dwarf2out.c (scompare_loc_descriptor): Likewise.
6792 (ucompare_loc_descriptor): Likewise.
6793 (minmax_loc_descriptor): Likewise.
6794 (mem_loc_descriptor): Likewise.
6795 (loc_descriptor): Likewise.
6796 * expmed.c (init_expmed_one_mode): Likewise.
6797 * lra-constraints.c (lra_constraint_offset): Likewise.
6798 * optabs.c (prepare_libcall_arg): Likewise.
6799 * postreload.c (move2add_note_store): Likewise.
6800 * reload.c (operands_match_p): Likewise.
6801 * rtl.h (load_extend_op): Likewise.
6802 * rtlhooks.c (gen_lowpart_general): Likewise.
6803 * simplify-rtx.c (simplify_truncation): Likewise.
6804 (simplify_unary_operation_1): Likewise.
6805 (simplify_binary_operation_1): Likewise.
6806 (simplify_const_binary_operation): Likewise.
6807 (simplify_const_relational_operation): Likewise.
6808 (simplify_subreg): Likewise.
6809 * stor-layout.c (bitwise_mode_for_mode): Likewise.
6810 * var-tracking.c (adjust_mems): Likewise.
6811 (prepare_call_arguments): Likewise.
6813 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6814 Alan Hayward <alan.hayward@arm.com>
6815 David Sherwood <david.sherwood@arm.com>
6817 * machmode.h (is_int_mode): New fuction.
6818 * combine.c (find_split_point): Use it.
6819 (combine_simplify_rtx): Likewise.
6820 (simplify_if_then_else): Likewise.
6821 (simplify_set): Likewise.
6822 (simplify_shift_const_1): Likewise.
6823 (simplify_comparison): Likewise.
6824 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
6825 * cse.c (notreg_cost): Likewise.
6826 (cse_insn): Likewise.
6827 * cselib.c (cselib_lookup_1): Likewise.
6828 * dojump.c (do_jump_1): Likewise.
6829 (do_compare_rtx_and_jump): Likewise.
6830 * dse.c (get_call_args): Likewise.
6831 * dwarf2out.c (rtl_for_decl_init): Likewise.
6832 (native_encode_initializer): Likewise.
6833 * expmed.c (emit_store_flag_1): Likewise.
6834 (emit_store_flag): Likewise.
6835 * expr.c (convert_modes): Likewise.
6836 (store_field): Likewise.
6837 (expand_expr_real_1): Likewise.
6838 * fold-const.c (fold_read_from_constant_string): Likewise.
6839 * gimple-ssa-sprintf.c (get_format_string): Likewise.
6840 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
6841 * optabs.c (expand_binop): Likewise.
6842 (expand_unop): Likewise.
6843 (expand_abs_nojump): Likewise.
6844 (expand_one_cmpl_abs_nojump): Likewise.
6845 * simplify-rtx.c (mode_signbit_p): Likewise.
6846 (val_signbit_p): Likewise.
6847 (val_signbit_known_set_p): Likewise.
6848 (val_signbit_known_clear_p): Likewise.
6849 (simplify_relational_operation_1): Likewise.
6850 * tree.c (vector_type_mode): Likewise.
6852 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6853 Alan Hayward <alan.hayward@arm.com>
6854 David Sherwood <david.sherwood@arm.com>
6856 * machmode.h (smallest_mode_for_size): Fix formatting.
6857 (smallest_int_mode_for_size): New function.
6858 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
6859 instead of smallest_mode_for_size.
6860 * combine.c (make_extraction): Likewise.
6861 * config/arc/arc.c (arc_expand_movmem): Likewise.
6862 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
6863 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
6864 * config/s390/s390.c (s390_expand_insv): Likewise.
6865 * config/sparc/sparc.c (assign_int_registers): Likewise.
6866 * config/spu/spu.c (spu_function_value): Likewise.
6867 (spu_function_arg): Likewise.
6868 * coverage.c (get_gcov_type): Likewise.
6869 (get_gcov_unsigned_t): Likewise.
6870 * dse.c (find_shift_sequence): Likewise.
6871 * expmed.c (store_bit_field_1): Likewise.
6872 * expr.c (convert_move): Likewise.
6873 (store_field): Likewise.
6874 * internal-fn.c (expand_arith_overflow): Likewise.
6875 * optabs-query.c (get_best_extraction_insn): Likewise.
6876 * optabs.c (expand_twoval_binop_libfunc): Likewise.
6877 * stor-layout.c (layout_type): Likewise.
6878 (initialize_sizetypes): Likewise.
6879 * targhooks.c (default_get_mask_mode): Likewise.
6880 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
6882 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6883 Alan Hayward <alan.hayward@arm.com>
6884 David Sherwood <david.sherwood@arm.com>
6886 * machmode.h (opt_mode::else_blk): New function.
6887 (int_mode_for_mode): Declare.
6888 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
6889 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
6891 * cfgexpand.c (expand_debug_expr): Likewise.
6892 * combine.c (gen_lowpart_or_truncate): Likewise.
6893 (gen_lowpart_for_combine): Likewise.
6894 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
6895 * config/avr/avr.c (avr_to_int_mode): Likewise.
6896 (avr_out_plus_1): Likewise.
6897 (avr_out_plus): Likewise.
6898 (avr_out_round): Likewise.
6899 * config/i386/i386.c (ix86_split_to_parts): Likewise.
6900 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
6901 (s390_expand_vcond): Likewise.
6902 * config/spu/spu.c (spu_split_immediate): Likewise.
6903 (spu_expand_mov): Likewise.
6904 * dse.c (get_stored_val): Likewise.
6905 * expmed.c (store_bit_field_1): Likewise.
6906 (convert_extracted_bit_field): Use int_mode_for_mode instead of
6908 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
6909 (extract_low_bits): Likewise.
6910 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
6911 handling rather than repeating the check.
6912 (emit_group_store): Likewise.
6913 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
6914 * optabs.c (expand_absneg_bit): Likewise.
6915 (expand_copysign_absneg): Likewise.
6916 (expand_copysign_bit): Likewise.
6917 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
6918 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
6919 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
6920 * var-tracking.c (prepare_call_arguments): Likewise.
6921 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
6922 int_mode_for_mode instead of mode_for_size.
6923 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
6925 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6926 Alan Hayward <alan.hayward@arm.com>
6927 David Sherwood <david.sherwood@arm.com>
6929 * machmode.h (int_mode_for_size): New function.
6930 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
6931 instead of mode_for_size.
6932 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
6934 * combine.c (expand_field_assignment): Use int_mode_for_size
6935 instead of mode_for_size.
6936 (make_extraction): Likewise.
6937 (simplify_shift_const_1): Likewise.
6938 (simplify_comparison): Likewise.
6939 * dojump.c (do_jump): Likewise.
6940 * dwarf2out.c (mem_loc_descriptor): Likewise.
6941 * emit-rtl.c (init_derived_machine_modes): Likewise.
6942 * expmed.c (flip_storage_order): Likewise.
6943 (convert_extracted_bit_field): Likewise.
6944 * expr.c (copy_blkmode_from_reg): Likewise.
6945 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
6946 * internal-fn.c (expand_mul_overflow): Likewise.
6947 * lower-subreg.c (simple_move): Likewise.
6948 * optabs-libfuncs.c (init_optabs): Likewise.
6949 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6950 * tree.c (vector_type_mode): Likewise.
6951 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
6952 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
6953 * tree-vect-generic.c (expand_vector_parallel): Likewise.
6954 * tree-vect-stmts.c (vectorizable_load): Likewise.
6955 (vectorizable_store): Likewise.
6957 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6958 Alan Hayward <alan.hayward@arm.com>
6959 David Sherwood <david.sherwood@arm.com>
6961 * coretypes.h (pod_mode): New type.
6962 (scalar_int_mode_pod): New typedef.
6963 * machmode.h (pod_mode): New class.
6964 (int_n_data_t::m): Change type to scalar_int_mode_pod.
6965 * genmodes.c (emit_mode_int_n): Update accordingly.
6966 * lower-subreg.h (target_lower_subreg): Change type to
6967 scalar_int_mode_pod.
6968 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
6969 scalar_int_mode_pod.
6971 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6972 Alan Hayward <alan.hayward@arm.com>
6973 David Sherwood <david.sherwood@arm.com>
6975 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
6976 machine_mode to scalar_int_mode.
6977 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
6978 (rs6000_option_override_internal): Remove cast to int.
6979 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
6980 machine_mode to scalar_int_mode.
6981 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
6982 (rs6000_option_override_internal): Remove cast to int.
6983 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
6984 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
6986 * config/s390/s390.c (s390_expand_builtin): Likewise.
6987 * coretypes.h (scalar_int_mode): New type.
6988 (opt_scalar_int_mode): New typedef.
6989 * machmode.h (scalar_int_mode): New class.
6990 (scalar_int_mode::includes_p): New function.
6991 (byte_mode): Change type to scalar_int_mode.
6992 (word_mode): Likewise.
6993 (ptr_mode): Likewise.
6994 * emit-rtl.c (byte_mode): Likewise.
6995 (word_mode): Likewise.
6996 (ptr_mode): Likewise.
6997 (init_derived_machine_modes): Update accordingly.
6998 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
6999 and MODE_PARTIAL_INT.
7000 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
7001 opt_scalar_int_mode.
7003 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7004 Alan Hayward <alan.hayward@arm.com>
7005 David Sherwood <david.sherwood@arm.com>
7007 * target.def (libgcc_floating_mode_supported_p): Take a
7009 * doc/tm.texi: Regenerate.
7010 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
7012 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
7013 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
7016 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7017 Alan Hayward <alan.hayward@arm.com>
7018 David Sherwood <david.sherwood@arm.com>
7020 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
7021 * doc/tm.texi: Regenerate.
7022 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
7023 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
7024 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
7025 * targhooks.h (default_floatn_mode): Likewise.
7026 * targhooks.c (default_floatn_mode): Likewise.
7027 * tree.c (build_common_tree_nodes): Update accordingly.
7029 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7030 Alan Hayward <alan.hayward@arm.com>
7031 David Sherwood <david.sherwood@arm.com>
7033 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
7034 (mode_iterator::iterate_p): Likewise.
7035 (mode_iterator::get_wider): Likewise.
7036 * expr.c (init_expr_target): Use opt_scalar_float_mode.
7038 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7039 Alan Hayward <alan.hayward@arm.com>
7040 David Sherwood <david.sherwood@arm.com>
7042 * coretypes.h (opt_scalar_float_mode): New typedef.
7043 * machmode.h (float_mode_for_size): New function.
7044 * emit-rtl.c (double_mode): Delete.
7045 (init_emit_once): Use float_mode_for_size.
7046 * stor-layout.c (layout_type): Likewise.
7047 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
7049 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7050 Alan Hayward <alan.hayward@arm.com>
7051 David Sherwood <david.sherwood@arm.com>
7053 * output.h (assemble_real): Take a scalar_float_mode.
7054 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
7055 * config/arm/arm.md (consttable_4): Likewise.
7056 (consttable_8): Likewise.
7057 (consttable_16): Likewise.
7058 * config/mips/mips.md (consttable_float): Likewise.
7059 * config/s390/s390.c (s390_output_pool_entry): Likewise.
7060 * varasm.c (assemble_real): Take a scalar_float_mode.
7061 (output_constant_pool_2): Update accordingly.
7062 (output_constant): Likewise.
7064 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7065 Alan Hayward <alan.hayward@arm.com>
7066 David Sherwood <david.sherwood@arm.com>
7068 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
7069 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
7070 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
7071 (native_encode_real): Likewise.
7072 (native_interpret_real): Likewise.
7073 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
7074 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7076 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7077 Alan Hayward <alan.hayward@arm.com>
7078 David Sherwood <david.sherwood@arm.com>
7080 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
7081 <scalar_float_mode>. Simplify.
7082 (gen_extend_conv_libfunc): Likewise.
7084 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7085 Alan Hayward <alan.hayward@arm.com>
7086 David Sherwood <david.sherwood@arm.com>
7088 * coretypes.h (scalar_float_mode): New type.
7089 * machmode.h (mode_traits::from_int): Use machine_mode if
7090 USE_ENUM_MODES is defined.
7091 (is_a): New function.
7093 (dyn_cast): Likewise.
7094 (scalar_float_mode): New class.
7095 (scalar_float_mode::includes_p): New function.
7096 (is_float_mode): Likewise.
7097 * gdbhooks.py (MachineModePrinter): New class.
7098 (build_pretty_printer): Use it for scalar_float_mode.
7099 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
7100 (format_helper::format_helper): Turn into a template.
7101 * genmodes.c (get_mode_class): New function.
7102 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
7103 or machine_mode if none.
7104 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
7105 as_a <scalar_float_mode>.
7106 * dwarf2out.c (mem_loc_descriptor): Likewise.
7107 (insert_float): Likewise.
7108 (add_const_value_attribute): Likewise.
7109 * simplify-rtx.c (simplify_immed_subreg): Likewise.
7110 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
7111 (expand_unop): Update accordingly.
7112 (expand_abs_nojump): Likewise.
7113 (expand_copysign_absneg): Take a scalar_float_mode.
7114 (expand_copysign_bit): Likewise.
7115 (expand_copysign): Update accordingly.
7117 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7118 Alan Hayward <alan.hayward@arm.com>
7119 David Sherwood <david.sherwood@arm.com>
7121 * coretypes.h (opt_mode): New class.
7122 * machmode.h (opt_mode): Likewise.
7123 (opt_mode::else_void): New function.
7124 (opt_mode::require): Likewise.
7125 (opt_mode::exists): Likewise.
7126 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
7127 (GET_MODE_2XWIDER_MODE): Likewise.
7128 (mode_iterator::get_wider): Update accordingly.
7129 (mode_iterator::get_2xwider): Likewise.
7130 (mode_iterator::get_known_wider): Likewise, turning into a template.
7131 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
7132 forcing a wider mode to exist.
7133 * config/cr16/cr16.h (LONG_REG_P): Likewise.
7134 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
7135 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
7136 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
7137 * lower-subreg.c (init_lower_subreg): Likewise.
7138 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
7139 on the final iteration.
7140 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
7141 a wider mode exists before asking for a move pattern.
7142 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
7143 forcing a wider mode to exist.
7144 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
7145 returning false if no such mode exists.
7146 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
7147 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
7148 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
7149 Avoid checking for a MODE_INT if we already know the mode is not a
7151 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
7152 forcing a wider mode to exist.
7153 (expmed_mult_highpart_optab): Likewise.
7154 (expmed_mult_highpart): Likewise.
7155 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
7157 * lto-streamer-in.c (lto_input_mode_table): Likewise.
7158 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
7159 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
7160 * internal-fn.c (expand_mul_overflow): Update use of
7161 GET_MODE_2XWIDER_MODE.
7162 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7163 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
7164 GET_MODE_WIDER_MODE.
7165 (convert_plusminus_to_widen): Likewise.
7166 * tree-switch-conversion.c (array_value_type): Likewise.
7167 * var-tracking.c (emit_note_insn_var_location): Likewise.
7168 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7169 Return false inside rather than outside the loop if no wider mode
7171 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
7172 and GET_MODE_2XWIDER_MODE
7173 (can_compare_p): Use else_void.
7174 * gdbhooks.py (OptMachineModePrinter): New class.
7175 (build_pretty_printer): Use it for opt_mode.
7177 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7178 Alan Hayward <alan.hayward@arm.com>
7179 David Sherwood <david.sherwood@arm.com>
7181 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
7182 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
7184 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7185 Alan Hayward <alan.hayward@arm.com>
7186 David Sherwood <david.sherwood@arm.com>
7188 * machmode.h (mode_traits): New structure.
7189 (get_narrowest_mode): New function.
7190 (mode_iterator::start): Likewise.
7191 (mode_iterator::iterate_p): Likewise.
7192 (mode_iterator::get_wider): Likewise.
7193 (mode_iterator::get_known_wider): Likewise.
7194 (mode_iterator::get_2xwider): Likewise.
7195 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
7196 (FOR_EACH_MODE): Likewise.
7197 (FOR_EACH_MODE_FROM): Likewise.
7198 (FOR_EACH_MODE_UNTIL): Likewise.
7199 (FOR_EACH_WIDER_MODE): Likewise.
7200 (FOR_EACH_2XWIDER_MODE): Likewise.
7201 * builtins.c (expand_builtin_strlen): Use new mode iterators.
7202 * combine.c (simplify_comparison): Likewise
7203 * config/i386/i386.c (type_natural_mode): Likewise.
7204 * cse.c (cse_insn): Likewise.
7205 * dse.c (find_shift_sequence): Likewise.
7206 * emit-rtl.c (init_derived_machine_modes): Likewise.
7207 (init_emit_once): Likewise.
7208 * explow.c (hard_function_value): Likewise.
7209 * expmed.c (extract_fixed_bit_field_1): Likewise.
7210 (extract_bit_field_1): Likewise.
7211 (expand_divmod): Likewise.
7212 (emit_store_flag_1): Likewise.
7213 * expr.c (init_expr_target): Likewise.
7214 (convert_move): Likewise.
7215 (alignment_for_piecewise_move): Likewise.
7216 (widest_int_mode_for_size): Likewise.
7217 (emit_block_move_via_movmem): Likewise.
7218 (copy_blkmode_to_reg): Likewise.
7219 (set_storage_via_setmem): Likewise.
7220 (compress_float_constant): Likewise.
7221 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7222 * optabs-query.c (get_best_extraction_insn): Likewise.
7223 * optabs.c (expand_binop): Likewise.
7224 (expand_twoval_unop): Likewise.
7225 (expand_twoval_binop): Likewise.
7226 (widen_leading): Likewise.
7227 (widen_bswap): Likewise.
7228 (expand_parity): Likewise.
7229 (expand_unop): Likewise.
7230 (prepare_cmp_insn): Likewise.
7231 (prepare_float_lib_cmp): Likewise.
7232 (expand_float): Likewise.
7233 (expand_fix): Likewise.
7234 (expand_sfix_optab): Likewise.
7235 * postreload.c (move2add_use_add2_insn): Likewise.
7236 * reg-stack.c (reg_to_stack): Likewise.
7237 * reginfo.c (choose_hard_reg_mode): Likewise.
7238 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
7239 * stor-layout.c (mode_for_size): Likewise.
7240 (smallest_mode_for_size): Likewise.
7241 (mode_for_vector): Likewise.
7242 (finish_bitfield_representative): Likewise.
7243 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
7244 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
7245 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7246 * var-tracking.c (prepare_call_arguments): Likewise.
7248 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7249 Alan Hayward <alan.hayward@arm.com>
7250 David Sherwood <david.sherwood@arm.com>
7252 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
7253 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
7254 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
7255 * machmode.h (mode_size): Move earlier in file.
7256 (mode_precision): Likewise.
7257 (mode_inner): Likewise.
7258 (mode_nunits): Likewise.
7259 (mode_unit_size): Likewise.
7260 (unit_unit_precision): Likewise.
7261 (mode_wider): Likewise.
7262 (mode_2xwider): Likewise.
7263 (machine_mode): New class.
7264 (mode_to_bytes): New function.
7265 (mode_to_bits): Likewise.
7266 (mode_to_precision): Likewise.
7267 (mode_to_inner): Likewise.
7268 (mode_to_unit_size): Likewise.
7269 (mode_to_unit_precision): Likewise.
7270 (mode_to_nunits): Likewise.
7271 (GET_MODE_SIZE): Use mode_to_bytes.
7272 (GET_MODE_BITSIZE): Use mode_to_bits.
7273 (GET_MODE_PRECISION): Use mode_to_precision.
7274 (GET_MODE_INNER): Use mode_to_inner.
7275 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
7276 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
7277 (GET_MODE_NUNITS): Use mode_to_nunits.
7278 * system.h (ALWAYS_INLINE): New macro.
7279 * config/powerpcspe/powerpcspe-c.c
7280 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
7281 int for arg1_mode and arg2_mode.
7283 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7284 Alan Hayward <alan.hayward@arm.com>
7285 David Sherwood <david.sherwood@arm.com>
7287 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
7288 Prefix mode names with E_ in case statements.
7289 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7290 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
7291 (aarch64_split_simd_move): Likewise.
7292 (aarch64_gen_storewb_pair): Likewise.
7293 (aarch64_gen_loadwb_pair): Likewise.
7294 (aarch64_gen_store_pair): Likewise.
7295 (aarch64_gen_load_pair): Likewise.
7296 (aarch64_get_condition_code_1): Likewise.
7297 (aarch64_constant_pool_reload_icode): Likewise.
7298 (get_rsqrte_type): Likewise.
7299 (get_rsqrts_type): Likewise.
7300 (get_recpe_type): Likewise.
7301 (get_recps_type): Likewise.
7302 (aarch64_gimplify_va_arg_expr): Likewise.
7303 (aarch64_simd_container_mode): Likewise.
7304 (aarch64_emit_load_exclusive): Likewise.
7305 (aarch64_emit_store_exclusive): Likewise.
7306 (aarch64_expand_compare_and_swap): Likewise.
7307 (aarch64_gen_atomic_cas): Likewise.
7308 (aarch64_emit_bic): Likewise.
7309 (aarch64_emit_atomic_swap): Likewise.
7310 (aarch64_emit_atomic_load_op): Likewise.
7311 (aarch64_evpc_trn): Likewise.
7312 (aarch64_evpc_uzp): Likewise.
7313 (aarch64_evpc_zip): Likewise.
7314 (aarch64_evpc_ext): Likewise.
7315 (aarch64_evpc_rev): Likewise.
7316 (aarch64_evpc_dup): Likewise.
7317 (aarch64_gen_ccmp_first): Likewise.
7318 (aarch64_gen_ccmp_next): Likewise.
7319 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
7320 (alpha_emit_xfloating_libcall): Likewise.
7321 (emit_insxl): Likewise.
7322 (alpha_arg_type): Likewise.
7323 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
7324 (arc_preferred_simd_mode): Likewise.
7325 (arc_secondary_reload): Likewise.
7326 (get_arc_condition_code): Likewise.
7327 (arc_print_operand): Likewise.
7328 (arc_legitimate_constant_p): Likewise.
7329 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7330 * config/arc/arc.md (casesi_load): Likewise.
7331 (casesi_compact_jump): Likewise.
7332 * config/arc/predicates.md (proper_comparison_operator): Likewise.
7333 (cc_use_register): Likewise.
7334 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7335 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
7336 (arm_init_iwmmxt_builtins): Likewise.
7337 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
7338 (neon_expand_vector_init): Likewise.
7339 (arm_attr_length_move_neon): Likewise.
7340 (maybe_get_arm_condition_code): Likewise.
7341 (arm_emit_vector_const): Likewise.
7342 (arm_preferred_simd_mode): Likewise.
7343 (arm_output_iwmmxt_tinsr): Likewise.
7344 (thumb1_output_casesi): Likewise.
7345 (thumb2_output_casesi): Likewise.
7346 (arm_emit_load_exclusive): Likewise.
7347 (arm_emit_store_exclusive): Likewise.
7348 (arm_expand_compare_and_swap): Likewise.
7349 (arm_evpc_neon_vuzp): Likewise.
7350 (arm_evpc_neon_vzip): Likewise.
7351 (arm_evpc_neon_vrev): Likewise.
7352 (arm_evpc_neon_vtrn): Likewise.
7353 (arm_evpc_neon_vext): Likewise.
7354 (arm_validize_comparison): Likewise.
7355 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
7356 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
7357 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
7358 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
7359 (c6x_preferred_simd_mode): Likewise.
7360 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
7361 (epiphany_rtx_costs): Likewise.
7362 * config/epiphany/predicates.md (proper_comparison_operator):
7364 * config/frv/frv.c (condexec_memory_operand): Likewise.
7365 (frv_emit_move): Likewise.
7366 (output_move_single): Likewise.
7367 (output_condmove_single): Likewise.
7368 (frv_hard_regno_mode_ok): Likewise.
7369 (frv_matching_accg_mode): Likewise.
7370 * config/h8300/h8300.c (split_adds_subs): Likewise.
7371 (h8300_rtx_costs): Likewise.
7372 (h8300_print_operand): Likewise.
7373 (compute_mov_length): Likewise.
7374 (output_logical_op): Likewise.
7375 (compute_logical_op_length): Likewise.
7376 (compute_logical_op_cc): Likewise.
7377 (h8300_shift_needs_scratch_p): Likewise.
7378 (output_a_shift): Likewise.
7379 (compute_a_shift_length): Likewise.
7380 (compute_a_shift_cc): Likewise.
7381 (expand_a_rotate): Likewise.
7382 (output_a_rotate): Likewise.
7383 * config/i386/i386.c (classify_argument): Likewise.
7384 (function_arg_advance_32): Likewise.
7385 (function_arg_32): Likewise.
7386 (function_arg_64): Likewise.
7387 (function_value_64): Likewise.
7388 (ix86_gimplify_va_arg): Likewise.
7389 (ix86_legitimate_constant_p): Likewise.
7390 (put_condition_code): Likewise.
7391 (split_double_mode): Likewise.
7392 (ix86_avx256_split_vector_move_misalign): Likewise.
7393 (ix86_expand_vector_logical_operator): Likewise.
7394 (ix86_split_idivmod): Likewise.
7395 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
7396 (ix86_build_const_vector): Likewise.
7397 (ix86_build_signbit_mask): Likewise.
7398 (ix86_match_ccmode): Likewise.
7399 (ix86_cc_modes_compatible): Likewise.
7400 (ix86_expand_branch): Likewise.
7401 (ix86_expand_sse_cmp): Likewise.
7402 (ix86_expand_sse_movcc): Likewise.
7403 (ix86_expand_int_sse_cmp): Likewise.
7404 (ix86_expand_vec_perm_vpermi2): Likewise.
7405 (ix86_expand_vec_perm): Likewise.
7406 (ix86_expand_sse_unpack): Likewise.
7407 (ix86_expand_int_addcc): Likewise.
7408 (ix86_split_to_parts): Likewise.
7409 (ix86_vectorize_builtin_gather): Likewise.
7410 (ix86_vectorize_builtin_scatter): Likewise.
7411 (avx_vpermilp_parallel): Likewise.
7412 (inline_memory_move_cost): Likewise.
7413 (ix86_tieable_integer_mode_p): Likewise.
7414 (x86_maybe_negate_const_int): Likewise.
7415 (ix86_expand_vector_init_duplicate): Likewise.
7416 (ix86_expand_vector_init_one_nonzero): Likewise.
7417 (ix86_expand_vector_init_one_var): Likewise.
7418 (ix86_expand_vector_init_concat): Likewise.
7419 (ix86_expand_vector_init_interleave): Likewise.
7420 (ix86_expand_vector_init_general): Likewise.
7421 (ix86_expand_vector_set): Likewise.
7422 (ix86_expand_vector_extract): Likewise.
7423 (emit_reduc_half): Likewise.
7424 (ix86_emit_i387_round): Likewise.
7425 (ix86_mangle_type): Likewise.
7426 (ix86_expand_round_sse4): Likewise.
7427 (expand_vec_perm_blend): Likewise.
7428 (canonicalize_vector_int_perm): Likewise.
7429 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
7430 (expand_vec_perm_1): Likewise.
7431 (expand_vec_perm_interleave3): Likewise.
7432 (expand_vec_perm_even_odd_pack): Likewise.
7433 (expand_vec_perm_even_odd_1): Likewise.
7434 (expand_vec_perm_broadcast_1): Likewise.
7435 (ix86_vectorize_vec_perm_const_ok): Likewise.
7436 (ix86_expand_vecop_qihi): Likewise.
7437 (ix86_expand_mul_widen_hilo): Likewise.
7438 (ix86_expand_sse2_abs): Likewise.
7439 (ix86_expand_pextr): Likewise.
7440 (ix86_expand_pinsr): Likewise.
7441 (ix86_preferred_simd_mode): Likewise.
7442 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
7443 * config/i386/sse.md (*andnot<mode>3): Likewise.
7444 (<mask_codefor><code><mode>3<mask_name>): Likewise.
7445 (*<code><mode>3): Likewise.
7446 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
7447 (ia64_expand_atomic_op): Likewise.
7448 (ia64_arg_type): Likewise.
7449 (ia64_mode_to_int): Likewise.
7450 (ia64_scalar_mode_supported_p): Likewise.
7451 (ia64_vector_mode_supported_p): Likewise.
7452 (expand_vec_perm_broadcast): Likewise.
7453 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
7454 (iq2000_function_arg_advance): Likewise.
7455 (iq2000_function_arg): Likewise.
7456 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
7457 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
7458 (m68k_libcall_value): Likewise.
7459 (m68k_function_value): Likewise.
7460 (sched_attr_op_type): Likewise.
7461 * config/mcore/mcore.c (mcore_output_move): Likewise.
7462 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
7464 (microblaze_function_arg): Likewise.
7465 * config/mips/mips.c (mips16_build_call_stub): Likewise.
7466 (mips_print_operand): Likewise.
7467 (mips_mode_ok_for_mov_fmt_p): Likewise.
7468 (mips_vector_mode_supported_p): Likewise.
7469 (mips_preferred_simd_mode): Likewise.
7470 (mips_expand_vpc_loongson_even_odd): Likewise.
7471 (mips_expand_vec_unpack): Likewise.
7472 (mips_expand_vi_broadcast): Likewise.
7473 (mips_expand_vector_init): Likewise.
7474 (mips_expand_vec_reduc): Likewise.
7475 (mips_expand_msa_cmp): Likewise.
7476 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
7477 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
7478 (cc_flags_for_mode): Likewise.
7479 * config/msp430/msp430.c (msp430_print_operand): Likewise.
7480 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
7481 (nds32_output_casesi_pc_relative): Likewise.
7482 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7483 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
7484 (nvptx_gen_unpack): Likewise.
7485 (nvptx_gen_pack): Likewise.
7486 (nvptx_gen_shuffle): Likewise.
7487 (nvptx_gen_wcast): Likewise.
7488 (nvptx_preferred_simd_mode): Likewise.
7489 * config/pa/pa.c (pa_secondary_reload): Likewise.
7490 * config/pa/predicates.md (base14_operand): Likewise.
7491 * config/powerpcspe/powerpcspe-c.c
7492 (altivec_resolve_overloaded_builtin): Likewise.
7493 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
7495 (rs6000_preferred_simd_mode): Likewise.
7496 (output_vec_const_move): Likewise.
7497 (rs6000_expand_vector_extract): Likewise.
7498 (rs6000_split_vec_extract_var): Likewise.
7499 (reg_offset_addressing_ok_p): Likewise.
7500 (rs6000_legitimate_offset_address_p): Likewise.
7501 (rs6000_legitimize_address): Likewise.
7502 (rs6000_emit_set_const): Likewise.
7503 (rs6000_const_vec): Likewise.
7504 (rs6000_emit_move): Likewise.
7505 (spe_build_register_parallel): Likewise.
7506 (rs6000_darwin64_record_arg_recurse): Likewise.
7507 (swap_selector_for_mode): Likewise.
7508 (spe_init_builtins): Likewise.
7509 (paired_init_builtins): Likewise.
7510 (altivec_init_builtins): Likewise.
7511 (do_load_for_compare): Likewise.
7512 (rs6000_generate_compare): Likewise.
7513 (rs6000_expand_float128_convert): Likewise.
7514 (emit_load_locked): Likewise.
7515 (emit_store_conditional): Likewise.
7516 (rs6000_output_function_epilogue): Likewise.
7517 (rs6000_handle_altivec_attribute): Likewise.
7518 (rs6000_function_value): Likewise.
7519 (emit_fusion_gpr_load): Likewise.
7520 (emit_fusion_p9_load): Likewise.
7521 (emit_fusion_p9_store): Likewise.
7522 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
7523 (fusion_gpr_mem_load): Likewise.
7524 (fusion_addis_mem_combo_load): Likewise.
7525 (fusion_addis_mem_combo_store): Likewise.
7526 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
7527 (fusion_gpr_mem_load): Likewise.
7528 (fusion_addis_mem_combo_load): Likewise.
7529 (fusion_addis_mem_combo_store): Likewise.
7530 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7532 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
7533 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
7534 (rs6000_preferred_simd_mode): Likewise.
7535 (output_vec_const_move): Likewise.
7536 (rs6000_expand_vector_extract): Likewise.
7537 (rs6000_split_vec_extract_var): Likewise.
7538 (reg_offset_addressing_ok_p): Likewise.
7539 (rs6000_legitimate_offset_address_p): Likewise.
7540 (rs6000_legitimize_address): Likewise.
7541 (rs6000_emit_set_const): Likewise.
7542 (rs6000_const_vec): Likewise.
7543 (rs6000_emit_move): Likewise.
7544 (rs6000_darwin64_record_arg_recurse): Likewise.
7545 (swap_selector_for_mode): Likewise.
7546 (paired_init_builtins): Likewise.
7547 (altivec_init_builtins): Likewise.
7548 (rs6000_expand_float128_convert): Likewise.
7549 (emit_load_locked): Likewise.
7550 (emit_store_conditional): Likewise.
7551 (rs6000_output_function_epilogue): Likewise.
7552 (rs6000_handle_altivec_attribute): Likewise.
7553 (rs6000_function_value): Likewise.
7554 (emit_fusion_gpr_load): Likewise.
7555 (emit_fusion_p9_load): Likewise.
7556 (emit_fusion_p9_store): Likewise.
7557 * config/rx/rx.c (rx_gen_move_template): Likewise.
7558 (flags_from_mode): Likewise.
7559 * config/s390/predicates.md (s390_alc_comparison): Likewise.
7560 (s390_slb_comparison): Likewise.
7561 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
7562 (s390_vector_mode_supported_p): Likewise.
7563 (s390_cc_modes_compatible): Likewise.
7564 (s390_match_ccmode_set): Likewise.
7565 (s390_canonicalize_comparison): Likewise.
7566 (s390_emit_compare_and_swap): Likewise.
7567 (s390_branch_condition_mask): Likewise.
7568 (s390_rtx_costs): Likewise.
7569 (s390_secondary_reload): Likewise.
7570 (__SECONDARY_RELOAD_CASE): Likewise.
7571 (s390_expand_cs): Likewise.
7572 (s390_preferred_simd_mode): Likewise.
7573 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
7574 * config/sh/sh.c (sh_print_operand): Likewise.
7575 (dump_table): Likewise.
7576 (sh_secondary_reload): Likewise.
7577 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7578 * config/sh/sh.md (casesi_worker_1): Likewise.
7579 (casesi_worker_2): Likewise.
7580 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
7581 (fcc_comparison_operator): Likewise.
7582 * config/sparc/sparc.c (sparc_expand_move): Likewise.
7583 (emit_soft_tfmode_cvt): Likewise.
7584 (sparc_preferred_simd_mode): Likewise.
7585 (output_cbranch): Likewise.
7586 (sparc_print_operand): Likewise.
7587 (sparc_expand_vec_perm_bmask): Likewise.
7588 (vector_init_bshuffle): Likewise.
7589 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
7590 (spu_vector_mode_supported_p): Likewise.
7591 (spu_expand_insv): Likewise.
7592 (spu_emit_branch_or_set): Likewise.
7593 (spu_handle_vector_attribute): Likewise.
7594 (spu_builtin_splats): Likewise.
7595 (spu_builtin_extract): Likewise.
7596 (spu_builtin_promote): Likewise.
7597 (spu_expand_sign_extend): Likewise.
7598 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
7599 (tilegx_simd_int): Likewise.
7600 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
7601 (tilepro_simd_int): Likewise.
7602 * config/v850/v850.c (const_double_split): Likewise.
7603 (v850_print_operand): Likewise.
7604 (ep_memory_offset): Likewise.
7605 * config/vax/vax.c (vax_rtx_costs): Likewise.
7606 (vax_output_int_move): Likewise.
7607 (vax_output_int_add): Likewise.
7608 (vax_output_int_subtract): Likewise.
7609 * config/visium/predicates.md (visium_branch_operator): Likewise.
7610 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
7611 (visium_print_operand_address): Likewise.
7612 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7613 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
7614 (xtensa_expand_conditional_branch): Likewise.
7615 (xtensa_copy_incoming_a7): Likewise.
7616 (xtensa_output_literal): Likewise.
7617 * dfp.c (decimal_real_maxval): Likewise.
7618 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
7620 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7621 Alan Hayward <alan.hayward@arm.com>
7622 David Sherwood <david.sherwood@arm.com>
7624 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
7625 (mode_nunits_inline): Likewise.
7626 (mode_inner_inline): Likewise.
7627 (mode_unit_size_inline): Likewise.
7628 (mode_unit_precision_inline): Likewise.
7629 (emit_insn_modes_h): Likewise. Also emit a #define of the
7631 (emit_mode_wider): Add an E_ prefix to mode names.
7632 (emit_mode_complex): Likewise.
7633 (emit_mode_inner): Likewise.
7634 (emit_mode_adjustments): Likewise.
7635 (emit_mode_int_n): Likewise.
7636 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
7637 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
7638 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
7639 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
7640 (CRC32_BUILTIN, ENTRY): Likewise.
7641 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
7642 (aarch64_pop_regs): Likewise.
7643 (aarch64_process_components): Likewise.
7644 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
7645 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
7646 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
7647 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
7648 * config/arm/arm.c (arm_init_libfuncs): Likewise.
7649 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
7651 * config/i386/i386-builtin.def (pcmpestr): Likewise.
7652 (pcmpistr): Likewise.
7653 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
7654 * config/mmix/mmix.c (mmix_output_condition): Likewise.
7655 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
7657 * config/rl78/rl78.c (mduc_regs): Likewise.
7658 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
7659 (htm_expand_builtin): Likewise.
7660 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
7661 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
7662 * config/xtensa/xtensa.c (print_operand): Likewise.
7663 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
7664 (NUM_MODE_VECTOR_INT): Likewise.
7665 * genoutput.c (null_operand): Likewise.
7666 (output_operand_data): Likewise.
7667 * genrecog.c (print_parameter_value): Likewise.
7668 * lra.c (debug_operand_data): Likewise.
7670 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7672 * dwarf2out.c (scompare_loc_descriptor_wide)
7673 (scompare_loc_descriptor_narrow): New functions, split out from...
7674 (scompare_loc_descriptor): ...here.
7675 * expmed.c (emit_store_flag_int): New function, split out from...
7676 (emit_store_flag): ...here.
7678 2017-08-30 Richard Biener <rguenther@suse.de>
7680 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
7681 (dwarf2out_early_finish): Move setting of AT_pubnames from
7682 early debug output to early finish.
7684 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
7686 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
7687 and -mdata-region to the assembler.
7689 2017-08-30 Richard Biener <rguenther@suse.de>
7691 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
7693 (gen_subprogram_die): Add DW_AT_object_pointer only early.
7694 (dwarf2out_early_global_decl): Only generate a DIE for the
7695 abstract origin if it doesn't already exist or is a declaration DIE.
7696 (resolve_addr): Do not add the linkage name twice when
7697 generating a stub DIE for the DW_TAG_GNU_call_site target.
7699 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7701 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7702 Use machine_mode rather than int for arg1_mode.
7704 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
7707 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
7708 that the second argument of the built-in functions to unpack
7709 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
7710 switch statement instead a lot of if statements.
7711 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
7712 Allow 64-bit values to be in Altivec registers as well as
7713 traditional floating point registers.
7714 (pack<mode>, FMOVE128_VSX iterator): Likewise.
7716 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
7718 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
7719 MAX_REGS_PER_ADDRESS == 1.
7721 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
7723 * config/i386/i386.opt (flag_fentry): Do not init to -1.
7724 * config/i386/i386.c (ix86_option_override_internal): Simplify
7725 setting of opts->x_flag_entry.
7727 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7728 Jakub Jelinek <jakub@redhat.com>
7729 Richard Biener <rguenther@suse.de>
7731 PR tree-optimization/81503
7732 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
7733 folded constant fits in the target type; reorder tests for clarity.
7735 2017-08-29 Martin Liska <mliska@suse.cz>
7737 * passes.def: Include pass_lower_switch.
7738 * stmt.c (dump_case_nodes): Remove and move to
7739 tree-switch-conversion.
7740 (case_values_threshold): Likewise.
7741 (expand_switch_as_decision_tree_p): Likewise.
7742 (emit_case_decision_tree): Likewise.
7743 (expand_case): Likewise.
7744 (balance_case_nodes): Likewise.
7745 (node_has_low_bound): Likewise.
7746 (node_has_high_bound): Likewise.
7747 (node_is_bounded): Likewise.
7748 (emit_case_nodes): Likewise.
7749 (struct simple_case_node): New struct.
7750 (add_case_node): Remove.
7751 (emit_case_dispatch_table): Use vector instead of case_list.
7752 (reset_out_edges_aux): Remove.
7753 (compute_cases_per_edge): Likewise.
7754 (expand_case): Build list of simple_case_node.
7755 (expand_sjlj_dispatch_table): Use it.
7756 * tree-switch-conversion.c (struct case_node): Moved from
7757 stmt.c and adjusted.
7758 (emit_case_nodes): Likewise.
7759 (node_has_low_bound): Likewise.
7760 (node_has_high_bound): Likewise.
7761 (node_is_bounded): Likewise.
7762 (case_values_threshold): Likewise.
7763 (reset_out_edges_aux): Likewise.
7764 (compute_cases_per_edge): Likewise.
7765 (add_case_node): Likewise.
7766 (dump_case_nodes): Likewise.
7767 (balance_case_nodes): Likewise.
7768 (expand_switch_as_decision_tree_p): Likewise.
7769 (emit_jump): Likewise.
7770 (emit_case_decision_tree): Likewise.
7771 (try_switch_expansion): Likewise.
7772 (do_jump_if_equal): Likewise.
7773 (emit_cmp_and_jump_insns): Likewise.
7774 (fix_phi_operands_for_edge): New function.
7775 (record_phi_operand_mapping): Likewise.
7776 (class pass_lower_switch): New pass.
7777 (pass_lower_switch::execute): New function.
7778 (make_pass_lower_switch): Likewise.
7779 (conditional_probability):
7780 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
7781 * tree-pass.h: Add make_pass_lower_switch.
7783 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
7786 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
7789 2017-08-29 Richard Biener <rguenther@suse.de>
7791 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
7792 we do not add a DW_AT_inline attribute twice.
7793 (gen_subprogram_die): Remove code setting DW_AT_inline on
7794 DECL_ABSTRACT_P nodes.
7796 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
7798 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
7799 calls to internal functions.
7800 (gimplify_modify_expr): Likewise.
7801 * tree-call-cdce.c (use_internal_fn): Likewise.
7802 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
7803 (convert_to_divmod): Set the nothrow flag.
7804 * tree-if-conv.c (predicate_mem_writes): Likewise.
7805 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7806 (vectorizable_call): Likewise.
7807 (vectorizable_store): Likewise.
7808 (vectorizable_load): Likewise.
7809 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
7810 (vect_recog_mask_conversion_pattern): Likewise.
7812 2017-08-29 Martin Liska <mliska@suse.cz>
7815 * gcc.c (driver_handle_option): Add new argument.
7816 * opts-common.c (handle_option): Pass
7817 target_option_override_hook.
7818 * opts-global.c (lang_handle_option): Add new option.
7819 (set_default_handlers): Add new argument.
7820 (decode_options): Likewise.
7821 * opts.c (target_handle_option): Likewise.
7822 (common_handle_option): Call target_option_override_hook.
7823 * opts.h (struct cl_option_handler_func): Add hook for
7824 target option override.
7825 (struct cl_option_handlers): Likewise.
7826 (set_default_handlers): Add new argument.
7827 (decode_options): Likewise.
7828 (common_handle_option): Likewise.
7829 (target_handle_option): Likewise.
7830 * toplev.c (toplev::main): Pass targetm.target_option.override
7833 2017-08-29 Richard Biener <rguenther@suse.de>
7834 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7836 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
7837 life to the active subtree.
7839 2017-08-28 Jeff Law <law@redhat.com>
7841 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
7842 derive_equivalences.
7843 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
7845 (edge_info::derive_equivalences): New private member function
7847 * tree-ssa-dom.c (class edge_info): Changed from a struct
7848 to a class. Add ctor/dtor, methods and data members.
7849 (edge_info::edge_info): Renamed from allocate_edge_info.
7850 Initialize additional members.
7851 (edge_info::~edge_info): New.
7852 (free_dom_edge_info): Delete the edge info.
7853 (record_edge_info): Use new class & associated member functions.
7854 Tighten forms for testing for edge equivalences.
7855 (record_temporary_equivalences): Iterate over the simple
7856 equivalences rather than assuming there's only one per edge.
7857 (cprop_into_successor_phis): Iterate over the simple
7858 equivalences rather than assuming there's only one per edge.
7859 (optimize_stmt): Use operand_equal_p rather than pointer
7860 equality for mini-DSE code.
7862 2017-08-28 Nathan Sidwell <nathan@acm.org>
7864 * gcc.c (execute): Fold SIGPIPE handling into switch
7865 statement. Adjust internal error message.
7867 2017-08-28 Richard Biener <rguenther@suse.de>
7870 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
7871 Do nothing for removed DIEs.
7873 2017-08-28 Richard Biener <rguenther@suse.de>
7875 PR tree-optimization/81977
7876 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
7879 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
7882 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
7883 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
7884 using targetm.gen_mem_thread_fence.
7886 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
7889 * config/i386/i386.md (*<btsc><mode>): Change operand 2
7890 predicate to register_operand. Reorder operands.
7891 (*btr<mode>): Ditto.
7892 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
7893 (*btr<mode>_mask): Ditto.
7895 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
7897 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
7898 * config/rs6000/xmmintrin.h: New file.
7899 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
7901 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7904 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
7905 parameter and_insn and return it.
7906 (recombine_lvx_pattern): Insert a copy to ensure availability of
7907 the base register of the copied masking operation at the point of
7908 the instruction replacement.
7909 (recombine_stvx_pattern): Likewise.
7911 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
7913 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
7914 undocumented switches.
7915 (-mpower9-dform-vector): Likewise.
7916 (-mpower9-dform): Likewise.
7917 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
7918 comments to delete references to -mpower9-dform* switches.
7919 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
7920 Delete reference to -mpower9-dform* switches, test for
7921 -mpower9-vector instead.
7922 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
7923 (OTHER_P9_VECTOR_MASKS): Likewise.
7924 (POWERPC_MASKS): Likewise.
7925 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
7926 tests against -mpower9-dform* to -mpower9-vector. Delete code
7927 that checked for -mpower9-dform* consistancy with other options.
7928 Add test for -mpower9-misc to enable other power9 switches.
7929 (rs6000_init_hard_regno_mode_ok): Likewise.
7930 (rs6000_option_override_internal): Likewise.
7931 (rs6000_emit_prologue): Likewise.
7932 (rs6000_emit_epilogue): Likewise.
7933 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
7934 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
7935 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
7937 (emit_fusion_p9_store): Likewise.
7938 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
7939 resetting these macros if the assembler does not support ISA 3.0
7941 (TARGET_P9_DFORM_VECTOR): Likewise.
7942 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
7943 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
7945 2017-08-25 Alan Modra <amodra@gmail.com>
7947 PR rtl-optimization/81747
7948 * cse.c (cse_extended_basic_block): Don't attempt to record
7949 equivalences for degenerate conditional branches that branch
7950 to their fall-through.
7952 2017-08-24 Martin Sebor <msebor@redhat.com>
7955 * gimple-fold.c (size_must_be_zero_p): New function.
7956 (gimple_fold_builtin_memory_op): Call it.
7958 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
7960 * config/rs6000/mm_malloc.h: New file.
7962 2017-08-24 Bin Cheng <bin.cheng@arm.com>
7964 PR tree-optimization/81913
7965 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
7966 analysis when either IVs in condition can wrap.
7968 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
7970 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
7971 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
7973 2017-08-24 Richard Biener <rguenther@suse.de>
7976 * targhooks.c (default_target_can_inline_p): Properly
7977 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
7978 is present and always compare.
7979 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
7980 infer -mfpmath=sse from TARGET_SSE_P.
7981 (ix86_can_inline_p): Properly use target_option_default_node when
7982 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
7984 2017-08-24 Richard Biener <rguenther@suse.de>
7987 * dwarf2out.c (output_die): Handle flag_generate_offload like
7989 (output_comp_unit): Likewise.
7990 (gen_array_type_die): Likewise.
7991 (dwarf2out_early_finish): Likewise.
7992 (note_variable_value_in_expr): Likewise.
7993 (dwarf2out_finish): Likewise. Adjust assert.
7994 * cgraphunit.c (symbol_table::compile): Move setting of
7995 flag_generate_offload earlier ...
7996 (symbol_table::finalize_compilation_unit): ... here, before
7997 early debug finalization.
7999 2017-08-24 Richard Biener <rguenther@suse.de>
8001 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
8002 and ipa-fnsummary.h.
8003 (ix86_can_inline_p): When ix86_fpmath flags do not match
8004 check whether the callee uses FP math at all.
8006 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
8009 * tree-ssanames.c (get_nonzero_bits): Use element_precision
8010 instead of TYPE_PRECISION.
8012 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
8013 Alan Hayward <alan.hayward@arm.com>
8014 David Sherwood <david.sherwood@arm.com>
8016 * combine.c (make_extraction): Use subreg_offset_from_lsb.
8018 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
8020 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
8022 (ix86_frame::stack_realign_allocate): New field.
8023 (struct machine_frame_state): Modify comments.
8024 (machine_frame_state::sp_realigned_fp_end): New field.
8025 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
8027 (sp_valid_at): Add assertion to assure no attempt to access invalid
8028 offset of a realigned stack.
8029 (fp_valid_at): Likewise.
8030 (choose_baseaddr): Modify comments.
8031 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
8032 ix86_expand_prologue.
8033 (ix86_expand_prologue): Modify stack realignment and allocation.
8034 (ix86_expand_epilogue): Modify comments.
8035 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
8036 avx2_runtime, avx512f, and avx512f_runtime.
8038 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
8040 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
8041 (mstackrealign): Do not init to -1.
8042 * config/i386/i386.c (ix86_option_override_internal):
8043 Check opts_set, not opts when setting default value of
8044 opts->x_ix86_force_align_arg_pointer.
8046 2017-08-23 Richard Biener <rguenther@suse.de>
8048 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
8049 lang_hooks.decl_printable_name.
8050 * print-rtl-function.c (print_rtx_function): Likewise.
8051 * tree-pretty-print.c (dump_function_header): Likewise.
8053 2017-08-23 Richard Biener <rguenther@suse.de>
8056 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
8057 -g0 at compile-time.
8059 2017-08-23 Tamar Christina <tamar.christina@arm.com>
8062 * doc/sourcebuild.texi (Other hardware attributes):
8065 2017-08-23 Tamar Christina <tamar.christina@arm.com>
8068 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
8069 Fix single-use check.
8071 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
8073 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
8075 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
8077 * doc/install.texi: Modify to add more details on running selected
8080 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
8082 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
8083 is combined with -mabi=ms.
8084 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
8087 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8089 PR tree-optimization/81488
8090 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
8091 and cached_basis fields.
8092 (MAX_SPREAD): New constant.
8093 (alloc_cand_and_find_basis): Initialize new fields.
8094 (clear_visited): New function.
8095 (create_phi_basis_1): Rename from create_phi_basis, set visited
8096 and cached_basis fields.
8097 (create_phi_basis): New wrapper function.
8098 (phi_add_costs_1): Rename from phi_add_costs, add spread
8099 parameter, set visited field, short-circuit when limits reached.
8100 (phi_add_costs): New wrapper function.
8101 (record_phi_increments_1): Rename from record_phi_increments, set
8103 (record_phi_increments): New wrapper function.
8104 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
8105 (phi_incr_cost): New wrapper function.
8106 (all_phi_incrs_profitable_1): Rename from
8107 all_phi_incrs_profitable, set visited field.
8108 (all_phi_incrs_profitable): New wrapper function.
8110 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
8111 Alan Hayward <alan.hayward@arm.com>
8112 David Sherwood <david.sherwood@arm.com>
8114 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
8115 that takes the outer and inner modes.
8116 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
8117 comparison as the canonical test for a paradoxical subreg.
8118 * combine.c (simplify_set): Use paradoxical_subreg_p.
8119 (make_extraction): Likewise.
8120 (force_to_mode): Likewise.
8121 (rtx_equal_for_field_assignment_p): Likewise.
8122 (gen_lowpart_for_combine): Likewise.
8123 (simplify_comparison): Likewise.
8124 * cse.c (equiv_constant): Likewise.
8125 * expmed.c (store_bit_field_1): Likewise.
8126 * final.c (alter_subreg): Likewise.
8127 * fwprop.c (propagate_rtx): Likewise.
8128 (forward_propagate_subreg): Likewise.
8129 * ira-conflicts.c (ira_build_conflicts): Likewise.
8130 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
8131 * lra-constraints.c (curr_insn_transform): Likewise.
8132 (split_reg): Likewise.
8133 * lra-eliminations.c (move_plus_up): Likewise.
8134 (lra_eliminate_regs_1): Likewise.
8135 * recog.c (general_operand): Likewise.
8136 * ree.c (combine_reaching_defs): Likewise.
8137 * reload.c (push_reload): Likewise.
8138 (find_reloads): Likewise.
8139 * reload1.c (elimination_effects): Likewise.
8140 (compute_reload_subreg_offset): Likewise.
8141 (choose_reload_regs): Likewise.
8142 * rtlanal.c (subreg_lsb_1): Likewise.
8143 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8144 (simplify_subreg): Likewise.
8145 * var-tracking.c (track_loc_p): Likewise.
8146 * emit-rtl.c (byte_lowpart_offset): Likewise.
8147 (paradoxical_subreg_p): Delete out-of-line definition.
8149 2017-08-22 Jeff Law <law@redhat.com>
8151 PR tree-optimization/81741
8152 PR tree-optimization/71947
8153 * tree-ssa-dom.c: Include tree-inline.h.
8154 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
8155 equivalences if one is more expensive to compute than the other.
8156 * tree-ssa-scopedtables.h (class const_or_copies): Make
8157 record_const_or_copy_raw method private.
8158 (class avail_exprs_stack): New method simplify_binary_operation.
8159 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
8160 avail_exprs_stack::simplify_binary_operation as needed.
8161 (avail_exprs_stack::simplify_binary_operation): New function.
8163 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
8165 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
8166 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
8167 (DOT_SYMBOLS): Likewise.
8168 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
8169 (RELOCATABLE_NEEDS_FIXUP): Likewise.
8170 (RS6000_ABI_NAME): Likewise.
8171 (TARGET_CMODEL): Likewise.
8172 (TOC_SECTION_ASM_OP): Likewise.
8173 (SET_CMODEL): New macro.
8174 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
8176 2017-08-22 Richard Biener <rguenther@suse.de>
8178 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
8179 to allow for free-lang-data replacements similar to verify_type_variant.
8181 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
8182 Alan Hayward <alan.hayward@arm.com>
8183 David Sherwood <david.sherwood@arm.com>
8185 * config/aarch64/aarch64.md (casesi): Use DImode rather than
8186 VOIDmode for the LABEL_REF.
8188 2017-08-22 Richard Biener <rguenther@suse.de>
8190 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
8192 2017-08-22 Richard Biener <rguenther@suse.de>
8194 * common.opt (feliminate-dwarf2-dups): Ignore.
8195 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
8196 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
8197 same_die_p_wrap, compute_section_prefix,
8198 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
8199 (comdat_symbol_id, comdat_symbol_number): Likewise.
8200 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
8202 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
8203 (output_die): Mark unreachable path unreachable.
8204 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
8205 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
8206 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
8207 (dwarf2out_early_finish): Likewise.
8209 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
8211 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
8213 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
8216 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
8217 not VAR_P. Filter attribute warnings with OPT_Wattributes.
8218 (avr_attribute_table) <io, io_low, address>: Initialize
8219 .decl_required with true.
8221 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
8223 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
8224 undocumented debugging options.
8225 (-mvsx-scalar-double): Likewise.
8226 (-mallow-df-permute): Likewise.
8227 (-mvectorize-builtins): Likewise.
8228 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8229 (rs6000_builtin_vectorized_function): Likewise.
8230 (rs6000_builtin_md_vectorized_function): Likewise.
8231 (rs6000_opt_vars): Likewise.
8233 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
8236 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
8237 (*btrq_imm): Rename from *btrq.
8238 (*btcq_imm): Rename from *btcq.
8239 (btsc): New code attribute.
8240 (*<btsc><mode>): New insn pattern.
8241 (*btr<mode>): Ditto.
8242 (*<btsc><mode>_mask): New insn_and_split pattern.
8243 (*btr<mode>_mask): Ditto.
8245 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8246 Alan Hayward <alan.hayward@arm.com>
8247 David Sherwood <david.sherwood@arm.com>
8249 * function.c (pad_below): Simplify padding calculation.
8251 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8252 Alan Hayward <alan.hayward@arm.com>
8253 David Sherwood <david.sherwood@arm.com>
8255 * target.def (function_prologue): Remove frame size argument.
8256 (function_epilogue): Likewise.
8257 * doc/tm.texi: Regenerate.
8258 * final.c (final_start_function): Update call to function_prologue.
8259 (final_end_function): Update call to function_epilogue.
8260 (default_function_pro_epilogue): Remove frame size argument.
8261 * output.h (default_function_pro_epilogue): Likewise.
8262 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
8263 (arm_output_function_prologue): Likewise.
8264 * config/frv/frv.c (frv_function_prologue): Likewise.
8265 (frv_function_epilogue): Likewise.
8266 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
8267 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
8268 (ia64_output_function_epilogue): Likewise.
8269 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
8270 (m32r_output_function_epilogue): Likewise.
8271 * config/microblaze/microblaze.c (microblaze_function_prologue)
8272 (microblaze_function_epilogue): Likewise.
8273 * config/mips/mips.c (mips_output_function_prologue): Likewise.
8274 (mips_output_function_epilogue): Likewise.
8275 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
8276 (mmix_target_asm_function_epilogue): Likewise.
8277 * config/msp430/msp430.c (msp430_start_function): Likewise.
8278 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
8279 (nds32_asm_function_epilogue): Likewise.
8280 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
8281 * config/pa/pa.c (pa_output_function_prologue): Likewise.
8282 (pa_output_function_epilogue): Likewise.
8283 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
8284 (rs6000_output_function_epilogue): Likewise.
8285 * config/rl78/rl78.c (rl78_start_function): Likewise.
8286 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
8287 (rs6000_output_function_epilogue): Likewise.
8288 * config/rx/rx.c (rx_output_function_prologue): Likewise.
8289 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
8290 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
8291 (sparc_asm_function_epilogue): Likewise.
8293 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8295 * tree.h (type_has_mode_precision_p): New function.
8296 * convert.c (convert_to_integer_1): Use it.
8297 * expr.c (expand_expr_real_2): Likewise.
8298 (expand_expr_real_1): Likewise.
8299 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
8300 * match.pd: Likewise.
8301 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
8302 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
8303 * tree-tailcall.c (process_assignment): Likewise.
8304 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8305 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
8306 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
8307 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8308 (vectorizable_assignment): Likewise.
8309 (vectorizable_shift): Likewise.
8310 (vectorizable_operation): Likewise.
8311 * tree-vrp.c (register_edge_assert_for_2): Likewise.
8313 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
8315 * match.pd: Add pow (C, x) simplification.
8317 2017-08-21 Richard Biener <rguenther@suse.de>
8319 PR tree-optimization/81900
8320 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
8321 for blocks with abnormal predecessors.
8322 (compute_antic): Do not set visited flag prematurely.
8324 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
8327 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
8329 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8331 * stor-layout.h (vector_type_mode): Move to...
8332 * tree.h (vector_type_mode): ...here.
8333 * stor-layout.c (vector_type_mode): Move to...
8334 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
8336 2017-08-21 Richard Biener <rguenther@suse.de>
8338 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
8339 register_external_die hooks.
8340 (debug_false_tree_charstarstar_uhwistar): Declare.
8341 (debug_nothing_tree_charstar_uhwi): Likewise.
8342 * debug.c (do_nothing_debug_hooks): Adjust.
8343 (debug_false_tree_charstarstar_uhwistar): New do nothing.
8344 (debug_nothing_tree_charstar_uhwi): Likewise.
8345 * dbxout.c (dbx_debug_hooks): Adjust.
8346 (xcoff_debug_hooks): Likewise.
8347 * sdbout.c (sdb_debug_hooks): Likewise.
8348 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
8349 * dwarf2out.c (macinfo_label_base): New global.
8350 (dwarf2out_register_external_die): New function for the
8351 register_external_die hook.
8352 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
8353 (dwarf2_debug_hooks): Use them.
8354 (dwarf2_lineno_debug_hooks): Adjust.
8355 (struct die_struct): Add with_offset flag.
8356 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
8357 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
8358 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
8359 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
8360 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
8361 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
8362 defining section names for the early LTO debug variants.
8363 (reset_indirect_string): New helper.
8364 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
8365 (print_dw_val): Add support for offsetted symbol references.
8366 (get_ultimate_context): Split out from is_cxx.
8367 (is_cxx): Use get_ultimate_context.
8368 (is_fortran): Add decl overload.
8369 (compute_comp_unit_symbol): Split out worker from
8370 compute_section_prefix.
8371 (compute_section_prefix): Call compute_comp_unit_symbol and
8372 set comdat_type_p here.
8373 (output_die): Skip DIE symbol output for the LTO added one.
8374 Handle DIE symbol references with offset.
8375 (output_comp_unit): Guard section name mangling properly.
8376 For LTO debug sections emit a symbol at the section beginning
8377 which we use to refer to its DIEs.
8378 (add_abstract_origin_attribute): For DIEs registered via
8379 dwarf2out_register_external_die directly refer to the early
8380 DIE rather than indirectly through the shadow one we created.
8381 Remove obsolete call to dwarf2out_abstract_function for
8382 non-function/block origins.
8383 (gen_array_type_die): When generating early LTO debug do
8384 not emit DW_AT_string_length.
8385 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
8386 late when in LTO. As suggested place a gcc_unreachable for
8387 the DECL_ABSTRACT_P case.
8388 (gen_subprogram_die): Avoid another specification DIE
8389 for early built declarations/definitions for the late LTO case.
8390 (gen_variable_die): Add type references for late duplicated VLA dies
8392 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
8393 we have the abstract instance already.
8394 (process_scope_var): Adjust decl DIE contexts in LTO which
8395 first puts them in limbo.
8396 (gen_decl_die): Do not generate type DIEs late apart from
8397 types for VLAs or for decls we do not yet have a DIE. Do not
8398 call dwarf2out_abstract_function late.
8399 (dwarf2out_early_global_decl): Make sure to create DIEs
8400 for abstract instances of a decl first.
8401 (dwarf2out_late_global_decl): Adjust comment.
8402 (output_macinfo_op): With multiple macro sections use
8403 macinfo_label_base to distinguish labels.
8404 (output_macinfo): Likewise. Update macinfo_label_base.
8405 Pass in the line info label.
8406 (note_variable_value_in_expr): When generating LTO resolve
8407 all variable values here by generating DIEs as needed.
8408 (init_sections_and_labels): Add early LTO debug flag parameter
8409 and generate different sections and names if set. Add generation
8410 counter for the labels so we can have multiple of them.
8411 (reset_dies): Helper to allow DIEs to be output multiple times.
8412 (dwarf2out_finish): When outputting DIEs to the fat part of an
8413 LTO object first reset DIEs.
8414 (dwarf2out_early_finish): Output early DIEs when generating LTO.
8415 (modified_type_die): Check for decl_ultimate_origin being self
8417 (gen_type_die_with_usage): Likewise.
8418 (gen_typedef_die): Allow decl_ultimate_origin being self.
8419 (set_decl_abstract_flags): Remove.
8420 (set_block_abstract_flags): Likewise.
8421 (dwarf2out_abstract_function): Treat the early generated DIEs
8422 as the abstract copy and only add DW_AT_inline and
8423 DW_AT_artificial here and call set_decl_origin_self.
8424 If the DIE has an abstract origin don't do anything.
8425 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
8426 if we have none yet (Go fails to build one, PR78628).
8427 (variably_modified_type_p): Prevent endless recursion for Ada
8428 cyclic pointer types.
8429 * lto-streamer-in.c: Include debug.h.
8430 (dref_queue): New global.
8431 (lto_read_tree_1): Stream in DIE references.
8432 (lto_input_tree): Register DIE references.
8433 (input_function): Stream DECL_DEBUG_ARGS.
8434 * lto-streamer-out.c: Include debug.h.
8435 (lto_write_tree_1): Output DIE references.
8436 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
8437 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
8438 (output_function): Stream DECL_DEBUG_ARGS.
8439 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
8440 Stream DECL_ABSTRACT_ORIGIN.
8441 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
8442 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
8443 DECL_CONTEXT for file-scope decls.
8444 * lto-streamer.h (struct dref_entry): Declare.
8445 (dref_queue): Likewise.
8446 * cfgexpand.c (pass_expand::execute): Do not call the
8447 outlining_inline_function hook here.
8448 * lto-wrapper.c (debug_obj): New global.
8449 (tool_cleanup): Unlink it if required.
8450 (debug_objcopy): New function.
8451 (run_gcc): Handle early debug sections in the IL files by
8452 extracting them to separate files, partially linkin them and
8453 feeding the result back as result to the linker.
8454 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
8455 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
8456 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
8457 sections into a separate segment.
8458 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
8460 (darwin_asm_dwarf_section): Likewise.
8461 (darwin_asm_output_dwarf_offset): Likewise.
8462 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
8464 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8465 Alan Hayward <alan.hayward@arm.com>
8466 David Sherwood <david.sherwood@arm.com>
8468 * read-md.h (md_reader::record_potential_iterator_use): Replace
8469 pointer argument with an rtx and an index.
8470 * read-rtl.c (iterator_group::apply_iterator): Likewise.
8471 (apply_mode_iterator): Likewise.
8472 (apply_code_iterator): Likewise.
8473 (apply_int_iterator): Likewise.
8474 (apply_subst_iterator): Likewise.
8475 (record_iterator_use): Likewise.
8476 (record_attribute_use): Likewise.
8477 (md_reader::record_potential_iterator_use): Likewise. Update calls
8478 to record_iterator_use and apply_iterator.
8479 (iterator_use): Replace ptr with x and index.
8480 (attribute_use): Likewise.
8481 (apply_attribute_uses): Update calls to apply_iterator.
8482 (apply_iterators): Likewise. Update initialization of iterator_use.
8483 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
8484 and record_potential_iterator_use.
8485 (rtx_reader::read_rtx_operand): Likewise.
8487 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8488 Alan Hayward <alan.hayward@arm.com>
8489 David Sherwood <david.sherwood@arm.com>
8491 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
8494 2017-08-21 Richard Biener <rguenther@suse.de>
8497 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
8498 at struct end conservatively when comparing common bases.
8500 2017-08-21 Richard Biener <rguenther@suse.de>
8502 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
8503 (mem_ref_in_stmt): Remove.
8504 (determine_max_movement): Use ref index to get at the reference.
8505 (invariantness_dom_walker::before_dom_children): Deal with
8506 lim data already initialized.
8507 (gather_mem_refs_stmt): Initialize lim data and record ref index.
8509 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
8511 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
8512 (TARGET_ISA_ROUND): Ditto.
8513 (TARGET_ROUND): Ditto.
8514 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
8515 * config/i386/i386.md: Ditto.
8516 * config/i386/sse.md: Ditto.
8517 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
8518 with OPTION_MASK_ISA_SSE4_1.
8520 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
8523 * doc/extend.texi (x86 Built-in Functions): Correct the name of
8524 __builtin_ia32_lzcnt_u16.
8526 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
8529 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
8530 (rs6000_set_current_function): Rewrite function to use it.
8532 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
8535 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
8536 and TYPE_WARN_IF_NOT_ALIGN.
8537 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
8538 (handle_warn_if_not_align): New.
8539 (place_union_field): Call handle_warn_if_not_align.
8540 (place_field): Call handle_warn_if_not_align.
8541 Copy TYPE_WARN_IF_NOT_ALIGN.
8542 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
8543 (layout_type): Likewise.
8544 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
8546 (tree_decl_common): Add warn_if_not_align.
8547 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
8548 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
8549 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
8550 (DECL_WARN_IF_NOT_ALIGN): Likewise.
8551 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
8552 * doc/extend.texi: Document warn_if_not_aligned attribute.
8553 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
8555 2017-08-17 Martin Liska <mliska@suse.cz>
8558 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
8559 (get_data_dependence): Use it as pointer type.
8560 (distribute_loop): Likewise.
8562 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8564 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
8565 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
8566 (p8_vmrgow_<mode>_direct): New define_insn.
8567 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
8568 handle endianness for vmrgew and vmrgow permute patterns.
8570 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
8572 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
8573 * config/rs6000/rs6000-cpus.def: Remove comment.
8574 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
8575 (POWERPC_MASKS): Likewise.
8576 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
8577 use of TARGET_VSX_TIMODE.
8578 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
8579 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
8580 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
8581 (rs6000_option_override_internal): Remove dead code.
8582 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
8583 (rs6000_legitimize_reload_address): Likewise.
8584 (rs6000_legitimate_address_p): Likewise.
8585 (rs6000_opt_masks): Delete "vsx-timode".
8586 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
8587 from function comment.
8588 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
8589 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
8590 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
8592 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
8593 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
8594 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
8597 (define_peephole2): Likewise.
8599 2017-08-17 Martin Sebor <msebor@redhat.com>
8602 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
8603 past the end of an array.
8604 (test_pp_format): Add test cases.
8606 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
8608 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
8609 missing ECF_NOTHROW flags.
8611 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
8614 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
8615 operands residing in integer registers.
8616 (*vsx_le_perm_load_<mode>): Likewise.
8617 (*vsx_le_perm_store_<mode>): Likewise.
8618 (define_peephole2): Add peepholes to optimize the above.
8620 2017-08-17 Marek Polacek <polacek@redhat.com>
8623 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
8624 to mimic what shorten_compare did. Change the return type to bool.
8625 (fold_cond_expr_with_comparison): Update call to
8626 operand_equal_for_comparison_p.
8627 (fold_ternary_loc): Likewise.
8629 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
8631 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
8633 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
8635 2017-08-17 Richard Biener <rguenther@suse.de>
8637 * tree-ssa-structalias.c (solve_graph): When propagating
8638 to successors update the graphs succ edges and avoid duplicate work.
8640 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
8643 * config/i386/i386.c (ix86_option_override_internal): Save target
8644 specific options after ix86_stack_protector_guard_reg was changed.
8646 2017-08-17 Richard Biener <rguenther@suse.de>
8648 PR tree-optimization/81827
8649 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
8650 (new_var_info): Initialize it conservatively.
8651 (get_call_vi): Mark register vars.
8652 (new_scalar_tmp_constraint_exp): Likewise.
8653 (handle_rhs_call): Likewise.
8654 (handle_const_call): Likewise.
8655 (create_function_info_for): Likewise.
8656 (solve_constraints): Sort varinfos to separate register from
8657 non-register vars to pack points-to solution bitmaps during
8660 2017-08-17 Marek Polacek <polacek@redhat.com>
8662 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
8664 2017-08-17 Richard Biener <rguenther@suse.de>
8666 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
8667 to true when overflow is undefined and we saturated the result.
8669 2017-08-17 Alan Modra <amodra@gmail.com>
8672 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
8673 Don't use store multiple if only one reg needs saving.
8674 (interesting_frame_related_regno): New function.
8675 (rs6000_frame_related): Don't emit frame info for regs that
8677 (rs6000_emit_epilogue): Likewise.
8679 2017-08-16 Nathan Sidwell <nathan@acm.org>
8681 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
8682 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
8683 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
8684 * tree.c (free_lang_data_in_type): Use else-if chain. Always
8685 clear TYPE_LANG_1. Remove obsolete member-function stripping.
8686 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
8687 (verify_type): Adjust for TYPE_BINFO move.
8688 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
8689 process TYPE_BINFO directly.
8690 (hash_tree): Likewise.
8691 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
8693 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
8696 2017-08-16 David Malcolm <dmalcolm@redhat.com>
8698 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
8700 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
8703 * config/i386/i386.md (*anddi_1_btr): Change predicates of
8704 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
8705 Add ix86_binary_operator_ok to insn constraint.
8706 (*iordi_1_bts): Ditto.
8707 (*xordi_1_btc): Ditto.
8708 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
8709 Update corresponding peephole2 pattern.
8713 2017-08-16 Bin Cheng <bin.cheng@arm.com>
8715 PR tree-optimization/81832
8716 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
8717 copy loop header which has IFN_LOOP_DIST_ALIAS call.
8719 2017-08-16 Marek Polacek <polacek@redhat.com>
8722 * fold-const.c (fold_indirect_ref_1): Restore original behavior
8723 regarding size_zero_node.
8725 2017-08-16 Martin Liska <mliska@suse.cz>
8728 * config.gcc: Respect previously set extra_objs in case
8731 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
8733 PR tree-optimization/81835
8734 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
8735 the phi SSA_NAME. Check that the condition in a COND_EXPR does
8736 not depend on the phi.
8738 2017-08-16 Alan Modra <amodra@gmail.com>
8740 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
8743 2017-08-16 Alan Modra <amodra@gmail.com>
8745 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
8746 (save_reg_p): ..into this. Update all callers.
8747 (first_reg_to_save): Simplify.
8749 2017-08-16 Alan Modra <amodra@gmail.com>
8751 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
8754 2017-08-15 Joseph Myers <joseph@codesourcery.com>
8758 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
8759 constant count if that count is less than 32.
8761 2017-08-15 Nathan Sidwell <nathan@acm.org>
8763 * gcc.c (execute): Emit friendlier message if inferior is killed
8764 by an external cause.
8766 2017-08-15 Richard Biener <rguenther@suse.de>
8768 PR tree-optimization/81790
8769 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
8770 CONSTRUCTORs from simplifying and VN.
8772 2017-08-14 Martin Sebor <msebor@redhat.com>
8774 * builtin-attrs.def: Add comments.
8776 2017-08-14 Martin Sebor <msebor@redhat.com>
8779 * doc/extend.texi (attribute nonstring): Document new attribute.
8781 2017-08-14 Martin Sebor <msebor@redhat.com>
8784 * tree-diagnostic.c (default_tree_printer): Handle %G.
8785 * gimple-pretty-print.h (percent_G_format): Declare new function.
8786 * gimple-pretty-print.c (percent_G_format): Define.
8787 * tree-pretty-print.c (percent_K_format): Add argument.
8789 2017-08-14 Martin Sebor <msebor@redhat.com>
8791 PR translation/79998
8792 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
8793 Remove a stray space.
8795 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
8798 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
8799 (*iordi_1_bts): Ditto.
8800 (*xordi_1_btc): Ditto.
8802 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8805 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
8806 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8808 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
8809 quoted strings, and make more translator-friendly.
8810 (darwin_rs6000_override_options): Likewise.
8811 (rs6000_option_override_internal): Likewise.
8812 (rs6000_return_in_memory): Fix overlong line.
8813 (init_cmulative_args): Use quoted strings, and make more
8814 translator-friendly.
8815 (rs6000_pass_by_reference): Fix overlong line.
8816 (def_builtin): Use quoted strings.
8817 (altivec_expand_predicate_builtin): Use quoted strings, and make
8818 more translator-friendly.
8819 (htm_expand_builtin): Use quoted strings.
8820 (cpu_expand_builtin): Use quoted strings, and make more
8821 translator-friendly.
8822 (altivec_expand_builtin): Likewise.
8823 (paired_expand_predicate_builtin): Likewise.
8824 (rs6000_invalid_builtin): Likewise.
8825 (builtin_function_type): Use quoted strings.
8826 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
8827 more translator-friendly.
8828 (rs6000_trampoline_init): Likewise.
8829 (rs6000_handle_altivec_attribute): Likewise.
8830 (rs6000_inner_target_options): Use quoted strings.
8831 (rs6000_disable_incompatible_switches): Likewise.
8832 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
8833 strings, and make more translator-friendly.
8834 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
8836 2017-08-14 Bin Cheng <bin.cheng@arm.com>
8838 PR tree-optimization/81799
8839 * tree-loop-distribution.c (version_loop_by_alias_check): Force
8840 cond_expr to simple gimple operand.
8842 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
8845 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
8847 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
8851 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
8852 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
8853 TARGET_GASISR_PROLOGUES.
8854 * config/avr/avr.c (avr_option_override): Same.
8855 (avr_pass_pre_proep::execute): Same.
8857 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
8860 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
8861 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
8863 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
8865 * config/i386/i386.md (*load_tp_<mode>): Redefine as
8866 define_insn_and_split. Split to a memory load from 0 in
8867 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
8868 using PTR mode iterator.
8869 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
8870 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
8871 (*add_tp_<mode>): Redefine as define_insn_and_split.
8872 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
8873 address space. Merge with *add_tp_x32 using PTR mode iterator.
8874 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
8875 Split to an add with a memory load from 0 in
8876 DEFAULT_TLS_SEG_REG address space.
8878 2017-08-12 Andrew Pinski <apinski@cavium.com>
8880 * config/aarch64/aarch64-option-extensions.def (rdma):
8881 Fix feature string to what Linux prints out in /proc/cpuinfo.
8883 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
8886 * dwarf2out.c (modified_type_die): For C typedef types that have
8887 an ultimate origin, process the ultimate origin instead of the
8889 (gen_typedef_die): Assert that input DECLs have no ultimate
8891 (gen_type_die_with_usage): For typedef variants that have an
8892 ultimate origin, just call gen_decl_die on the original DECL.
8893 (process_scope_var): Avoid creating DIEs for local typedefs and
8894 concrete static variables.
8896 2017-08-12 Alan Modra <amodra@gmail.com>
8900 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
8901 match gnu-user.h startfile.
8902 (ENDFILE_LINUX_SPEC): Similarly.
8904 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
8907 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
8909 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
8911 2017-08-11 Tamar Christina <tamar.christina@arm.com>
8912 * config/aarch64/aarch64.md (mov<mode>): Change.
8913 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
8914 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
8915 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
8917 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
8919 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
8920 for storage order barriers.
8922 2017-08-11 Martin Liska <mliska@suse.cz>
8925 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
8926 variables of void type.
8928 2017-08-11 Martin Liska <mliska@suse.cz>
8930 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
8931 TARGET_SUPPORTS_ALIASES.
8932 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
8933 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
8934 (optimize_weakref): Likewise.
8935 * symtab.c (symtab_node::noninterposable_alias): Likewise.
8936 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
8937 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
8939 2017-08-11 Martin Liska <mliska@suse.cz>
8942 * config/i386/i386.c (make_resolver_func): Do complete
8943 refactoring of the function.
8945 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
8948 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
8949 * config/i386/i386.c (ix86_stack_protect_guard): Use
8950 ix86_stack_protect_guard_symbol_str to generate varible declaration.
8951 * doc/invoke.texi (x86 Options): Document
8952 -mstack-protector-guard-symbol= option.
8954 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
8956 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
8957 * config/i386/i386.c (ix86_split_stack_guard): New function.
8958 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
8959 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
8960 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
8961 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
8962 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
8963 (split_stack_space_check): Call ix86_split_stack_guard.
8965 2017-08-10 Martin Sebor <msebor@redhat.com>
8967 * print-tree.c (print_node): Print location using the established
8969 Replace spaces with colons.
8970 (debug_raw, debug): Ditto.
8972 2017-08-10 Martin Sebor <msebor@redhat.com>
8975 * pretty-print.c (pp_format): Correct the handling of %s precision.
8977 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
8980 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
8982 (ix86_finalize_stack_frame_flags): This. Also clear
8983 frame_pointer_needed if -fno-omit-frame-pointer is used without
8985 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
8986 with ix86_finalize_stack_frame_flags.
8987 (ix86_expand_epilogue): Likewise.
8988 (ix86_expand_split_stack_prologue): Likewise.
8989 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
8991 2017-08-10 Martin Liska <mliska@suse.cz>
8994 * c-attribs.c (handle_target_attribute):
8995 Report warning for an empty string argument of target attribute.
8997 2017-08-09 Jakub Jelinek <jakub@redhat.com>
9000 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
9002 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
9003 or DECL_NONLOCAL labels.
9004 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
9005 or DECL_NONLOCAL labels here.
9007 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
9009 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
9010 to indicate when early gimple folding has been disabled.
9011 (rs6000_gimple_fold_builtin): Add debug content.
9012 (rs6000_invalid_builtin): Fix whitespace.
9013 (rs6000_expand_builtin): Fix whitespace.
9014 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
9016 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
9019 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
9020 SAVE_MULTIPLE if not all the registers that saves, should be saved.
9022 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
9024 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
9025 (qdf24xx): Likewise.
9026 * config/aarch64/aarch64.md: Include falkor.md.
9027 * config/aarch64/falkor.md: New.
9029 2017-08-09 Marek Polacek <polacek@redhat.com>
9032 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
9033 * diagnostic.c (emit_diagnostic): Add a comment.
9034 (emit_diagnostic_valist): New function.
9036 2017-08-09 Marek Polacek <polacek@redhat.com>
9039 * input.c (make_location): New overload.
9040 * input.h (make_location): Declare.
9042 2017-08-08 Alan Modra <amodra@gmail.com>
9043 H.J. Lu <hongjiu.lu@intel.com>
9046 * gcc.c (NO_PIE_SPEC): Delete.
9047 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
9049 (LINK_PIE_SPEC): ..to here.
9050 (LINK_COMMAND_SPEC): Support -no-pie.
9051 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
9052 chain of crtbegin*.o selection, update for PIE_SPEC changes and
9054 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
9055 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
9056 (ENDFILE_CRTEND_SPEC): Similarly.
9058 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
9061 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
9062 (mstack-protector-guard-offset=): Ditto.
9063 * config/i386/i386.c (ix86_option_override): Handle
9064 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
9066 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
9067 ix86_stack_protect_guard_offset variables.
9068 (TARGET_STACK_PROTECT_GUARD): Always define.
9069 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
9070 and -mstack-protector-guard-offset= options.
9072 2017-08-08 Bin Cheng <bin.cheng@arm.com>
9074 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
9075 boundary case for the last candidate.
9077 2017-08-08 Bin Cheng <bin.cheng@arm.com>
9079 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
9080 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
9082 2017-08-08 Tamar Christina <tamar.christina@arm.com>
9085 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
9086 * config/aarch64/aarch64-builtins.c
9087 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
9088 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
9089 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
9091 2017-08-08 Tamar Christina <tamar.christina@arm.com>
9092 Andrew Pinski <pinskia@gmail.com>
9095 * internal-fn.def (XORSIGN): New.
9096 * optabs.def (xorsign_optab): New.
9097 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
9098 (convert_expand_mult_copysign): New.
9099 (pass_optimize_widening_mul::execute): Call
9100 convert_expand_mult_copysign.
9102 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9104 PR tree-optimization/81354
9105 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
9106 Insert on edges rather than explicitly creating landing pads.
9107 (analyze_candidates_and_replace): Commit edge inserts.
9109 2017-08-08 Richard Biener <rguenther@suse.de>
9112 * tree-ssa-loop-niter.c: Include tree-dfa.h.
9113 (expand_simple_operations): Also look through ADDR_EXPRs with
9114 MEM_REF bases treating them as POINTER_PLUS_EXPR.
9116 2017-08-08 Richard Biener <rguenther@suse.de>
9118 PR tree-optimization/81723
9119 * tree-vect-slp.c (struct bst_traits): New hash traits.
9120 (bst_fail): New global.
9121 (vect_build_slp_tree_2): New worker, split out from ...
9122 (vect_build_slp_tree): ... this now wrapping it with using
9123 bst_fail set to cache SLP tree build fails. Properly handle
9125 (vect_analyze_slp_instance): Allocate and free bst_fail.
9127 2017-08-08 Martin Liska <mliska@suse.cz>
9130 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
9131 LABEL_DECLs that can be from a different function.
9133 2017-08-08 Bin Cheng <bin.cheng@arm.com>
9135 PR tree-optimization/81744
9136 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
9137 loop's number of iterations.
9139 2017-08-08 Martin Liska <mliska@suse.cz>
9141 * asan.c: Include header files.
9142 * attribs.c (build_decl_attribute_variant): New function moved
9144 (build_type_attribute_qual_variant): Likewise.
9145 (cmp_attrib_identifiers): Likewise.
9146 (simple_cst_list_equal): Likewise.
9147 (omp_declare_simd_clauses_equal): Likewise.
9148 (attribute_value_equal): Likewise.
9149 (comp_type_attributes): Likewise.
9150 (build_type_attribute_variant): Likewise.
9151 (lookup_ident_attribute): Likewise.
9152 (remove_attribute): Likewise.
9153 (merge_attributes): Likewise.
9154 (merge_type_attributes): Likewise.
9155 (merge_decl_attributes): Likewise.
9156 (merge_dllimport_decl_attributes): Likewise.
9157 (handle_dll_attribute): Likewise.
9158 (attribute_list_equal): Likewise.
9159 (attribute_list_contained): Likewise.
9160 * attribs.h (lookup_attribute): New function moved from tree.[ch].
9161 (lookup_attribute_by_prefix): Likewise.
9162 * bb-reorder.c: Include header files.
9163 * builtins.c: Likewise.
9164 * calls.c: Likewise.
9165 * cfgexpand.c: Likewise.
9166 * cgraph.c: Likewise.
9167 * cgraphunit.c: Likewise.
9168 * convert.c: Likewise.
9169 * dwarf2out.c: Likewise.
9170 * final.c: Likewise.
9171 * fold-const.c: Likewise.
9172 * function.c: Likewise.
9173 * gimple-expr.c: Likewise.
9174 * gimple-fold.c: Likewise.
9175 * gimple-pretty-print.c: Likewise.
9176 * gimple.c: Likewise.
9177 * gimplify.c: Likewise.
9178 * hsa-common.c: Likewise.
9179 * hsa-gen.c: Likewise.
9180 * internal-fn.c: Likewise.
9181 * ipa-chkp.c: Likewise.
9182 * ipa-cp.c: Likewise.
9183 * ipa-devirt.c: Likewise.
9184 * ipa-fnsummary.c: Likewise.
9185 * ipa-inline.c: Likewise.
9186 * ipa-visibility.c: Likewise.
9188 * lto-cgraph.c: Likewise.
9189 * omp-expand.c: Likewise.
9190 * omp-general.c: Likewise.
9191 * omp-low.c: Likewise.
9192 * omp-offload.c: Likewise.
9193 * omp-simd-clone.c: Likewise.
9194 * opts-global.c: Likewise.
9195 * passes.c: Likewise.
9196 * predict.c: Likewise.
9197 * sancov.c: Likewise.
9198 * sanopt.c: Likewise.
9199 * symtab.c: Likewise.
9200 * toplev.c: Likewise.
9201 * trans-mem.c: Likewise.
9202 * tree-chkp.c: Likewise.
9203 * tree-eh.c: Likewise.
9204 * tree-into-ssa.c: Likewise.
9205 * tree-object-size.c: Likewise.
9206 * tree-parloops.c: Likewise.
9207 * tree-profile.c: Likewise.
9208 * tree-ssa-ccp.c: Likewise.
9209 * tree-ssa-live.c: Likewise.
9210 * tree-ssa-loop.c: Likewise.
9211 * tree-ssa-sccvn.c: Likewise.
9212 * tree-ssa-structalias.c: Likewise.
9213 * tree-ssa.c: Likewise.
9214 * tree-streamer-in.c: Likewise.
9215 * tree-vectorizer.c: Likewise.
9216 * tree-vrp.c: Likewise.
9218 * ubsan.c: Likewise.
9219 * varasm.c: Likewise.
9220 * varpool.c: Likewise.
9221 * tree.c: Remove functions moved to attribs.[ch].
9223 * config/aarch64/aarch64.c: Add attrs.h header file.
9224 * config/alpha/alpha.c: Likewise.
9225 * config/arc/arc.c: Likewise.
9226 * config/arm/arm.c: Likewise.
9227 * config/avr/avr.c: Likewise.
9228 * config/bfin/bfin.c: Likewise.
9229 * config/c6x/c6x.c: Likewise.
9230 * config/cr16/cr16.c: Likewise.
9231 * config/cris/cris.c: Likewise.
9232 * config/darwin.c: Likewise.
9233 * config/epiphany/epiphany.c: Likewise.
9234 * config/fr30/fr30.c: Likewise.
9235 * config/frv/frv.c: Likewise.
9236 * config/ft32/ft32.c: Likewise.
9237 * config/h8300/h8300.c: Likewise.
9238 * config/i386/winnt.c: Likewise.
9239 * config/ia64/ia64.c: Likewise.
9240 * config/iq2000/iq2000.c: Likewise.
9241 * config/lm32/lm32.c: Likewise.
9242 * config/m32c/m32c.c: Likewise.
9243 * config/m32r/m32r.c: Likewise.
9244 * config/m68k/m68k.c: Likewise.
9245 * config/mcore/mcore.c: Likewise.
9246 * config/microblaze/microblaze.c: Likewise.
9247 * config/mips/mips.c: Likewise.
9248 * config/mmix/mmix.c: Likewise.
9249 * config/mn10300/mn10300.c: Likewise.
9250 * config/moxie/moxie.c: Likewise.
9251 * config/msp430/msp430.c: Likewise.
9252 * config/nds32/nds32-isr.c: Likewise.
9253 * config/nds32/nds32.c: Likewise.
9254 * config/nios2/nios2.c: Likewise.
9255 * config/nvptx/nvptx.c: Likewise.
9256 * config/pa/pa.c: Likewise.
9257 * config/pdp11/pdp11.c: Likewise.
9258 * config/powerpcspe/powerpcspe.c: Likewise.
9259 * config/riscv/riscv.c: Likewise.
9260 * config/rl78/rl78.c: Likewise.
9261 * config/rx/rx.c: Likewise.
9262 * config/s390/s390.c: Likewise.
9263 * config/sh/sh.c: Likewise.
9264 * config/sol2.c: Likewise.
9265 * config/sparc/sparc.c: Likewise.
9266 * config/spu/spu.c: Likewise.
9267 * config/stormy16/stormy16.c: Likewise.
9268 * config/tilegx/tilegx.c: Likewise.
9269 * config/tilepro/tilepro.c: Likewise.
9270 * config/v850/v850.c: Likewise.
9271 * config/vax/vax.c: Likewise.
9272 * config/visium/visium.c: Likewise.
9273 * config/xtensa/xtensa.c: Likewise.
9275 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9278 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
9279 constraints since the -mupper-regs-* switches have been
9281 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
9282 into a vector from a double word element that was extracted from
9283 another vector, and eliminate extra XXPERMDI instructions.
9284 (vsx_concat_<mode>_2): Likewise.
9285 (vsx_concat_<mode>_3): Likewise.
9286 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
9287 concat to allow optimizing inserts from previous extracts.
9289 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
9291 * config/i386/i386.c (ix86_stack_protect_guard): Generate
9292 memory reference to a SSP offset in TLS address space.
9293 (ix86_print_operand) <case '@'>: Remove.
9294 (ix86_print_operand_punct_valid_p): Remove '@' code.
9295 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
9297 (stack_tls_protect_set_<mode>): Remove.
9298 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
9299 (stack_tls_protect_test_<mode>): Remove.
9300 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
9302 2017-08-07 Olivier Hainque <hainque@adacore.com>
9305 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
9307 2017-08-07 Douglas Rupp <rupp@adacore.com>
9309 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
9310 variable was referenced as multidir in command.
9312 2017-08-07 Jakub Jelinek <jakub@redhat.com>
9315 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
9318 2017-08-07 Martin Liska <mliska@suse.cz>
9320 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
9321 * config/rl78/rl78.c: Add include of attribs.h.
9322 * config/sh/sh.c: Likewise.
9323 * config/v850/v850.c: Likewise.
9325 2017-08-07 Tom de Vries <tom@codesourcery.com>
9328 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
9330 2017-08-07 Martin Liska <mliska@suse.cz>
9332 * config/mips/mips.c: Include attribs.h.
9334 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
9337 * doc/invoke.texi: Document change in behvaior for -Ofast for
9340 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
9342 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
9344 (aarch64_pop_regs): Likewise.
9345 (aarch64_gen_load_pair): Likewise.
9346 (aarch64_save_callee_saves): Likewise.
9347 (aarch64_restore_callee_saves): Likewise.
9349 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
9351 * config/i386/i386.c: Revert the last change.
9353 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
9356 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
9358 (ix86_finalize_stack_frame_flags): This. Also clear
9359 frame_pointer_needed if -fno-omit-frame-pointer is used without
9361 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
9362 with ix86_finalize_stack_frame_flags.
9363 (ix86_expand_epilogue): Likewise.
9364 (ix86_expand_split_stack_prologue): Likewise.
9366 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
9369 * config/i386/i386.c (get_builtin_code_for_version): Set priority
9370 to P_AES for Westmere.
9372 2017-08-07 Jonathan Yong <10walls@gmail.com>
9374 * config/i386/mingw.opt (fset-stack-executable): Removed.
9375 * config/i386/cygming.opt (fset-stack-executable): Moved
9377 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
9379 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
9381 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
9383 2017-08-07 Marek Polacek <polacek@redhat.com>
9386 * fold-const.c (fold_indirect_ref_1): Check type_domain.
9388 2017-08-07 Martin Liska <mliska@suse.cz>
9390 * attribs.h (canonicalize_attr_name): New function.
9391 (cmp_attribs): Move from c-format.c and adjusted.
9392 (is_attribute_p): Moved from tree.h.
9393 * tree-inline.c: Add new includes.
9394 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
9395 (private_is_attribute_p): Remove.
9396 (private_lookup_attribute): Likewise.
9397 (private_lookup_attribute_by_prefix): Simplify.
9398 (remove_attribute): Use is_attribute_p.
9399 * tree.h: Remove removed declarations.
9401 2017-08-07 Jakub Jelinek <jakub@redhat.com>
9404 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
9405 instead of computing it in the function. Formatting fix.
9406 (expand_case): Don't rely on default_edge being the first edge,
9407 clear it if removing it, pass default_edge to
9408 emit_case_dispatch_table.
9409 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
9412 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
9414 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
9415 insn in the function, emit NOP after the insn.
9417 2017-08-06 Tom de Vries <tom@codesourcery.com>
9419 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
9422 2017-08-06 Tom de Vries <tom@codesourcery.com>
9424 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
9427 2017-08-04 Yury Gribov <tetra2005@gmail.com>
9429 PR tree-optimization/57371
9430 * match.pd: New pattern.
9432 2017-08-04 Marek Polacek <polacek@redhat.com>
9435 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
9436 perform the computation in offset_int.
9438 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
9440 PR tree-optimization/81136
9441 * tree-vectorizer.h: Include tree-hash-traits.h.
9442 (vec_base_alignments): New typedef.
9443 (vec_info): Add a base_alignments field.
9444 (vect_record_base_alignments): Declare.
9445 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
9447 (DR_IS_CONDITIONAL_IN_STMT): New macro.
9448 (create_data_ref): Add an is_conditional_in_stmt argument.
9449 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
9450 the is_conditional_in_stmt field.
9451 (data_ref_loc): Add an is_conditional_in_stmt field.
9452 (get_references_in_stmt): Set the is_conditional_in_stmt field.
9453 (find_data_references_in_stmt): Update call to create_data_ref.
9454 (graphite_find_data_references_in_stmt): Likewise.
9455 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
9456 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
9457 (vect_record_base_alignment): New function.
9458 (vect_record_base_alignments): Likewise.
9459 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
9460 for nested statements even if we fail to compute a misalignment.
9461 Use pooled base alignments for unconditional references.
9462 (vect_find_same_alignment_drs): Compare base addresses instead
9464 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
9465 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
9467 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
9469 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
9470 Add an explicit name for the enum. Use auto_vec for slp_instances
9472 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
9474 (_bb_vec_info): Add a constructor and destructor.
9475 (vinfo_for_stmt): Return NULL for uids of -1 as well.
9476 (destroy_loop_vec_info): Delete.
9477 (vect_destroy_datarefs): Likewise.
9478 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
9479 (vec_info::vec_info): New function.
9480 (vec_info::~vec_info): Likewise.
9481 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
9482 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
9483 destroy_loop_vec_info.
9484 * tree-vect-loop.c (new_loop_vec_info): Replace with...
9485 (_loop_vec_info::_loop_vec_info): ...this.
9486 (destroy_loop_vec_info): Replace with...
9487 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
9488 the stmt_vec_infos. Leave handling of vec_info information to its
9489 destructor. Remove explicit vector releases.
9490 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
9491 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
9492 * tree-vect-slp.c (new_bb_vec_info): Replace with...
9493 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
9494 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
9495 (destroy_bb_vec_info): Replace with...
9496 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
9497 information to its destructor.
9498 (vect_slp_analyze_bb_1): Use new and delete instead of
9499 new_bb_vec_info and destroy_bb_vec_info.
9500 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
9503 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
9505 * tree-data-ref.h (subscript): Add access_fn field.
9506 (data_dependence_relation): Add could_be_independent_p.
9507 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
9508 (same_access_functions): Move to tree-data-ref.c.
9509 * tree-data-ref.c (ref_contains_union_access_p): New function.
9510 (access_fn_component_p): Likewise.
9511 (access_fn_components_comparable_p): Likewise.
9512 (dr_analyze_indices): Add a reference to access_fn_component_p.
9513 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
9515 (constant_access_functions): Likewise.
9516 (add_other_self_distances): Likewise.
9517 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
9518 (initialize_data_dependence_relation): Use XCNEW and remove
9519 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
9520 of access functions that have the same type. Allow the
9521 subsequence to end with different bases in some circumstances.
9522 Record the chosen access functions in SUB_ACCESS_FN.
9523 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
9524 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
9525 (subscript_dependence_tester_1): Likewise dra and drb.
9526 (build_classic_dist_vector): Update calls accordingly.
9527 (subscript_dependence_tester): Likewise.
9528 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
9529 DDR_COULD_BE_INDEPENDENT_P.
9530 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
9531 comp_alias_ddrs instead of may_alias_ddrs.
9532 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
9534 (vect_analyze_data_ref_dependence): Use it if
9535 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
9536 distance vectors if that fails.
9537 (dependence_distance_ge_vf): New function.
9538 (vect_prune_runtime_alias_test_list): Use it. Don't clear
9539 LOOP_VINFO_MAY_ALIAS_DDRS.
9541 2017-08-04 Richard Biener <rguenther@suse.de>
9544 * fold-const.c (fold_binary_loc): Properly restrict
9545 minus_var0 && minus_var1 case when associating undefined overflow
9548 2017-08-04 Tom de Vries <tom@codesourcery.com>
9550 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
9552 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9554 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9555 Don't start diagnostic messages with a capital letter.
9556 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9558 (rs6000_invalid_builtin): Likewise.
9559 (rs6000_trampoline_init): Likewise.
9561 2017-08-03 Jakub Jelinek <jakub@redhat.com>
9564 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
9565 after setting changeable df flags.
9567 2017-08-03 Richard Biener <rguenther@suse.de>
9569 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
9570 up if the use is in USE - X.
9572 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
9574 * toplev.c (dumpfile.h): New include.
9575 (internal_error_reentered): New static function. Use it...
9576 (internal_error_function): ...here to handle reentered internal_error.
9578 2017-08-03 Richard Biener <rguenther@suse.de>
9581 * fold-const.c (split_tree): Add minus_var and minus_con
9582 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
9583 here but always use minus_*.
9584 (associate_trees): Assert we never associate with MINUS_EXPR
9585 and NULL first operand. Do not recurse for PLUS_EXPR operands
9586 when associating as MINUS_EXPR either.
9587 (fold_binary_loc): Track minus_var and minus_con.
9589 2017-08-03 Tom de Vries <tom@codesourcery.com>
9592 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
9593 ACCEL_COMPILER, apply finish_options on
9594 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
9596 2017-08-03 Tom de Vries <tom@codesourcery.com>
9599 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
9600 function_entry_patch_area_size > 0.
9602 2017-08-03 Jakub Jelinek <jakub@redhat.com>
9605 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
9606 instead of 10??LU, perform unit multiplication in wide_int,
9607 don't change alloc_object_size_limit if the limit is larger
9610 PR tree-optimization/81655
9611 PR tree-optimization/81588
9612 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
9613 the case when ranges[i].low and high are 1 for unsigned type with
9617 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
9618 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
9620 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9622 * tree-vrp.h: Add include guard.
9624 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
9627 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
9628 (ud2): New insn pattern.
9629 * config/i386/i386.c (ix86_expand_epilogue):
9630 For naked functions, generate ud2 instead of trap insn.
9632 2017-08-02 Marek Polacek <polacek@redhat.com>
9635 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
9637 2017-08-02 Tom de Vries <tom@codesourcery.com>
9638 Cesar Philippidis <cesar@codesourcery.com>
9640 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
9641 Add missing edge probabilities.
9643 2017-08-02 Tamar Christina <tamar.christina@arm.com>
9645 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
9648 2017-08-02 Jakub Jelinek <jakub@redhat.com>
9651 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
9652 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
9653 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
9654 prologue_seq sequences - if any.
9656 2017-08-02 Richard Biener <rguenther@suse.de>
9658 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
9659 via vectors if supported, integer extracts via punning if supported
9660 or otherwise vector extracts.
9662 2017-08-02 Richard Biener <rguenther@suse.de>
9664 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
9666 (bitmap_insert_into_set): ... this.
9668 2017-08-02 Richard Biener <rguenther@suse.de>
9670 PR tree-optimization/81633
9672 2015-08-17 Alan Hayward <alan.hayward@arm.com>
9674 PR tree-optimization/71752
9675 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
9677 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
9679 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
9680 (machine_function::call_ms2sysv_pad_out): Remove field.
9681 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
9682 (ix86_compute_frame_layout): Likewise.
9684 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
9687 * config/i386/i386.c (ix86_set_func_type): Disallow naked
9688 attribute with interrupt attribute.
9690 2017-08-01 Andrew Pinski <apinski@cavium.com>
9692 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
9693 BIT_INSERT_EXPR's operand 1
9694 to see if the types precision matches.
9696 2017-08-01 Martin Liska <mliska@suse.cz>
9699 * builtins.c (expand_builtin_memcpy_args): Remove.
9700 (expand_builtin_memcpy): Call newly added function
9701 expand_builtin_memory_copy_args.
9702 (expand_builtin_memcpy_with_bounds): Likewise.
9703 (expand_builtin_mempcpy): Remove last argument.
9704 (expand_builtin_mempcpy_with_bounds): Likewise.
9705 (expand_builtin_memory_copy_args): New function created from
9706 expand_builtin_mempcpy_args with small modifications.
9707 (expand_builtin_mempcpy_args): Remove.
9708 (expand_builtin_stpcpy): Remove unused argument.
9709 (expand_builtin): Likewise.
9710 (expand_builtin_with_bounds): Likewise.
9712 2017-08-01 Martin Liska <mliska@suse.cz>
9715 Make mempcpy more optimal (PR middle-end/70140).
9717 2017-08-01 Jakub Jelinek <jakub@redhat.com>
9720 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9721 __builtin_vec_cmpne verify both arguments are compatible vectors
9722 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
9723 verify arg1_type is a pointer or array type. For __builtin_vec_st,
9724 move computation of aligned to after checking the argument types.
9728 * config/rs6000/vsx.md (vextract_fp_from_shorth,
9729 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
9732 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
9733 Doug Rupp <rupp@adacore.com>
9734 Olivier Hainque <hainque@adacore.com>
9736 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
9737 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
9739 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
9740 for TARGET_OS_CPP_BUILTIN.
9741 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
9742 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
9744 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
9745 passing required abi options to the assembler for EABI configurations.
9746 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
9747 of .text.hot and .text.unlikely sections for kernel modules when
9748 using ARM style exceptions.
9749 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
9750 options. Add EXTRA_CC1_SPEC.
9751 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
9753 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
9755 (ARM_TARGET2_DWARF_FORMAT): Define.
9756 * config/arm/t-vxworks: Adjust multilib control to removal of the
9757 Diab command line options.
9759 2017-08-01 Martin Liska <mliska@suse.cz>
9761 PR gcov-profile/81561
9762 * gcov.c (unblock): Make unblocking safe as we need to preserve
9763 index correspondence of blocks and block_lists.
9765 2017-08-01 Richard Biener <rguenther@suse.de>
9767 PR tree-optimization/81181
9768 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9769 (compute_antic): ... end of iteration here.
9771 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
9773 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
9774 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
9775 (ftree-slp-vectorize): Likewise.
9776 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
9777 can no longer be set independent of flag_tree_loop_vectorize.
9778 * omp-general.c (emp_max_vf): Likewise.
9779 * opts.c (enable_fdo_optimizations): Remove references to
9780 flag_tree_vectorize, these are now implicit.
9781 (common_handle_option): Remove handling for OPT_ftree_vectorize,
9782 and leave it for the options machinery.
9784 2017-08-01 Martin Liska <mliska@suse.cz>
9787 * builtins.c (expand_builtin_memcpy_args): Remove.
9788 (expand_builtin_memcpy): Call newly added function
9789 expand_builtin_memory_copy_args.
9790 (expand_builtin_memcpy_with_bounds): Likewise.
9791 (expand_builtin_mempcpy): Remove last argument.
9792 (expand_builtin_mempcpy_with_bounds): Likewise.
9793 (expand_builtin_memory_copy_args): New function created from
9794 expand_builtin_mempcpy_args with small modifications.
9795 (expand_builtin_mempcpy_args): Remove.
9796 (expand_builtin_stpcpy): Remove unused argument.
9797 (expand_builtin): Likewise.
9798 (expand_builtin_with_bounds): Likewise.
9800 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
9803 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
9804 print "ds:" only for immediates in generic address space.
9806 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
9809 * config/i386/i386.c (ix86_funciton_naked): New prototype.
9810 (ix86_function_ok_for_sibcall): Return false for naked functions.
9812 2017-08-01 Richard Biener <rguenther@suse.de>
9814 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
9815 (compute_antic): Seed worklist with exit block predecessors.
9816 * cfganal.c (dfs_find_deadend): For a cycle return the source
9817 of the edge closing it.
9819 2017-08-01 Tamar Christina <tamar.christina@arm.com>
9821 * config/aarch64/aarch64.c
9822 (aarch64_can_const_movi_rtx_p): Move 0 check.
9824 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9826 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
9827 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
9829 * match.pd: Ditto in address comparison pattern.
9831 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9833 PR tree-optimization/81627
9834 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
9835 closed ssa form for store-store chain.
9837 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9839 PR tree-optimization/81620
9840 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
9841 for store-store chain.
9843 2017-08-01 Jakub Jelinek <jakub@redhat.com>
9845 PR tree-optimization/81588
9846 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
9847 ranges[i].in_p, invert comparison code ccode. For >/>=,
9848 swap rhs1 and rhs2 and comparison code unconditionally,
9849 for </<= don't do that. Don't swap rhs1/rhs2 again if
9850 ranges[i].in_p, instead invert comparison code ccode if
9851 opcode or oe->rank is BIT_IOR_EXPR.
9854 * optabs.def (vec_extract_optab, vec_init_optab): Change from
9855 a direct optab to conversion optab.
9856 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
9857 with GET_MODE_INNER as last argument instead of optab_handler.
9858 * expmed.c (extract_bit_field_1): Likewise. Use vector from
9859 vector extraction if possible and optab is available.
9860 * expr.c (store_constructor): Use convert_optab_handler instead
9861 of optab_handler. Use vector initialization from smaller
9862 vectors if possible and optab is available.
9863 * tree-vect-stmts.c (vectorizable_load): Likewise.
9864 * doc/md.texi (vec_extract, vec_init): Document that the optabs
9866 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
9867 of vec_init from half-sized vectors with the same element mode.
9868 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
9869 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
9870 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
9871 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
9872 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
9873 after mode in gen_vec_extract* calls.
9874 (vec_extract<mode>): Renamed to ...
9875 (vec_extract<mode><ssescalarmodelower>): ... this.
9876 (vec_extract<mode><ssehalfvecmodelower>): New expander.
9877 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
9878 element mode after mode in gen_vec_init* calls.
9879 (VEC_INIT_HALF_MODE): New mode iterator.
9880 (vec_init<mode>): Renamed to ...
9881 (vec_init<mode><ssescalarmodelower>): ... this.
9882 (vec_init<mode><ssehalfvecmodelower>): New expander.
9883 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
9884 (vec_extractv2sfsf): ... this.
9885 (vec_initv2sf): Renamed to ...
9886 (vec_initv2sfsf): ... this.
9887 (vec_extractv2si): Renamed to ...
9888 (vec_extractv2sisi): ... this.
9889 (vec_initv2si): Renamed to ...
9890 (vec_initv2sisi): ... this.
9891 (vec_extractv4hi): Renamed to ...
9892 (vec_extractv4hihi): ... this.
9893 (vec_initv4hi): Renamed to ...
9894 (vec_initv4hihi): ... this.
9895 (vec_extractv8qi): Renamed to ...
9896 (vec_extractv8qiqi): ... this.
9897 (vec_initv8qi): Renamed to ...
9898 (vec_initv8qiqi): ... this.
9899 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
9900 (vec_init<mode>): Renamed to ...
9901 (vec_init<mode><VEC_base_l>): ... this.
9902 (vec_extract<mode>): Renamed to ...
9903 (vec_extract<mode><VEC_base_l>): ... this.
9904 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
9905 (vec_initv2sfsf): ... this.
9906 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
9907 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
9908 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
9909 element mode after mode in gen_vec_init* calls.
9910 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
9911 (vec_init<mode><Vel>): ... this.
9912 (vec_extract<mode>): Renamed to ...
9913 (vec_extract<mode><Vel>): ... this.
9914 * config/aarch64/iterators.md (Vel): New mode attribute.
9915 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
9916 Add element mode after mode in gen_vec_extract* calls.
9917 * config/s390/vector.md (non_vec_l): New mode attribute.
9918 (vec_extract<mode>): Renamed to ...
9919 (vec_extract<mode><non_vec_l>): ... this.
9920 (vec_init<mode>): Renamed to ...
9921 (vec_init<mode><non_vec_l>): ... this.
9922 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
9923 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
9925 * config/arm/iterators.md (V_elem_l): New mode attribute.
9926 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
9927 (vec_extract<mode><V_elem_l>): ... this.
9928 (vec_extractv2di): Renamed to ...
9929 (vec_extractv2didi): ... this.
9930 (vec_init<mode>): Renamed to ...
9931 (vec_init<mode><V_elem_l>): ... this.
9932 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
9933 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
9934 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
9935 Add element mode after gen_vec_extract* calls.
9936 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
9937 (vec_init<mode><unitmode>): ... this.
9938 (vec_extract<mode>): Renamed to ...
9939 (vec_extract<mode><unitmode>): ... this.
9940 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
9941 (vec_init<mode><unitmode>): ... this.
9942 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
9943 (vec_initv2sfsf): ... this.
9944 (vec_extractv2sf): Renamed to ...
9945 (vec_extractv2sfsf): ... this.
9946 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
9947 Add element mode after gen_vec_extract* calls.
9948 * config/mips/mips.md (unitmode): New mode iterator.
9949 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
9950 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
9951 * config/spu/spu.md (inner_l): New mode attribute.
9952 (vec_init<mode>): Renamed to ...
9953 (vec_init<mode><inner_l>): ... this.
9954 (vec_extract<mode>): Renamed to ...
9955 (vec_extract<mode><inner_l>): ... this.
9956 * config/sparc/sparc.md (veltmode): New mode iterator.
9957 (vec_init<VMALL:mode>): Renamed to ...
9958 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
9959 * config/ia64/vect.md (vec_initv2si): Renamed to ...
9960 (vec_initv2sisi): ... this.
9961 (vec_initv2sf): Renamed to ...
9962 (vec_initv2sfsf): ... this.
9963 (vec_extractv2sf): Renamed to ...
9964 (vec_extractv2sfsf): ... this.
9965 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
9966 (vec_init<mode>): Renamed to ...
9967 (vec_init<mode><VEC_base_l>): ... this.
9968 (vec_extract<mode>): Renamed to ...
9969 (vec_extract<mode><VEC_base_l>): ... this.
9970 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
9971 (vec_initv2sfsf): ... this.
9972 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
9973 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
9974 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
9975 gen_vec_init* calls.
9977 2017-08-01 Richard Biener <rguenther@suse.de>
9979 PR tree-optimization/81297
9980 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
9981 TREE_OVERFLOW from INTEGER_CSTs.
9983 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
9985 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
9987 2017-07-31 Carl Love <cel@us.ibm.com>
9989 * config/rs6000/rs6000-c: Add support for built-in functions
9990 vector signed char vec_xl_be (signed long long, signed char *);
9991 vector unsigned char vec_xl_be (signed long long, unsigned char *);
9992 vector signed int vec_xl_be (signed long long, signed int *);
9993 vector unsigned int vec_xl_be (signed long long, unsigned int *);
9994 vector signed long long vec_xl_be (signed long long, signed long long *);
9995 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
9996 vector signed short vec_xl_be (signed long long, signed short *);
9997 vector unsigned short vec_xl_be (signed long long, unsigned short *);
9998 vector double vec_xl_be (signed long long, double *);
9999 vector float vec_xl_be (signed long long, float *);
10000 * config/rs6000/altivec.h (vec_xl_be): Add #define.
10001 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
10002 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
10004 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
10005 (altivec_expand_builtin): Add switch statement to call
10006 altivec_expand_xl_be for each builtin.
10007 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
10008 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
10009 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
10010 __builtin_vsx_le_be_v16qi.
10011 * doc/extend.texi: Update the built-in documentation file for the
10012 new built-in functions.
10014 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
10017 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
10019 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
10021 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10023 * config.gcc: Add z14.
10024 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
10025 CPU model numbers for z13s and z14.
10026 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
10028 * config/s390/s390-opts.h (enum processor_type): Rename
10029 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
10030 * config/s390/s390.c (processor_table): Add field for CPU name to
10031 be passed to Binutils.
10032 (s390_asm_output_machine_for_arch): Use the new field in
10033 processor_table for Binutils.
10034 (s390_expand_builtin): Replace arch12 with z14.
10035 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
10036 (s390_get_sched_attrmask): Likewise.
10037 (s390_get_unit_mask): Likewise.
10038 * config/s390/s390.opt: Add z14 to processor_type enum.
10040 2017-07-31 Martin Jambor <mjambor@suse.cz>
10043 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
10044 regardless of optimization level.
10046 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
10047 Martin Liska <mliska@suse.cz>
10049 * predict.def: Remove old comment and adjust probability.
10050 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
10051 PREDICT statements.
10053 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
10056 * config/i386/i386.c (ix86_function_naked): New function.
10057 (ix86_can_use_return_insn_p): Return false for naked functions.
10058 (ix86_expand_prologue): Skip prologue for naked functions.
10059 (ix86_expand_epilogue): Skip epilogue for naked functions
10060 and emit trap instruction.
10061 (ix86_warn_func_return): New function.
10062 (ix86_attribute_table): Add "naked" attribute specification.
10063 (TARGET_WARN_FUNC_RETURN): Define.
10064 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
10066 2017-07-31 Martin Liska <mliska@suse.cz>
10068 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
10069 (dump_gimple_bb_header): Always dump BB info.
10070 (pp_cfg_jump): Do not append info about BB when dumping a jump.
10072 2017-07-31 Martin Liska <mliska@suse.cz>
10075 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
10076 also with current_function_decl non-null equality.
10078 2017-07-31 Jakub Jelinek <jakub@redhat.com>
10081 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
10082 change type to the element type, instead add eltype variable and
10083 use it where we are interested in the element type.
10085 PR tree-optimization/81603
10086 * ipa-polymorphic-call.c
10087 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
10088 offset arithmetic in offset_int, bail out if the resulting bit offset
10089 doesn't fit into shwi.
10091 2017-07-31 Martin Liska <mliska@suse.cz>
10093 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
10094 (gimplify_save_expr): Fix comment.
10096 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
10099 * config/i386/i386.c (ix86_function_arg): Update arguments for
10101 (ix86_compute_frame_layout): Set the initial stack offset to
10102 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
10103 INCOMING_FRAME_SP_OFFSET.
10104 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
10105 stack before exception handler returns.
10106 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
10107 the 'ERROR_CODE' for exception handler.
10109 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
10111 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
10112 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
10113 (ASM_OUTPUT_REG_POP): Ditto.
10114 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
10115 instead of asm_fprintf to output pure string.
10117 2017-07-29 Jakub Jelinek <jakub@redhat.com>
10119 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
10120 to imported_module_or_decl hook.
10121 (debug_nothing_tree_tree_tree_bool): Remove.
10122 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
10123 * debug.c (do_nothing_debug_hooks): Use
10124 debug_nothing_tree_tree_tree_bool_bool instead of
10125 debug_nothing_tree_tree_tree_bool.
10126 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
10127 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
10128 * sdbout.c (sdb_debug_hooks): Likewise.
10129 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
10130 (gen_namespace_die): Add DW_AT_export_symbols attribute if
10132 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
10133 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
10134 attribute, don't add anything.
10136 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10138 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
10139 (fold_build2_stat_loc): Likewise.
10140 (fold_build3_stat_loc): Likewise.
10141 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
10142 (fold_build1_loc): Remove macro.
10143 (fold_build2_loc): Likewise.
10144 (fold_build3_loc): Likewise.
10146 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10148 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
10149 (gimple_build_debug_bind_source_stat): Likewise.
10150 * gimple.h (gimple_build_debug_bind): Remove macro.
10151 (gimple_build_debug_bind_source): Likewise.
10153 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10155 * bitmap.c (bitmap_alloc): Adjust.
10156 (bitmap_gc_alloc): Likewise.
10157 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
10159 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10161 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
10162 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
10163 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
10164 (bitmap_gc_alloc_stat): Likewise.
10165 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
10167 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10169 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
10170 * rtl.h (shallow_copy_rtx): Remove macro.
10172 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10174 * emit-rtl.c (gen_raw_REG): Adjust.
10175 * gengenrtl.c (gendef): Likewise.
10176 * rtl.c (rtx_alloc_stat): Remove _stat from name.
10177 * rtl.h (rtx_alloc): Remove macro.
10179 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10181 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
10182 (build_tree_list_stat): Likewise.
10183 * tree.h (build_tree_list): Remove macro.
10184 (build_tree_list_vec): Likewise.
10186 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10188 * tree.c (make_vector_stat): Remove _stat from name.
10189 (build_vector_stat): Likewise.
10190 * tree.h (make_vector_stat): Remove macro.
10191 (build_vector_stat): Likewise.
10193 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10195 * tree.h (build_var_debug_value): Remove prototype.
10197 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10199 * tree.c (tree_cons_stat): Remove _stat from name.
10200 * tree.h (tree_cons): Remove macro.
10202 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10204 * tree.c (build_vl_exp_stat): Remove _stat from name.
10205 * tree.h (build_vl_exp): Remove macro.
10207 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10209 * tree.c (build_decl_stat): Remove _stat from name.
10210 * tree.h (build_decl): Remove macro.
10212 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10214 * gimple.c (gimple_build_with_ops_stat): Adjust.
10215 (gimple_alloc_stat): Remove _stat from name.
10216 * gimple.h (gimple_alloc): Remove macro.
10218 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10220 * tree.c (make_tree_vec_stat): Remove _stat from name.
10221 (grow_tree_vec_stat): Likewise.
10222 * tree.h (make_tree_vec_stat): Adjust prototype.
10223 (grow_tree_vec_stat): Likewise.
10224 (make_tree_vec): Remove macro.
10225 (grow_tree_vec): Likewise.
10227 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10229 * fold-const.c (fold_build1_stat_loc): Adjust.
10230 (fold_build2_stat_loc): Likewise.
10231 (fold_build3_stat_loc): Likewise.
10232 * tree.c (build0_stat): Remove _stat from name.
10233 (build1_stat): Likewise.
10234 (build2_stat): Likewise.
10235 (build3_stat): Likewise.
10236 (build4_stat): Likewise.
10237 (build5_stat): Likewise.
10238 * tree.h (build1_loc): Remove macro, and rename _stat function
10240 (build2_loc): Likewise.
10241 (build3_loc): Likewise.
10242 (build4_loc): Likewise.
10243 (build5_loc): Likewise.
10245 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10247 * tree.c (make_int_cst_stat): Remove _stat from name.
10248 * tree.h (make_int_cst_stat): Adjust prototype.
10249 (make_int_cst): Remove macro.
10251 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10253 * tree.c (make_tre_binfo_stat): Remove _stat from name.
10254 * tree.h (make_tree_binfo_stat): Adjust prototype.
10255 (make_tree_binfo): Remove.
10257 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10259 * tree.c (copy_node_stat): Rename to copy_node.
10260 (build_distinct_type_copy): Adjust.
10261 * tree.h (copy_node_stat): Adjust prototype.
10262 (copy_node): Remove macro.
10264 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10266 * tree.c (make_node_stat): rename to make_node.
10267 (build_tree_list_stat): Adjust.
10268 (build0_stat): Likewise.
10269 (build2_stat): Likewise.
10270 (build3_stat): Likewise.
10271 (build4_stat): Likewise.
10272 (build5_stat): Likewise.
10273 (build_decl_stat): Likewise.
10274 * tree.h (make_node_stat): Adjust prototype.
10275 (make_node): remove macro.
10277 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
10279 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
10280 (PPC_FEATURE2_SCV): Likewise.
10281 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
10283 2017-07-28 Tamar Christina <tamar.christina@arm.com>
10285 * config/aarch64/aarch64.c
10286 (aarch64_internal_mov_immediate): Add new special pattern.
10287 * config/aarch64/aarch64.md (*movdi_aarch64):
10288 Add reg/32bit const mov case.
10290 2017-07-28 Tamar Christina <tamar.christina@arm.com>
10291 Richard Sandiford <richard.sandiford@linaro.org>
10293 * config/aarch64/aarch64.md (mov<mode>): Generalize.
10294 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
10295 Add integer and movi cases.
10296 (movi-split-hf-df-sf split, fp16): New.
10297 (enabled): Added TARGET_FP_F16INST.
10298 * config/aarch64/iterators.md (GPF_HF): New.
10299 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
10301 2017-07-28 Tamar Christina <tamar.christina@arm.com>
10303 * config/aarch64/aarch64.c
10304 (aarch64_simd_container_mode): Add prototype.
10305 (aarch64_expand_mov_immediate): Add HI support.
10306 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
10307 (aarch64_can_const_movi_rtx_p): New.
10308 (aarch64_preferred_reload_class):
10309 Remove restrictions of using FP registers for certain SIMD operations.
10310 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
10311 (aarch64_valid_floating_const): Add integer move validation.
10312 (aarch64_simd_imm_scalar_p): Remove.
10313 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
10314 (aarch64_legitimate_constant_p): Expand list of supported cases.
10315 * config/aarch64/aarch64-protos.h
10316 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
10317 (aarch64_reinterpret_float_as_int): New.
10318 (aarch64_simd_imm_scalar_p): Remove.
10319 * config/aarch64/constraints.md (Uvi): New.
10320 (Dd): Split into Ds and new Dd.
10321 * config/aarch64/aarch64.md (*movsi_aarch64):
10323 (*movdi_aarch64): Add SIMD mov case.
10325 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10327 * tree-predcom.c: (struct chain): Handle store-store chain in which
10328 stores for elimination only store loop invariant values.
10329 (execute_pred_commoning_chain): Ditto.
10330 (prepare_initializers_chain_store_elim): Ditto.
10331 (prepare_finalizers): Ditto.
10332 (is_inv_store_elimination_chain): New function.
10333 (initialize_root_vars_store_elim_1): New function.
10335 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10337 * tree-predcom.c: Revise general description of the pass.
10338 (enum chain_type): New enum type for store elimination.
10339 (struct chain): New field supporting store elimination.
10340 (struct component): Ditto.
10341 (dump_chain): Dump store-stores chain.
10342 (release_chain): Release resources.
10343 (split_data_refs_to_components): Compute and create component
10344 contains only stores for elimination.
10345 (get_chain_last_ref_at): New function.
10346 (make_invariant_chain): Initialization.
10347 (make_rooted_chain): Specify chain type in parameter and record it.
10348 (add_looparound_copies): Skip for store-stores chain.
10349 (determine_roots_comp): Compute type of chain and pass it to
10351 (initialize_root_vars_store_elim_2): New function.
10352 (finalize_eliminated_stores): New function.
10353 (remove_stmt): Handle store for elimination.
10354 (execute_pred_commoning_chain): Execute predictive commoning on
10355 store-store chains.
10356 (determine_unroll_factor): Skip unroll for store-stores chain.
10357 (prepare_initializers_chain_store_elim): New function.
10358 (prepare_initializers_chain): Hanlde store-store chain.
10359 (prepare_finalizers_chain, prepare_finalizers): New function.
10360 (tree_predictive_commoning_loop): Return integer value indicating
10361 if loop is unrolled or lcssa form is corrupted.
10362 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
10364 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10366 * tree-predcom.c (initialize_root): Delete.
10367 (execute_pred_commoning_chain): Initialize root vars and replace
10368 reference of non-combined chain directly, rather than call above
10371 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10373 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
10374 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
10376 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10378 * tree-predcom.c (struct chain): New field init_seq.
10379 (release_chain): Release init_seq.
10380 (prepare_initializers_chain): Record intialization stmts in above
10382 (insert_init_seqs): New function.
10383 (tree_predictive_commoning_loop): Call insert_init_seqs.
10385 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10387 * tree-predcom.c (determine_roots_comp): Skip trivial components.
10389 2017-07-28 Richard Biener <rguenther@suse.de>
10391 * match.pd: Remove superfluous :c.
10392 * genmatch.c (simplify::id): Add member.
10393 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
10395 (current_id): New global.
10396 (dt_node::parent): Move from ...
10397 (dt_operand::parent): ... here. Add for_id member.
10398 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
10399 (decision_tree::find_node): Relax order requirement when
10400 merging DT_TRUE nodes to ones inbetween the current simplify
10401 and the one we try to merge with. Add diagnostic whenever
10402 we need to enforce pattern order by not merging.
10403 (decision_tree::insert): Set current_id.
10404 (decision_tree::print_node): Dump parent node and for_id.
10405 (parser::last_id): Add member.
10406 (parser::push_simplify): Assign unique id.
10407 (parser::parser): Initialize last_id.
10409 2017-07-28 Martin Liska <mliska@suse.cz>
10412 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
10413 gimple_build_debug_bind.
10415 2017-07-28 Richard Biener <rguenther@suse.de>
10417 PR tree-optimization/81502
10418 * match.pd: Add pattern combining BIT_INSERT_EXPR with
10420 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
10422 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
10423 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
10424 for BIT_FIELD_REF args.
10425 * fold-const.c (make_bit_field_ref): Likewise.
10426 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
10428 2017-07-28 Jakub Jelinek <jakub@redhat.com>
10431 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
10432 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
10433 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
10434 Or it into SANITIZER_UNDEFINED.
10435 * ubsan.c: Include gimple-fold.h and varasm.h.
10436 (ubsan_expand_ptr_ifn): New function.
10437 (instrument_pointer_overflow): New function.
10438 (maybe_instrument_pointer_overflow): New function.
10439 (instrument_object_size): Formatting fix.
10440 (pass_ubsan::execute): Call instrument_pointer_overflow
10441 and maybe_instrument_pointer_overflow.
10442 * internal-fn.c (expand_UBSAN_PTR): New function.
10443 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
10444 * sanitizer.def (__ubsan_handle_pointer_overflow,
10445 __ubsan_handle_pointer_overflow_abort): New builtins.
10446 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
10447 * internal-fn.def (UBSAN_PTR): New internal function.
10448 * opts.c (sanitizer_opts): Add pointer-overflow.
10449 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
10450 * fold-const.c (build_range_check): Compute pointer range check in
10451 integral type if pointer arithmetics would be needed. Formatting
10454 2017-07-28 Martin Liska <mliska@suse.cz>
10457 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
10458 parameters that are of a variable-length.
10460 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
10462 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
10464 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
10465 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
10466 (CRT_CALL_STATIC_FUNCTION): Likewise.
10467 (ASM_DEFAULT_SPEC): New define.
10468 (ASM_SPEC32): Likewise.
10469 (ASM_SPEC64): Likewise.
10470 (ASM_SPEC_COMMON): Likewise.
10471 (ASM_SPEC): Likewise.
10472 (INVALID_64BIT): Likewise.
10473 (LINK_OS_DEFAULT_SPEC): Likewise.
10474 (LINK_OS_SPEC32): Likewise.
10475 (LINK_OS_SPEC64): Likewise.
10476 (POWERPC_LINUX): Likewise.
10477 (PTRDIFF_TYPE): Likewise.
10478 (RESTORE_FP_PREFIX): Likewise.
10479 (RESTORE_FP_SUFFIX): Likewise.
10480 (SAVE_FP_PREFIX): Likewise.
10481 (SAVE_FP_SUFFIX): Likewise.
10482 (SIZE_TYPE): Likewise.
10483 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
10484 (TARGET_64BIT): Likewise.
10485 (TARGET_64BIT): Likewise.
10486 (TARGET_AIX): Likewise.
10487 (WCHAR_TYPE_SIZE): Likewise.
10488 (WCHAR_TYPE): Undefine.
10489 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
10490 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
10491 (CPP_OS_RTEMS_SPEC): Delete.
10492 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
10493 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
10495 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
10497 2017-07-28 Jakub Jelinek <jakub@redhat.com>
10499 PR tree-optimization/81578
10500 * tree-parloops.c (build_new_reduction): Bail out if
10501 reduction_code isn't one of the standard OpenMP reductions.
10502 Move the details printing after that decision.
10504 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
10506 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
10507 related to reload_in_progress.
10508 (splat_input_operand): Likewise.
10509 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
10511 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
10513 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
10514 (TARGET_INSTANTIATE_DECLS): Likewise.
10515 (legitimate_indexed_address_p): Delete reload_in_progress code.
10516 (rs6000_debug_legitimate_address_p): Likewise.
10517 (rs6000_eliminate_indexed_memrefs): Likewise.
10518 (rs6000_emit_le_vsx_store): Likewise.
10519 (rs6000_emit_move_si_sf_subreg): Likewise.
10520 (rs6000_emit_move): Likewise.
10521 (register_to_reg_type): Likewise.
10522 (rs6000_pre_atomic_barrier): Likewise.
10523 (rs6000_machopic_legitimize_pic_address): Likewise.
10524 (rs6000_allocate_stack_temp): Likewise.
10525 (rs6000_address_for_fpconvert): Likewise.
10526 (rs6000_address_for_altivec): Likewise.
10527 (rs6000_secondary_memory_needed_rtx): Delete function.
10528 (rs6000_check_sdmode): Likewise.
10529 (rs6000_alloc_sdmode_stack_slot): Likewise.
10530 (rs6000_instantiate_decls): Likewise.
10531 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
10532 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
10533 Delete reload_in_progress.
10534 (*vec_reload_and_plus_<mptrsize>): Likewise.
10535 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
10536 (vsx_div_v2di): Likewise.
10537 (vsx_udiv_v2di): Likewise.
10539 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
10541 * config/rs6000/rs6000.opt (mlra): Replace with stub.
10542 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
10543 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
10544 (rs6000_debug_reg_global): Delete print of LRA status.
10545 (rs6000_option_override_internal): Delete dead LRA related code.
10546 (rs6000_lra_p): Delete function.
10547 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
10549 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
10551 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
10552 * config/riscv/rtems.h: New file.
10554 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10555 Sudakshina Das <sudi.das@arm.com>
10557 * config/aarch64/aarch64.md
10558 (define_split for and<mode>3nr_compare): Move
10559 non aarch64_logical_operand to a register.
10560 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
10561 register immediate operand to a register.
10562 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
10564 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
10566 PR middle-end/81564
10567 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
10569 2017-07-27 Richard Biener <rguenther@suse.de>
10571 PR tree-optimization/81573
10572 PR tree-optimization/81494
10573 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
10574 multi defuse cycle case.
10576 2017-07-27 Richard Biener <rguenther@suse.de>
10578 PR tree-optimization/81571
10579 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
10582 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
10584 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
10585 earlier and only if MASK_FPU is set. Adjust formatting.
10587 2017-07-27 Martin Liska <mliska@suse.cz>
10589 * opt-functions.awk: Add validation of value of Init.
10590 * optc-gen.awk: Pass new argument.
10592 2017-07-27 Martin Liska <mliska@suse.cz>
10594 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
10595 Fix wrong condition.
10597 2017-07-27 Martin Liska <mliska@suse.cz>
10599 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
10600 BBs and edges seen by autoFDO.
10602 2017-07-27 Richard Biener <rguenther@suse.de>
10604 PR tree-optimization/81502
10605 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
10606 with incompatible but same sized type.
10607 (execute_update_addresses_taken): Likewise.
10609 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
10611 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
10612 flag_tree_loop_vectorize rather than flag_tree_vectorize.
10614 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10617 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
10618 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
10619 Change s_operand to memory_operand.
10621 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
10623 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
10624 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
10625 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
10626 Emit instructions rather than returning an expression. Handle TFmode
10627 and KFmode by casting to TImode.
10628 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
10629 (rs6000_emit_le_vsx_store): Likewise.
10630 * config/rs6000/vsx.md (VSX_TI): New iterator.
10631 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
10632 (*vsx_le_undo_permute_<mode>): Likewise.
10633 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
10634 emit the split sequence.
10635 (*vsx_le_perm_store_<mode>): Likewise.
10637 2017-07-27 Jakub Jelinek <jakub@redhat.com>
10639 PR tree-optimization/81555
10640 PR tree-optimization/81556
10641 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
10642 if true, force CHANGED for the recursive invocation.
10643 (reassociate_bb): Remember original length of ops array, pass
10644 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
10646 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
10647 attributes for noipa attribute. For naked attribute use
10648 lookup_attribute first before lookup_attribute_spec.
10649 * final.c (rest_of_handle_final): Disable IPA RA for functions with
10651 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
10652 for functions with noipa attribute.
10653 (cgraph_externally_visible_p): Return true for functions with noipa
10655 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
10656 for functions with noipa attribute.
10657 * doc/extend.texi: Document noipa function attribute.
10658 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
10659 also for functions with noipa attribute.
10660 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
10662 2017-07-26 Andrew Pinski <apinski@cavium.com>
10664 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
10665 vec_unalign_load_cost and vec_unalign_store_cost.
10667 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10669 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
10670 -mvsx-small-integer option.
10671 (ISA_3_0_MASKS_IEEE): Likewise.
10672 (OTHER_VSX_VECTOR_MASKS): Likewise.
10673 (POWERPC_MASKS): Likewise.
10674 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
10675 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
10676 code, only testing for DImode being allowed in non-VSX floating
10678 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
10679 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
10681 (rs6000_option_override_internal): Delete -mvsx-small-integer.
10682 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
10683 TARGET_P8_VECTOR test.
10684 (rs6000_secondary_reload_simple_move): Likewise.
10685 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
10686 since TARGET_P9_VECTOR was already tested.
10687 (rs6000_opt_masks): Remove -mvsx-small-integer.
10688 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
10689 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
10691 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
10692 test for TARGET_VEXTRACTUB was used, and that uses
10694 (p9 extract splitter): Likewise.
10695 (vsx_extract_<mode>_di_p9): Likewise.
10696 (vsx_extract_<mode>_store_p9): Likewise.
10697 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
10698 for TARGET_P9_VECTOR was used. Delete code that is now dead with
10699 the elimination of TARGET_VSX_SMALL_INTEGER.
10700 (vsx_extract_<mode>_p8): Likewise.
10701 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
10702 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
10703 (vsx_set_<mode>_p9): Likewise.
10704 (vsx_set_v4sf_p9): Likewise.
10705 (vsx_set_v4sf_p9_zero): Likewise.
10706 (vsx_insert_extract_v4sf_p9): Likewise.
10707 (vsx_insert_extract_v4sf_p9_2): Likewise.
10708 * config/rs6000/rs6000.md (sign extend splitter): Change
10709 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
10710 (floatsi<mode>2_lfiwax_mem): Likewise.
10711 (floatunssi<mode>2_lfiwzx_mem): Likewise.
10712 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
10713 since a test for TARGET_P9_VECTOR was used.
10714 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10715 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
10716 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10717 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
10718 TARGET_P8_VECTOR test.
10719 (fix_trunc<mode>si2_stfiwx): Likewise.
10720 (fix_trunc<mode>si2_internal): Likewise.
10721 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
10722 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
10724 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
10725 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
10726 TARGET_P8_VECTOR test.
10727 (fixuns_trunc<mode>si2_stfiwx): Likewise.
10728 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
10729 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
10731 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
10732 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
10733 since a test for TARGET_P9_VECTOR was used.
10734 (splitter for loading small constants): Likewise.
10736 2017-07-26 Andrew Pinski <apinski@cavium.com>
10738 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
10741 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
10744 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
10745 (fp_valid_at): Likewise.
10747 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
10749 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
10750 (qdf24xx_addrcost_table): Likewise.
10751 (cortexa57_tunings): Update to use generic_branch_cost.
10752 (cortexa72_tunings): Likewise.
10753 (cortexa73_tunings): Likewise.
10754 (qdf24xx_tunings): Likewise.
10756 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
10758 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
10759 (thunderx2t99_branch_cost): Likewise.
10760 (cortexa35_tunings): Update to use generic_branch_cost.
10761 (cortexa53_tunings): Likewise.
10762 (cortexa57_tunings): Likewise.
10763 (cortexa72_tunings): Likewise.
10764 (cortexa73_tunings): Likewise.
10765 (thunderx2t99_tunings): Likewise.
10767 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
10769 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
10770 (sparc_option_override): Honour MASK_FSMULD.
10771 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
10772 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
10773 * config/sparc/sparc.opt (mfsmuld): New option.
10774 * doc/invoke.texi (mfsmuld): Document option.
10776 2017-07-26 Marek Polacek <polacek@redhat.com>
10778 PR middle-end/70992
10779 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
10781 2017-07-26 Richard Biener <rguenther@suse.de>
10783 * gimple-match-head.c (do_valueize): Return OP if valueize
10785 (get_def): New helper to get at the def stmt of a SSA name
10786 if valueize allows.
10787 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
10788 do_valueize to get at the def stmt.
10789 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
10791 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
10793 PR middle-end/46932
10794 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
10796 2017-07-26 Martin Liska <mliska@suse.cz>
10799 * function.c (expand_function_start): Make expansion of
10800 nonlocal_goto_save_area after parm_birth_insn.
10802 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
10804 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
10805 from all CPU target flags enable members.
10807 2017-07-26 Richard Biener <rguenther@suse.de>
10809 * genmatch.c (dt_simplify::gen): Make iterator vars const.
10810 (decision_tree::gen): Make 'type' const.
10811 (write_predicate): Likewise.
10813 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
10815 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
10816 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
10817 (rs6000_option_override_internal): Likewise.
10818 (rs6000_expand_vector_set): Likewise.
10819 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
10820 (TARGET_UPPER_REGS_SF): Likewise.
10821 (TARGET_UPPER_REGS_DI): Likewise.
10822 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
10823 (TARGET_DIRECT_MOVE_64BIT): Likewise.
10824 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
10825 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10826 (Splitters for DI constants in Altivec registers): Likewise.
10827 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
10828 (vsx_set_v4sf_p9): Likewise.
10829 (vsx_set_v4sf_p9_zero): Likewise.
10830 (vsx_insert_extract_v4sf_p9): Likewise.
10831 (vsx_insert_extract_v4sf_p9_2): Likewise.
10833 2017-07-25 Carl Love <cel@us.ibm.com>
10835 * doc/extend.texi: Update the built-in documentation file for the
10836 existing built-in functions
10837 vector signed char vec_cnttz (vector signed char);
10838 vector unsigned char vec_cnttz (vector unsigned char);
10839 vector signed short vec_cnttz (vector signed short);
10840 vector unsigned short vec_cnttz (vector unsigned short);
10841 vector signed int vec_cnttz (vector signed int);
10842 vector unsigned int vec_cnttz (vector unsigned int);
10843 vector signed long long vec_cnttz (vector signed long long);
10844 vector unsigned long long vec_cnttz (vector unsigned long long);
10846 2017-07-25 Andrew Pinski <apinski@cavium.com>
10848 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
10849 accesses where the use is for the first operand of a BIT_INSERT.
10851 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
10854 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
10855 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
10857 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
10859 * config/i386/gstabs.h: Delete.
10860 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
10862 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
10864 * config/i386/i386.c (ix86_decompose_address): Do not check for
10865 register RTX when looking at index_reg or base_reg.
10866 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
10868 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
10870 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
10871 to update EH info here.
10873 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10875 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
10877 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10879 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
10881 2017-07-25 Torsten Duwe <duwe@suse.de>
10883 * common.opt: Introduce -fpatchable-function-entry
10884 command line option, and its variables function_entry_patch_area_size
10885 and function_entry_patch_area_start.
10886 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
10887 including a two-value parser.
10888 * target.def (print_patchable_function_entry): New target hook.
10889 * targhooks.h (default_print_patchable_function_entry): New function.
10890 * targhooks.c (default_print_patchable_function_entry): Likewise.
10891 * toplev.c (process_options): Switch off IPA-RA if
10892 patchable function entries are being generated.
10893 * varasm.c (assemble_start_function): Look at the
10894 patchable-function-entry command line switch and current
10895 function attributes and maybe generate NOP instructions by
10896 calling the print_patchable_function_entry hook.
10897 * doc/extend.texi: Document patchable_function_entry attribute.
10898 * doc/invoke.texi: Document -fpatchable_function_entry
10899 command line option.
10900 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
10902 * doc/tm.texi: Re-generate.
10904 2017-07-25 Jakub Jelinek <jakub@redhat.com>
10907 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
10908 TARGET_AVX512DQ rather than TARGET_AVX512BW.
10910 2017-07-25 Tamar Christina <tamar.christina@arm.com>
10912 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
10914 2017-07-25 Richard Biener <rguenther@suse.de>
10916 PR tree-optimization/81455
10917 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
10918 not walk in cycles when looking for guards.
10920 2017-07-25 Richard Biener <rguenther@suse.de>
10922 PR tree-optimization/81529
10923 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
10924 when optimizing backedge uses.
10926 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
10928 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
10930 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
10931 to dl_section_ref. On AIX, append an expression to subtract
10932 the size of the section length to dl_section_ref.
10934 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
10936 * configure.ac: If any of the config.* scripts fail, exit 1.
10937 * configure: Regenerate.
10939 2017-07-25 Richard Biener <rguenther@suse.de>
10941 PR middle-end/81546
10942 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
10943 of immediate uses, be more verbose on errors.
10945 2017-07-25 Richard Biener <rguenther@suse.de>
10947 PR tree-optimization/81510
10948 * tree-vect-loop.c (vect_is_simple_reduction): When the
10949 reduction stmt is not inside the loop bail out.
10951 2017-07-25 Richard Biener <rguenther@suse.de>
10953 PR tree-optimization/81303
10954 * tree-vect-loop-manip.c (vect_loop_versioning): Build
10955 profitability check against LOOP_VINFO_NITERSM1.
10957 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10959 * domwalk.c (cmp_bb_postorder): Simplify.
10960 (sort_bbs_postorder): New function. Use it...
10961 (dom_walker::walk): ...here to optimize common cases.
10963 2017-07-25 Martin Liska <mliska@suse.cz>
10966 * ipa-visibility.c (function_and_variable_visibility): Make the
10967 redirection just on target that supports aliasing.
10968 Fix GNU coding style.
10970 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
10973 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
10975 * config/arm/rtems.h: Likewise.
10976 * config/bfin/rtems.h: Likewise.
10977 * config/i386/rtemself.h: Likewise.
10978 * config/lm32/rtems.h: Likewise.
10979 * config/m32c/rtems.h: Likewise.
10980 * config/m68k/rtemself.h: Likewise.
10981 * config/microblaze/rtems.h: Likewise.
10982 * config/mips/rtems.h: Likewise.
10983 * config/moxie/rtems.h: Likewise.
10984 * config/nios2/rtems.h: Likewise.
10985 * config/powerpcspe/rtems.h: Likewise.
10986 * config/rs6000/rtems.h: Likewise.
10987 * config/rtems.h: Likewise.
10988 * config/sh/rtems.h: Likewise.
10989 * config/sh/rtemself.h: Likewise.
10990 * config/sparc/rtemself.h: Likewise.
10992 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
10995 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
10996 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
10997 * tree-ssa-structalias.c (alias_get_name): Same.
10999 2017-07-25 Bin Cheng <bin.cheng@arm.com>
11002 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
11003 instructions if no du chain is found.
11005 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
11007 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
11009 2017-07-25 Richard Biener <rguenther@suse.de>
11011 PR middle-end/81505
11012 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
11015 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
11017 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
11018 upper-regs options.
11019 (ISA_2_7_MASKS_SERVER): Likewise.
11020 (ISA_3_0_MASKS_IEEE): Likewise.
11021 (OTHER_P8_VECTOR_MASKS): Likewise.
11022 (OTHER_VSX_VECTOR_MASKS): Likewise.
11023 (POWERPC_MASKS): Likewise.
11024 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
11025 duplicate list of options.
11026 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
11027 explicit -mupper-regs options.
11028 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
11029 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
11030 alias for -mupper-regs-df.
11031 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
11032 (rs6000_init_hard_regno_mode_ok): Likewise.
11033 (rs6000_option_override_internal): Likewise.
11034 (rs6000_opt_masks): Likewise.
11035 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
11036 options in terms of whether -mvsx or -mpower8-vector was used.
11037 (TARGET_UPPER_REGS_DI): Likewise.
11038 (TARGET_UPPER_REGS_SF): Likewise.
11039 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
11040 -mupper-regs-* options.
11042 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
11044 * passes.c (emergency_dump_function): Print some empty lines and a
11045 header before the RTL dump.
11047 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
11049 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
11051 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
11054 * config/aarch64/aarch64.c (aarch64_classify_symbol):
11055 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
11057 2017-07-24 Carl Love <cel@us.ibm.com>
11059 * config/rs6000/rs6000-c.c: Add support for built-in functions
11060 vector float vec_extract_fp32_from_shorth (vector unsigned short);
11061 vector float vec_extract_fp32_from_shortl (vector unsigned short);
11062 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
11063 vec_extract_fp_from_shortl): Add defines for the two builtins.
11064 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
11065 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
11067 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
11068 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
11069 * doc/extend.texi: Update the built-in documentation file for the
11070 new built-in function.
11072 2017-07-24 Jakub Jelinek <jakub@redhat.com>
11075 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
11077 * doc/generic.texi: Likewise.
11078 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
11079 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
11081 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
11083 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
11084 (aarch64_mls_elt_merge<mode>): Likewise.
11086 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
11088 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
11089 having __cxa_atexit.
11091 2017-07-23 Michael Collison <michael.collison@arm.com>
11093 * config/arm/arm.c (arm_option_override): Deprecate
11094 use of -mstructure-size-boundary.
11095 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
11096 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
11098 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11101 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
11102 Reduce cost estimate for direct moves.
11104 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
11107 * config/i386/i386.c (ix86_option_override_internal): Disable
11108 BMI, BMI2 and TBM instructions for -m16.
11110 2017-07-21 Carl Love <cel@us.ibm.com>
11112 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11113 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11114 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11115 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11116 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11117 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11118 VMULOSW): New enum "unspec" values.
11119 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11120 altivec_vmulosw): New patterns.
11121 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11122 VMULOSW): Add definitions.
11124 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
11126 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
11127 (qdf24xx): Likewise.
11128 * config/aarch64/aarch64-options-extensions.def (rdma); New.
11129 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
11130 (AARCH64_FL_V8_1): Renumber.
11131 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
11132 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
11133 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
11134 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
11135 rdma to feature modifiers list.
11137 2017-07-21 Yury Gribov <tetra2005@gmail.com>
11139 PR middle-end/56727
11140 * ipa-visibility (function_and_variable_visibility): Convert
11141 recursive PLT call to direct call if appropriate.
11143 2017-07-21 Andrew Pinski <apinski@cavium.com>
11145 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
11146 operand 1 to see if the types precision matches.
11147 * fold-const.c (operand_equal_p): Likewise.
11149 2017-07-21 Richard Biener <rguenther@suse.de>
11151 PR tree-optimization/81303
11152 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
11153 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
11154 (vect_peeling_hash_get_lowest_cost): Adjust.
11155 (vect_enhance_data_refs_alignment): Likewise. Use
11156 vect_get_peeling_costs_all_drs to compute the penalty for no
11157 peeling to match up costs.
11159 2017-07-21 Richard Biener <rguenther@suse.de>
11161 PR tree-optimization/81500
11162 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
11163 we didn't identify a reduction path.
11165 2017-07-21 Tom de Vries <tom@codesourcery.com>
11166 Cesar Philippidis <cesar@codesourcery.com>
11168 PR gcov-profile/81442
11169 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
11172 2017-07-21 Tom de Vries <tom@codesourcery.com>
11175 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
11177 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
11178 nvptx_override_options_after_change.
11180 2017-07-21 Ulrich Drepper <drepper@redhat.com>
11182 * dwarf2out.c (output_file_names): Avoid double testing for
11183 dwarf_version >= 5.
11185 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
11187 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
11189 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
11191 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
11193 (try_crossjump_to_edge): Do not punt on partitioned functions.
11195 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
11197 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
11198 Put all BBs reachable only via paths crossing cold region to cold
11200 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
11202 2016-07-21 Richard Biener <rguenther@suse.de>
11204 PR tree-optimization/81303
11205 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
11206 into account prologue and epilogue iterations when raising
11207 min_profitable_iters to sth at least covering one vector iteration.
11209 2017-07-21 Tamar Christina <tamar.christina@arm.com>
11211 * config/arm/arm.c (arm_test_cpu_arch_dat):
11214 2017-07-20 Nathan Sidwell <nathan@acm.org>
11216 Remove TYPE_METHODS.
11217 * tree.h (TYPE_METHODS): Delete.
11218 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
11219 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
11220 (dbxout_type_methods): Scan TYPE_FIELDS.
11221 (dbxout_type): Don't check TYPE_METHODS here.
11222 * function.c (use_register_for_decl): Always ignore register for
11223 class types when not optimizing.
11224 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
11225 * tree.c (free_lang_data_in_type): Stitch out member functions and
11226 templates from TYPE_FIELDS.
11227 (build_distinct_type_copy, verify_type_variant,
11228 verify_type): Member fns are on TYPE_FIELDS.
11229 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
11230 * tree-pretty-print.c (dump_generic_node): Likewise.
11232 2017-07-20 Jakub Jelinek <jakub@redhat.com>
11235 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
11236 V2TImode and V4TImode.
11237 (ix86_expand_vector_extract): Likewise.
11238 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
11239 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
11240 (ssescalarmode): Handle V4TImode and V2TImode.
11241 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
11242 (*vec_extractv2ti, *vec_extractv4ti): New insns.
11243 (VEXTRACTI128_MODE): New mode iterator.
11244 (splitter for *vec_extractv?ti first element): New.
11245 (VEC_INIT_MODE): New mode iterator.
11246 (vec_init<mode>): Consolidate 3 expanders into one using
11247 VEC_INIT_MODE mode iterator.
11249 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
11251 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
11252 non_spilled_static_chain_regno_p.
11254 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
11256 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
11258 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
11260 * bb-reorder.c (connect_traces): Allow copying of blocks within
11263 2017-07-20 Richard Biener <rguenther@suse.de>
11265 * gimple.h (gimple_phi_result): Add gphi * overload.
11266 (gimple_phi_result_ptr): Likewise.
11267 (gimple_phi_arg): Likewise. Adjust index assert to only
11268 allow actual argument accesses rather than all slots available
11270 (gimple_phi_arg_def): Add gphi * overload.
11271 * tree-phinodes.c (make_phi_node): Initialize only actual
11273 (resize_phi_node): Clear memory not covered by old node,
11274 do not initialize excess argument slots.
11275 (reserve_phi_args_for_new_edge): Initialize new argument slot
11278 2017-07-20 Bin Cheng <bin.cheng@arm.com>
11280 PR tree-optimization/81388
11282 2016-07-21 Bin Cheng <bin.cheng@arm.com>
11284 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
11285 by removing computation of may_be_zero.
11287 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
11288 Tom de Vries <tom@codesourcery.com>
11290 PR middle-end/81030
11291 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
11292 when gimple level profile disagrees with what RTL expander did.
11294 2017-07-20 Richard Biener <rguenther@suse.de>
11296 PR tree-optimization/61171
11297 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
11298 (vect_analyze_stmt): Add slp instance parameter.
11299 (vectorizable_reduction): Likewise.
11300 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
11301 (vect_is_simple_reduction): Deal with chains not detected
11302 as SLP reduction chain, specifically not properly associated
11303 chains containing a mix of plus/minus.
11304 (get_reduction_op): Remove.
11305 (get_initial_defs_for_reduction): Simplify, pass in whether
11306 this is a reduction chain, pass in the SLP node for the PHIs.
11307 (vect_create_epilog_for_reduction): Get the SLP instance as
11309 (vectorizable_reduction): Get the SLP instance as arg.
11310 During analysis remember the SLP node with the PHIs in the
11311 instance. Simplify getting at the vectorized reduction PHIs.
11312 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
11313 through SLP instance.
11314 (vect_slp_analyze_operations): Likewise.
11315 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
11316 (vect_transform_stmt): Likewise.
11318 2017-07-20 Tom de Vries <tom@codesourcery.com>
11320 PR tree-optimization/81489
11321 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
11322 read of phi arg location to before loop that modifies phi.
11324 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11326 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
11329 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
11331 PR middle-end/81331
11332 * except.c (execute): Fix ordering issue.
11334 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
11336 PR rtl-optimization/81423
11337 * combine.c (make_compound_operation_int): Don't try to optimize
11338 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
11340 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
11342 PR rtl-optimization/81423
11343 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
11344 with a constant that is -1 in the truncated to mode.
11346 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
11348 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
11349 (determine_unlikely_bbs): ... here.
11350 * predict.h (propagate_unlikely_bbs_forward): Declare.
11351 * cfgexpand.c (pass_expand::execute): Use it.
11352 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
11354 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
11355 propagate_unlikely_bbs_forward.
11357 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
11359 PR middle-end/81331
11360 * except.c (maybe_add_nop_after_section_switch): New function.
11363 2017-07-19 Tom de Vries <tom@codesourcery.com>
11365 * gimple.h (gimple_phi_set_arg): Make assert more strict.
11367 2017-07-19 Tom de Vries <tom@codesourcery.com>
11369 * gimple.h (gimple_phi_arg): Make assert more strict.
11371 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
11373 * config.gcc (powerpc*-*-*): Add mmintrin.h.
11374 * config/rs6000/mmintrin.h: New file.
11375 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
11377 2017-07-19 Jakub Jelinek <jakub@redhat.com>
11379 PR tree-optimization/81346
11380 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
11382 2017-07-19 Tom de Vries <tom@codesourcery.com>
11384 * config/nvptx/nvptx.md (VECIM): Add V2DI.
11386 2017-07-19 Tom de Vries <tom@codesourcery.com>
11388 * config/nvptx/nvptx-modes.def: Add V2DImode.
11389 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
11390 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
11391 (nvptx_output_mov_insn): Handle lack of mov.b128.
11392 (nvptx_print_operand): Handle 'H' and 'L' codes.
11393 (nvptx_vector_mode_supported): Allow V2DImode.
11394 (nvptx_preferred_simd_mode): New function.
11395 (nvptx_data_alignment): New function.
11396 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
11397 nvptx_preferred_simd_mode.
11398 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
11400 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
11402 2017-07-19 Tom de Vries <tom@codesourcery.com>
11404 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
11405 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
11406 (nvptx_vector_mode_supported): New function. Allow V2SImode.
11407 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
11408 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
11409 (mov<VECIM>_insn): New define_insn.
11410 (define_expand "mov<VECIM>): New define_expand.
11412 2017-07-19 Tom de Vries <tom@codesourcery.com>
11414 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
11416 2017-07-19 Jakub Jelinek <jakub@redhat.com>
11418 PR tree-optimization/81346
11419 * fold-const.h (fold_div_compare, range_check_type): Declare.
11420 * fold-const.c (range_check_type): New function.
11421 (build_range_check): Use range_check_type.
11422 (fold_div_compare): No longer static, rewritten into
11423 a match.pd helper function.
11424 (fold_comparison): Don't call fold_div_compare here.
11425 * match.pd (X / C1 op C2): New optimization using fold_div_compare
11426 as helper function.
11428 2017-07-19 Nathan Sidwell <nathan@acm.org>
11430 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
11431 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
11432 * tree.c (find_decls_types_r, verify_type): Use
11433 TYPE_{MIN,MAX}_VALUE_RAW.
11434 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
11435 (hash_tree): Likewise.
11436 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
11438 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
11441 2017-07-18 Tom de Vries <tom@codesourcery.com>
11443 PR middle-end/81464
11444 * omp-expand.c (expand_omp_for_static_chunk): Handle
11445 equal-argument loop exit phi.
11447 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
11450 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
11451 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
11452 operand 2 predicate.
11453 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
11454 operand 2 predicate.
11455 (ror,rol -> rorx splitters): Use const_int_operand as
11456 operand 2 predicate.
11458 2017-06-18 Richard Biener <rguenther@suse.de>
11460 PR tree-optimization/81410
11461 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
11462 the gap in the ! slp_perm SLP case after each group.
11464 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
11466 PR middle-end/81463
11467 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
11470 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
11472 PR middle-end/81462
11473 * predict.c (set_even_probabilities): Cleanup; do not affect
11474 probabilities that are already known.
11475 (combine_predictions_for_bb): Call even when count is set.
11477 2017-07-18 Nathan Sidwell <nathan@acm.org>
11479 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
11482 2017-07-18 Bin Cheng <bin.cheng@arm.com>
11485 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
11486 optimization for loop niter analysis.
11488 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
11491 * config/avr/avr.c (avr_optimize_casesi): Don't use
11492 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
11494 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11496 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
11497 body_cost_vec from _vect_peel_extended_info.
11498 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
11499 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
11502 2017-07-18 Bin Cheng <bin.cheng@arm.com>
11504 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
11506 2017-07-18 Richard Biener <rguenther@suse.de>
11508 PR tree-optimization/80620
11509 PR tree-optimization/81403
11510 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
11511 info when re-using a VN table entry.
11513 2017-07-18 Richard Biener <rguenther@suse.de>
11515 PR tree-optimization/81418
11516 * tree-vect-loop.c (vectorizable_reduction): Properly compute
11517 vectype_in. Verify that with lane-reducing reduction operations
11518 we have a single def-use cycle.
11520 2017-07-17 Carl Love <cel@us.ibm.com>
11522 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
11524 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11525 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11526 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11527 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11528 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11529 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11530 VMULOSW): New enum "unspec" values.
11531 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
11532 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
11533 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11534 altivec_vmulosw): New patterns.
11535 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11536 VMULOSW): Add definitions.
11537 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
11539 * config/alpha/alpha.c: Include predict.h.
11541 2017-07-17 Yury Gribov <tetra2005@gmail.com>
11543 * tree-vrp.c (compare_assert_loc): Fix comparison function
11544 to return predictable results.
11546 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11548 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
11550 (subdi3): Likewise.
11551 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
11552 * doc/invoke.texi (mexpand-adddi): Update text.
11554 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11556 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
11557 that also clobbers the CC register. The old expand code is moved
11559 (*arc_clzsi2): ... here.
11560 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
11561 the CC register. The old expand code is moved to ...
11562 (arc_ctzsi2): ... here.
11564 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11566 * config/arc/arc.opt (mindexed-loads): Use initial value
11567 TARGET_INDEXED_LOADS_DEFAULT.
11568 (mauto-modify-reg): Use initial value
11569 TARGET_AUTO_MODIFY_REG_DEFAULT.
11570 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
11571 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
11572 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
11573 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
11575 2017-07-17 Martin Liska <mliska@suse.cz>
11578 * opts.c (finish_options): Do not allow -fgnu-tm
11579 w/ -fsanitize={kernel-,}address. Say sorry.
11581 2017-07-17 Bin Cheng <bin.cheng@arm.com>
11584 * tree-loop-distribution.c (classify_partition): Only assert on
11585 numer of iterations.
11586 (merge_dep_scc_partitions): Delete prameter. Update function call.
11587 (distribute_loop): Remove code handling loop with unknown niters.
11588 (pass_loop_distribution::execute): Skip loop with unknown niters.
11590 2017-07-17 Bin Cheng <bin.cheng@arm.com>
11593 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
11594 function sort_partitions_by_post_order.
11596 2017-07-17 Bin Cheng <bin.cheng@arm.com>
11598 PR tree-optimization/81374
11599 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
11600 the max index of basic blocks, rather than number of basic blocks.
11602 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11604 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
11606 (arc_legitimate_pic_operand_p): Likewise.
11607 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
11609 (arc_needs_pcl_p): Likewise.
11610 (arc_legitimate_pc_offset_p): Likewise.
11611 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
11612 function is also used in constrains.md.
11613 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
11614 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
11615 PLUS. Only return true/false in known cases, otherwise assert.
11616 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
11617 is already called in arc_legitimate_constant_p.
11618 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
11620 (LEGITIMATE_PIC_OPERAND_P): Use
11621 arc_raw_symbolic_reference_mentioned_p function.
11622 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
11627 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11628 Andrew Burgess <andrew.burgess@embecosm.com>
11630 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
11631 (arc_return_address_register): New function.
11632 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
11633 (arc_handle_fndecl_attribute): Add naked attribute.
11634 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
11635 (TARGET_WARN_FUNC_RETURN): Likewise.
11636 (arc_allocate_stack_slots_for_args): New function.
11637 (arc_warn_func_return): Likewise.
11638 (machine_function): Change type fn_type.
11639 (arc_compute_function_type): Consider new naked function type,
11640 change function return type.
11641 (arc_must_save_register): Adapt to handle new
11642 arc_compute_function_type's return type.
11643 (arc_expand_prologue): Likewise.
11644 (arc_expand_epilogue): Likewise.
11645 (arc_return_address_regs): Delete.
11646 (arc_return_address_register): New function.
11647 (arc_epilogue_uses): Use above function.
11648 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
11649 (arc_function_type): Change encoding, add naked type.
11650 (ARC_INTERRUPT_P): Change to handle the new encoding.
11651 (ARC_FAST_INTERRUPT_P): Likewise.
11652 (ARC_NORMAL_P): Define.
11653 (ARC_NAKED_P): Likewise.
11654 (arc_compute_function_type): Delete prototype.
11655 * config/arc/arc.md (in_ret_delay_slot): Use
11656 arc_return_address_register function.
11657 (simple_return): Likewise.
11658 (p_return_i): Likewise.
11660 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11662 PR tree-optimization/81428
11663 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
11664 can't be built for those types.
11666 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
11668 Remove stuff dead since r239246.
11670 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
11671 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
11672 (avr_inform_devices): Remove dead stuff.
11674 2017-07-17 Tamar Christina <tamar.christina@arm.com>
11676 * config/arm/arm_neon.h: Fix softp typo.
11678 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11680 PR tree-optimization/81365
11681 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
11682 aggregate moves onto bb predecessor edges, make sure there are no
11683 loads that could alias the lhs in between the start of bb and the
11686 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
11689 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
11690 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
11691 [LSHIFTRT, outer_code = TRUNCATE]: Same.
11693 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11695 PR tree-optimization/81396
11696 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
11697 (init_symbolic_number): Initialize it to 1.
11698 (perform_symbolic_merge): Add n_ops from both operands into the new
11700 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
11701 without base_addr as useless if they need more than one operation.
11702 (bswap_replace): Handle !bswap case for NULL base_addr.
11704 2017-07-17 Tom de Vries <tom@codesourcery.com>
11707 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
11710 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
11712 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
11713 conditional builtin define __FIX_LEON3FT_B2BST.
11715 2017-07-17 Daniel Cederman <cederman@gaisler.com>
11717 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
11718 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
11721 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
11723 PR rtl-optimization/81424
11724 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
11725 to remove potential trapping from operands if -fnon-call-exceptions.
11727 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11729 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
11730 profile_proability for scalling.
11731 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
11733 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11735 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
11737 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11739 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
11740 fixpoint arithmetics.
11742 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11744 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
11745 fixpoint arithmetics.
11747 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11749 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
11750 fixpoint arithmetics.
11752 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11754 * profile-count.h (profile_probability::from_reg_br_prob_note,
11755 profile_probability::to_reg_br_prob_note): New functions.
11756 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
11757 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
11758 * predict.c (probability_reliable_p): Update.
11759 (edge_probability_reliable_p): Update.
11760 (br_prob_note_reliable_p): Update.
11761 (invert_br_probabilities): Update.
11762 (add_reg_br_prob_note): New function.
11763 (combine_predictions_for_insn): Update.
11764 * asan.c (asan_clear_shadow): Update.
11765 * cfgbuild.c (compute_outgoing_frequencies): Update.
11766 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
11767 (update_br_prob_note): Update.
11768 (rtl_verify_edges): Update.
11769 (purge_dead_edges): Update.
11770 (fixup_reorder_chain): Update.
11771 * emit-rtl.c (try_split): Update.
11772 * ifcvt.c (cond_exec_process_insns): Update.
11773 (cond_exec_process_if_block): Update.
11774 (dead_or_predicable): Update.
11775 * internal-fn.c (expand_addsub_overflow): Update.
11776 (expand_neg_overflow): Update.
11777 (expand_mul_overflow): Update.
11778 * loop-doloop.c (doloop_modify): Update.
11779 * loop-unroll.c (compare_and_jump_seq): Update.
11780 * optabs.c (emit_cmp_and_jump_insn_1): Update.
11781 * predict.h: Update.
11782 * reorg.c (mostly_true_jump): Update.
11784 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
11785 * config/alpha/alpha.c (emit_unlikely_jump): Update.
11786 * config/arc/arc.c: (emit_unlikely_jump): Update.
11787 * config/arm/arm.c: (emit_unlikely_jump): Update.
11788 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
11789 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
11790 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
11791 (ix86_print_operand): Update.
11792 (ix86_split_fp_branch): Update.
11793 (predict_jump): Update.
11794 * config/ia64/ia64.c (ia64_print_operand): Update.
11795 * config/mmix/mmix.c (mmix_print_operand): Update.
11796 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
11797 (rs6000_expand_split_stack_prologue): Update.
11798 * config/rs6000/rs6000.c: Update.
11799 * config/s390/s390.c (s390_expand_vec_strlen): Update.
11800 (s390_expand_vec_movstr): Update.
11801 (s390_expand_cs_tdsi): Update.
11802 (s390_expand_split_stack_prologue): Update.
11803 * config/sh/sh.c (sh_print_operand): Update.
11804 (expand_cbranchsi4): Update.
11805 (expand_cbranchdi4): Update.
11806 * config/sparc/sparc.c (output_v9branch): Update.
11807 * config/spu/spu.c (get_branch_target): Update.
11808 (ea_load_store_inline): Update.
11809 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
11810 * config/tilepro/tilepro.c: Update.
11812 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
11814 * gimplify.c (mostly_copy_tree_r): Revert latest change.
11815 (gimplify_save_expr): Likewise.
11817 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11819 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
11821 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11823 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
11825 * timevar.def (TV_IPA_FNSUMMARY): Define.
11827 2017-07-16 Daniel Cederman <cederman@gaisler.com>
11829 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
11830 to back store errata sensitive sequence from being generated.
11831 (sqrtdf2_fix): Likewise.
11833 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11835 * tree-ssa-threadupdate.c (compute_path_counts,
11836 update_joiner_offpath_counts): Use profile_probability.
11838 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
11841 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11843 * config/arm/arm-c.c (arm_cpu_builtins): Define
11844 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
11846 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
11848 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11849 array entries to represent __ieee128 versions of the
11850 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
11851 scalar_extract_sig, and scalar_insert_exp built-in functions.
11852 (altivec_resolve_overloaded_builtin): Add special case handling
11853 for the __builtin_scalar_insert_exp function, as represented by
11854 the P9V_BUILTIN_VEC_VSIEDP constant.
11855 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
11856 exponent support for __ieee128 argument.
11857 (VSESQP): Add scalar extract signature support for __ieee128
11859 (VSTDCNQP): Add scalar test negative support for __ieee128
11861 (VSIEQP): Add scalar insert exponent support for __int128 argument
11862 with __ieee128 result.
11863 (VSIEQPF): Add scalar insert exponent support for __ieee128
11864 argument with __ieee128 result.
11865 (VSTDCQP): Add scalar test data class support for __ieee128
11867 (VSTDCNQP): Add overload support for scalar test negative with
11868 __ieee128 argument.
11869 (VSTDCQP): Add overload support for scalar test data class
11870 __ieee128 argument.
11871 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
11872 UNSPEC_VSX_SXSIGDP.
11873 (UNSPEC_VSX_SIEXPQP): New constant.
11874 (xsxexpqp): New insn for VSX scalar extract exponent quad
11876 (xsxsigqp): New insn for VSX scalar extract significand quad
11878 (xsiexpqpf): New insn for VSX scalar insert exponent quad
11879 precision with floating point argument.
11880 (xststdcqp): New expand for VSX scalar test data class quad
11882 (xststdcnegqp): New expand for VSX scalar test negative quad
11884 (xststdcqp): New insn to match expansions for VSX scalar test data
11885 class quad precision and VSX scalar test negative quad precision.
11886 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
11887 special case operand checking to enforce that second operand of
11888 VSX scalar test data class with quad precision argument is a 7-bit
11890 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
11891 prototypes and descriptions of __ieee128 versions of
11892 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
11893 scalar_test_data_class, and scalar_test_neg built-in functions.
11895 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11897 PR tree-optimization/81162
11898 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
11899 replace a negate with an add.
11901 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
11903 * doc/invoke.texi (arm/-mcpu): Document +crypto.
11905 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11907 * config/arm/arm-c.c (arm_cpu_builtins): Define
11908 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
11910 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11912 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
11913 (armv8-r): Set ARM Cortex-R52 as default CPU.
11914 * config/arm/arm-tables.opt: Regenerate.
11915 * config/arm/arm-tune.md: Regenerate.
11916 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
11918 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
11919 extension for -mcpu=cortex-r52.
11921 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11923 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
11924 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
11925 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
11926 (fp-armv8): Define it as FP_ARMv8 only.
11927 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
11928 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
11930 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
11931 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
11932 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
11933 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
11934 than TARGET_FPU_ARMV8.
11935 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
11936 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
11937 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
11939 * config/arm/neon.md (neon_vrint): Likewise.
11940 (neon_vcvt): Likewise.
11941 (neon_<fmaxmin_op><mode>): Likewise.
11942 (<fmaxmin><mode>3): Likewise.
11943 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
11944 * config/arm/predicates.md (arm_cond_move_operator): Check against
11945 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
11947 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
11949 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
11950 to top of function.
11952 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11954 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
11955 loop in comment with memset.
11957 2017-07-14 Martin Liska <mliska@suse.cz>
11959 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
11960 * dwarf2out.c (is_java): Remove the function.
11961 (output_pubname): Remove usage of the function.
11962 (lower_bound_default): Remove usage of DW_LANG_Java.
11963 (gen_compile_unit_die): Likewise.
11964 * gcc.c: Remove compiler defaults for .java and .zip files.
11965 * gimple-expr.c (remove_suffix): Change as there's no longer
11966 extension than 4-letter one.
11967 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
11968 (gimplify_save_expr): Likewise.
11969 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
11970 as it's possible even for other languages than Java.
11971 * langhooks.h (struct lang_hooks): Remove Java from a comment.
11972 * lto-opts.c (lto_write_options): Remove reference to Java.
11973 * opts.c (strip_off_ending): Update file extension handling.
11974 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
11975 * tree-eh.c (lower_resx): Likewise.
11976 * tree.c (free_lang_data_in_type): Remove dead code.
11977 (find_decls_types_r): Likewise.
11978 (build_common_builtin_nodes): Remove Java from a comment.
11979 (verify_type): Remove dead code.
11980 * varasm.c (assemble_external): Remove Java from a comment.
11982 2017-07-14 Martin Liska <mliska@suse.cz>
11984 * opts.c (finish_options): Add quotes.
11985 (common_handle_option): Likewise.
11987 2017-07-14 Martin Liska <mliska@suse.cz>
11989 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
11990 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
11991 Remove N_SO_PASCAL.
11992 * dwarf2out.c (lower_bound_default): Do not handle
11994 (gen_compile_unit_die): Likewise.
11995 * gcc.c: Remove default extension binding for GNU Pascal.
11996 * stmt.c: Remove Pascal language from a comment.
11997 * xcoffout.c: Likewise.
11999 2017-07-13 David Malcolm <dmalcolm@redhat.com>
12002 * diagnostic-show-locus.c (fixit_cmp): New function.
12003 (layout::layout): Sort m_fixit_hints.
12004 (column_range::column_range): Assert that the values are valid.
12005 (struct char_span): New struct.
12006 (correction::overwrite): New method.
12007 (struct source_line): New struct.
12008 (line_corrections::add_hint): Add assertions. Reimplement memcpy
12009 calls in terms of classes source_line and char_span, and
12010 correction::overwrite.
12011 (selftest::test_overlapped_fixit_printing_2): New function.
12012 (selftest::diagnostic_show_locus_c_tests): Call it.
12014 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
12016 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
12017 early if there is no lhs.
12019 2017-07-13 Martin Liska <mliska@suse.cz>
12021 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
12022 (gen_reference_type_die): Likewise.
12023 * stor-layout.c: Remove Pascal-related comment.
12025 2017-07-13 Martin Liska <mliska@suse.cz>
12027 * opts.c (finish_options): Add quotes to error messages.
12028 (parse_sanitizer_options): Likewise.
12030 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
12032 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
12034 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
12036 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
12038 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
12040 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
12042 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
12044 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
12047 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
12048 provides the hardware capability bits, define the macro
12049 __BUILTIN_CPU_SUPPORTS__.
12050 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
12051 if GLIBC does not provide the hardware capability bits. Add a
12052 gcc_unreachable call if the built-in cpu function is neither
12053 __builtin_cpu_is nor __builtin_cpu_supports.
12054 (rs6000_get_function_versions_dispatcher): Change the warning
12055 that an old GLIBC is used which does not export the capability
12056 bits to be an error.
12057 * doc/extend.texi (target_clones attribute): Document the
12058 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
12059 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
12060 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
12061 the macros defined by GCC if the newer GLIBC is available.
12063 2017-07-12 Jeff Law <law@redhat.com>
12065 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
12066 remaining includes slightly.
12067 * config/riscv/riscv-builtins.c: Include profile-count.h.
12069 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
12072 * config/avr/avr.c (avr_set_current_function): In diagnostic
12073 messages: Quote keywords and (parts of) identifiers.
12074 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
12077 2017-07-12 Carl Love <cel@us.ibm.com>
12079 * config/rs6000/rs6000-c.c: Add support for built-in functions
12080 vector bool char vec_revb (vector bool char);
12081 vector bool short vec_revb (vector short char);
12082 vector bool int vec_revb (vector bool int);
12083 vector bool long long vec_revb (vector bool long long);
12084 * doc/extend.texi: Update the built-in documentation file for the
12085 new built-in functions.
12087 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12089 * config/s390/s390.md: Remove movcc splitter.
12091 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12093 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
12094 load/store on condition.
12096 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
12099 * config/avr/avr.c (avr_encode_section_info)
12100 [progmem && !TREE_READONLY]: Error if progmem object needs
12103 2017-07-11 Michael Collison <michael.collison@arm.com>
12105 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
12108 2017-07-11 Carl Love <cel@us.ibm.com>
12110 * config/rs6000/rs6000-c.c: Add support for builtins
12111 vector unsigned int vec_parity_lsbb (vector signed int);
12112 vector unsigned int vec_parity_lsbb (vector unsigned int);
12113 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
12114 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
12115 vector unsigned long long vec_parity_lsbb (vector signed long long);
12116 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
12117 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
12118 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
12119 * doc/extend.texi: Update the built-in documentation file for the
12120 new built-in functions.
12122 2017-07-11 David Malcolm <dmalcolm@redhat.com>
12124 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
12125 (layout::m_primary_loc): New field.
12126 (layout::layout): Initialize new field. Move location filtering
12127 logic from here to...
12128 (layout::maybe_add_location_range): ...this new method. Add
12129 support for filtering to just the lines already specified by other
12131 (layout::will_show_line_p): New method.
12132 (gcc_rich_location::add_location_if_nearby): New method.
12133 (selftest::test_add_location_if_nearby): New test function.
12134 (selftest::diagnostic_show_locus_c_tests): Call it.
12135 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
12138 2017-07-11 Tom de Vries <tom@codesourcery.com>
12140 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
12141 (bb_first_real_insn): New function.
12142 (nvptx_single): Add extra initialization of broadcasted condition
12145 2017-07-11 Nathan Sidwell <nathan@acm.org>
12147 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
12149 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
12151 * doc/extend.texi (AVR Function Attributes): Remove weblink to
12152 Binutils doc as TEXI will mess them up.
12153 * doc/invoke.texi (AVR Options): Same here.
12155 2017-07-11 Daniel Cederman <cederman@gaisler.com>
12157 * config/sparc/sparc.opt (mfix-ut700): New option.
12158 (mfix-gr712rc): Likewise.
12159 (sparc_fix_b2bst): New variable.
12160 * doc/invoke.texi (SPARC options): Document them.
12161 (ARM options): Fix warnings.
12162 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
12163 instructions to prevent sequences that can trigger the store-store
12164 errata for certain LEON3FT processors.
12165 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
12166 (sparc_option_override): Set sparc_fix_b2bst appropriately.
12167 * config/sparc/sparc.md (fix_b2bst): New attribute.
12168 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
12170 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
12173 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
12175 (*rsqrtsf2_sse): Ditto.
12177 (div<mode>3): Macroize insn from divdf3 and divsf3
12178 using MODEF mode iterator.
12180 2017-07-10 Martin Sebor <msebor@redhat.com>
12182 PR tree-optimization/80397
12183 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
12184 instead of testing for equality to INTEGER_TYPE.
12186 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
12188 * config.gcc: Remove uclibc from arc target spec.
12190 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
12192 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
12194 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12197 * lto-wrapper.c (remove_option): New function.
12198 (merge_and_complain): Merge PIC/PIE options more realistically.
12200 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
12202 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
12206 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
12207 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
12208 * config.in: Regenerate.
12209 * configure: Regenerate.
12210 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
12211 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
12212 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
12213 (TARGET_GASISR_PROLOGUES): ...target mask.
12214 * common/config/avr/avr-common.c
12215 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
12216 Set -mgas-isr-prologues.
12217 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
12218 INSERT_PASS_BEFORE for it.
12219 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
12220 * config/avr/avr.c (avr_option_override)
12221 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
12222 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
12223 (avr_attribute_table) <no_gccisr>: Add new function attribute.
12224 (avr_set_current_function) <is_no_gccisr>: Init machine field.
12225 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
12227 (make_avr_pass_pre_proep): New function.
12228 (emit_push_sfr) <treg>: Add argument to function and use it
12229 instead of TMP_REG.
12230 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
12231 and set machine->gasisr.yes.
12232 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
12233 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
12234 __gcc_isr.n_pushed to .L__stack_usage.
12235 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
12236 (avr_asm_final_postscan_insn): ...this new static function.
12237 * config/avr/avr.h (machine_function)
12238 <is_no_gccisr, use_L__stack_usage>: New fields.
12239 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
12240 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
12241 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
12242 (gasisr, *gasisr): New expander and insn.
12243 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
12244 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
12245 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
12247 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
12249 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
12252 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
12254 Move jump-tables out of .text again.
12257 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
12258 (ASM_OUTPUT_ADDR_VEC): New function.
12259 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
12260 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
12261 INSN_ADDRESSes as asm comment.
12262 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
12263 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
12264 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
12265 * config/avr/avr.md (*tablejump): Adjust comment.
12266 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
12267 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
12269 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
12270 (avr_output_addr_vec): New proto.
12271 (avr_log_t) <insn_addresses>: New field.
12273 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
12276 * config/i386/i386.c (ix86_function_arg_advance): Set
12277 outgoing_args_on_stack to true if there are outgoing arguments
12279 (ix86_function_arg): Likewise.
12280 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
12281 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
12282 * config/i386/i386.h (machine_function): Add
12283 outgoing_args_on_stack.
12285 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
12287 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
12288 supporting pthreds.
12289 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
12291 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
12293 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
12294 (REAL_H): Remove $(MACHMODE_H).
12295 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
12297 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
12298 $(MACHMODE_H) and double-int.h.
12299 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
12301 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
12304 2017-07-07 Andrew Pinski <apinski@cavium.com>
12306 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
12307 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
12309 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
12311 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
12312 Add warning if GCC was not configured to link against a GLIBC that
12313 exports the hardware capability bits.
12314 (make_resolver_func): Make resolver function private and not a
12315 COMDAT function. Create the name with clone_function_name instead
12316 of make_unique_name.
12319 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
12320 correct operand in doing the split.
12322 2017-07-07 Carl Love <cel@us.ibm.com>
12324 * config/rs6000/rs6000-c: Add support for built-in function
12325 vector unsigned short vec_pack_to_short_fp32 (vector float,
12327 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
12328 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
12329 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
12330 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
12331 (convert_4f32_8i16): Add define_expand.
12332 * doc/extend.texi: Update the built-in documentation file for the
12333 new built-in function.
12335 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12337 * config/sparc/m8.md: New file.
12338 * config/sparc/sparc.md: Include m8.md.
12340 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12342 * config/sparc/sparc.opt: New option -mvis4b.
12343 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
12344 (sparc_option_override): Handle VIS4B.
12345 (enum sparc_builtins): Define
12346 SPARC_BUILTIN_DICTUNPACK{8,16,32},
12347 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
12348 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
12349 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
12350 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
12351 (check_constant_argument): New function.
12352 (sparc_vis_init_builtins): Define builtins
12353 __builtin_vis_dictunpack{8,16,32},
12354 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
12355 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
12356 __builtin_vis_fpcmpde{8,16,32}shl and
12357 __builtin_vis_fpcmpur{8,16,32}shl.
12358 (sparc_expand_builtin): Check that the constant operands to
12359 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
12360 constant and in range.
12361 * config/sparc/sparc-c.c (sparc_target_macros): Handle
12363 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
12364 (SPARC_IMM5_P): Likewise.
12365 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
12366 (enabled): Handle vis4b.
12367 (UNSPEC_DICTUNPACK): New unspec.
12368 (UNSPEC_FPCMPSHL): Likewise.
12369 (UNSPEC_FPUCMPSHL): Likewise.
12370 (UNSPEC_FPCMPDESHL): Likewise.
12371 (UNSPEC_FPCMPURSHL): Likewise.
12372 (cpu_feature): New CPU feature `vis4b'.
12373 (dictunpack{8,16,32}): New insns.
12374 (FPCSMODE): New mode iterator.
12375 (fpcscond): New code iterator.
12376 (fpcsucond): Likewise.
12377 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
12378 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
12379 (fpcmpde{8,16,32}{si,di}shl): Likewise.
12380 (fpcmpur{8,16,32}{si,di}shl): Likewise.
12381 * config/sparc/constraints.md: Define constraints `q' for unsigned
12382 2-bit integer constants and `t' for unsigned 5-bit integer
12384 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
12386 (imm5_operand_dictunpack16): Likewise.
12387 (imm5_operand_dictunpack32): Likewise.
12388 (imm2_operand): Likewise.
12389 * doc/invoke.texi (SPARC Options): Document -mvis4b.
12390 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
12391 ditunpack* and fpcmp*shl builtins.
12393 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12395 * config.gcc: Handle m8 in --with-{cpu,tune} options.
12396 * config.in: Add HAVE_AS_SPARC6 define.
12397 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
12399 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
12401 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
12402 (CPP_CPU_SPEC): Handle m8.
12403 (ASM_CPU_SPEC): Likewise.
12404 * config/sparc/sparc-opts.h (enum processor_type): Add
12406 * config/sparc/sparc.c (m8_costs): New struct.
12407 (sparc_option_override): Handle TARGET_CPU_m8.
12408 (sparc32_initialize_trampoline): Likewise.
12409 (sparc64_initialize_trampoline): Likewise.
12410 (sparc_issue_rate): Likewise.
12411 (sparc_register_move_cost): Likewise.
12412 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
12413 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
12414 (ASM_CPU64_DEFAULT_SPEC): Likewise.
12415 (CPP_CPU_SPEC): Handle M8.
12416 (ASM_CPU_SPEC): Likewise.
12417 (AS_M8_FLAG): Define.
12418 * config/sparc/sparc.md: Add m8 to the cpu attribute.
12419 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
12420 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
12422 * configure: Regenerate.
12423 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
12426 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12428 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
12430 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
12431 ("*movdi_insn_sp32"): Do not set v3pipe.
12432 ("*movsi_insn"): Likewise.
12433 ("*movdi_insn_sp64"): Likewise.
12434 ("*movsf_insn"): Likewise.
12435 ("*movdf_insn_sp32"): Likewise.
12436 ("*movdf_insn_sp64"): Likewise.
12437 ("*zero_extendsidi2_insn_sp64"): Likewise.
12438 ("*sign_extendsidi2_insn"): Likewise.
12439 ("*mov<VM32:mode>_insn"): Likewise.
12440 ("*mov<VM64:mode>_insn_sp64"): Likewise.
12441 ("*mov<VM64:mode>_insn_sp32"): Likewise.
12442 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
12443 ("<vlop:code><VL:mode>3"): Likewise.
12444 ("*not_<vlop:code><VL:mode>3"): Likewise.
12445 ("*nand<VL:mode>_vis"): Likewise.
12446 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
12447 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
12448 ("one_cmpl<VL:mode>2"): Likewise.
12449 ("faligndata<VM64:mode>_vis"): Likewise.
12450 ("alignaddrsi_vis"): Likewise.
12451 ("alignaddrdi_vis"): Likweise.
12452 ("alignaddrlsi_vis"): Likewise.
12453 ("alignaddrldi_vis"): Likewise.
12454 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
12455 ("bmaskdi_vis"): Likewise.
12456 ("bmasksi_vis"): Likewise.
12457 ("bshuffle<VM64:mode>_vis"): Likewise.
12458 ("cmask8<P:mode>_vis"): Likewise.
12459 ("cmask16<P:mode>_vis"): Likewise.
12460 ("cmask32<P:mode>_vis"): Likewise.
12461 ("pdistn<P:mode>_vis"): Likewise.
12462 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
12464 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12466 * config/sparc/sparc.md ("subtype"): New insn attribute.
12467 ("*wrgsr_sp64"): Set insn subtype.
12468 ("*rdgsr_sp64"): Likewise.
12469 ("alignaddrsi_vis"): Likewise.
12470 ("alignaddrdi_vis"): Likewise.
12471 ("alignaddrlsi_vis"): Likewise.
12472 ("alignaddrldi_vis"): Likewise.
12473 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
12474 ("fexpand_vis"): Likewise.
12475 ("fpmerge_vis"): Likewise.
12476 ("faligndata<VM64:mode>_vis"): Likewise.
12477 ("bshuffle<VM64:mode>_vis"): Likewise.
12478 ("cmask8<P:mode>_vis"): Likewise.
12479 ("cmask16<P:mode>_vis"): Likewise.
12480 ("cmask32<P:mode>_vis"): Likewise.
12481 ("fchksm16_vis"): Likewise.
12482 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
12483 ("fmean16_vis"): Likewise.
12484 ("fp<plusminus_insn>64_vis"): Likewise.
12485 ("<plusminus_insn>v8qi3"): Likewise.
12486 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
12487 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
12488 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
12489 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
12490 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
12491 ("*movqi_insn"): Likewise.
12492 ("*movhi_insn"): Likewise.
12493 ("*movsi_insn"): Likewise.
12494 ("movsi_pic_gotdata_op"): Likewise.
12495 ("*movdi_insn_sp32"): Likewise.
12496 ("*movdi_insn_sp64"): Likewise.
12497 ("movdi_pic_gotdata_op"): Likewise.
12498 ("*movsf_insn"): Likewise.
12499 ("*movdf_insn_sp32"): Likewise.
12500 ("*movdf_insn_sp64"): Likewise.
12501 ("*zero_extendhisi2_insn"): Likewise.
12502 ("*zero_extendqihi2_insn"): Likewise.
12503 ("*zero_extendqisi2_insn"): Likewise.
12504 ("*zero_extendqidi2_insn"): Likewise.
12505 ("*zero_extendhidi2_insn"): Likewise.
12506 ("*zero_extendsidi2_insn_sp64"): Likewise.
12507 ("ldfsr"): Likewise.
12508 ("prefetch_64"): Likewise.
12509 ("prefetch_32"): Likewise.
12510 ("tie_ld32"): Likewise.
12511 ("tie_ld64"): Likewise.
12512 ("*tldo_ldub_sp32"): Likewise.
12513 ("*tldo_ldub1_sp32"): Likewise.
12514 ("*tldo_ldub2_sp32"): Likewise.
12515 ("*tldo_ldub_sp64"): Likewise.
12516 ("*tldo_ldub1_sp64"): Likewise.
12517 ("*tldo_ldub2_sp64"): Likewise.
12518 ("*tldo_ldub3_sp64"): Likewise.
12519 ("*tldo_lduh_sp32"): Likewise.
12520 ("*tldo_lduh1_sp32"): Likewise.
12521 ("*tldo_lduh_sp64"): Likewise.
12522 ("*tldo_lduh1_sp64"): Likewise.
12523 ("*tldo_lduh2_sp64"): Likewise.
12524 ("*tldo_lduw_sp32"): Likewise.
12525 ("*tldo_lduw_sp64"): Likewise.
12526 ("*tldo_lduw1_sp64"): Likewise.
12527 ("*tldo_ldx_sp64"): Likewise.
12528 ("*mov<VM32:mode>_insn"): Likewise.
12529 ("*mov<VM64:mode>_insn_sp64"): Likewise.
12530 ("*mov<VM64:mode>_insn_sp32"): Likewise.
12532 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12534 * config/sparc/sparc.md ("type"): New insn type viscmp.
12535 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
12537 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
12538 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
12539 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
12540 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
12542 ("n7_vis_logical_11cycle"): Likewise.
12543 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
12544 * config/sparc/niagara2.md ("niag3_vis": Likewise.
12545 * config/sparc/niagara.md ("niag_vis"): Likewise.
12546 * config/sparc/ultra3.md ("us3_fga"): Likewise.
12547 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
12549 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12551 * config/sparc/sparc.md: New instruction type `bmask'.
12552 (bmaskdi_vis): Use the `bmask' type.
12553 (bmasksi_vis): Likewise.
12554 * config/sparc/ultra3.md (us3_array): Likewise.
12555 * config/sparc/niagara7.md (n7_array): Likewise.
12556 * config/sparc/niagara4.md (n4_array): Likewise.
12557 * config/sparc/niagara2.md (niag2_vis): Likewise.
12558 (niag3_vis): Likewise.
12559 * config/sparc/niagara.md (niag_vis): Likewise.
12561 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12563 * ipa-comdats.c: Remove optimize check from gate.
12564 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
12565 for functions not optimized.
12566 (ipa_fn_summary_read): Skip optimize check.
12567 (ipa_fn_summary_write): Likewise.
12568 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
12570 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
12572 (can_inline_edge_p): Check flag_pcc_struct_return for match.
12573 (check_callers): Give up on caller which is not optimized.
12574 (inline_small_functions): Likewise.
12575 (ipa_inline): Do not give up when not optimizing.
12576 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
12577 away unoptimizes cdtors.
12578 (whole_program_function_and_variable_visibility): Do
12579 ipa_discover_readonly_nonaddressable_vars in LTO mode.
12580 * ipa.c (process_references): Do not check optimize.
12581 (symbol_table::remove_unreachable_nodes): Update optimize check.
12582 (set_writeonly_bit): Update optimize check.
12583 (pass_ipa_cdtor_merge::gate): Do not check optimize.
12584 (pass_ipa_single_use::gate): Remove.
12586 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12588 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
12589 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
12590 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
12591 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
12592 permute_load, permute_store, adjust_extract, adjust_splat,
12593 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
12594 replace_swap_with_copy, dump_swap_insn_table,
12595 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
12596 recombine_lvx_pattern, recombine_stvx_pattern,
12597 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
12598 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
12599 to file rs6000-p8swap.c.
12600 * config/rs6000/rs6000-p8swap.c: New file.
12601 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
12602 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
12603 and rs6000*-*-* targets.
12605 2017-07-06 David Malcolm <dmalcolm@redhat.com>
12607 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
12609 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12611 * lto-wrapper.c (merge_and_complain): Do not merge
12612 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
12613 fsigned_zeros, ftrapping_math, fwrapv.
12614 (append_compiler_options): Do not track these options.
12615 (append_linker_options): Likewie
12617 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12619 * cgraphunit.c (cgraph_node::finalize_function): When
12620 !flag_toplevel_reorde set no_reorder flag.
12621 (varpool_node::finalize_decl): Likewise.
12622 (symbol_table::compile): Drop no toplevel reorder path.
12624 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12626 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
12627 edges; zero probability is not better than uninitialized.
12629 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
12631 * asan.h (asan_sanitize_allocas_p): Declare.
12632 * asan.c (asan_sanitize_allocas_p): New function.
12633 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
12634 (handle_builtin_alloca): Likewise.
12635 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
12636 if !asan_sanitize_allocas_p.
12637 * params.def (asan-instrument-allocas): Add new option.
12638 * params.h (ASAN_PROTECT_ALLOCAS): Define.
12639 * opts.c (common_handle_option): Disable allocas sanitization for
12642 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
12644 * asan.c: Include gimple-fold.h.
12645 (get_last_alloca_addr): New function.
12646 (handle_builtin_stackrestore): Likewise.
12647 (handle_builtin_alloca): Likewise.
12648 (asan_emit_allocas_unpoison): Likewise.
12649 (get_mem_refs_of_builtin_call): Add new parameter, remove const
12650 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
12651 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
12652 (instrument_builtin_call): Pass gimple iterator to
12653 get_mem_refs_of_builtin_call.
12654 (last_alloca_addr): New global.
12655 * asan.h (asan_emit_allocas_unpoison): Declare.
12656 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
12657 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
12658 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
12659 if function calls alloca.
12660 * gimple-fold.c (replace_call_with_value): Remove static keyword.
12661 * gimple-fold.h (replace_call_with_value): Declare.
12662 * internal-fn.c: Include asan.h.
12663 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
12664 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
12666 2017-07-06 David Malcolm <dmalcolm@redhat.com>
12668 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
12669 (C_SELFTEST_FLAGS): New.
12670 (CPP_SELFTEST_FLAGS): New.
12671 (SELFTEST_DEPS): New, from deps of s-selftest.
12672 (C_SELFTEST_DEPS): New, from deps of s-selftest.
12673 (CPP_SELFTEST_DEPS): New.
12674 (selftest): Add dependency on s-selftest-c++.
12675 (s-selftest): Rename to...
12676 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
12677 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
12678 than SELFTEST_FLAGS.
12679 (selftest-gdb): Rename to...
12680 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
12682 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
12683 (selftest-valgrind): Rename to...
12684 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
12686 (selftest-valgrind): Reintroduce as an alias for
12687 selftest-c-valgrind.
12688 (s-selftest-c++): New.
12689 (selftest-c++-gdb): New.
12690 (selftest-c++-valgrind): New.
12692 2017-07-06 Olivier Hainque <hainque@adacore.com>
12694 * gcc.c (process_command): When deciding if undefined variables
12695 should be ignored when processing specs, accept "gcc -v" as well.
12697 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12699 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
12700 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
12702 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
12704 * config/arm/arm-cpus.in (armv8-r): Add new entry.
12705 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
12706 * config/arm/arm-tables.opt: Regenerate.
12707 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
12709 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
12711 2017-07-06 Carl Love <cel@us.ibm.com>
12713 * ChangeLog: Clean up from mid air collision
12715 2017-07-06 Carl Love <cel@us.ibm.com>
12717 * config/rs6000/rs6000-c.c: Add support for built-in functions
12718 vector signed int vec_subc (vector signed int, vector signed int);
12719 vector signed __int128 vec_subc (vector signed __int128,
12720 vector signed __int128);
12721 vector unsigned __int128 vec_subc (vector unsigned __int128,
12722 vector unsigned __int128);
12723 vector signed int vec_sube (vector signed int, vector signed int,
12724 vector signed int);
12725 vector unsigned int vec_sube (vector unsigned int,
12726 vector unsigned int,
12727 vector unsigned int);
12728 vector signed __int128 vec_sube (vector signed __int128,
12729 vector signed __int128,
12730 vector signed__int128);
12731 vector unsigned __int128 vec_sube (vector unsigned __int128,
12732 vector unsigned __int128,
12733 vector unsigned __int128);
12734 vector signed int vec_subec (vector signed int, vector signed int,
12735 vector signed int);
12736 vector unsigned int vec_subec (vector unsigned int,
12737 vector unsigned int,
12738 vector unsigned int);
12739 vector signed __int128 vec_subec (vector signed __int128,
12740 vector signed __int128,
12741 vector signed__int128);
12742 vector unsigned __int128 vec_subec (vector unsigned __int128,
12743 vector unsigned __int128,
12744 vector unsigned __int128);
12745 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
12746 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
12747 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
12748 BU_ALTIVEC_OVERLOAD_X definitions.
12749 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
12750 * doc/extend.texi: Update the built-in documentation file for the new
12751 built-in functions.
12753 2017-07-06 David Malcolm <dmalcolm@redhat.com>
12756 * diagnostic-show-locus.c (layout::layout): Use start and finish
12757 spelling location for the start and finish of each range.
12758 * genmatch.c (linemap_client_expand_location_to_spelling_point):
12759 Add unused aspect param.
12760 * input.c (expand_location_1): Add "aspect" param, and use it
12761 to access the correct part of the location.
12762 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
12764 (expand_location_to_spelling_point): Likewise.
12765 (linemap_client_expand_location_to_spelling_point): Add "aspect"
12766 param, and pass it to expand_location_1.
12768 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
12770 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
12771 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
12772 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
12773 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
12774 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
12775 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
12776 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
12777 _mm_maskz_getmant_ss): New intrinsics.
12778 (__builtin_ia32_getexpss128_mask): Changed to ...
12779 __builtin_ia32_getexpss128_round ... this.
12780 (__builtin_ia32_getexpsd128_mask): Changed to ...
12781 __builtin_ia32_getexpsd128_round ... this.
12782 * config/i386/i386-builtin-types.def
12783 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
12784 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
12785 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
12786 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
12787 __builtin_ia32_getmantss_mask_round): New builtins.
12788 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
12789 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
12790 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
12791 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
12792 * config/i386/sse.md
12793 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
12794 avx512f_sgetexp<mode><mask_scalar_name>
12795 <round_saeonly_scalar_name> ... this.
12796 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
12797 %0, %1, %2<round_saeonly_op3>}): Changed to ...
12798 vgetexp<ssescalarmodesuffix>
12799 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12800 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
12801 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
12802 avx512f_vgetmant<mode><mask_scalar_name>
12803 <round_saeonly_scalar_name> ... this.
12804 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
12805 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
12806 vgetmant<ssescalarmodesuffix>
12807 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
12808 %0<mask_scalar_operand4>, %1, %2
12809 <round_saeonly_scalar_mask_op4>, %3} ... this.
12810 * config/i386/subst.md (mask_scalar_operand4,
12811 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
12812 round_saeonly_scalar_nimm_predicate): New subst attributes.
12814 2017-07-06 Julia Koval <julia.koval@intel.com>
12816 * config/i386/i386.c (ix86_erase_embedded_rounding):
12817 Remove code for old rounding pattern.
12819 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
12821 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
12823 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
12825 * doc/sourcebuild.texi (Test Directives, Variants of
12826 dg-require-support): Add documentation for dg-require-stack-check.
12828 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
12830 * config/i386/subst.md (mask_scalar, round_scalar,
12831 round_saeonly_scalar): New meta-templates.
12832 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
12833 round_scalar_mask_operand3, round_scalar_mask_op3,
12834 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
12835 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
12836 round_saeonly_scalar_constraint,
12837 round_saeonly_scalar_prefix): New subst attribute.
12838 * config/i386/sse.md
12839 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
12840 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
12841 <round_scalar_name> ... this.
12842 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
12843 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
12844 <round_scalar_name> ... this.
12845 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
12846 <sse>_vm<code><mode>3<mask_scalar_name>
12847 <round_saeonly_scalar_name> ... this.
12848 (v<plusminus_mnemonic><ssescalarmodesuffix>
12849 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
12850 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
12851 v<plusminus_mnemonic><ssescalarmodesuffix>
12852 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12853 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
12854 (v<multdiv_mnemonic><ssescalarmodesuffix>
12855 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
12856 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
12857 v<multdiv_mnemonic><ssescalarmodesuffix>
12858 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12859 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
12860 (v<maxmin_float><ssescalarmodesuffix>
12861 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
12862 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
12863 v<maxmin_float><ssescalarmodesuffix>
12864 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12865 %0<mask_scalar_operand3>, %1, %<iptr>2
12866 <round_saeonly_scalar_mask_op3>} ... this.
12868 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
12870 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
12871 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
12873 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
12874 Alan Hayward <alan.hayward@arm.com>
12875 David Sherwood <david.sherwood@arm.com>
12877 * combine.c (simplify_if_then_else): Remove "enum" before
12879 * compare-elim.c (can_eliminate_compare): Likewise.
12880 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
12882 (aarch64_lookup_simd_builtin_type): Likewise.
12883 (aarch64_simd_builtin_type): Likewise.
12884 (aarch64_init_simd_builtin_types): Likewise.
12885 (aarch64_simd_expand_args): Likewise.
12886 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
12888 (aarch64_reverse_mask): Likewise.
12889 (aarch64_simd_emit_reg_reg_move): Likewise.
12890 (aarch64_gen_adjusted_ldpstp): Likewise.
12891 (aarch64_ccmp_mode_to_code): Likewise.
12892 (aarch64_operands_ok_for_ldpstp): Likewise.
12893 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12894 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
12896 (aarch64_min_divisions_for_recip_mul): Likewise.
12897 (aarch64_reassociation_width): Likewise.
12898 (aarch64_get_condition_code_1): Likewise.
12899 (aarch64_simd_emit_reg_reg_move): Likewise.
12900 (aarch64_simd_attr_length_rglist): Likewise.
12901 (aarch64_reverse_mask): Likewise.
12902 (aarch64_operands_ok_for_ldpstp): Likewise.
12903 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12904 (aarch64_gen_adjusted_ldpstp): Likewise.
12905 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
12907 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
12908 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
12909 (arm_lookup_simd_builtin_type): Likewise.
12910 (arm_simd_builtin_type): Likewise.
12911 (arm_init_simd_builtin_types): Likewise.
12912 (arm_expand_builtin_args): Likewise.
12913 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
12914 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
12915 (ft32_setup_incoming_varargs): Likewise.
12916 (ft32_function_arg): Likewise.
12917 (ft32_function_arg_advance): Likewise.
12918 (ft32_pass_by_reference): Likewise.
12919 (ft32_arg_partial_bytes): Likewise.
12920 (ft32_valid_pointer_mode): Likewise.
12921 (ft32_addr_space_pointer_mode): Likewise.
12922 (ft32_addr_space_legitimate_address_p): Likewise.
12923 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
12925 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
12926 (ix86_emit_outlined_ms2sysv_restore): Likewise.
12927 (iamcu_alignment): Likewise.
12928 (canonicalize_vector_int_perm): Likewise.
12929 (ix86_noce_conversion_profitable_p): Likewise.
12930 (ix86_mpx_bound_mode): Likewise.
12931 (ix86_operands_ok_for_move_multiple): Likewise.
12932 * config/microblaze/microblaze-protos.h
12933 (microblaze_expand_conditional_branch_reg): Likewise.
12934 * config/microblaze/microblaze.c
12935 (microblaze_expand_conditional_branch_reg): Likewise.
12936 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12938 (rs6000_reassociation_width): Likewise.
12939 (rs6000_invalid_binary_op): Likewise.
12940 (fusion_p9_p): Likewise.
12941 (emit_fusion_p9_load): Likewise.
12942 (emit_fusion_p9_store): Likewise.
12943 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
12945 (riscv_hard_regno_mode_ok_p): Likewise.
12946 (riscv_address_insns): Likewise.
12947 (riscv_split_symbol): Likewise.
12948 (riscv_legitimize_move): Likewise.
12949 (riscv_function_value): Likewise.
12950 (riscv_hard_regno_nregs): Likewise.
12951 (riscv_expand_builtin): Likewise.
12952 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
12953 (riscv_build_integer): Likewise.
12954 (riscv_split_integer): Likewise.
12955 (riscv_legitimate_constant_p): Likewise.
12956 (riscv_cannot_force_const_mem): Likewise.
12957 (riscv_regno_mode_ok_for_base_p): Likewise.
12958 (riscv_valid_base_register_p): Likewise.
12959 (riscv_valid_offset_p): Likewise.
12960 (riscv_valid_lo_sum_p): Likewise.
12961 (riscv_classify_address): Likewise.
12962 (riscv_legitimate_address_p): Likewise.
12963 (riscv_address_insns): Likewise.
12964 (riscv_load_store_insns): Likewise.
12965 (riscv_force_binary): Likewise.
12966 (riscv_split_symbol): Likewise.
12967 (riscv_force_address): Likewise.
12968 (riscv_legitimize_address): Likewise.
12969 (riscv_move_integer): Likewise.
12970 (riscv_legitimize_const_move): Likewise.
12971 (riscv_legitimize_move): Likewise.
12972 (riscv_address_cost): Likewise.
12973 (riscv_subword): Likewise.
12974 (riscv_output_move): Likewise.
12975 (riscv_canonicalize_int_order_test): Likewise.
12976 (riscv_emit_int_order_test): Likewise.
12977 (riscv_function_arg_boundary): Likewise.
12978 (riscv_pass_mode_in_fpr_p): Likewise.
12979 (riscv_pass_fpr_single): Likewise.
12980 (riscv_pass_fpr_pair): Likewise.
12981 (riscv_get_arg_info): Likewise.
12982 (riscv_function_arg): Likewise.
12983 (riscv_function_arg_advance): Likewise.
12984 (riscv_arg_partial_bytes): Likewise.
12985 (riscv_function_value): Likewise.
12986 (riscv_pass_by_reference): Likewise.
12987 (riscv_setup_incoming_varargs): Likewise.
12988 (riscv_print_operand): Likewise.
12989 (riscv_elf_select_rtx_section): Likewise.
12990 (riscv_save_restore_reg): Likewise.
12991 (riscv_for_each_saved_reg): Likewise.
12992 (riscv_register_move_cost): Likewise.
12993 (riscv_hard_regno_mode_ok_p): Likewise.
12994 (riscv_hard_regno_nregs): Likewise.
12995 (riscv_class_max_nregs): Likewise.
12996 (riscv_memory_move_cost): Likewise.
12997 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
12998 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
12999 (rl78_addr_space_address_mode): Likewise.
13000 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13002 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13003 (rs6000_reassociation_width): Likewise.
13004 (rs6000_invalid_binary_op): Likewise.
13005 (fusion_p9_p): Likewise.
13006 (emit_fusion_p9_load): Likewise.
13007 (emit_fusion_p9_store): Likewise.
13008 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
13009 (ok_for_simple_move_operands): Likewise.
13010 (ok_for_simple_move_strict_operands): Likewise.
13011 (ok_for_simple_arith_logic_operands): Likewise.
13012 (visium_legitimize_reload_address): Likewise.
13013 (visium_select_cc_mode): Likewise.
13014 (output_cbranch): Likewise.
13015 (visium_split_double_move): Likewise.
13016 (visium_expand_copysign): Likewise.
13017 (visium_expand_int_cstore): Likewise.
13018 (visium_expand_fp_cstore): Likewise.
13019 * config/visium/visium.c (visium_pass_by_reference): Likewise.
13020 (visium_function_arg): Likewise.
13021 (visium_function_arg_advance): Likewise.
13022 (visium_libcall_value): Likewise.
13023 (visium_setup_incoming_varargs): Likewise.
13024 (visium_legitimate_constant_p): Likewise.
13025 (visium_legitimate_address_p): Likewise.
13026 (visium_legitimize_address): Likewise.
13027 (visium_secondary_reload): Likewise.
13028 (visium_register_move_cost): Likewise.
13029 (visium_memory_move_cost): Likewise.
13030 (prepare_move_operands): Likewise.
13031 (ok_for_simple_move_operands): Likewise.
13032 (ok_for_simple_move_strict_operands): Likewise.
13033 (ok_for_simple_arith_logic_operands): Likewise.
13034 (visium_function_value_1): Likewise.
13035 (rtx_ok_for_offset_p): Likewise.
13036 (visium_legitimize_reload_address): Likewise.
13037 (visium_split_double_move): Likewise.
13038 (visium_expand_copysign): Likewise.
13039 (visium_expand_int_cstore): Likewise.
13040 (visium_expand_fp_cstore): Likewise.
13041 (visium_split_cstore): Likewise.
13042 (visium_select_cc_mode): Likewise.
13043 (visium_split_cbranch): Likewise.
13044 (output_cbranch): Likewise.
13045 (visium_print_operand_address): Likewise.
13046 * expmed.c (flip_storage_order): Likewise.
13047 * expmed.h (emit_cstore): Likewise.
13048 (flip_storage_order): Likewise.
13049 * genrecog.c (validate_pattern): Likewise.
13050 * hsa-gen.c (gen_hsa_addr): Likewise.
13051 * internal-fn.c (expand_arith_overflow): Likewise.
13052 * ira-color.c (allocno_copy_cost_saving): Likewise.
13053 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13054 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
13055 (process_invariant_for_inheritance): Likewise.
13056 * lra-eliminations.c (move_plus_up): Likewise.
13057 * omp-low.c (lower_oacc_reductions): Likewise.
13058 * simplify-rtx.c (simplify_subreg): Likewise.
13059 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
13060 (TARGET_CHKP_BOUND_MODE): Likewise..
13061 * targhooks.c (default_chkp_bound_mode): Likewise.
13062 (default_setup_incoming_vararg_bounds): Likewise.
13063 * targhooks.h (default_chkp_bound_mode): Likewise.
13064 (default_setup_incoming_vararg_bounds): Likewise.
13065 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
13066 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
13067 (have_whole_vector_shift): Likewise.
13068 * tree-vect-stmts.c (vectorizable_load): Likewise.
13069 * doc/tm.texi: Regenerate.
13071 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
13073 Graceful degrade if Binutils PR21472 is not available.
13076 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
13077 .rodata in flash test fails.
13078 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
13079 * confgure: Regenerate.
13080 * config.in: Regenerate.
13081 * config/avr/avr.c (avr_asm_named_section)
13082 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
13083 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
13084 (avr_asm_init_sections): Same.
13086 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13088 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
13089 (fma<VH:mode>4_intrinsic): Likewise.
13090 (*fmsub<VCVTF:mode>4): Likewise.
13091 (*fmsub<VH:mode>4_intrinsic): Likewise.
13093 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
13096 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
13097 Don't depend on "optimize > 0".
13098 (out_movhi_r_mr, out_movqi_mr_r): Same.
13099 (out_movhi_mr_r, out_movqi_r_mr): Same.
13100 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
13101 io_address_operand on "optimize > 0".
13103 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13105 * tree-loop-distribution.c: Add general explanantion on the pass.
13106 (generate_loops_for_partition): Mark distributed loop.
13107 (pg_add_dependence_edges): New parameter. Handle alias data
13108 dependence specially and record it in the parameter if asked.
13109 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
13110 (init_partition_graph_vertices, add_partition_graph_edge): New.
13111 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
13112 (free_partition_graph_vdata, build_partition_graph): New.
13113 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
13114 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
13115 (data_ref_segment_size, latch_dominated_by_data_ref): New.
13116 (compute_alias_check_pairs, version_loop_by_alias_check): New.
13117 (version_for_distribution_p, finalize_partitions): New.
13118 (distribute_loop): Handle alias data dependence specially. Factor
13119 out loop fusion code as functions and call these functions.
13121 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13123 * tree-loop-distribution.c (classify_partition): New parameter and
13124 better handle reduction statement.
13125 (rdg_build_partitions): Revise comment.
13126 (distribute_loop): Compute statements in all partitions and pass it
13127 to classify_partition.
13129 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13131 * tree-loop-distribution.c (enum partition_type): New.
13132 (struct partition): New field type.
13133 (partition_merge_into): Add parameter. Update partition type.
13134 (data_dep_in_cycle_p, update_type_for_merge): New functions.
13135 (build_rdg_partition_for_vertex): Compute partition type.
13136 (rdg_build_partitions): Dump partition type.
13137 (distribute_loop): Update calls to partition_merge_into.
13139 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13141 * tree-loop-distribution.c (struct ddr_hasher): New.
13142 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
13144 (classify_partition): Call get_data_dependence.
13145 (pg_add_dependence_edges): Ditto.
13146 (distribute_loop): Release data dependence hash table.
13148 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13150 * tree-loop-distribution.c (ref_base_address): Delete.
13151 (similar_memory_accesses): Rename ...
13152 (share_memory_accesses): ... to this. Check if partitions access
13153 the same memory reference.
13154 (distribute_loop): Call share_memory_accesses.
13156 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13158 * tree-loop-distribution.c (struct partition): New field recording
13159 its data reference.
13160 (partition_alloc, partition_free): Init and release data refs.
13161 (partition_merge_into): Merge data refs.
13162 (build_rdg_partition_for_vertex): Collect data refs for partition.
13163 (pg_add_dependence_edges): Change parameters from vector to bitmap.
13165 (distribute_loop): Remve data refs from vertice data of partition
13168 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13170 * tree-loop-distribution.c (params.h): Include header file.
13171 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
13172 (datarefs_vec): New global var.
13173 (create_rdg_vertices): Use datarefs_vec directly.
13174 (free_rdg): Don't free data references.
13175 (build_rdg): Update use. Don't free data references.
13176 (distribute_loop): Compute global variable for data references.
13177 Bail out if there are too many data references.
13179 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13181 * tree-loop-distribution.c (loop_nest): New global var.
13182 (build_rdg): Use loop directly, rather than loop nest.
13183 (pg_add_dependence_edges): Remove loop nest parameter. Use global
13185 (distribute_loop): Compute global variable loop nest. Update use.
13187 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13189 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
13190 (partition_merge_into): New parameter. Dump reason for fusion.
13191 (distribute_loop): Update use of partition_merge_into.
13193 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13195 * tree-loop-distribution.c (bb_top_order_index): New.
13196 (bb_top_order_index_size, bb_top_order_cmp): New.
13197 (stmts_from_loop): Use topological order.
13198 (pass_loop_distribution::execute): Compute and release topological
13199 order for basic blocks.
13201 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13203 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
13206 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13208 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
13209 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
13210 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
13211 * internal-fn.def (LOOP_DIST_ALIAS): New.
13212 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
13213 (fold_loop_internal_call): ... this.
13214 (vect_loop_dist_alias_call): New function.
13215 (set_uid_loop_bbs): Call fold_loop_internal_call.
13216 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
13219 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
13222 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
13223 Require dead FLAGS_REG at the beginning of a peephole.
13225 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
13228 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
13229 arguments in the call to __builtin_ia32_sbb_u32.
13230 (_subborrow_u64): Swap _X and _Y arguments in the call to
13231 __builtin_ia32_sbb_u64.
13233 2017-07-04 Jakub Jelinek <jakub@redhat.com>
13236 * tree-vrp.c (compare_assert_loc): Turn into a function template
13237 with stable template parameter. Only test if a->e is NULL,
13238 !a->e == !b->e has been verified already. Use e == NULL or
13239 e != NULL instead of e or ! e tests. If stable is true, don't use
13240 iterative_hash_expr, on the other side allow a or b or both NULL
13241 and sort the NULLs last.
13242 (process_assert_insertions): Sort using compare_assert_loc<false>
13243 instead of compare_assert_loc, later sort using
13244 compare_assert_loc<true> before calling process_assert_insertions_for
13245 in a loop. Use break instead of continue once seen NULL pointer.
13247 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
13249 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
13250 Cortex-R7 and Cortex-R8 processors.
13252 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
13254 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
13255 uninitialized while src is not.
13257 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
13259 * common/config/arm/arm-common.c: Adjust include path for
13261 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
13262 (arm-cpu.h): Create in build directory. Adjust dependency rules.
13263 (arm-cpu-data.h): Likewise.
13264 (arm-cpu-cdata.h): Likewise.
13265 * config/arm/arm-cpu.h: Delete.
13266 * config/arm/arm-cpu-cdata.h: Delete.
13267 * config/arm/arm-cpu-data.h: Delete.
13269 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
13271 * config/arm/arm-cpus.in (cortex-a55): New.
13272 (cortex-a75): Likewise.
13273 (cortex-a75.cortex-a55): Likewise.
13274 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
13276 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
13277 * config/arm/arm-cpu-cdata.h: Regenerate.
13278 * config/arm/arm-cpu-data.h: Regenerate.
13279 * config/arm/arm-cpu.h: Regenerate.
13280 * config/arm/arm-tables.opt: Regenerate.
13281 * config/arm/arm-tune.md: Regenerate.
13283 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
13285 * haifa-sched.c (sched_create_recovery_edges): Update profile.
13287 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
13289 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
13292 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
13294 PR tree-optimization/81292
13295 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
13296 full_string_p, also call adjust_related_strinfos if the adjustment
13297 is simple, otherwise invalidate related strinfos.
13299 2017-07-04 Martin Liska <mliska@suse.cz>
13302 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
13303 newly created variable as DECL_IGNORED_P.
13305 2017-07-04 Martin Liska <mliska@suse.cz>
13308 * ipa-inline.c (inline_small_functions):
13309 Use xstrdup_for_dump.
13311 2017-07-04 Tom de Vries <tom@codesourcery.com>
13313 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
13315 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
13318 * config/darwin.c (darwin_function_switched_text_sections):
13321 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
13323 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
13325 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
13327 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
13329 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13331 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
13332 min_profitable_iters, and th as inclusive lower bounds.
13333 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
13334 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
13335 for min_profitable_iters and min_profitable_estimate.
13336 (vect_transform_loop): Treat th as an inclusive lower bound.
13337 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
13339 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
13342 * config/darwin.c (darwin_function_switched_text_sections):
13343 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
13344 in two pieces, and suppress the use of buf.
13346 2017-07-03 Nathan Sidwell <nathan@acm.org>
13348 * hash-table.h (hash_table_mod1): Fix indentation.
13350 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13352 PR middle-end/81290
13353 * predict.c (force_edge_cold): Be more careful about propagation
13355 * profile-count.h (profile_probability::guessed,
13356 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
13358 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
13360 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
13362 * doc/invoke.texi (rcpc architecture extension): Document it.
13364 2017-07-03 Richard Biener <rguenther@suse.de>
13366 PR tree-optimization/60510
13367 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
13368 the scalar reduction PHI and use it.
13369 (vectorizable_reduction): Properly guard the single_defuse_cycle
13370 path for non-SLP reduction chains where we cannot use it.
13371 Rework reduc_def/index and vector type deduction. Rework
13372 vector operand gathering during reduction op code-gen.
13373 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
13374 chains dissolve the chain and leave it to non-SLP reduction
13377 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13379 * tree-data-ref.h (dr_alignment): Declare.
13380 * tree-data-ref.c (dr_alignment): New function.
13381 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
13382 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
13384 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
13386 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13388 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
13389 and base_misalignment fields.
13390 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
13391 * tree-data-ref.c: Include builtins.h.
13392 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
13393 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
13394 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
13395 * tree-vect-data-refs.c: Include tree-cfg.h.
13396 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
13397 fields instead of calculating an alignment here.
13398 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
13399 innermost_loop_behavior fields.
13401 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13403 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
13405 (DR_STEP_ALIGNMENT): New macro.
13406 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
13407 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
13408 (create_data_ref): Print it.
13409 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
13410 to tell whether the step preserves vector (mis)alignment.
13411 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
13412 Move the check for an integer step and generalise to all INTEGER_CST.
13413 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
13414 Print the outer step alignment.
13416 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13418 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
13419 with offset_alignment.
13420 (DR_ALIGNED_TO): Delete.
13421 (DR_OFFSET_ALIGNMENT): New macro.
13422 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
13423 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
13424 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
13425 (create_data_ref): Likewise.
13426 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
13427 (vect_analyze_data_refs): Likewise.
13428 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
13429 creating dummy innermost behavior.
13431 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13433 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
13434 with a "innermost_loop_behavior *" and refeence tree.
13435 * tree-data-ref.c (dr_analyze_innermost): Likewise.
13436 (create_data_ref): Update call accordingly.
13437 * tree-predcom.c (find_looparound_phi): Likewise.
13439 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13441 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
13442 fields with dr_wrt_vec_loop.
13443 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
13444 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
13445 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
13446 (vect_dr_behavior): New function.
13447 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
13448 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
13449 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
13450 track whether the step preserves the misalignment.
13451 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
13452 Use vect_dr_behavior.
13453 (vect_setup_realignment): Update call accordingly.
13454 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
13455 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
13456 call to vect_create_addr_base_for_vector_ref.
13457 (vect_create_cond_for_align_checks): Likewise.
13458 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
13459 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
13460 (vect_recog_mask_conversion_pattern): Likewise.
13461 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
13462 (new_stmt_vec_info): Remove redundant zeroing.
13464 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
13466 * common/config/arm/arm-common.c (arm_be8_option): New function.
13467 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
13468 (ISA_ARMv6): Add isa_bit_be8.
13469 * config/arm/arm.h (arm_be8_option): Add prototype.
13470 (BE8_SPEC_FUNCTION): New define.
13471 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
13472 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
13473 (mlittle-endian): Similarly.
13474 (mbe8, mbe32): New options.
13475 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
13476 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
13478 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13480 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
13482 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13484 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
13485 (cleanup_tree_cfg_bb): Use it.
13486 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
13488 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
13490 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13493 * loop-doloop.c (add_test): Update profile.
13495 2017-07-03 Martin Liska <mliska@suse.cz>
13498 * sanopt.c (rewrite_usage_of_param): New function.
13499 (sanitize_rewrite_addressable_params): Likewise.
13500 (pass_sanopt::execute): Call rewrite_usage_of_param.
13502 2017-07-03 Richard Biener <rguenther@suse.de>
13504 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
13505 back to using VIEW_CONVERT_EXPR.
13507 2017-07-03 Martin Liska <mliska@suse.cz>
13510 * doc/extend.texi: Document when a resolver function is
13511 generated for target_clones.
13513 2017-07-03 Martin Liska <mliska@suse.cz>
13515 * asan.c (asan_emit_stack_protection): Unpoison just red zones
13516 and shadow memory of auto variables which are subject of
13517 use-after-scope sanitization.
13518 (asan_expand_mark_ifn): Add do set only when is_poison.
13520 2016-07-03 Richard Biener <rguenther@suse.de>
13522 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
13524 (vect_force_simple_reduction): Record reduction def -> phi mapping.
13525 (vectorizable_reduction): Perform reduction PHI creation when
13526 visiting a reduction PHI and adjust and simplify code generation
13527 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
13528 (vect_transform_loop): Visit reduction PHIs.
13529 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
13530 defs into the SLP tree.
13531 (vect_build_slp_tree): Reduction defs terminate the recursion.
13532 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
13534 (vect_get_vec_defs_for_stmt_copy): Export.
13535 (vect_get_vec_defs): Likewise.
13536 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
13538 (vect_get_vec_defs_for_stmt_copy): Declare.
13539 (vect_get_vec_defs): Likewise.
13541 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13543 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
13544 parameter with a "loop" parameter and use it instead of the
13545 loop containing DR_STMT. Don't check simple_iv when doing
13546 BB analysis. Describe the two analysis modes in the comment.
13548 2017-07-03 Tom de Vries <tom@codesourcery.com>
13550 PR tree-optimization/69468
13551 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
13552 (find_same_succ_bb): Handle ignore_edge_flags.
13554 2017-07-03 Tom de Vries <tom@codesourcery.com>
13556 PR tree-optimization/81192
13557 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
13559 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
13561 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
13563 2017-07-03 Tom de Vries <tom@codesourcery.com>
13565 PR tree-optimization/81192
13566 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
13567 BB_SAME_SUCC (bb) == NULL.
13569 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13571 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
13574 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13576 * dumpfile.c: Include profile-count.h
13577 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
13579 (insert_cond_bb): Update profile.
13580 * tree-cfg.h (insert_cond_bb): Update prototype.
13581 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
13582 * tree-dump.c: Do not include tree-cfg.
13584 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13586 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
13588 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13590 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
13593 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13595 * tree-complex.c (expand_complex_div_wide): update profile.
13597 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13598 Alan Hayward <alan.hayward@arm.com>
13599 David Sherwood <david.sherwood@arm.com>
13601 * Makefile.in (MACHMODE_H): Remove insn-modes.h
13602 (CORETYPES_H): New define.
13603 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
13604 (insn-modes-inline.h, s-modes-inline-h): New rules.
13605 (generated_files): Add insn-modes-inline.h.
13606 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
13607 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
13608 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
13609 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
13610 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
13611 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
13612 (build/gencodes.o, build/genconditions.o): Likewise.
13613 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
13614 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
13615 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
13616 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
13617 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
13618 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
13619 * coretypes.h: Include everything up to real.h for generators.
13620 Include insn-modes.h first. Include wide-int-print.h after
13621 wide-int.h. Include insn-modes-inline.h and then machmode.h.
13622 * machmode.h: Don't include insn-modes.h here.
13623 * function-tests.c: Remove includes of signop.h, machmode.h,
13624 double-int.h and wide-int.h.
13626 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
13628 * optc-save-gen.awk: Likewise.
13629 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
13630 * godump.c: Remove include of wide-int-print.h.
13631 * pretty-print.h: Likewise.
13632 * wide-int-print.cc: Likewise.
13633 * wide-int.cc: Likewise.
13634 * hash-map-tests.c: Remove include of signop.h.
13635 * hash-set-tests.c: Likewise.
13636 * rtl-tests.c: Likewise.
13637 * mkconfig.sh: Remove include of machmode.h.
13638 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
13640 (emit_insn_modes_inline_h): ...this new function. Emit the code
13641 into an insn-modes-inline.h header file, adding appropriate
13642 include guards and end comments.
13643 (emit_insn_modes_c_header): Remove include of machmode.h.
13644 (emit_min_insn_modes_c_header): Include coretypes.h rather than
13646 (main): Handle -i flag and call emit_insn_modes_inline_h when
13649 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13651 * tree-ssa-strlen.c (strinfo): Rename the length field to
13652 nonzero_chars. Add a full_string_p field.
13653 (compare_nonzero_chars, zero_length_string_p): New functions.
13654 (get_addr_stridx): Add an offset_out parameter.
13655 Use compare_nonzero_chars.
13656 (get_stridx): Update accordingly. Use compare_nonzero_chars.
13657 (new_strinfo): Update after above changes to strinfo.
13658 (set_endptr_and_length): Set full_string_p.
13659 (get_string_length): Update after above changes to strinfo.
13660 (unshare_strinfo): Update call to new_strinfo.
13661 (maybe_invalidate): Likewise.
13662 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
13663 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
13664 as a uhwi instead of an shwi. Update after above changes to
13665 strinfo and new_strinfo.
13666 (zero_length_string): Assert that chainsi contains full strings.
13667 Use zero_length_string_p. Update call to new_strinfo.
13668 (adjust_related_strinfos): Update after above changes to strinfo.
13669 Copy full_string_p from origsi.
13670 (adjust_last_stmt): Use zero_length_string_p.
13671 (handle_builtin_strlen): Update after above changes to strinfo and
13672 new_strinfo. Install the lhs as the string length if the previous
13673 entry didn't describe a full string.
13674 (handle_builtin_strchr): Update after above changes to strinfo
13676 (handle_builtin_strcpy): Likewise.
13677 (handle_builtin_strcat): Likewise.
13678 (handle_builtin_malloc): Likewise.
13679 (handle_pointer_plus): Likewise.
13680 (handle_builtin_memcpy): Likewise. Track nonzero characters
13681 that aren't necessarily followed by a nul terminator.
13682 (handle_char_store): Likewise.
13684 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13686 PR tree-optimization/80769
13687 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
13688 for malloc and calloc. Document the new invariant that all related
13689 strinfos have delayed lengths or none do.
13690 (verify_related_strinfos): Move earlier in file.
13691 (set_endptr_and_length): New function, split out from...
13692 (get_string_length): ...here. Also set the lengths of related
13694 (zero_length_string): Assert that chainsi has known (rather than
13696 (adjust_related_strinfos): Likewise.
13698 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13700 PR tree-optimization/81136
13701 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
13702 assert that two references with the same misalignment have the same
13703 compile-time misalignment if those compile-time misalignments
13706 2017-07-01 Andi Kleen <ak@linux.intel.com>
13708 * print-tree.c (print_node): Print all attributes.
13710 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13712 * cfg.c (scale_bbs_frequencies): New function.
13713 * cfg.h (scale_bbs_frequencies): Declare it.
13714 * cfgloopanal.c (single_likely_exit): Cleanup.
13715 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
13717 (scale_loop_profile): Likewise.
13718 (loop_version): Likewise.
13719 (create_empty_loop_on_edge): Update.
13720 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
13721 scale_loop_frequencies, scale_loop_profile, loopify,
13722 loop_version): Update prototypes.
13723 * modulo-sched.c (sms_schedule): Update.
13724 * predict.c (unlikely_executed_edge_p): Also check probability.
13725 (probably_never_executed_edge_p): Fix typo.
13726 * tree-if-conv.c (version_loop_for_if_conversion): Update.
13727 * tree-parloops.c (gen_parallel_loop): Update.
13728 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
13729 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13730 * tree-ssa-loop-split.c (split_loop): Update.
13731 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
13732 * tree-vect-loop-manip.c (vect_do_peeling): Update.
13733 (vect_loop_versioning): Update.
13734 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13736 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13738 * trans-mem.c (split_bb_make_tm_edge): Update profile.
13740 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13742 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
13743 to keep profile consistent.
13745 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13747 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
13748 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
13749 * profile-count.h (max_safe_multiplier): Make unsigned.
13750 (profile_count::guessed_zero): New.
13752 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13754 * bb-reorder.c (fix_up_crossing_landing_pad,
13755 fix_crossing_conditional_branches): Use make_single_succ_edge
13756 to keep profile consistent.
13758 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13760 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
13763 2017-07-01 Jakub Jelinek <jakub@redhat.com>
13766 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
13767 the right scopes, make sure cond_jump isn't preserved between multiple
13768 iterations. Search for fallthru edge whenever there are 3+ edges and
13769 use find_fallthru_edge for it.
13771 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13773 Patch by Alexander Monakov <amonakov@ispras.ru>
13774 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
13775 probabilities consistently.
13777 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13779 * pa.c (pa_expand_compare_and_swap_loop): Update call of
13780 emit_cmp_and_jump_insns.
13782 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13785 * tree-inline.c (expand_call_inline): Combine profile statuses.
13787 2017-06-30 Andrew Pinski <apinski@cavium.com>
13789 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
13790 fold_stmt returned true.
13792 2017-06-30 Nathan Sidwell <nathan@acm.org>
13794 * ggc.h (empty_string): Delete.
13795 * cfgexpand.c (expand_asm_stmt): Use plain "".
13796 * optabs.c (expand_asm_memory_barrier): Likewise.
13797 * stringpool.c (empty_string): Delete.
13798 (digit_vector, digit_string): Delete.
13799 (ggc_alloc_string): Use plain "", don't optimize single digit
13800 strings. Use ggc_alloc_atomic.
13802 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
13804 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
13805 comparison set and one other set, use the cost of the non-comparison
13808 2017-06-30 Nathan Sidwell <nathan@acm.org>
13810 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
13813 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
13815 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
13816 loops. Remove now unneeded calls to gimple_switch_set_label() that
13817 just set removed labels to NULL_TREE.
13819 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
13821 * tree-ssanames.c (set_range_info_raw): Abstract from ...
13822 (set_range_info): ...here. Only call set_range_info_raw if domain
13824 (set_nonzero_bits): Call set_range_info_raw.
13825 * tree-ssanames.h (set_range_info_raw): New.
13827 2017-06-30 Jakub Jelinek <jakub@redhat.com>
13830 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
13831 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
13832 of nonimmediate_operand and <store_mask_constraint> instead of m
13833 for the input operand. For V8FI iterator, always split if input
13834 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
13835 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
13836 <store_mask_predicate> instead of register_operand and
13837 <store_mask_constraint> instead of v for the input operand. Make
13838 sure both operands aren't MEMs for if not <mask_applied>.
13840 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
13842 * lto-wrapper.c (copy_file) Close both file descriptors before
13845 2017-06-30 Martin Liska <mliska@suse.cz>
13848 * multiple_target.c (create_dispatcher_calls): Make ifunc
13849 also for function that don't have calls or are not referenced.
13851 2017-06-30 Richard Biener <rguenther@suse.de>
13853 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
13854 analyze the first scalar stmt. Move vector type computation
13855 for the BB case here from ...
13856 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
13857 live operation processing in the SLP case properly.
13859 2017-06-30 Richard Biener <rguenther@suse.de>
13861 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
13863 2017-06-30 Martin Liska <mliska@suse.cz>
13866 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
13867 before BUILT_IN_UNWIND_RESUME when ASAN is used.
13869 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
13871 * doc/invoke.texi (AArch64): Add missing options and remove redundant
13874 2017-06-30 Richard Biener <rguenther@suse.de>
13876 PR tree-optimization/81249
13877 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
13878 condition reduction result to original scalar type.
13880 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13882 * profile-count.h (enum profile_quality): Fix typos and whitespace
13885 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13887 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
13888 type for branch probabilities.
13890 2017-06-29 Julian Brown <julian@codesourcery.com>
13891 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13893 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
13894 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
13895 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
13896 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
13898 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13900 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
13901 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
13902 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
13903 CC usage from generic code to here.
13904 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
13905 CC usage into the target macros.
13907 2017-06-29 Maya Rashish <coypu@sdf.org>
13909 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
13912 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13914 * arm/arm-builtins.c: Include profile-count.h
13915 * except.c (sjlj_emit_function_enter): Use
13916 profile_probability::unlikely.
13918 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13920 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
13921 and tocrel_offset be pointer args rather than implicitly using
13923 (legitimate_constant_pool_address_p, rs6000_emit_move,
13924 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
13925 tocrel_offset and use in toc_relative_expr_p call.
13926 (print_operand, print_operand_address): Use static tocrel_base_oac
13927 and tocrel_offset_oac.
13928 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
13931 2017-06-29 Maya Rashish <coypu@sdf.org>
13933 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
13935 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
13937 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
13938 objects, take into account only the alignment of 'op0' and 'mode1' if
13941 2017-06-29 Steve Ellcey <sellcey@cavium.com>
13943 * ccmp.c (ccmp_tree_comparison_p): New function.
13944 (ccmp_candidate_p): Update to use above function.
13945 (get_compare_parts): New function.
13946 (expand_ccmp_next): Update to use new functions.
13947 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
13949 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
13950 take mode as argument.
13951 * ccmp.h (expand_ccmp_expr): Add mode as argument.
13952 * expr.c (expand_expr_real_1): Pass mode as argument.
13954 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
13956 * combine.c (combine_instructions): Print insns to dump_file, together
13959 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13961 * asan.c (asan_emit_stack_protection): Update.
13962 (create_cond_insert_point): Update.
13963 * auto-profile.c (afdo_propagate_circuit): Update.
13964 * basic-block.h (struct edge_def): Turn probability to
13965 profile_probability.
13966 (EDGE_FREQUENCY): Update.
13967 * bb-reorder.c (find_traces_1_round): Update.
13968 (better_edge_p): Update.
13969 (sanitize_hot_paths): Update.
13970 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
13971 (make_single_succ_edge): Update.
13972 (check_bb_profile): Update.
13973 (dump_edge_info): Update.
13974 (update_bb_profile_for_threading): Update.
13975 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
13976 probabilitycount to 0.
13977 * cfgbuild.c (compute_outgoing_frequencies): Update.
13978 * cfgcleanup.c (try_forward_edges): Update.
13979 (outgoing_edges_match): Update.
13980 (try_crossjump_to_edge): Update.
13981 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
13982 (expand_gimple_tailcall): Update.
13983 (construct_init_block): Use make_single_succ_edge.
13984 (construct_exit_block): Use make_single_succ_edge.
13985 * cfghooks.c (verify_flow_info): Update.
13986 (redirect_edge_succ_nodup): Update.
13987 (split_edge): Update.
13988 (account_profile_record): Update.
13989 * cfgloopanal.c (single_likely_exit): Update.
13990 * cfgloopmanip.c (scale_loop_profile): Update.
13991 (set_zero_probability): Remove.
13992 (duplicate_loop_to_header_edge): Update.
13993 * cfgloopmanip.h (loop_version): Update prototype.
13994 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
13995 (force_nonfallthru_and_redirect): Update.
13996 (update_br_prob_note): Update.
13997 (rtl_verify_edges): Update.
13998 (purge_dead_edges): Update.
13999 (rtl_lv_add_condition_to_bb): Update.
14000 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
14001 * cgraphunit.c (init_lowered_empty_function): Update.
14002 (cgraph_node::expand_thunk): Update.
14003 * cilk-common.c: Include profile-count.h
14004 * dojump.c (inv): Remove.
14005 (jumpifnot): Update.
14006 (jumpifnot_1): Update.
14007 (do_jump_1): Update.
14009 (do_jump_by_parts_greater_rtx): Update.
14010 (do_compare_rtx_and_jump): Update.
14011 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
14012 do_jump_1. do_compare_rtx_and_jump): Update prototype.
14013 * dwarf2cfi.c: Include profile-count.h
14014 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
14015 (sjlj_emit_dispatch_table): Likewise.
14016 * explow.c: Include profile-count.h
14017 * expmed.c (emit_store_flag_force): Update.
14018 (do_cmp_and_jump): Update.
14019 * expr.c (compare_by_pieces_d::generate): Update.
14020 (compare_by_pieces_d::finish_mode): Update.
14021 (emit_block_move_via_loop): Update.
14022 (store_expr_with_bounds): Update.
14023 (store_constructor): Update.
14024 (expand_expr_real_2): Update.
14025 (expand_expr_real_1): Update.
14026 * expr.h (try_casesi, try_tablejump): Update prototypes.
14027 * gimple-pretty-print.c (dump_probability): Update.
14028 (dump_profile): New.
14029 (dump_gimple_label): Update.
14030 (dump_gimple_bb_header): Update.
14031 * graph.c (draw_cfg_node_succ_edges): Update.
14032 * hsa-gen.c (convert_switch_statements): Update.
14033 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
14034 (find_if_case_1): Update.
14035 (find_if_case_2): Update.
14036 * internal-fn.c (expand_arith_overflow_result_store): Update.
14037 (expand_addsub_overflow): Update.
14038 (expand_neg_overflow): Update.
14039 (expand_mul_overflow): Update.
14040 (expand_vector_ubsan_overflow): Update.
14041 * ipa-cp.c (good_cloning_opportunity_p): Update.
14042 * ipa-split.c (split_function): Use make_single_succ_edge.
14043 * ipa-utils.c (ipa_merge_profiles): Update.
14044 * loop-doloop.c (add_test): Update.
14045 (doloop_modify): Update.
14046 * loop-unroll.c (compare_and_jump_seq): Update.
14047 (unroll_loop_runtime_iterations): Update.
14048 * lra-constraints.c (lra_inheritance): Update.
14049 * lto-streamer-in.c (input_cfg): Update.
14050 * lto-streamer-out.c (output_cfg): Update.
14051 * mcf.c (adjust_cfg_counts): Update.
14052 * modulo-sched.c (sms_schedule): Update.
14053 * omp-expand.c (expand_omp_for_init_counts): Update.
14054 (extract_omp_for_update_vars): Update.
14055 (expand_omp_ordered_sink): Update.
14056 (expand_omp_for_ordered_loops): Update.
14057 (expand_omp_for_generic): Update.
14058 (expand_omp_for_static_nochunk): Update.
14059 (expand_omp_for_static_chunk): Update.
14060 (expand_cilk_for): Update.
14061 (expand_omp_simd): Update.
14062 (expand_omp_taskloop_for_outer): Update.
14063 (expand_omp_taskloop_for_inner): Update.
14064 * omp-simd-clone.c (simd_clone_adjust): Update.
14065 * optabs.c (expand_doubleword_shift): Update.
14066 (expand_abs): Update.
14067 (emit_cmp_and_jump_insn_1): Update.
14068 (expand_compare_and_swap_loop): Update.
14069 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
14070 * predict.c (predictable_edge_p): Update.
14071 (edge_probability_reliable_p): Update.
14072 (set_even_probabilities): Update.
14073 (combine_predictions_for_insn): Update.
14074 (combine_predictions_for_bb): Update.
14075 (propagate_freq): Update.
14076 (estimate_bb_frequencies): Update.
14077 (force_edge_cold): Update.
14078 * profile-count.c (profile_count::dump): Add missing space into dump.
14079 (profile_count::debug): Add newline.
14080 (profile_count::differs_from_p): Explicitly convert to unsigned.
14081 (profile_count::stream_in): Update.
14082 (profile_probability::dump): New member function.
14083 (profile_probability::debug): New member function.
14084 (profile_probability::differs_from_p): New member function.
14085 (profile_probability::differs_lot_from_p): New member function.
14086 (profile_probability::stream_in): New member function.
14087 (profile_probability::stream_out): New member function.
14088 * profile-count.h (profile_count_quality): Rename to ...
14089 (profile_quality): ... this one.
14090 (profile_probability): New.
14091 (profile_count): Update.
14092 * profile.c (compute_branch_probabilities): Update.
14093 * recog.c (peep2_attempt): Update.
14094 * sched-ebb.c (schedule_ebbs): Update.
14095 * sched-rgn.c (find_single_block_region): Update.
14096 (compute_dom_prob_ps): Update.
14097 (schedule_region): Update.
14098 * sel-sched-ir.c (compute_succs_info): Update.
14099 * stmt.c (struct case_node): Update.
14100 (do_jump_if_equal): Update.
14101 (get_outgoing_edge_probs): Update.
14102 (conditional_probability): Update.
14103 (emit_case_dispatch_table): Update.
14104 (expand_case): Update.
14105 (expand_sjlj_dispatch_table): Update.
14106 (emit_case_nodes): Update.
14107 * targhooks.c: Update.
14108 * tracer.c (better_p): Update.
14109 (find_best_successor): Update.
14110 * trans-mem.c (expand_transaction): Update.
14111 * tree-call-cdce.c: Update.
14112 * tree-cfg.c (gimple_split_edge): Upate.
14113 (move_sese_region_to_fn): Upate.
14114 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
14115 * tree-eh.c (lower_resx): Upate.
14116 (cleanup_empty_eh_move_lp): Upate.
14117 * tree-if-conv.c (version_loop_for_if_conversion): Update.
14118 * tree-inline.c (copy_edges_for_bb): Update.
14119 (copy_cfg_body): Update.
14120 * tree-parloops.c (gen_parallel_loop): Update.
14121 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
14122 (gimple_gen_time_profiler): Update.
14123 * tree-ssa-dce.c (remove_dead_stmt): Update.
14124 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
14125 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
14126 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
14127 (unloop_loops): Update.
14128 (try_peel_loop): Update.
14129 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14130 * tree-ssa-loop-split.c (connect_loops): Update.
14131 (split_loop): Update.
14132 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
14133 (hoist_guard): Update.
14134 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
14135 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
14136 (value_replacement): Update.
14137 * tree-ssa-reassoc.c (branch_fixup): Update.
14138 * tree-ssa-tail-merge.c (replace_block_by): Update.
14139 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
14140 (create_edge_and_update_destination_phis): Update.
14141 (compute_path_counts): Update.
14142 (recompute_probabilities): Update.
14143 (update_joiner_offpath_counts): Update.
14144 (freqs_to_counts_path): Update.
14145 (duplicate_thread_path): Update.
14146 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
14147 (struct switch_conv_info): Update.
14148 (gen_inbound_check): Update.
14149 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
14150 (vect_do_peeling): Update.
14151 (vect_loop_versioning): Update.
14152 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14153 (optimize_mask_stores): Update.
14154 * ubsan.c (ubsan_expand_null_ifn): Update.
14155 * value-prof.c (gimple_divmod_fixed_value): Update.
14156 (gimple_divmod_fixed_value_transform): Update.
14157 (gimple_mod_pow2): Update.
14158 (gimple_mod_pow2_value_transform): Update.
14159 (gimple_mod_subtract): Update.
14160 (gimple_mod_subtract_transform): Update.
14161 (gimple_ic): Update.
14162 (gimple_stringop_fixed_value): Update.
14163 (gimple_stringops_transform): Update.
14164 * value-prof.h: Update.
14166 2017-06-29 Carl Love <cel@us.ibm.com>
14168 * config/rs6000/rs6000-c.c: Add support for built-in functions
14169 vector signed int vec_signed (vector float);
14170 vector signed long long vec_signed (vector double);
14171 vector signed int vec_signed2 (vector double, vector double);
14172 vector signed int vec_signede (vector double);
14173 vector signed int vec_signedo (vector double);
14174 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
14175 instruction generator.
14176 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
14177 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
14178 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
14180 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
14181 vunsignede_v2df): Add define_expands.
14182 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
14183 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
14184 VEC_UNSIGNEDO): Add definitions.
14185 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
14186 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
14187 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
14188 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
14189 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
14190 * config/rs6000/altivec.h (vec_signed, vec_signed2,
14191 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
14192 vec_unsignede, vec_unsignedo): Add builtin defines.
14193 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
14195 * doc/extend.texi: Update the built-in documentation file for the
14196 new built-in functions.
14198 2017-06-29 Richard Biener <rguenther@suse.de>
14200 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
14201 reduction chains to LOOP_VINFO_REDUCTIONS.
14202 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
14203 SLP reductions after processing reduction chains.
14205 2017-06-29 Nathan Sidwell <nathan@acm.org>
14207 * builtins.c (fold_builtin_FUNCTION): Use
14208 lang_hooks.decl_printable_name.
14210 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
14212 PR middle-end/81194
14213 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
14214 with only one label.
14215 * stmt.c (expand_case): Assert NCASES is greater than one.
14217 2017-06-29 Richard Biener <rguenther@suse.de>
14219 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
14221 (group_case_labels): Likewise.
14222 (find_taken_edge): Push sanity checking on val to workers...
14223 (find_taken_edge_cond_expr): ... here
14224 (find_taken_edge_switch_expr): ... and here, handle cases
14225 with just a default label.
14226 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
14227 (group_case_labels): Likewise.
14228 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
14229 group_case_labels does anything cleanup the CFG again.
14231 2017-06-29 Bin Cheng <bin.cheng@arm.com>
14233 PR tree-optimization/81196
14234 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
14235 exit condition comparing two IVs.
14237 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
14239 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
14240 profile to the dummy entry at the end of the list of architectures.
14241 * config/arm/arm-cpu-cdata.h: Regenerated.
14243 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14244 Michael Collison <michael.collison@arm.com>
14247 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
14249 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
14250 (*aarch64_reg_<mode>3_minus_mask): New pattern.
14251 (*aarch64_<optab>_reg_di3_mask2): New pattern.
14252 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
14253 of shift when the shift amount is masked with constant equal to
14254 the size of the mode.
14255 * config/aarch64/predicates.md (subreg_lowpart_operator): New
14258 2017-06-29 Martin Liska <mliska@suse.cz>
14260 * config/i386/i386.opt: Change range from [1,5] to [0,5].
14262 2017-06-29 Yury Gribov <tetra2005@gmail.com>
14265 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
14268 (edge_growth_cache_entry::edge_growth_cache_entry): New
14270 (reset_edge_growth_cache): Update to use constructor.
14272 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14274 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
14275 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
14276 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
14278 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
14280 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
14281 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
14283 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
14285 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
14286 (*-linux-uclibc*): Add t-uclibc tmake_file.
14287 * config/t-musl: New.
14288 * config/t-uclibc: New.
14290 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
14292 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
14294 (gen_comm_data): Emit architectural setting of arch_prof.
14295 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
14297 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
14298 (armv8-m.base, armv8-m.main): Likewise.
14299 * arm-protos.h (arm_build_target): Add profile field.
14300 (arch_option): Likewise.
14301 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
14303 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
14304 arm_active_target.profile.
14306 2017-06-28 Richard Biener <rguenther@suse.de>
14308 PR middle-end/81227
14309 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
14310 TYPE_OVERFLOW_WRAPS.
14311 * match.pd (negate_expr_p): Likewise.
14312 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
14313 fold_build2, not fold_binary.
14315 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
14317 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
14318 Convert memory address to Pmode.
14319 (aarch64_print_operand): Assert MEM operands are always Pmode.
14321 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
14324 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
14325 Remove redundant if.
14326 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
14327 * config/arm/aarch-common-protos.h
14328 (aarch_forward_to_shift_is_not_shifted_re): Remove.
14329 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
14331 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
14334 * multiple_target.c (create_dispatcher_calls): Set the default
14335 clone to be static, not public.
14337 2017-06-28 Richard Biener <rguenther@suse.de>
14339 * tree-vect-loop.c (vectorizable_reduction): Move special
14340 cond reduction IV var creation ...
14341 (vect_create_epilog_for_reduction): ... here. Remove induction_index
14342 parameter. Use STMT_VINFO_VECTYPE.
14343 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
14346 2017-06-28 Martin Liska <mliska@suse.cz>
14349 * ipa-visibility.c (non_local_p): Handle visibility.
14351 2017-06-28 Martin Liska <mliska@suse.cz>
14354 * common.opt: Add IntegerRange to various options.
14355 * opt-functions.awk (integer_range_info): New function.
14356 * optc-gen.awk: Add integer_range_info to cl_options struct.
14357 * opts-common.c (decode_cmdline_option): Handle
14358 CL_ERR_INT_RANGE_ARG.
14359 (cmdline_handle_error): Likewise.
14360 * opts.c (print_filtered_help): Show valid interval in
14361 when --help is provided.
14362 * opts.h (struct cl_option): Add range_min and range_max fields.
14363 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
14365 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
14367 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
14368 (x * C EQ/NE y * C): New transformation.
14370 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
14372 * genmultilib (combination_space): Accept '+' in option names.
14374 2017-06-28 Martin Liska <mliska@suse.cz>
14377 * asan.c (instrument_derefs): Bail out inner references
14378 that are hard register variables.
14380 2017-06-28 Jakub Jelinek <jakub@redhat.com>
14383 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
14384 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
14386 2017-06-28 Richard Biener <rguenther@suse.de>
14388 * tree-vectorizer.h (vect_get_vec_defs): Remove.
14389 (vect_get_slp_defs): Adjust.
14390 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
14392 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
14394 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
14395 get_initial_defs_for_reduction instead of vect_get_vec_defs.
14396 (vectorizable_reduction): Adjust.
14397 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
14399 (vect_get_slp_defs): Likewise.
14400 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
14401 (vectorizable_bswap): Adjust.
14402 (vectorizable_call): Likewise.
14403 (vectorizable_conversion): Likewise.
14404 (vectorizable_assignment): Likewise.
14405 (vectorizable_shift): Likewise.
14406 (vectorizable_operation): Likewise.
14407 (vectorizable_store): Likewise.
14408 (vectorizable_condition): Likewise.
14409 (vectorizable_comparison): Likewise.
14411 2017-06-28 Michael Collison <michael.collison@arm.com>
14414 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
14416 (arm_gen_movmemqi): Removed unused variable 'i'.
14417 Convert 'for' loop into 'while' loop.
14418 (arm_expand_prologue): Remove last unnecessary set of insn.
14419 (thumb_pop): Remove unused variable 'pushed_words'.
14420 (thumb_exit): Remove last unnecessary set of regs_to_pop.
14422 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14424 * config/s390/predicates.md: Use s390_rel_address_ok_p.
14425 * config/s390/s390-protos.h: Add prototype of
14426 s390_rel_address_ok_p.
14427 * config/s390/s390.c (s390_got_symbol): New function.
14428 (s390_rel_address_ok_p): New function.
14429 (legitimize_pic_address): Use s390_rel_address_ok_p.
14430 (s390_load_got): Use s390_got_symbol.
14431 (s390_option_override): Issue error if
14432 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
14433 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
14435 * config/s390/s390.opt: New option mpic-data-is-text-relative.
14437 2017-06-27 Andrew Pinski <apinski@cavium.com>
14439 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
14440 (X * copysign (1.0, X)): New pattern.
14441 (X * copysign (1.0, -X)): New pattern.
14442 (copysign (-1.0, CST)): New pattern.
14444 2017-06-27 Joseph Myers <joseph@codesourcery.com>
14446 * genmultilib (combination_space): Remove variable.
14447 Validate reuse rules against regular expression for any sequence
14448 of multilib options in any order.
14450 2017-06-27 Michael Collison <michael.collison@arm.com>
14452 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
14453 call aarch64_split_simd_combine.
14454 * (aarch64_combine_internal<mode>): Delete pattern.
14455 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
14456 Allow register and subreg operands.
14458 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14460 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
14461 specific need, just fallback on defaults.
14462 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
14464 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14465 Olivier Hainque <hainque@adacore.com>
14467 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
14469 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
14470 targets. Pick a default if no particular attempt applied.
14471 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
14474 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14476 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
14477 (x86_64-wrs-vxworks7): Likewise.
14479 2017-06-27 Marek Polacek <polacek@redhat.com>
14482 * ubsan.c (instrument_null): Check get_base_address's result for null.
14484 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
14486 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
14488 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
14490 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
14491 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
14492 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
14493 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
14494 New function types.
14495 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
14496 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
14497 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
14498 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
14499 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
14500 BUILT_IN_FEUPDATEENV): New builtins.
14501 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
14502 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
14503 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
14504 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
14506 (builtin_structptr_types): Adjust size.
14507 * tree.c (builtin_structptr_types): Add four entries.
14509 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14510 Olivier Hainque <hainque@adacore.com>
14512 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
14513 (TLS_SYM): New local macro, forcing reference to __tls__ on
14514 link command lines for VxWorks 7 RTPs, triggering initialization
14516 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
14517 OS features TLS support, true for RTPs on VxWorks 7.
14518 * config/vxworks.c (vxworks_override_options): Setup emutls
14521 2017-06-27 Jakub Jelinek <jakub@redhat.com>
14523 * predict.c (test_prediction_value_range): Use -1U instead of -1
14524 to avoid narrowing conversion warning.
14525 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
14526 to avoid narrowing conversion warning.
14527 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
14529 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
14531 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14533 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
14534 64bit configurations.
14535 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
14536 (SIZE_TYPE): Likewise.
14537 * config/vxworks.c (vxworks_emutls_var_fields): Use
14538 long_unsigned_type_node instead of unsigned_type_node as the offset
14539 field type, which is "pointer" mode in emutls.c.
14541 2017-06-27 Jakub Jelinek <jakub@redhat.com>
14544 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
14546 PR middle-end/81207
14547 * gimple-fold.c (replace_call_with_call_and_fold): Handle
14548 gimple_vuse copying separately from gimple_vdef copying.
14550 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14552 * value-prof.c (free_hist): Remove call to memset and the enclosing if
14555 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
14556 Olivier Hainque <hainque@adacore.com>
14558 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
14559 for all vxworks7 targets.
14560 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
14561 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
14562 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
14563 variations for VX6/VX7 and 32/64bits later on in ...
14564 (VXWORKS_LIB_SPEC): Leverage new macros.
14565 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
14566 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
14568 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
14570 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
14571 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
14573 2017-06-26 Carl Love <cel@us.ibm.com>
14575 * config/rs6000/rs6000-c.c: Add support for built-in functions
14576 vector bool char vec_reve (vector bool char);
14577 vector signed char vec_reve (vector signed char);
14578 vector unsigned char vec_reve (vector unsigned char);
14579 vector bool int vec_reve (vector bool int);
14580 vector signed int vec_reve (vector signed int);
14581 vector unsigned int vec_reve (vector unsigned int);
14582 vector bool long long vec_reve (vector bool long long);
14583 vector signed long long vec_reve (vector signed long long);
14584 vector unsigned long long vec_reve (vector unsigned long long);
14585 vector bool short vec_reve (vector bool short);
14586 vector signed short vec_reve (vector signed short);
14587 vector double vec_reve (vector double);
14588 vector float vec_reve (vector float);
14589 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
14590 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
14591 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
14592 (altivec_vreve): New pattern.
14593 * config/rs6000/altivec.h (vec_reve): New define.
14594 * doc/extend.texi (vec_rev): Update the built-in documentation file
14595 for the new built-in functions.
14597 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14599 PR tree-optimization/71815
14600 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
14602 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
14604 (slsr_process_phi): Likewise.
14605 (replace_uncond_cands_and_profitable_phis): Don't replace a
14606 multiply candidate with a stride of 1 (copy or cast).
14607 (phi_incr_cost): Call uses_consumed_by_stmt rather than
14609 (lowest_cost_path): Likewise.
14610 (total_savings): Likewise.
14612 2017-06-26 Richard Biener <rguenther@suse.de>
14615 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
14616 Use def_builtin_pure for all gather builtins.
14618 2017-06-26 Richard Biener <rguenther@suse.de>
14620 PR tree-optimization/81203
14621 * tree-tailcall.c (find_tail_calls): Do not move stmts into
14622 non-dominating BBs.
14624 2017-06-26 Marek Polacek <polacek@redhat.com>
14627 * doc/invoke.texi: Document -Wmultistatement-macros.
14629 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
14631 * doc/sourcebuild.texi (ARM-specific attributes): Document new
14632 arm_neon_ok_no_float_abi effective target.
14634 2017-06-26 Richard Biener <rguenther@suse.de>
14636 PR tree-optimization/80928
14637 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
14638 (copy_bbs): Set BB_DUPLICATED flag early.
14639 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
14641 (execute_on_shrinking_pred): Likewise.
14642 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
14643 BB_DUPLICATED blocks.
14644 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
14645 iterate over all PHIs considering removal of *gsi.
14647 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
14649 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
14652 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14654 * config/rs6000/rs6000-string.c: (expand_block_clear,
14655 do_load_for_compare, select_block_compare_mode,
14656 compute_current_alignment, expand_block_compare,
14657 expand_strncmp_align_check, expand_strn_compare,
14658 expand_block_move, rs6000_output_load_multiple)
14659 Move functions related to string/block move/compare
14660 to a separate file.
14661 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
14662 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
14663 for this function which is now used in two files.
14664 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
14665 * config.gcc: Add rs6000-string.o to extra_objs for
14666 targets powerpc*-*-* and rs6000*-*-*.
14668 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
14671 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
14672 32-bit, since indexed is not valid for DImode.
14673 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
14674 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
14675 (define_peephole2 for Altivec d-form load): Add 32-bit support.
14676 (define_peephole2 for Altivec d-form store): Likewise.
14679 * multiple_target.c (create_dispatcher_calls): Only create the
14680 dispatcher call if the function is the default clone of a
14681 versioned function.
14683 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
14685 PR middle-end/80902
14686 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
14687 a call, force the call to not be a tail call.
14689 2017-06-23 Jeff Law <law@redhat.com>
14691 * doc/contrib.texi: Add entry for Steven Pemberton's work on
14694 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
14696 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
14697 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
14698 handling for early expansion of vector shifts (sl,sr,sra,rl).
14699 (builtin_function_type): Add vector shift right instructions
14700 to the unsigned argument list.
14702 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
14704 rtl-optimizatoin/79286
14705 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
14706 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
14707 trap. PIC register plus a const unspec without offset can never trap.
14709 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
14711 * tree.h (builtin_structptr_type): New type.
14712 (builtin_structptr_types): Declare new array.
14713 * tree.c (builtin_structptr_types): New array.
14714 (free_lang_data, build_common_tree_nodes): Use it.
14716 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
14719 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
14722 2017-06-22 Matt Turner <mattst88@gmail.com>
14724 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
14725 Lake models to skylake case. Assume skylake for unknown
14726 models with clflushopt.
14728 2017-06-22 Jeff Law <law@redhat.com>
14730 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
14731 frame sizes that do not satisfy aarch64_uimm12_shift.
14733 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
14735 * profile-count.h (apply_probability,
14736 apply_scale, probability_in): Fix checks for zero.
14738 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14740 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
14741 * doc/cppdiropts.texi (-I @var{dir}): Document it.
14743 2016-06-22 Richard Biener <rguenther@suse.de>
14745 * tree-vect-loop.c (vect_model_reduction_cost): Handle
14746 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
14747 REDUC_MAX_EXPR support.
14748 (vectorizable_reduction): Likewise.
14749 (vect_create_epilog_for_reduction): Likewise.
14751 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
14753 * match.pd (A / (1 << B) -> A >> B): New.
14754 * generic-match-head.c: Include optabs-tree.h.
14755 * gimple-match-head.c: Likewise.
14756 * optabs-tree.h (target_supports_op_p): New.
14757 * optabs-tree.c (target_supports_op_p): New.
14759 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14761 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
14762 $gcc_cv_ld --help output.
14763 (gcc_cv_ld_demangle): Likewise.
14764 (gcc_cv_ld_eh_frame_hdr): Likewise.
14765 (gcc_cv_ld_pie): Likewise.
14766 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
14767 (gcc_cv_ld_buildid): Likewise.
14768 (gcc_cv_ld_sysroot): Likewise.
14769 (ld_bndplt_support): Likewise.
14770 (ld_pushpopstate_support): Likewise.
14771 * configure: Regenerate.
14772 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
14774 2017-06-21 Jakub Jelinek <jakub@redhat.com>
14777 * config/i386/sse.md (round<mode>2): Renumber match_dup and
14778 operands indexes to avoid gap between operands and match_dups.
14780 2017-06-21 Andrew Pinski <apinski@cavium.com>
14782 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
14783 Increment Arith_shift and Arith_shift_reg by 1.
14784 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
14786 * config/aarch64/aarch64.c (thunderx_tunings): Enable
14787 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
14788 (aarch64_strip_extend): Add new argument and test for it.
14789 (aarch64_cheap_mult_shift_p): New function.
14790 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
14791 add a cost if it is true.
14792 Update calls to aarch64_strip_extend.
14793 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
14795 2017-06-21 Andrew Pinski <apinski@cavium.com>
14797 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
14799 (thunderxt88): Likewise.
14800 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
14801 (thunderx_prefetch_tune): New variable.
14802 (thunderx2t99_prefetch_tune): Update for the correct values.
14803 (thunderxt88_tunings): New variable.
14804 (thunderx_tunings): Use thunderx_prefetch_tune instead of
14805 generic_prefetch_tune.
14806 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
14808 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14810 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
14811 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
14812 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
14813 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
14814 (aarch64_atomic_cas<mode>, GPI): Likewise.
14816 2017-06-21 Martin Liska <mliska@suse.cz>
14818 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
14819 statements on cold and hot labels.
14820 * predict.c (tree_estimate_probability_bb): Remove the
14821 prediction from this place.
14823 2017-06-21 Martin Liska <mliska@suse.cz>
14825 PR tree-optimization/79489
14826 * gimplify.c (maybe_add_early_return_predict_stmt): New
14828 (gimplify_return_expr): Call the function.
14829 * predict.c (tree_estimate_probability_bb): Remove handling
14831 * predict.def: Update comment about early return predictor.
14832 * gimple-predict.h (is_gimple_predict): New function.
14833 * predict.def: Change default value of early return to 66.
14834 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
14836 * passes.def: Put pass_strip_predict_hints to the beginning of
14839 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
14841 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
14842 FUNCTION_DECL declarations.
14843 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
14845 (dwaf2out_decl): Likewise.
14846 * godump.c (go_early_global_decl): Skip call to the real debug hook
14847 for FUNCTION_DECL declarations.
14848 * passes.c (rest_of_decl_compilation): Skip call to the
14849 early_global_decl debug hook for FUNCTION_DECL declarations, unless
14850 -fdump-go-spec is passed.
14852 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
14854 * config/i386/i386.c (struct builtin_isa): New field pure_p.
14855 Reorder for compactness.
14856 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
14857 (def_builtin_pure, def_builtin_pure2): New functions.
14858 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
14860 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
14862 * match.pd (nop_convert): New predicate.
14863 ((A +- CST1) +- CST2): Allow some NOP conversions.
14865 2017-06-21 Jakub Jelinek <jakub@redhat.com>
14868 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
14869 with ctors/dtors if GOVD_SHARED is set.
14871 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14873 * config/aarch64/aarch64.md (movti_aarch64):
14874 Emit mov rather than orr.
14875 (movtf_aarch64): Likewise.
14876 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
14877 Emit mov rather than orr.
14879 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14881 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
14882 Swap alternatives, make integer dup more expensive.
14884 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14886 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
14887 Return true for non-tls symbols.
14889 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
14891 * config/aarch64/aarch64-cores.def (cortex-a55): New.
14892 (cortex-a75): Likewise.
14893 (cortex-a75.cortex-a55): Likewise.
14894 * config/aarch64/aarch64-tune.md: Regenerate.
14895 * doc/invoke.texi (-mtune): Document new values for -mtune.
14897 2017-06-21 Tom de Vries <tom@codesourcery.com>
14899 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
14900 stack_size feature.
14901 (Effective-Target Keywords, Other attributes): Suggest using
14902 dg-add-options stack_size feature to get stack limit in stack_size
14903 effective target documentation.
14905 2017-06-21 Julian Brown <julian@codesourcery.com>
14906 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14908 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
14909 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
14910 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
14912 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
14913 attribute type list for neon_multiply.
14914 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
14915 attribute type list for neon_multiply.
14916 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
14917 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
14918 attribute type list for neon_multiply.
14919 * config/arm/types.md (crypto_pmull): Add.
14920 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
14921 attribute type list.
14923 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
14925 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
14928 2017-06-20 Jakub Jelinek <jakub@redhat.com>
14930 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
14931 to make sure not to dereference a NULL cost_classes_ptr pointer.
14933 2017-06-20 Carl Love <cel@us.ibm.com>
14935 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14936 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14937 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14938 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14939 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14940 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
14941 VMULOSW): New enum "unspec" values.
14942 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
14943 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
14944 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
14945 altivec_vmulosw): New patterns.
14946 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14947 VMULOSW): Add definitions.
14949 2017-06-20 Julia Koval <julia.koval@intel.com>
14951 * config/i386/i386.c: Fix rounding expand for new pattern.
14952 * config/i386/subst.md: Fix pattern (parallel -> unspec).
14954 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14956 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
14957 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
14959 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14961 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
14964 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14966 * config/aarch64/aarch64-cores.def: Rearrange to sort by
14967 architecture, then by implementer ID.
14968 * config/aarch64/aarch64-tune.md: Regenerate.
14970 2017-06-20 Richard Biener <rguenther@suse.de>
14972 PR middle-end/81097
14973 * fold-const.c (split_tree): Fold to type before negating.
14975 2017-06-20 David Malcolm <dmalcolm@redhat.com>
14977 * diagnostic-show-locus.c
14978 (selftest::test_fixit_deletion_affecting_newline): New function.
14979 (selftest::diagnostic_show_locus_c_tests): Call it.
14981 2017-06-20 Andreas Schwab <schwab@suse.de>
14984 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
14987 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14989 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
14990 __ARM_FEATURE_COPROC according to support.
14992 2017-06-20 Jakub Jelinek <jakub@redhat.com>
14994 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
14995 Rewritten to avoid overflow for > 32-bit pointers.
14998 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
14999 by removing enum keyword.
15000 (ubsan_type_descriptor): Likewise. Formatting fix.
15003 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
15004 splitter): Require TARGET_SSE2 in the condition.
15006 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
15009 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
15010 for doing vector set of SFmode on ISA 3.0.
15011 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
15012 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
15014 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
15015 SFmode value into a V4SF variable that was extracted from another
15016 V4SF variable without converting the element to double precision
15017 and back to single precision vector format.
15018 (vsx_insert_extract_v4sf_p9_2): Likewise.
15020 2017-06-19 Jakub Jelinek <jakub@redhat.com>
15022 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
15023 in UWHI to avoid undefined overflow.
15026 * ubsan.h (enum ubsan_encode_value_phase): New.
15027 (ubsan_encode_value): Change second argument to
15028 enum ubsan_encode_value_phase with default value of
15029 UBSAN_ENCODE_VALUE_GENERIC.
15030 * ubsan.c (ubsan_encode_value): Change second argument to
15031 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
15032 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
15033 create_tmp_var_raw instead of create_tmp_var and use a
15035 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
15036 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
15037 ubsan_encode_value callers.
15040 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
15041 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
15042 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
15044 2017-06-19 Richard Biener <rguenther@suse.de>
15046 PR middle-end/81118
15047 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
15048 estimates if we changed anything.
15050 2017-06-19 Richard Biener <rguenther@suse.de>
15052 PR tree-optimization/80887
15053 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
15054 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
15055 simplified lookups, then reset mprts_hook.
15056 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
15058 (try_to_simplify): Likewise.
15060 2017-06-19 Martin Liska <mliska@suse.cz>
15063 * gimplify.c (gimplify_switch_expr):
15064 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
15066 2017-06-19 Martin Liska <mliska@suse.cz>
15068 * doc/install.texi: Document that PGO runs in 4 stages.
15070 2017-06-19 Martin Liska <mliska@suse.cz>
15073 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
15074 to dispatcher function name.
15075 * multiple_target.c (replace_function_decl): New function.
15076 (create_dispatcher_calls): Redirect both edges and references.
15078 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
15080 * profile-count.c (profile_count::dump): Dump quality.
15081 (profile_count::differs_from_p): Update for unsigned val.
15082 * profile-count.h (profile_count_quality): New enum.
15083 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
15085 2017-06-19 Richard Biener <rguenther@suse.de>
15087 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
15088 struct function as arg.
15089 (estimate_numbers_of_iterations): Export overload with loop arg.
15090 (free_numbers_of_iterations_estimates_loop): Use an overload of
15091 free_numbers_of_iterations_estimates instead.
15092 * tree-cfg.c (remove_bb): Adjust.
15093 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
15094 * tree-parloops.c (gen_parallel_loop): Likewise.
15095 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
15097 (tree_unroll_loops_completely): Likewise.
15098 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
15099 Use an overload instead and export.
15100 (estimated_loop_iterations): Adjust.
15101 (max_loop_iterations): Likewise.
15102 (likely_max_loop_iterations): Likewise.
15103 (estimate_numbers_of_iterations): Take struct function as arg
15105 (loop_exits_before_overflow): Adjust.
15106 (free_numbers_of_iterations_estimates_loop): Use an overload.
15107 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
15108 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
15110 2017-06-19 Richard Biener <rguenther@suse.de>
15113 * ipa-prop.c (find_constructor_constant_at_offset): Handle
15114 RANGE_EXPR conservatively.
15116 2017-06-16 Carl Love <cel@us.ibm.com>
15118 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15119 definitions for vec_float, vec_float2, vec_floato,
15120 vec_floate built-ins.
15121 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
15122 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
15124 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
15125 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
15126 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
15127 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
15128 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
15129 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
15130 vec_floato): Add builtin defines.
15131 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
15132 Update the built-in documentation file for the new built-in
15135 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15137 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
15138 (mthumb): Mark as the negative of -marm.
15140 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15142 * doc/invoke.texi (ARM Options, -mcpu): Document supported
15144 (ARM Options, -mtune): Document that this accepts the same
15145 extension options as -mcpu.
15146 (ARM Options, -mfpu): Document addition of -mfpu=auto.
15148 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15150 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
15151 permitted extensions.
15153 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15155 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
15156 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
15157 (armv8-m.main): Add option +nodsp.
15158 * config/arm/arm-cpu-cdata.h: Regenerated.
15160 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15162 * config/arm/t-fuchsia: New file.
15163 * config.gcc (arm*-*-fuchsia*): Use it.
15165 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15167 * config/arm/t-symbian: Rewrite for new option infrastructure.
15169 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15171 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
15172 (MULTILIB_REQUIRED): Likewise.
15174 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15176 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
15177 (MULTILIB_RESUE): Likewise.
15178 (MULTILIB_MATCHES): Likewise.
15179 (MULTLIB_REQUIRED): Likewise.
15181 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15183 * config/arm/t-rtems: Rewrite for new option framework.
15185 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15187 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
15188 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
15189 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
15190 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
15191 * config/arm/t-multilib: ... here.
15192 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
15193 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
15194 armv7-a and armv8*-a when A-profile libraries have not been built.
15195 * config/arm/t-rmprofile: Rewrite.
15197 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15199 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
15200 with a backslash. Remove the backslash after substituting unescaped
15202 * doc/fragments.texi (MULTILIB_REUSE): Document it.
15204 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15206 * config.gcc: (arm*-*-*): When building a-profile libraries, force
15207 the driver to pass through the default setting of -mfloat-abi.
15208 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
15210 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
15211 (all_feat_combs): New rule.
15212 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
15214 * config/arm/t-aprofile: Rewrite.
15216 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15218 * config/arm/arm.h (FPUTYPE_AUTO): Define.
15219 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
15220 fpu is not specified by the user/command-line.
15221 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
15222 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
15223 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
15224 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
15225 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
15226 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
15228 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15230 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
15231 * config/arm/t-arm-elf: Rewritten.
15233 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15235 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
15236 have some floating-point instructions.
15237 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
15238 (TARGET_MAYBE_HARD_FLOAT): New macro.
15239 * config/arm/arm-builtins.c (arm_init_builtins): Use
15240 TARGET_MAYBE_HARD_FLOAT.
15241 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
15243 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15245 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
15246 (configargs.h): Include it.
15247 (arm_print_hint_for_fpu_option): New function.
15248 (arm_parse_fpu_option): New function.
15249 (candidate_extension): New class.
15250 (arm_canon_for_multilib): New function.
15251 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
15252 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
15253 (ARCH_CANONICAL_SPECS): New macro.
15254 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
15256 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15258 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
15259 are set after handling multilib fragments. Set target_cpu_default2
15262 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15264 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
15266 (arm*-*-*): Set target_cpu_default2 to a quoted string.
15267 * config/arm/parsecpu.awk (check_cpu): Validate any extension
15269 (check_arch): Likewise.
15270 * config/arm/arm.c (arm_configure_build_target): Handle
15271 TARGET_CPU_DEFAULT being a string constant. Scan any feature
15272 options in the default.
15274 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15276 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
15277 when an option is an alias of another.
15278 * config/arm/parsecpu.awk (optalias): New parser token.
15279 (gen_comm_data): Mark non-alias options as such. Emit entries
15280 for extension aliases.
15281 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
15282 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
15283 (armv6kz, armv6zk, armv6t2): Likewise.
15284 (armv7): Make vfpv3-d16 an alias.
15285 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
15287 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
15288 Sort in canonical order.
15289 (armv8-a): Sort in canonical order.
15290 (armv8.1-a, armv8.2-a): Likewise.
15291 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
15293 (cortex-a9): Sort in canonical order.
15294 * config/arm/arm.c (selftests.h): Include it.
15295 (arm_test_cpu_arch_data): New function.
15296 (arm_run_self_tests): New function.
15297 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
15298 (targetm): Move declaration to the end of the file.
15299 * arm-cpu-cdata.h: Regenerated.
15301 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15303 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
15304 call to target_mode_check describing the type of option passed.
15305 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
15306 (arm_target_thumb_only): Use arm_parse_arch_option_name or
15307 arm_parse_cpu_option_name to match parameters against list of
15309 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
15310 arm_arch_core_flags data structure.
15311 * config/arm/arm-cpu_cdata.h: Regenerated.
15313 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15315 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
15317 (arm_print_hint_for_cpu_option): Likewise.
15318 (arm_print_hint_for_arch_option): Likewise.
15319 (arm_parse_cpu_option_name): Likewise.
15320 (arm_parse_arch_option_name): Likewise.
15321 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
15322 of entries in the all_fpus list.
15323 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
15324 (arm_parse_cpu_option_name): Declare.
15325 (arm_parse_arch_option_name): Declare.
15326 (arm_parse_option_features): Declare.
15327 (arm_intialize_isa): Declare.
15328 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
15330 (gen_comm_data): ... here. Make definitions non-static.
15331 * config/arm/arm-cpu-data.h: Regenerated.
15332 * config/arm/arm-cpu-cdata.h: Regenerated.
15334 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15336 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
15337 (cpu_arch_extension): New structure.
15338 (cpu_arch_option, arch_option, cpu_option): New structures.
15339 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
15340 architecture types.
15341 (gen_data): Generate new format data tables.
15342 * config/arm/arm.c (cpu_tune): New structure.
15343 (cpu_option, processors): Delete.
15344 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
15345 (arm_print_hint_for_cpu_option): ... this and ...
15346 (arm_print_hint_for_arch_option): ... this.
15347 (arm_parse_arch_cpu_name): Delete. Replace with ...
15348 (arm_parse_cpu_option_name): ... this and ...
15349 (arm_parse_arch_option_name): ... this.
15350 (arm_unrecognized_feature): Change type of target parameter to
15352 (arm_parse_arch_cpu_features): Delete. Replace with ...
15353 (arm_parse_option_features): ... this.
15354 (arm_configure_build_target): Rework to use new configuration data
15356 (arm_print_tune_info): Rework for new configuration data tables.
15357 * config/arm/arm-cpu-data.h: Regenerated.
15358 * config/arm/arm-cpu.h: Regenerated.
15360 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15362 * Makefile.in (OBJS): Move sbitmap.o from here ...
15363 (OBJS-libcommon): ... to here.
15365 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15367 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
15368 (ISA_ALL_CRYPTO): New macro.
15369 (ISA_ALL_SIMD): New macro
15370 (ISA_ALL_FP): New macro.
15371 * config/arm/arm.c (fpu_bitlist): Update initializer.
15372 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
15374 (arm9e): Add fpu. Add option for nofp
15375 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
15376 (arm926ej-s, arm1026ej-s): Likewise.
15377 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
15378 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
15379 neon-fp16, neon-vfpv4, nofp and nosimd.
15380 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
15381 (cortex-a8): Add fpu. Add option for nofp.
15382 (cortex-a9): Add fpu. Add options for nosimd and nofp.
15383 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
15384 (cortex-r4f): Add fpu.
15385 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
15386 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
15388 (cortex-r8): Likewise.
15389 (cortex-m4): Add fpu. Add option for nofp.
15390 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
15391 (cortex-a17.cortex-a7): Likewise.
15392 (cortex-a32): Add fpu. Add options for crypto and nofp.
15393 (cortex-a35, cortex-a53): Likewise.
15394 (cortex-a57): Add fpu. Add option for crypto.
15395 (cortex-a72, cortex-a73): Likewise.
15396 (exynos-m1): Likewise.
15397 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
15398 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
15399 (cortex-m33): Add fpu. Add option for nofp.
15400 * config/arm/arm-cpu-cdata.h: Regenerated
15401 * config/arm/arm-cpu-data.h: Regenerated.
15403 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15405 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
15406 (armv5te, armv5tej): Likewise.
15407 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
15408 (armv7): Add options fp and vfpv3-d16.
15409 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
15410 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
15412 (armv7ve): Likewise.
15413 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
15414 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
15415 (armv8-a): Add nocrypto option.
15416 (armv8.1-a, armv8.2-a): Likewise.
15417 (armv8-m.main): add options fp, fp.dp and nofp.
15419 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15421 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
15423 (armv8-a+crc): Delete.
15424 (armv8.1-a): Add options simd, crypto and nofp.
15425 (armv8.2-a): Add options fp16, simd, crypto and nofp.
15426 (armv8.2-a+fp16): Delete.
15427 (armv8-m.main): Add option dsp.
15428 (armv8-m.main+dsp): Delete.
15429 (cortex-a8): Add fpu. Add nofp option.
15430 (cortex-a9): Add fpu. Add nofp and nosimd options.
15431 * config/arm/parsecpu.awk (gen_data): Generate option tables and
15432 link to main cpu and architecture data structures.
15433 (gen_comm_data): Only put isa attributes from the main architecture
15435 (option): New statement for architecture and CPU entries.
15436 * arm.c (struct cpu_option): New structure.
15437 (struct processors): Add entry for options.
15438 (arm_unrecognized_feature): New function.
15439 (arm_parse_arch_cpu_name): Ignore any characters after the first
15441 (arm_parse_arch_cpu_feature): New function.
15442 (arm_configure_build_target): Separate out any CPU and architecture
15443 features and parse separately. Don't error out if -mfpu=auto is
15444 used with only an architecture string.
15445 (arm_print_asm_arch_directives): New function.
15446 (arm_file_start): Call it.
15447 * config/arm/arm-cpu-cdata.h: Regenerated.
15448 * config/arm/arm-cpu-data.h: Likewise.
15449 * config/arm/arm-tables.opt: Likewise.
15451 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15453 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
15454 assembler when it is not -mfpu=auto.
15456 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15458 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
15459 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
15460 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
15461 (ASM_CPU_SPEC): Rewrite.
15462 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
15463 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
15464 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
15465 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
15466 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
15467 copied string is NUL-terminated. Also strip any characters prefixed
15469 (arm_rewrite_selected_arch): New function.
15470 (arm_rewrite_march): New function.
15472 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15474 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
15475 (x_arm_cpu_string, x_arm_tune_string): Likewise.
15476 (march, mcpu, mtune): Convert to string-based options.
15477 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
15478 (arm_parse_arch_cpu_name): New function.
15479 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
15480 identify selected architecture or CPU.
15481 (arm_option_save): New function.
15482 (TARGET_OPTION_SAVE): Redefine.
15483 (arm_option_restore): Restore string options.
15484 (arm_option_print): Print string options.
15486 2017-06-16 Martin Sebor <msebor@redhat.com>
15488 PR tree-optimization/80933
15489 PR tree-optimization/80934
15490 * builtins.c (fold_builtin_3): Do not handle bcmp here.
15491 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
15492 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
15493 (gimple_fold_builtin): Call them.
15495 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15497 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
15498 as unlikely; update profile.
15500 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15502 * predict.c (force_edge_cold): Handle declaring edges impossible
15505 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15507 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
15509 (try_unroll_loop_completely): Fix reporting.
15511 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15513 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
15515 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
15518 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
15519 if given a non-constant argument for an intrinsic which requires a
15522 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15524 * profile.c (compare_freqs): New function.
15525 (branch_prob): Sort edge list.
15526 (find_spanning_tree): Assume that the list is priority sorted.
15528 2017-06-16 Richard Biener <rguenther@suse.de>
15530 PR tree-optimization/81090
15531 * passes.def (pass_record_bounds): Remove.
15532 * tree-pass.h (make_pass_record_bounds): Likewise.
15533 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
15534 make_pass_record_bounds): Likewise.
15535 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
15536 not free niter estimates at the beginning but at the end.
15537 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
15539 2017-06-16 Richard Biener <rguenther@suse.de>
15541 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
15542 initializer to workaround ICE in host GCC 4.8.
15544 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15546 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
15548 (clone_inlined_nodes): Update.
15550 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15552 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
15553 prefetch settings, and enable prefetching by default at -O3.
15555 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15557 * config/aarch64/aarch64.c (aarch64_override_options_internal):
15558 Set flag_prefetch_loop_arrays according to tuning data.
15560 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15562 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
15563 New tune structure.
15564 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
15565 [Unrelated to main purpose of the patch] Place the pointer field last
15566 to enable type checking errors when tune structure are wrongly merged.
15567 * config/aarch64/aarch64.c (generic_prefetch_tune,)
15568 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
15569 (thunderx2t99_prefetch_tune): New tune constants.
15570 (tune_params *_tunings): Update all tunings (no functional change).
15571 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
15572 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
15573 from tunings structures.
15575 2017-06-16 Jakub Jelinek <jakub@redhat.com>
15578 * ubsan.c (instrument_null): Add T argument, use it instead
15579 of computing it based on IS_LHS.
15580 (instrument_object_size): Likewise.
15581 (pass_ubsan::execute): Adjust instrument_null and
15582 instrument_object_size callers to pass gimple_get_lhs or
15583 gimple_assign_rhs1 result to it. Use instrument_null instead of
15584 calling get_base_address and instrument_mem_ref. Handle
15585 aggregate call arguments for object-size sanitization.
15587 2017-06-16 Yury Gribov <tetra2005@gmail.com>
15589 PR tree-optimization/81089
15590 * tree-vrp.c (is_masked_range_test): Validate operands of
15593 2017-06-15 Martin Sebor <msebor@redhat.com>
15596 * dumpfile.c (dump_register): Avoid calling memset to initialize
15597 a class with a default ctor.
15598 * gcc.c (struct compiler): Remove const qualification.
15599 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
15600 * hash-table.h: Ditto.
15601 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
15603 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
15604 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
15606 * params.h (struct param_info): Make struct members non-const.
15607 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
15608 with default initialization.
15609 * vec.h (vec_copy_construct, vec_default_construct): New helper
15611 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
15612 with vec_copy_construct.
15613 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
15614 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
15615 * doc/invoke.texi (-Wclass-memaccess): Document.
15617 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15619 * emit-rtl.h (is_leaf): Update comment about local
15620 register allocator.
15622 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15625 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
15626 for a variable to have a section before checking if the section has a
15628 Set section to.persistent if persistent attribute is set.
15629 Warn if .persistent attribute is used on an automatic variable.
15631 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
15633 PR rtl-optimization/80474
15634 * reorg.c (update_block): Do not ignore instructions in a delay slot.
15636 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
15638 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
15641 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
15643 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
15644 (casesi): Emit bounds checking as RTL.
15645 (casesi_internal_mips16_<mode>): Remove bounds checking.
15647 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
15649 * config/xtensa/xtensa.c (xtensa_option_override): Append
15650 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
15651 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
15652 xtensa_doloop_hooks): Define unconditionally.
15653 (xtensa_reorg_loops): Only call reorg_loops in the presence of
15655 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
15656 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
15657 for it in xtensa_option_override.
15658 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
15659 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
15661 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
15663 * doc/cppopts.texi: Document '-' special value to -MF.
15665 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
15667 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
15668 (cortex_a53_fconst): Likewise.
15669 (cortex_a53_fpmul): Likewise.
15670 (cortex_a53_f_load_64): Likewise.
15671 (cortex_a53_f_load_many): Likewise.
15672 (cortex_a53_advsimd_alu): Likewise.
15673 (cortex_a53_advsimd_alu_q): Likewise.
15674 (cortex_a53_advsimd_mul): Likewise.
15675 (cortex_a53_advsimd_mul_q): Likewise.
15676 (fpmac bypass): Add new bypass for fpmac-fpmac case.
15677 Add missing fmul, r2f_cvt and fconst cases.
15679 2017-06-14 Richard Biener <rguenther@suse.de>
15681 PR middle-end/81088
15682 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
15684 (fold_binary_loc): When associating do not treat pre-existing
15685 TREE_OVERFLOW on literal constants as a reason to allow
15686 TREE_OVERFLOW on associated literal constants.
15688 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15690 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
15691 (MASK_FEATURES): New macro.
15692 * config/sparc/sparc.c (sparc_option_override): Remove the special
15693 handling of -mfpu and generalize it to all MASK_FEATURES switches.
15695 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15697 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
15698 a division of 0 if non-call exceptions are enabled.
15700 2017-06-14 Andrew Pinski <apinski@cavium.com>
15701 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15704 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
15705 Improve vector initialization code gen for only variable case.
15707 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15709 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
15711 2017-06-14 Richard Biener <rguenther@suse.de>
15713 PR tree-optimization/81083
15714 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
15717 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15719 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
15720 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
15721 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
15722 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
15723 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
15724 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
15726 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15728 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
15729 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
15731 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15733 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
15735 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15737 * config/rs6000/t-rtems: Don't handle SPE.
15739 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15741 * config/rs6000/t-linux: Don't handle SPE.
15743 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15745 * config/rs6000/eabispe.h: Delete file.
15747 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15749 * config/rs6000/t-spe: Delete file.
15751 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15753 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
15754 (rs6000_legitimate_offset_address_p): Return false for anything in
15755 V2SImode or V2SFmode.
15757 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15759 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
15760 except V2SF and V2SI. Rearrange the vector modes, and add comments.
15761 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
15763 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
15764 (rs6000_legitimate_offset_address_p): Ditto.
15765 (rs6000_emit_move): Ditto.
15766 (rs6000_init_builtins): Remove V4HI_type_node.
15768 2017-06-13 Martin Liska <mliska@suse.cz>
15771 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
15772 (gate_asan): Likewise.
15773 * asan.h (asan_no_sanitize_address_p): Remove the function.
15774 (sanitize_flags_p): New function.
15775 * builtins.def: Fix coding style.
15776 * common.opt: Use renamed enum value.
15777 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
15778 * doc/extend.texi: Document no_sanitize attribute.
15779 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
15780 to SANITIZE_UNDEFINED_NONDEFAULT.
15781 * gcc.c (sanitize_spec_function): Use the renamed enum value.
15782 * gimple-fold.c (optimize_atomic_compare_exchange_p):
15783 Use sanitize_flags_p.
15784 * gimplify.c (gimplify_function_tree): Likewise.
15785 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
15786 * opts.c (parse_no_sanitize_attribute): New function.
15787 (common_handle_option): Use renamed enum value.
15788 * opts.h (parse_no_sanitize_attribute): Declare.
15789 * tree.c (sanitize_flags_p): New function.
15790 * tree.h: Declared here.
15791 * tsan.c: Use sanitize_flags_p.
15792 * ubsan.c (ubsan_expand_null_ifn): Likewise.
15793 (instrument_mem_ref): Likewise.
15794 (instrument_bool_enum_load): Likewise.
15795 (do_ubsan_in_current_function): Remove the function.
15796 (pass_ubsan::execute): Use sanitize_flags_p.
15797 * ubsan.h: Remove do_ubsan_in_current_function
15798 * tree-cfg.c (print_no_sanitize_attr_value): New function.
15799 (dump_function_to_file): Use it here.
15801 2017-06-13 Martin Jambor <mjambor@suse.cz>
15803 PR tree-optimization/80803
15804 PR tree-optimization/81063
15805 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
15806 (propagate_subaccesses_across_link): Enqueue subtree whenever
15807 necessary instead of relying on the caller.
15809 2017-06-13 Martin Jambor <mjambor@suse.cz>
15811 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
15812 that have a first_link.
15813 (sort_and_splice_var_accesses): Do not check first_link before
15815 (subtree_mark_written_and_enqueue): Likewise.
15816 (propagate_all_subaccesses): Likewise and do not stop at first
15817 parent with a first_link.
15819 2017-06-13 Martin Jambor <mjambor@suse.cz>
15821 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
15824 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15826 * match.pd: New pattern.
15828 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15830 * tree-vrp.c (is_masked_range_test): New function.
15831 (register_edge_assert_for): Determine ranges for
15834 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15836 PR tree-optimization/67328
15837 * fold-const.c (maskable_range_p): New function.
15838 (build_range_check): Generate bittests if possible.
15840 2017-06-13 Martin Liska <mliska@suse.cz>
15842 * gimple-pretty-print.c (dump_probability): Add new argument.
15843 (dump_edge_probability): Dump both probability and count.
15844 (dump_gimple_label): Likewise.
15845 (dump_gimple_bb_header): Likewise.
15847 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
15850 * config/avr/avr-devices.c: Fix indentation.
15851 * config/avr/gen-avr-mmcu-specs.c: Dito.
15853 2017-06-13 Richard Biener <rguenther@suse.de>
15855 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
15856 instead get vector type from stmt_info.
15857 (vectorizable_reduction): Adjust. Remove dead code.
15859 2017-06-13 Richard Biener <rguenther@suse.de>
15861 PR middle-end/81065
15862 * fold-const.c (extract_muldiv_1): Remove bogus distribution
15863 case of C * (x * C2 + C3).
15864 (fold_addr_of_array_ref_difference): Properly fold index difference.
15866 2017-06-12 David S. Miller <davem@davemloft.net>
15869 * config/sparc/sparc.md (return expander): Emit frame blockage if
15870 function uses alloca.
15872 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
15874 * combine.c (make_field_assignment): Check len rather than the mode
15875 precision when calling force_to_mode.
15877 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
15879 Support multilibs and devices that see flash in RAM address range.
15882 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
15883 (avr_mcu_t) <flash_pm_offset>: New field.
15884 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
15885 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
15886 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
15887 (AVR_TINY_PM_OFFSET): Remove macro.
15888 * config/avr/avr.opt (-mshort-calls): New option.
15889 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
15890 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
15891 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
15892 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
15893 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
15894 instead of avr_arch->have_jmp_call.
15895 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
15896 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
15897 avr_arch->flash_pm_offset to define.
15898 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
15899 new field flash_pm_offset. Add entry for avrxmega3.
15900 (avr_texinfo): Add entry for avrxmega3.
15901 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
15902 attiny212, attiny214,
15903 attiny412, attiny414, attiny416, attiny417,
15904 attiny814, attiny816, attiny817,
15905 attiny1614, attiny1616, attiny1617,
15906 attiny3214, attiny3216, attiny3217.
15907 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
15908 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
15909 (avr_print_operand_address) [AVR_TINY]: Same.
15910 (avr_asm_init_sections) <readonly_data_section>: Only patch
15911 callback if avr_arch->flash_pm_offset = 0.
15912 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
15913 for rodata if avr_arch->flash_pm_offset != 0.
15914 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
15915 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
15916 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
15917 (m_options): Append opt_rcall.
15918 (m_dirnames): Append dir_rcall.
15919 * config/avr/t-multilib: Regenerate.
15921 * configure.ac [target=avr]: Check whether avrxmega3 default
15922 linker description file works as needed.
15923 * configure: Regenerate.
15924 * doc/avr-mmcu.texi: Regenerate.
15925 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
15926 <__AVR_ARCH__>: Document avrxmega3 and 103.
15927 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
15928 <__AVR_SHORT_CALLS__>: Document it.
15929 <__AVR_PM_BASE_ADDRESS__>: Document it.
15930 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
15931 (AVR Variable Attributes) <progmem>: Document this is
15932 not needed for avrxmega3.
15933 (AVR Named Address Spaces) <__flash>: Dito.
15935 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
15937 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
15939 2017-06-12 Doug Rupp <rupp@adacore.com>
15941 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
15942 Append vxworks-stdint.h to the tm_file list.
15943 * config/vxworks-stdint.h: New file.
15945 2017-06-12 Martin Liska <mliska@suse.cz>
15947 PR tree-optimization/81041
15948 * tree-profile.c (gimple_gen_ic_func_profiler):
15949 Create an extra BB in profile-generate
15950 (gimple_gen_time_profiler): Likewise.
15952 2017-06-12 Jakub Jelinek <jakub@redhat.com>
15954 PR tree-optimization/81003
15955 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
15956 (update_range_test): Use it instead of force_gimple_operand_gsi.
15958 2017-06-12 Richard Biener <rguenther@suse.de>
15960 PR tree-optimization/81053
15961 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
15962 with backedge value not defined in loop. Simplify def stmt
15965 2017-06-11 Tom de Vries <tom@codesourcery.com>
15968 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
15970 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
15971 nvptx_cannot_force_const_mem.
15973 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15975 * opts.c (finish_options): Move test for flag_split_stack after
15976 it has been initialized.
15978 2017-06-11 Jason Merrill <jason@redhat.com>
15980 * tree.h (id_equal): New.
15981 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
15982 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
15983 instead of strcmp of IDENTIFIER_POINTER.
15985 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15987 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
15988 (mark_all_inlined_calls_cdtor): Fix formating.
15989 (inline_transform): Rescale profile before inlining.
15991 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15993 * cgraph.h (cgraph_edge::clone): Update prototype.
15994 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
15995 (cgraph_node::create_clone): Update.
15996 (cgraph_node::create_version_clone): Update.
15997 * tree-inline.c (copy_bb): Update.
15998 (expand_call_inline): Update.
16000 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
16002 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
16003 factored out from ...
16004 (rs6000_emit_prologue): ... here.
16006 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
16008 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
16009 factored out from ...
16010 (rs6000_emit_prologue): ... here.
16012 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16014 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
16016 (handle_missing_profiles): Fix computation of tp_first_run.
16017 (counts_to_freqs): Do not touch freqs when count is 0.
16019 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16021 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
16024 2017-06-10 Tom de Vries <tom@codesourcery.com>
16026 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
16027 attributes): Document signal effective target.
16029 2017-06-10 Tom de Vries <tom@codesourcery.com>
16031 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
16032 Document effective target stack_size.
16034 2017-06-09 David Malcolm <dmalcolm@redhat.com>
16036 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
16037 to the edit_context if they can be auto-applied.
16039 2017-06-9 Ian Lance Taylor <iant@golang.org>
16041 * opts.c (finish_options): If -fsplit-stack, disable implicit
16042 -forder-blocks-and-partition.
16043 * doc/invoke.texi (Optimize Options): Document that when using
16044 -fsplit-stack -forder-blocks-and-partition is not implicitly
16047 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16049 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
16050 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
16051 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
16052 * builtins.def (abort, trap, unreachable): Declare cold.
16053 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
16054 * tree-core.h (ECF_COLD): New.
16055 * tree.c (set_call_expr_flags): Handle ECF_COLD.
16056 (build_common_builtin_nodes): Mark unreachable and abort as cold.
16058 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16060 * predict.c (unlikely_executed_stmt_p): Cleanup.
16062 2017-06-09 Richard Biener <rguenther@suse.de>
16064 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
16065 model if the ref is always written to.
16067 2017-06-09 Tamar Christina <tamar.christina@arm.com>
16069 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
16071 2017-06-09 Tamar Christina <tamar.christina@arm.com>
16073 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
16076 2017-06-09 Tom de Vries <tom@codesourcery.com>
16079 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
16080 "target cannot support label values" when encountering LABEL_REF.
16082 2017-06-09 Martin Liska <mliska@suse.cz>
16084 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
16085 (gimple_gen_ic_func_profiler): Emit direct comparison
16086 of __gcov_indirect_call_callee with NULL.
16087 (gimple_gen_time_profiler): Change probability from
16088 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
16090 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16092 * profile.c (edge_gcov_counts): Turn to pointer.
16093 (compute_branch_probabilities, compute_branch_probabilities): Update.
16094 (branch_prob): Do not clear edge_gcov_count.
16095 * profile.h (edge_gcov_counts): Turn to pointer.
16096 (edge_gcov_count): Update.
16098 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16100 * gimple.h (gimple_check_failed): Mark cold.
16102 2017-06-09 Richard Biener <rguenther@suse.de>
16104 PR tree-optimization/66623
16105 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
16106 refactor check_reduction into two parts, properly computing
16107 whether we have to check reduction validity for outer loop
16110 2017-06-09 Richard Biener <rguenther@suse.de>
16112 PR tree-optimization/79483
16113 * graphite-scop-detection.c (order): New global.
16114 (get_order): Compute bb to order mapping that satisfies code
16115 generation constraints.
16116 (cmp_pbbs): New helper.
16117 (build_scops): Start domwalk at entry block, sort generated
16120 2017-06-09 Richard Biener <rguenther@suse.de>
16122 PR middle-end/81007
16123 * ipa-polymorphic-call.c
16124 (ipa_polymorphic_call_context::restrict_to_inner_class):
16125 Skip FIELD_DECLs with error_mark_node type.
16126 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
16129 2017-06-09 Martin Liska <mliska@suse.cz>
16131 * predict.c (struct branch_predictor): New struct.
16132 (test_prediction_value_range): New test.
16133 (predict_c_tests): New function.
16134 * selftest-run-tests.c (selftest::run_tests): Run the function.
16135 * selftest.h: Declare new tests.
16137 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
16140 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
16141 gen_add3_insn did not fail.
16142 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
16143 r0, construct that number in a temporary reg and add that reg to r0.
16144 If asked to put the result in r0 as well, fail.
16146 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
16148 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16149 for early expansion of vec_eqv.
16151 2017-06-08 Jakub Jelinek <jakub@redhat.com>
16153 PR middle-end/81005
16154 * ubsan.c (instrument_null): Avoid pointless code temporary.
16155 (pass_ubsan::execute): Instrument aggregate arguments of calls.
16157 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
16161 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
16164 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
16167 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16169 * predict.c (unlikely_executed_edge_p): Move ahead.
16170 (probably_never_executed_edge_p): Use it.
16172 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
16174 PR middle-end/79988
16175 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
16176 gimple_call_builtin_p call.
16178 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16180 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
16181 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
16182 rtl_check_failed_type2, rtl_check_failed_code1,
16183 rtl_check_failed_code2, rtl_check_failed_code_mode,
16184 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
16185 rtvec_check_failed_bounds, rtl_check_failed_flag,
16186 _fatal_insn_not_found, _fatal_insn): Likewise.
16187 * tree.h (tree_contains_struct_check_failed,
16188 tree_check_failed, tree_not_check_failed,
16189 tree_class_check_failed, tree_range_check_failed,
16190 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
16191 tree_vec_elt_check_failed, phi_node_elt_check_failed,
16192 tree_operand_check_failed, omp_clause_check_failed,
16193 omp_clause_operand_check_failed, omp_clause_range_check_failed):
16196 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16198 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
16199 flag_branch_probabilities.
16200 * ipa-inline.c (edge_badness): Likewise.
16201 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
16202 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
16203 * predict.c (maybe_hot_frequency_p): Likewise.
16204 (probably_never_executed): Likewise.
16205 * sched-ebb.c (schedule_ebbs): Likewise.
16206 * sched-rgn.c (find_single_block_region): Likewise.
16207 * tracer.c (tail_duplicate): Likewise.
16209 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16211 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
16212 longer requires x_flag_profile_use.
16214 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16216 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
16217 instead of flag_reorder_blocks_and_partition.
16218 * dbxout.c (dbxout_function_end): Likewise.
16219 * dwarf2out.c (gen_subprogram_die): Likewise.
16220 * haifa-sched.c (sched_create_recovery_edges): Likewise.
16221 * hw-doloop.c (reorg_loops): Likewise.
16222 * varasm.c (assemble_start_function,
16223 assemble_end_function): Likewise.
16224 (decide_function_section): Do not check for
16225 flag_reorder_blocks_and_partition.
16227 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
16229 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
16231 (chkp_get_hard_register_fake_addr_expr): Ditto.
16232 (chkp_build_addr_expr): Add check for hard reg case.
16233 (chkp_parse_array_and_component_ref): Ditto.
16234 (chkp_find_bounds_1): Ditto.
16235 (chkp_process_stmt): Don't generate bounds store for
16238 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16240 * predict.c (maybe_hot_bb_p): Do not check profile status.
16241 (maybe_hot_edge_p): Likewise.
16242 (probably_never_executed): Check for zero counts even if profile
16244 (unlikely_executed_edge_p): New function.
16245 (unlikely_executed_stmt_p): New function.
16246 (unlikely_executed_bb_p): New function.
16247 (set_even_probabilities): Use unlikely predicates.
16248 (combine_predictions_for_bb): Likewise.
16249 (predict_paths_for_bb): Likewise.
16250 (predict_paths_leading_to_edge): Likewise.
16251 (determine_unlikely_bbs): New function.
16252 (estimate_bb_frequencies): Use it.
16253 (compute_function_frequency): Use zero counts even if profile is
16255 * profile-count.h: Fix typo.
16257 2017-08-08 Julia Koval <julia.koval@intel.com>
16259 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
16260 _mm512_mask_cvtsepi16_storeu_epi8,
16261 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
16262 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
16263 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
16264 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
16265 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
16266 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
16267 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
16268 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
16269 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
16270 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
16271 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
16272 __builtin_ia32_pmovuswb256mem_mask,
16273 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
16274 __builtin_ia32_pmovwb512mem_mask): New builtins.
16276 2017-08-08 Julia Koval <julia.koval@intel.com>
16278 PR target/73350,80862
16279 * config/i386/subst.md (round): Fix round pattern.
16280 * config/i386/i386.c (ix86_erase_embedded_rounding):
16281 Fix erasing rounding for the fixed pattern.
16283 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16285 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
16287 2017-06-08 Martin Liska <mliska@suse.cz>
16289 PR gcov-profile/80911
16290 * gcov.c (block_info::block_info): New constructor.
16292 2017-06-07 Carl Love <cel@us.ibm.com>
16294 * config/rs6000/rs6000-c: The return type of the following
16295 built-in functions was implemented as int not long long. Fix sign
16296 of return value for the unsigned version of vec_mulo and vec_mule.
16297 vector unsigned long long vec_bperm (vector unsigned long long,
16298 vector unsigned char)
16299 vector signed long long vec_mule (vector signed int,
16301 vector unsigned long long vec_mule (vector unsigned int,
16302 vector unsigned int)
16303 vector signed long long vec_mulo (vector signed int,
16305 vector unsigned long long vec_mulo (vector unsigned int,
16306 vector unsigned int)
16307 * doc/extend.texi: Fix the documentation for the built-in
16310 2017-06-07 Carl Love <cel@us.ibm.com>
16313 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
16316 2017-06-07 Carl Love <cel@us.ibm.com>
16318 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
16319 support, Generate doublehv for signed int/float for BE case only.
16321 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
16323 * doc/invoke.texi (mcx16): Rewrite.
16325 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16327 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
16328 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
16329 *mov<mode>_softfloat, and an anonymous splitter): Use
16330 nonimmediate_operand instead of rs6000_nonimmediate_operand.
16332 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16334 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
16336 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
16337 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
16338 (rs6000_debug_reg_global): Adjust.
16339 (rs6000_init_hard_regno_mode_ok): Adjust.
16340 (rs6000_dbx_register_number): Adjust.
16341 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
16342 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
16343 Remove SPE_ACC and SPEFSCR.
16344 (REG_ALLOC_ORDER): Ditto.
16345 (FRAME_POINTER_REGNUM): Change to 111.
16346 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
16347 (REG_CLASS_NAMES): Ditto.
16348 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
16349 (REGISTER_NAMES): Ditto.
16350 (ADDITIONAL_REG_NAMES): Ditto.
16351 (rs6000_reg_names): Ditto.
16352 * config/rs6000/rs6000.md: Renumber some register number
16355 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16357 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
16359 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
16360 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
16362 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
16363 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
16364 Delete the SPE high registers.
16365 (REG_ALLOC_ORDER): Ditto.
16366 (enum reg_class): Remove SPE_HIGH_REGS.
16367 (REG_CLASS_NAMES): Ditto.
16368 (REG_CLASS_CONTENTS): Delete the SPE high registers.
16369 (REGISTER_NAMES): Ditto.
16370 (rs6000_reg_names): Ditto.
16371 * doc/tm.texi.in: Remove SPE as example.
16372 * doc/tm.texi: Regenerate.
16374 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16376 * config/rs6000/8540.md (ppc8540_brinc): Delete.
16377 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
16378 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
16379 * config/rs6000/rs6000.md (type): Remove "brinc".
16381 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16383 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
16384 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
16385 * config/rs6000/linuxspe.h: Delete file.
16386 * config/rs6000/rs6000.md: Don't include spe.md.
16387 * config/rs6000/spe.h: Delete file.
16388 * config/rs6000/spe.md: Delete file.
16389 * config/rs6000/t-rs6000: Remove spe.md.
16391 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16393 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
16394 (reg_or_none500mem_operand): Delete.
16395 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
16396 instead of reg_or_none500mem_operand.
16398 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16400 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
16401 handling of SPE flags.
16402 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
16404 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16406 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
16408 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
16409 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
16410 paired_divv2sf3): Similar.
16411 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
16412 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
16413 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
16415 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
16416 Rename the paired_* instruction patterns.
16417 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
16419 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
16420 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
16421 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
16422 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
16423 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
16424 PAIRED_VECTOR_MODE.
16425 (struct machine_function): Delete field spe_insn_chain_scanned_p.
16426 (spe_func_has_64bit_regs_p): Delete.
16427 (spe_expand_predicate_builtin): Delete.
16428 (spe_expand_evsel_builtin): Delete.
16429 (TARGET_DWARF_REGISTER_SPAN): Do not define.
16430 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
16431 (invalid_e500_subreg): Delete.
16432 (rs6000_legitimize_address): Always force_reg op2 as well, for
16433 paired single memory accesses.
16434 (rs6000_member_type_forces_blk): Delete.
16435 (rs6000_spe_function_arg): Delete.
16436 (rs6000_expand_unop_builtin): Delete SPE handling.
16437 (rs6000_expand_binop_builtin): Ditto.
16438 (spe_expand_stv_builtin): Delete.
16439 (bdesc_2arg_spe): Delete.
16440 (spe_expand_builtin): Delete.
16441 (spe_expand_predicate_builtin): Delete.
16442 (spe_expand_evsel_builtin): Delete.
16443 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
16444 (spe_init_builtins): Delete.
16445 (spe_func_has_64bit_regs_p): Delete.
16446 (savres_routine_name): Delete "info" parameter. Adjust callers.
16447 (rs6000_emit_stack_reset): Ditto.
16448 (rs6000_dwarf_register_span): Delete.
16449 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
16450 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
16451 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
16453 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
16455 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
16456 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
16457 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
16458 mulv2sf3, divv2sf3): Delete expanders.
16460 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16462 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
16464 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16466 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
16467 * config/rs6000/rs6000.c: Ditto.
16469 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16471 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
16472 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
16473 comparison_operator.
16475 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16477 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
16478 * config/rs6000/rs6000.opt: Ditto.
16479 * config/rs6000/t-rtems: Ditto.
16481 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16483 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
16484 TARGET_E500_SINGLE by 0, simplify.
16485 * config/rs6000/rs6000.c: Ditto.
16486 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
16487 (spe_build_register_parallel): Delete.
16488 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
16489 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
16490 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
16491 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
16492 (E500_CONVERT): Delete.
16493 * config/rs6000/spe.md: Remove many patterns and all define_constants.
16495 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16497 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
16498 * config/rs6000/dfp.md: Ditto.
16499 (negdd2, *negdd2_fpr): Merge.
16500 (absdd2, *absdd2_fpr): Merge.
16501 (negtd2, *negtd2_fpr): Merge.
16502 (abstd2, *abstd2_fpr): Merge.
16503 * config/rs6000/e500.h: Delete file.
16504 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
16505 TARGET_FPRS by 1 and simplify.
16506 * config/rs6000/rs6000-c.c: Ditto.
16507 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
16508 TARGET_DF_SPE by 0.
16509 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
16511 * config/rs6000/rs6000.md: Ditto.
16512 (floatdidf2, *floatdidf2_fpr): Merge.
16513 (move_from_CR_gt_bit): Delete.
16514 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
16515 (E500_CR_IOR_COMPARE): Delete.
16516 (All patterns that require !TARGET_FPRS): Delete.
16517 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
16519 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16521 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
16523 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16525 * graphds.c (add_edge): Intitialize edge's attached data.
16526 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
16527 pointer parameter. Call pointed function on each edge during
16528 graph traversing. Skip traversing the edge when the function
16530 (graphds_dfs, graphds_scc): Ditto.
16531 (for_each_edge): New parameter. Pass the new parameter to callback
16533 * graphds.h (skip_edge_callback): New function pointer type.
16534 (graphds_dfs, graphds_scc): New function pointer parameter.
16535 (graphds_edge_callback, for_each_edge): New parameter.
16537 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16539 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
16540 out code checking if runtime alias check is possible to below ...
16541 Call the new function.
16542 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
16543 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
16545 2017-06-07 Marek Polacek <polacek@redhat.com>
16548 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
16549 TYPE_OVERFLOW_WRAPS check.
16551 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16553 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
16554 if versioning is required.
16555 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
16556 peeling with the check for versioning.
16558 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16560 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
16561 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
16562 Set true to new parameter if new ssa variable is defined.
16563 (vect_gen_vector_loop_niters): Refactor. Set range information
16564 for the new vector loop bound variable.
16565 (vect_do_peeling): Ditto.
16567 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16569 * tree-affine.c (ssa.h): Include header file.
16570 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
16571 has wrapping overflow behavior.
16573 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16575 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
16577 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16579 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
16580 (tree_to_aff_combination): ... here.
16582 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16584 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
16585 reg_pressure model function.
16586 (ivopts_global_cost_for_size): Delete.
16587 (determine_set_costs, iv_ca_recount_cost): Call new model function
16588 ivopts_estimate_reg_pressure.
16590 2017-06-07 Tamar Christina <tamar.christina@arm.com>
16592 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
16593 expensive than udiv. Remove floating point cases from mod.
16595 2017-06-07 Tamar Christina <tamar.christina@arm.com>
16597 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
16598 Increase idiv cost.
16600 2017-06-07 Tamar Christina <tamar.christina@arm.com>
16602 * config/aarch64/aarch64.md
16603 (copysignsf3): Fix mask generation.
16605 2017-06-07 Jakub Jelinek <jakub@redhat.com>
16607 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
16609 (class dump_manager): Add register_dumps method.
16610 * dumpfile.c: Include langhooks.h.
16611 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
16612 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
16613 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
16614 (dump_manager::dump_register): Start with 512 entries instead of 32.
16615 (dump_manager::register_dumps): New method.
16616 * toplev.c (general_init): Instead of invoking register_dumps
16617 langhook, invoke register_dumps method on the dump manager.
16618 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
16621 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
16623 * doc/md.texi: Clarify the restrictions on a define_insn condition.
16624 Say that # requires an associated define_split to exist, and that
16625 the define_split must be suitable for use after register allocation.
16627 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16629 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
16630 (compute_outgoing_frequencies): Also initialize zero counts.
16631 (find_many_sub_basic_blocks): Do not produce uninitialized profile
16632 around loops; preserve more of profile when nothing changes.
16634 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
16636 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
16638 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
16639 * config/arm/arm-cpu-cdata.h: Regenerate.
16640 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
16641 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
16642 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
16643 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
16644 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
16646 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
16648 * config/arm/t-rmprofile: Likewise.
16649 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
16651 2017-06-06 David S. Miller <davem@davemloft.net>
16654 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
16655 blockage if function uses alloca.
16657 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16659 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
16660 New "uid" fields to hold pretty-print IDs of group and ref.
16661 Memory references are now identified as <group_id>:<ref_id>
16662 instead of using [random] addresses.
16663 (dump_mem_details): Simplify, no functional change.
16664 (dump_mem_ref): Simplify and make output more concise.
16665 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
16666 (find_or_create_group): Initialize group uid.
16667 (record_ref): Initialize ref uid. Improve debug output.
16668 (prune_group_by_reuse, should_issue_prefetch_p,)
16669 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
16670 (mark_nontemporal_store, determine_loop_nest_reuse):
16671 Improve debug output.
16673 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16675 * dbgcnt.def (prefetch): New debug counter.
16676 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
16677 (schedule_prefetches): Stop issueing prefetches if debug counter
16680 2017-06-06 Tom de Vries <tom@codesourcery.com>
16682 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
16683 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
16685 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16687 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
16688 Use aarch64_reg_or_zero predicate for operand 4.
16689 (aarch64_compare_and_swap<mode> define_insn_and_split):
16690 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
16691 (aarch64_store_exclusive<mode>): Likewise for operand 2.
16693 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
16695 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
16696 (arm_compute_save_core_reg_mask): This.
16697 (thumb1_compute_save_reg_mask): Rename into ...
16698 (thumb1_compute_save_core_reg_mask): This.
16699 (arm_compute_save_reg0_reg12_mask): Adapt comment.
16700 (arm_compute_frame_layout): Likewise.
16702 2017-06-06 Richard Biener <rguenther@suse.de>
16704 PR tree-optimization/80974
16705 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
16706 keep or clear leaders SSA info.
16708 2017-06-06 Tom de Vries <tom@codesourcery.com>
16710 * config/nvptx/nvptx.c (split_mode_p): New function.
16711 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
16713 2017-06-06 Tom de Vries <tom@codesourcery.com>
16715 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
16717 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16720 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
16723 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16725 * shrink-wrap.c (handle_simple_exit): Update profile.
16726 (try_shrink_wrapping): Upate profile.
16728 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16730 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
16731 (tree_guess_outgoing_edge_probabilities): New.
16732 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
16733 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
16735 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16737 * ipa-split.c (split_function): Initialize return bb profile.
16739 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16741 * profile.c (compute_branch_probabilities): Also initialize
16742 EXIT_BLOCK profile.
16744 2017-06-06 Richard Biener <rguenther@suse.de>
16746 PR tree-optimization/80928
16747 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
16748 (vect_analyze_loop_operations): Properly guard analysis for
16750 (vect_transform_loop): Likewise.
16751 (vect_analyze_loop_2): Also reset SLP type on PHIs.
16752 (vect_model_induction_cost): Do not cost for pure SLP.
16753 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
16754 of induction in inner loop vectorization.
16755 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
16756 (vect_get_and_check_slp_defs): Handle vect_induction_def.
16757 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
16759 (vect_analyze_slp_cost_1): Cost induction.
16760 (vect_detect_hybrid_slp_stmts): Handle PHIs.
16761 (vect_get_slp_vect_defs): Likewise.
16762 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
16763 (vect_transform_stmt): Handle SLP reductions.
16764 * tree-vectorizer.h (vectorizable_induction): Adjust.
16766 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16768 * config/rs6000/rs6000.c (make_resolver_func): Update
16769 init_lowered_empty_function call.
16771 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
16773 * doc/invoke.texi: Document the -fprofile-abs-path option.
16774 * common.opt (fprofile-abs-path): New option.
16775 * gcov-io.h (gcov_write_filename): Declare.
16776 * gcov-io.c (gcov_write_filename): New function.
16777 * coverage.c (coverage_begin_function): Use gcov_write_filename.
16778 * profile.c (output_location): Likewise.
16780 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16782 * shring-wrap.c: Revert accidental commit.
16784 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
16786 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
16788 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16790 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
16792 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
16793 profile in callgraph edge.
16794 * profile-count.h (apply_probability): If THIS is 0, then result is 0
16795 (apply_scale): Likewise.
16796 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
16797 Also scale profile when inlining function with zero profile.
16798 (initialize_cfun): Update exit block profile even when it is zero.
16799 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
16800 when profile is read.
16802 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16804 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
16805 (CLONE_*): New constants to define the processors we can generate
16806 code for with the target_clone attribute.
16807 (rs6000_clone_map): New array to identify which clone processors
16808 the current program is running on.
16809 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
16810 target_clone attribute.
16811 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
16812 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
16813 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
16814 (cpu_expand_builtin): Add support for target_clone attribute.
16815 (rs6000_valid_attribute_p): Allow "default" attribute.
16816 (get_decl_name): New debug function to simplify printing the
16817 current function name in debugging statements.
16818 (rs6000_clone_priority): New functions to support the target_clone
16819 attribute, and be able to generate code to switch between ISA 2.05
16820 through ISA 3.0 (power6 through power9).
16821 (rs6000_compare_version_priority): Likewise.
16822 (rs6000_get_function_versions_dispatcher): Likewise.
16823 (make_resolver_func): Likewise.
16824 (add_condition_to_bb): Likewise.
16825 (dispatch_function_versions): Likewise.
16826 (rs6000_generate_version_dispatcher_body): Likewise.
16827 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
16828 (fusion_gpr_load_p): Fix a spacing issue.
16829 * doc/extend.texi (Common Function Attributes): Document that the
16830 PowerPC supports the target_clone attribute.
16832 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
16834 * config/arm/arm.h: explain F symbol found in description of ARM
16835 register allocation in its legend.
16837 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16839 * config/mips/frame-header-opt.c: Include profile-count.h.
16840 * config/riscv/riscv.c: Include profile-count.h
16842 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16844 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
16846 (sm_set_flag_if_changed): Add bbs field.
16847 (execute_sm_if_changed_flag_set): Pass BBS.
16848 (execute_sm): Update.
16850 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16852 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
16855 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16857 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
16858 (peephole2): New peephole2 to emit the above.
16859 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
16861 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16863 * config/aarch64/aarch64.c (define_peephole2 above
16864 *sub_<shift>_<mode>): New peephole.
16866 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16868 * config/i386/i386.c (make_resolver_func): Update.
16869 * Makefile.in: Add profile-count.h and profile-count.o
16870 * auto-profile.c (afdo_indirect_call): Update to new API.
16871 (afdo_set_bb_count): Update.
16872 (afdo_propagate_edge): Update.
16873 (afdo_propagate_circuit): Update.
16874 (afdo_calculate_branch_prob): Update.
16875 (afdo_annotate_cfg): Update.
16876 * basic-block.h: Include profile-count.h
16877 (struct edge_def): Turn count to profile_count.
16878 (struct basic_block_def): Likewie.
16879 (REG_BR_PROB_BASE): Move to profile-count.h
16880 (RDIV): Move to profile-count.h
16881 * bb-reorder.c (max_entry_count): Turn to profile_count.
16882 (find_traces): Update.
16883 (rotate_loop):Update.
16884 (connect_traces):Update.
16885 (sanitize_hot_paths):Update.
16886 * bt-load.c (migrate_btr_defs): Update.
16887 * cfg.c (RDIV): Remove.
16888 (init_flow): Use alloc_block.
16889 (alloc_block): Uninitialize count.
16890 (unchecked_make_edge): Uninitialize count.
16891 (check_bb_profile): Update.
16892 (dump_edge_info): Update.
16893 (dump_bb_info): Update.
16894 (update_bb_profile_for_threading): Update.
16895 (scale_bbs_frequencies_int): Update.
16896 (scale_bbs_frequencies_gcov_type): Update.
16897 (scale_bbs_frequencies_profile_count): New.
16898 * cfg.h (update_bb_profile_for_threading): Update.
16899 (scale_bbs_frequencies_profile_count): Declare.
16900 * cfgbuild.c (compute_outgoing_frequencies): Update.
16901 (find_many_sub_basic_blocks): Update.
16902 * cfgcleanup.c (try_forward_edges): Update.
16903 (try_crossjump_to_edge): Update.
16904 * cfgexpand.c (expand_gimple_tailcall): Update.
16905 (construct_exit_block): Update.
16906 * cfghooks.c (verify_flow_info): Update.
16907 (dump_bb_for_graph): Update.
16908 (split_edge): Update.
16909 (make_forwarder_block): Update.
16910 (duplicate_block): Update.
16911 (account_profile_record): Update.
16912 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
16913 (get_estimated_loop_iterations): Update.
16914 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
16915 (single_likely_exit): Update.
16916 * cfgloopmanip.c (scale_loop_profile): Update.
16918 (set_zero_probability): Update.
16919 (lv_adjust_loop_entry_edge): Update.
16920 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16921 (purge_dead_edges): Update.
16922 (rtl_account_profile_record): Update.
16923 * cgraph.c (cgraph_node::create): Uninitialize count.
16924 (symbol_table::create_edge): Uninitialize count.
16925 (cgraph_update_edges_for_call_stmt_node): Update.
16926 (cgraph_edge::dump_edge_flags): Update.
16927 (cgraph_node::dump): Update.
16928 (cgraph_edge::maybe_hot_p): Update.
16929 * cgraph.h: Include profile-count.h
16930 (create_clone), create_edge, create_indirect_edge): Update.
16931 (cgraph_node): Turn count to profile_count.
16932 (cgraph_edge0: Likewise.
16933 (make_speculative, clone): Update.
16934 (create_edge): Update.
16935 (init_lowered_empty_function): Update.
16936 * cgraphclones.c (cgraph_edge::clone): Update.
16937 (duplicate_thunk_for_node): Update.
16938 (cgraph_node::create_clone): Update.
16939 * cgraphunit.c (cgraph_node::analyze): Update.
16940 (cgraph_node::expand_thunk): Update.
16941 * final.c (dump_basic_block_info): Update.
16942 * gimple-streamer-in.c (input_bb): Update.
16943 * gimple-streamer-out.c (output_bb): Update.
16944 * graphite.c (print_global_statistics): Update.
16945 (print_graphite_scop_statistics): Update.
16946 * hsa-brig.c: Include basic-block.h.
16947 * hsa-dump.c: Include basic-block.h.
16948 * hsa-gen.c (T sum_slice): Update.
16949 (convert_switch_statements):Update.
16950 * hsa-regalloc.c: Include basic-block.h.
16951 * ipa-chkp.c (chkp_produce_thunks): Update.
16952 * ipa-cp.c (struct caller_statistics): Update.
16953 (init_caller_stats): Update.
16954 (gather_caller_stats): Update.
16955 (ipcp_cloning_candidate_p): Update.
16956 (good_cloning_opportunity_p): Update.
16957 (get_info_about_necessary_edges): Update.
16958 (dump_profile_updates): Update.
16959 (update_profiling_info): Update.
16960 (update_specialized_profile): Update.
16961 (perhaps_add_new_callers): Update.
16962 (decide_about_value): Update.
16963 (ipa_cp_c_finalize): Update.
16964 * ipa-devirt.c (struct odr_type_warn_count): Update.
16965 (struct decl_warn_count): Update.
16966 (struct final_warning_record): Update.
16967 (possible_polymorphic_call_targets): Update.
16968 (ipa_devirt): Update.
16969 * ipa-fnsummary.c (redirect_to_unreachable): Update.
16970 * ipa-icf.c (sem_function::merge): Update.
16971 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
16972 * ipa-inline.c (compute_uninlined_call_time): Update.
16973 (compute_inlined_call_time): Update.
16974 (want_inline_small_function_p): Update.
16975 (want_inline_self_recursive_call_p): Update.
16976 (edge_badness): Update.
16977 (lookup_recursive_calls): Update.
16978 (recursive_inlining): Update.
16979 (inline_small_functions): Update.
16980 (dump_overall_stats): Update.
16981 (dump_inline_stats): Update.
16982 * ipa-profile.c (ipa_profile_generate_summary): Update.
16983 (ipa_propagate_frequency): Update.
16984 (ipa_profile): Update.
16985 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
16986 * ipa-utils.c (ipa_merge_profiles): Update.
16987 * loop-doloop.c (doloop_modify): Update.
16988 * loop-unroll.c (report_unroll): Update.
16989 (unroll_loop_runtime_iterations): Update.
16990 * lto-cgraph.c (lto_output_edge): Update.
16991 (lto_output_node): Update.
16992 (input_node): Update.
16993 (input_edge): Update.
16994 (merge_profile_summaries): Update.
16995 * lto-streamer-in.c (input_cfg): Update.
16996 * lto-streamer-out.c (output_cfg): Update.
16997 * mcf.c (create_fixup_graph): Update.
16998 (adjust_cfg_counts): Update.
16999 (sum_edge_counts): Update.
17000 * modulo-sched.c (sms_schedule): Update.
17001 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
17002 * predict.c (maybe_hot_count_p): Update.
17003 (probably_never_executed): Update.
17004 (dump_prediction): Update.
17005 (combine_predictions_for_bb): Update.
17006 (propagate_freq): Update.
17007 (handle_missing_profiles): Update.
17008 (counts_to_freqs): Update.
17009 (rebuild_frequencies): Update.
17010 (force_edge_cold): Update.
17011 * predict.h: Include profile-count.h
17012 (maybe_hot_count_p, counts_to_freqs): UPdate.
17013 * print-rtl-function.c: Do not include cfg.h
17014 * print-rtl.c: Include basic-block.h
17015 * profile-count.c: New file.
17016 * profile-count.h: New file.
17017 * profile.c (is_edge_inconsistent): Update.
17018 (correct_negative_edge_counts): Update.
17019 (is_inconsistent): Update.
17020 (set_bb_counts): Update.
17021 (read_profile_edge_counts): Update.
17022 (compute_frequency_overlap): Update.
17023 (compute_branch_probabilities): Update; Initialize and deinitialize
17025 (branch_prob): Update.
17026 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
17027 (edge_gcov_count): New.
17028 (bb_gcov_count): New.
17029 * shrink-wrap.c (try_shrink_wrapping): Update.
17030 * tracer.c (better_p): Update.
17031 * trans-mem.c (expand_transaction): Update.
17032 (ipa_tm_insert_irr_call): Update.
17033 (ipa_tm_insert_gettmclone_call): Update.
17034 * tree-call-cdce.c: Update.
17035 * tree-cfg.c (gimple_duplicate_sese_region): Update.
17036 (gimple_duplicate_sese_tail): Update.
17037 (gimple_account_profile_record): Update.
17038 (execute_fixup_cfg): Update.
17039 * tree-inline.c (copy_bb): Update.
17040 (copy_edges_for_bb): Update.
17041 (initialize_cfun): Update.
17042 (freqs_to_counts): Update.
17043 (copy_cfg_body): Update.
17044 (expand_call_inline): Update.
17045 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
17046 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
17047 (try_unroll_loop_completely): Update.
17048 (try_peel_loop): Update.
17049 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17050 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
17051 * tree-ssa-loop-split.c (connect_loops): Update.
17052 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
17053 * tree-ssa-reassoc.c (branch_fixup): Update.
17054 * tree-ssa-tail-merge.c (replace_block_by): Update.
17055 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
17056 (compute_path_counts): Update.
17057 (update_profile): Update.
17058 (recompute_probabilities): Update.
17059 (update_joiner_offpath_counts): Update.
17060 (estimated_freqs_path): Update.
17061 (freqs_to_counts_path): Update.
17062 (clear_counts_path): Update.
17063 (ssa_fix_duplicate_block_edges): Update.
17064 (duplicate_thread_path): Update.
17065 * tree-switch-conversion.c (case_bit_test_cmp): Update.
17066 (struct switch_conv_info): Update.
17067 * tree-tailcall.c (decrease_profile): Update.
17068 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
17069 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17070 * value-prof.c (check_counter): Update.
17071 (gimple_divmod_fixed_value): Update.
17072 (gimple_mod_pow2): Update.
17073 (gimple_mod_subtract): Update.
17074 (gimple_ic_transform): Update.
17075 (gimple_stringop_fixed_value): Update.
17076 * value-prof.h (gimple_ic): Update.
17078 2017-06-02 Carl Love <cel@us.ibm.com>
17080 * config/rs6000/rs6000-c: Add support for built-in functions
17081 vector double vec_doublee (vector signed int);
17082 vector double vec_doublee (vector unsigned int);
17083 vector double vec_doublee (vector float);
17084 vector double vec_doubleh (vector signed int);
17085 vector double vec_doubleh (vector unsigned int);
17086 vector double vec_doubleh (vector float);
17087 vector double vec_doublel (vector signed int);
17088 vector double vec_doublel (vector unsigned int);
17089 vector double vec_doublel (vector float);
17090 vector double vec_doubleo (vector signed int);
17091 vector double vec_doubleo (vector unsigned int);
17092 vector double vec_doubleo (vector float);.
17093 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
17094 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
17096 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
17097 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
17098 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
17100 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
17101 vec_doublel, vec_doubleh.
17102 * doc/extend.texi: Update the built-in documentation file for the
17103 new built-in functions.
17105 2017-06-02 David Malcolm <dmalcolm@redhat.com>
17108 * ipa-inline-analysis.c (free_growth_caches): Set
17109 edge_removal_hook_holder to NULL after removing it.
17111 2017-06-02 Sudakshina Das <sudi.das@arm.com>
17113 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
17114 comparision with zero.
17116 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
17117 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
17118 for early expansion of vec_min and vec_max builtins.
17119 (builtin_function_type): Add min/max unsigned variants to those
17120 identified as having unsigned arguments.
17122 2017-06-02 Olivier Hainque <hainque@adacore.com>
17124 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
17126 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17128 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
17129 Use VALL_F16 iterator rather than VALL.
17131 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17133 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
17134 Emit CBNZ inside loop when doing a strong exchange and comparing
17135 against zero. Generate the CC flags after the loop.
17137 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
17139 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
17140 (dl_section_ref): New.
17141 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
17142 On AIX, append an expression to subtract the size of the
17143 section length to dl_section_ref.
17145 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
17147 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
17148 for early expansion of vector absolute builtins.
17150 2017-06-02 Richard Biener <rguenther@suse.de>
17152 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
17153 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
17155 2017-06-02 Richard Biener <rguenther@suse.de>
17157 PR tree-optimization/80948
17158 * tree-tailcall.c (find_tail_calls): Track stmts to move in
17159 stmt order as well.
17161 2017-06-02 Richard Biener <rguenther@suse.de>
17163 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
17165 * tree-vect-stmts.c (process_use): Do not mark backedge defs
17166 for inductions as relevant.
17168 2017-06-02 Richard Biener <rguenther@suse.de>
17170 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
17171 (vectorizable_induction): ... this. Remove dead code.
17173 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
17175 * builtins. (expand_builtin_alloca): Remove second parameter and
17176 infer its value from the first parameter instead.
17177 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
17179 2017-06-02 Jakub Jelinek <jakub@redhat.com>
17181 PR rtl-optimization/80903
17182 * loop-doloop.c (add_test): Unshare sequence.
17184 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
17186 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
17188 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
17190 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
17192 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
17193 xlogue_layout::get_instance, logue_layout::xlogue_layout,
17194 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
17195 (xlogue_layout::get_stub_rtx): Make static.
17196 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
17197 (xlogue_layout::compute_stub_managed_regs): Rename to...
17198 (xlogue_layout::count_stub_managed_regs): ...this.
17199 (xlogue_layout::is_stub_managed_reg): New function.
17200 (xlogue_layout::m_stub_names): Rename to...
17201 (xlogue_layout::s_stub_names): ...this, make static.
17202 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
17203 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
17204 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
17205 xlogue_layout::s_stub_names): Instantiate statics.
17206 (stub_managed_regs): Remove.
17207 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
17208 (disable_call_ms2sysv_xlogues): Rename to...
17209 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
17210 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
17212 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
17213 change after reload_completed.
17214 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
17216 (ix86_expand_prologue): Likewise.
17217 (ix86_expand_epilogue): Likewise.
17218 (ix86_expand_split_stack_prologue): Likewise.
17219 (ix86_compute_frame_layout): Remove frame parameter ...
17220 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
17221 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
17223 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
17224 (ix86_frame): Move from here ...
17225 * config/i386/i386.h (ix86_frame): ... to here.
17226 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
17227 complete ix86_frame data structure instead. Remove some_ld_name.
17229 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
17231 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
17232 symbols that hold a DECL_VALUE_EXPR.
17234 2017-06-01 Martin Jambor <mjambor@suse.cz>
17236 PR tree-optimization/80898
17237 * tree-sra.c (process_subtree_disqualification): Removed.
17238 (disqualify_candidate): Do not acll
17239 process_subtree_disqualification.
17240 (subtree_mark_written_and_enqueue): New function.
17241 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
17242 RHS has been disqualified and re-queue LHS if necessary. Apart
17243 from that, ignore disqualified RHS.
17245 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17247 * config/s390/s390.c (s390_emit_epilogue): Disable early return
17248 address fetch for z10 or later.
17250 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17252 * config/arc/arc.md (tst_movb): Add guard when splitting.
17254 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17256 * config/arc/arc.c (arc_can_eliminate): Test against
17257 arc_frame_pointer_needed.
17259 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17261 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
17262 to prevent store reordering.
17263 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
17264 (type): Add block type.
17265 (stack_tie): Define special instruction to be used in
17268 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17270 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
17271 constraint. It is not valid for the pattern.
17272 (noncommutative_binary_comparison): Likewise.
17274 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17276 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
17279 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17281 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
17282 be used by the reg-alloc.
17284 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17286 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
17287 reg-alloc when having mul64 or mul32x16 instructions.
17288 (mulsidi3): Likewise.
17289 (umulsidi3): Likewise.
17290 (mulsi32x16): New pattern.
17291 (mulsi64): Likewise.
17292 (mulsidi64): Likewise.
17293 (umulsidi64): Likewise.
17294 (MUL32x16_REG): Define.
17295 (mul64_600): Use MUL32x16_REG.
17296 (mac64_600): Likewise.
17297 (umul64_600): Likewise.
17298 (umac64_600): Likewise.
17300 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17302 * config/arc/arc.md (mulsi3_700): Make it commutative.
17304 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
17306 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
17308 (*sign_extendsidi2_insn): Likewise for movstosw.
17310 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
17312 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
17313 the type of the input discriminant value. Convert the
17314 discriminant value of signedness vary.
17316 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
17318 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
17319 Add to -Wall section.
17321 2017-06-01 Richard Biener <rguenther@suse.de>
17323 PR middle-end/66313
17324 * fold-const.c (fold_plusminus_mult_expr): If the factored
17325 factor may be zero use a wrapping type for the inner operation.
17326 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
17327 and handle moved defs.
17328 (process_assignment): Properly guard the unary op case. Return a
17329 tri-state indicating that moving the stmt before the call may allow
17330 to continue. Pass through to_move.
17331 (find_tail_calls): Handle moving unrelated defs before
17334 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
17337 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
17338 splitter result in the canonical way.
17340 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
17342 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
17343 also for 32bit target. Update insn attributes.
17344 (zero-extendsidi2 splitter): Allow all registers for operand 1.
17346 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
17348 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
17349 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
17350 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
17351 (_mm_maskz_min_ss): New intrinsics.
17353 2017-05-31 Martin Liska <mliska@suse.cz>
17355 * tree-vect-loop.c (vect_create_epilog_for_reduction):
17356 Change comment style to one we normally use.
17357 (vectorizable_reduction): Likewise.
17358 (vectorizable_induction): Likewise.
17359 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
17360 (vectorizable_call): Likewise.
17361 (vectorizable_simd_clone_call): Likewise.
17362 (vectorizable_conversion): Likewise.
17363 (vectorizable_assignment): Likewise.
17364 (vectorizable_shift): Likewise.
17365 (vectorizable_operation): Likewise.
17366 (vectorizable_store): Likewise.
17367 (vectorizable_load): Likewise.
17368 * tree-vectorizer.h: Likewise.
17370 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
17372 * passes.c (emergency_dump_function): New.
17373 * tree-pass.h (emergency_dump_function): Declare.
17374 * plugin.c (plugins_internal_error_function): Remove.
17375 * plugin.h (plugins_internal_error_function): Remove declaration.
17376 * toplev.c (internal_error_function): New static function. Use it...
17377 (general_init): ...here.
17379 2017-05-31 Graham Markall <graham.markall@embecosm.com>
17381 * config/arc/arc.c (arc_print_operand): Handle constant operands.
17382 (arc_rtx_costs): Add costs for new patterns.
17383 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
17384 * config/arc/predicates.md: Add _1_2_3_operand predicate.
17386 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
17388 * tree-ssa-strlen.c (get_next_strinfo): New function.
17389 (get_stridx_plus_constant): Use it.
17390 (zero_length_string): Likewise.
17391 (adjust_related_strinfos): Likewise.
17392 (adjust_last_stmt): Likewise.
17394 2017-05-31 Richard Biener <rguenther@suse.de>
17397 * config/i386/i386.c (ix86_expand_builtin): Remove assert
17398 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
17400 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
17402 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
17403 loop_vinfo argument and use of dependence distance vectors.
17404 Check instead whether the two references differ only in their
17405 initial value and assume that they have the same alignment if the
17406 difference is a multiple of the vector alignment.
17407 (vect_analyze_data_refs_alignment): Update call accordingly.
17409 2017-05-31 Martin Liska <mliska@suse.cz>
17412 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
17414 2017-05-31 Bin Cheng <bin.cheng@arm.com>
17416 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
17417 (create_intersect_range_checks): Move from ...
17418 * tree-data-ref.c (create_intersect_range_checks_index)
17419 (create_intersect_range_checks): ... to here.
17420 (create_runtime_alias_checks): New function factored from ...
17421 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
17422 here. Call above function.
17423 * tree-data-ref.h (create_runtime_alias_checks): New function.
17425 2017-05-31 Bin Cheng <bin.cheng@arm.com>
17427 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
17428 segment length for dr_b and compute it in wide_int.
17430 2017-05-31 Richard Biener <rguenther@suse.de>
17432 PR tree-optimization/80906
17433 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
17434 and pass through iv_map.
17435 (copy_bb_and_scalar_dependences): Adjust.
17436 (translate_pending_phi_nodes): Likewise.
17437 (copy_loop_close_phi_args): Handle code-generating IVs instead
17440 2017-05-30 David Malcolm <dmalcolm@redhat.com>
17442 * diagnostic-color.c (color_dict): Add "type-diff".
17443 (parse_gcc_colors): Update comment.
17444 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
17445 -fdiagnostics-show-template-tree and -fno-elide-type.
17446 (GCC_COLORS): Add type-diff to example.
17448 (-fdiagnostics-show-template-tree): New.
17449 (-fno-elide-type): New.
17450 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
17451 the pp_format_decoder callback. Call any m_format_postprocessor's
17453 (pretty_printer::pretty_printer): Initialize
17454 m_format_postprocessor.
17455 (pretty_printer::~pretty_printer): Delete any
17456 m_format_postprocessor.
17457 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
17458 (class format_postprocessor): New class.
17459 (struct pretty_printer::format_decoder): Document the new parameters.
17460 (struct pretty_printer::m_format_postprocessor): New field.
17461 * tree-diagnostic.c (default_tree_printer): Update for new
17462 bool and const char ** params.
17463 * tree-diagnostic.h (default_tree_printer): Likewise.
17465 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
17467 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
17468 (lwa_operand): Delete rs6000_gen_cell_microcode test.
17469 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
17470 rs6000_gen_cell_microcode code.
17471 (rs6000_final_prescan_insn): Delete.
17472 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
17473 "warn-cell-microcode" entries.
17474 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
17475 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
17476 throughout. Change cc_reg_not_micro_cr0_operand to
17477 cc_reg_not_cr0_operand throughout.
17478 (*extendhi<mode>2_noload): Delete.
17479 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
17480 (mwarn-cell-microcode): Delete.
17481 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
17482 -mgen-cell-microcode and -mwarn-cell-microcode.
17484 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
17487 * config/i386/constraints.md (Yd): New constraint.
17489 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
17490 and (?Yd, r) alternatives. Update insn attributes.
17491 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
17492 and (?*Yd, r) alternatives. Update insn attributes.
17493 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
17495 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
17497 * gimplify.c (gimplify_modify_expr): Don't create a
17498 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
17501 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
17503 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
17505 2017-05-30 Richard Biener <rguenther@suse.de>
17507 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
17508 and reduc_def fields.
17509 (STMT_VINFO_REDUC_TYPE): New define.
17510 (STMT_VINFO_REDUC_DEF): Likewise.
17511 (vect_force_simple_reduction): Adjust prototype.
17512 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
17513 (vect_is_simple_reduction): Remove check_reduction argument.
17514 (vect_force_simple_reduction): Adjust and set
17515 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
17516 (vectorizable_reduction): Do not re-do reduction analysis
17517 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
17518 * tree-parloops.c (gather_scalar_reductions): Adjust.
17520 2017-05-30 Richard Biener <rguenther@suse.de>
17522 PR middle-end/80901
17523 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
17526 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
17528 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
17529 Introduce unknown_misalignment parameter and remove vf.
17530 (vect_peeling_hash_get_lowest_cost):
17531 Pass unknown_misalignment parameter.
17532 (vect_enhance_data_refs_alignment):
17533 Fix unsupportable data ref treatment.
17535 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17537 * tree-vect-data-refs.c (vect_get_data_access_cost):
17538 Workaround for SLP handling.
17539 (vect_enhance_data_refs_alignment):
17540 Compute costs for doing no peeling at all, compare to the best
17541 peeling costs so far and avoid peeling if cheaper.
17543 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17545 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
17546 Return peeling info and set costs to zero for unlimited cost
17548 (vect_enhance_data_refs_alignment): Also inspect all datarefs
17549 with unknown misalignment. Compute and costs for unknown
17550 misalignment, compare them to the costs for known misalignment
17551 and choose the cheapest for peeling.
17553 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17555 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
17556 (vect_get_peeling_costs_all_drs): Create function.
17557 (vect_peeling_hash_get_lowest_cost):
17558 Use vect_get_peeling_costs_all_drs.
17559 (vect_peeling_supportable): Create function.
17560 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
17562 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17564 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
17565 DR_HAS_NEGATIVE_STEP.
17566 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
17567 (vect_enhance_data_refs_alignment): Use.
17568 (vect_duplicate_ssa_name_ptr_info): Use.
17569 * tree-vectorizer.h (dr_misalignment): Use.
17570 (known_alignment_for_access_p): Use.
17572 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17575 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
17577 (has_section_name): New function.
17579 2017-05-30 Martin Liska <mliska@suse.cz>
17582 * auto-profile.c (get_function_decl_from_block): Fix
17585 2017-05-30 Richard Biener <rguenther@suse.de>
17587 PR middle-end/80876
17588 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
17590 2017-05-30 Martin Liska <mliska@suse.cz>
17592 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
17593 * dumpfile.h (struct dump_file_info): Remove ctors.
17595 2017-05-30 Martin Liska <mliska@suse.cz>
17597 * predict.def: Fix GNU coding style.
17599 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
17601 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
17602 Mark 'to' argument with ATTRIBUTE_UNUSED.
17604 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
17606 * config/xtensa/xtensa.c (xtensa_emit_call): Use
17607 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
17608 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
17611 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
17613 * doc/install.texi (Options specification): Restore entry of
17614 --enable-sjlj-exceptions.
17616 2017-05-27 Michael Eager <eager@eagercon.com>
17619 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
17621 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
17623 * config/microblaze/microblaze.h
17624 (FIXED_REGISTERS): Update in macro.
17625 (CALL_USED_REGISTERS): Update in macro.
17627 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
17629 * doc/install.texi: Add links to macOS binary distributions.
17631 2017-05-27 Jakub Jelinek <jakub@redhat.com>
17635 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17637 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
17639 2017-05-26 Martin Liska <mliska@suse.cz>
17641 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
17643 2017-05-26 Martin Liska <mliska@suse.cz>
17645 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
17646 always leading ';; '.
17647 (dump_bb_info): Likewise.
17648 (brief_dump_cfg): Likewise.
17649 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
17650 * dumpfile.c: Remove usage of TDF_VERBOSE.
17651 * dumpfile.h (enum dump_kind): Likewise.
17652 (dump_gimple_bb_header): Do not use TDF_COMMENT.
17653 * print-tree.c (debug_verbose): Remove.
17654 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
17655 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
17656 * tree-diagnostic.c (default_tree_printer): Replace
17657 TDF_DIAGNOSTIC with TDF_SLIM.
17659 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17661 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
17662 in parameter loop, rather than loop_vinfo.
17663 (create_intersect_range_checks): Ditto.
17664 (vect_create_cond_for_alias_checks): Update call to above functions.
17666 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17668 PR tree-optimization/80815
17669 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
17670 for merging runtime alias checks. Handle negative DR_STEPs.
17672 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17674 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
17676 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
17677 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
17678 out code pruning runtime alias checks.
17679 * tree-data-ref.c (prune_runtime_alias_test_list): New function
17680 factored out from above.
17681 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
17683 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
17685 (prune_runtime_alias_test_list): New decalaration.
17687 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17689 * tree-vect-data-refs.c (compare_tree): Rename and move ...
17690 * tree-data-ref.c (data_ref_compare_tree): ... to here.
17691 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
17692 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
17693 (operator==, comp_dr_with_seg_len_pair): Ditto.
17694 (vect_prune_runtime_alias_test_list): Ditto.
17696 2017-05-26 Martin Liska <mliska@suse.cz>
17699 * params.def: Bound partial-inlining-entry-probability param.
17701 2017-05-26 Marek Polacek <polacek@redhat.com>
17704 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
17707 2017-05-26 Richard Biener <rguenther@suse.de>
17709 PR tree-optimization/80842
17710 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
17713 2017-05-26 Richard Biener <rguenther@suse.de>
17715 PR tree-optimization/80844
17716 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
17718 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
17720 * doc/md.texi (Machine Constraints): Update x86 family
17721 machine constraints section to match 'config/i386/constraints.md'.
17723 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
17725 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
17727 2017-05-25 Nathan Sidwell <nathan@acm.org>
17729 * doc/invoke.texi (--enable-languages): Update documentation.
17731 2017-05-25 Martin Liska <mliska@suse.cz>
17733 * dumpfile.c: Add TDF_FOLDING.
17734 * dumpfile.h (enum dump_kind): Likewise.
17735 * genmatch.c (dt_simplify::gen_1): Use it.
17737 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17739 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
17741 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17743 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
17744 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
17746 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17748 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
17749 * match.pd (X == C): Rewrite it here.
17750 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
17751 with_certain_nonzero_bits2): New predicates.
17752 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
17754 2017-05-24 Nathan Sidwell <nathan@acm.org>
17756 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
17759 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
17762 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
17764 * config/powerpcspe: New port. Files are copied from the rs6000
17765 port, with "rs6000" in filenames replaced by "powerpcspe".
17767 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
17769 PR rtl-optimization/80754
17770 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
17772 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
17774 * config/sparc/sparc.md (length): Return the correct value for -mflat
17775 sibcalls to match output_sibcall.
17777 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
17781 * config/rs6000/rs6000.c (struct machine_function): Add new field
17783 (rs6000_get_separate_components): Init that field, use it.
17784 (rs6000_components_for_bb): Use the field.
17786 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
17788 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
17790 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
17792 PR middle-end/80823
17793 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
17795 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17798 * config/s390/s390.c (s390_check_qrst_address): Check incoming
17799 address against address_operand predicate.
17800 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
17802 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
17804 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
17805 parameters passed indirectly.
17807 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
17809 * config/i386/i386.md (*movdi_internal): Remove SSE4
17810 alternative 18 (?r, *v). Update insn attributes.
17811 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
17812 Update insn attributes.
17813 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
17814 Update insn attributes.
17815 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
17816 alternative 1 (r, v). Remove isa attribute.
17817 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
17818 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
17819 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
17821 2017-05-23 Tom de Vries <tom@codesourcery.com>
17823 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
17826 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17828 * cgraphunit.c (symbol_table::process_new_functions): Update.
17829 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
17830 (inline_generate_summary): Rename to ...
17831 (ipa_fn_summary_generate): ... this one.
17832 (inline_read_summary): Rename to ...
17833 (ipa_fn_summary_read): ... this one.
17834 (inline_write_summary): Rename to ...
17835 (ipa_fn_summary_write): ... this one.
17836 (inline_free_summary): Rename to ...
17837 (ipa_free_fn_summary): ... this one.
17838 (pass_data_local_fn_summary, pass_local_fn_summary,
17839 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
17840 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
17841 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
17842 make_pass_ipa_fn_summary): New.
17843 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
17844 inline_write_summary, inline_free_summary): Remove.
17845 (ipa_free_fn_summary) : New.
17846 * ipa-inline.c (ipa_inline): Update.
17847 (pass_ipa_inline): Do not generate summaries.
17848 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
17850 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
17851 and add pass_ipa_fn_summary.
17852 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
17854 (make_pass_inline_parameters): Remove.
17856 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
17858 * omp-low.c (struct omp_context): Remove "default_kind" member.
17861 * omp-offload.c (execute_oacc_device_lower): Remove the
17862 parallelism dimensions function attributes for unparallelized
17863 OpenACC kernels constructs.
17865 2017-05-23 Martin Liska <mliska@suse.cz>
17867 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
17869 (cgraph_edge::make_speculative): Likewise.
17870 (cgraph_edge::resolve_speculation): Likewise.
17871 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
17872 (cgraph_node::dump): Likewise.
17873 * cgraph.h: Likewise.
17874 * cgraphunit.c (analyze_functions): Likewise.
17875 (symbol_table::compile): Likewise.
17876 * ipa-cp.c (print_all_lattices): Likewise.
17877 (determine_versionability): Likewise.
17878 (initialize_node_lattices): Likewise.
17879 (ipcp_verify_propagated_values): Likewise.
17880 (estimate_local_effects): Likewise.
17881 (update_profiling_info): Likewise.
17882 (create_specialized_node): Likewise.
17883 (perhaps_add_new_callers): Likewise.
17884 (decide_about_value): Likewise.
17885 (decide_whether_version_node): Likewise.
17886 (identify_dead_nodes): Likewise.
17887 (ipcp_store_bits_results): Likewise.
17888 * ipa-devirt.c (dump_targets): Likewise.
17889 (ipa_devirt): Likewise.
17890 * ipa-icf.c (sem_item::dump): Likewise.
17891 (sem_function::equals): Likewise.
17892 (sem_variable::equals): Likewise.
17893 (sem_item_optimizer::read_section): Likewise.
17894 (sem_item_optimizer::execute): Likewise.
17895 (congruence_class::dump): Likewise.
17896 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
17897 (dump_inline_summary): Likewise.
17898 (estimate_node_size_and_time): Likewise.
17899 (inline_analyze_function): Likewise.
17900 * ipa-inline-transform.c (inline_call): Likewise.
17901 * ipa-inline.c (report_inline_failed_reason): Likewise.
17902 (want_early_inline_function_p): Likewise.
17903 (edge_badness): Likewise.
17904 (update_edge_key): Likewise.
17905 (inline_small_functions): Likewise.
17906 * ipa-profile.c (ipa_profile): Likewise.
17907 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
17908 (ipa_make_edge_direct_to_target): Likewise.
17909 (remove_described_reference): Likewise.
17910 (ipa_impossible_devirt_target): Likewise.
17911 (propagate_controlled_uses): Likewise.
17912 (ipa_print_node_params): Likewise.
17913 (ipcp_transform_function): Likewise.
17914 * ipa-pure-const.c (pure_const_read_summary): Likewise.
17915 (propagate_pure_const): Likewise.
17916 * ipa-reference.c (generate_summary): Likewise.
17917 (read_write_all_from_decl): Likewise.
17918 (propagate): Likewise.
17919 (ipa_reference_read_optimization_summary): Likewise.
17920 * ipa-utils.c (ipa_merge_profiles): Likewise.
17921 * ipa.c (walk_polymorphic_call_targets): Likewise.
17922 (symbol_table::remove_unreachable_nodes): Likewise.
17923 (ipa_single_use): Likewise.
17924 * passes.c (execute_todo): Likewise.
17925 * predict.c (drop_profile): Likewise.
17926 * symtab.c (symtab_node::get_dump_name): New function.
17927 (symtab_node::dump_name): Likewise.
17928 (symtab_node::dump_asm_name): Likewise.
17929 (symtab_node::dump_references): Likewise.
17930 (symtab_node::dump_referring): Likewise.
17931 (symtab_node::dump_base): Likewise.
17932 (symtab_node::debug_symtab): Likewise.
17933 * tree-sra.c (convert_callers_for_node): Likewise.
17934 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
17935 * value-prof.c (init_node_map): Likewise.
17937 2017-05-23 Martin Liska <mliska@suse.cz>
17939 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
17940 and symtab_node::debug_symtab to symbol_table::debug.
17941 * cgraphunit.c (analyze_functions): Use the renamed function.
17942 (symbol_table::compile): Likewise.
17943 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
17944 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
17945 * passes.c (execute_todo): Likewise.
17946 * symtab.c (symbol_table::dump): New function.
17947 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
17949 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17951 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
17952 that nonconst implies exec.
17954 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17956 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
17957 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
17958 (inline_edge_summary_vec): Turn into ...
17959 (ipa_call_summaries): ... this one.
17960 (redirect_to_unreachable, edge_set_predicate,
17961 evaluate_properties_for_edge, inline_summary_alloc,
17962 reset_ipa_call_summary, reset_inline_summary,
17963 inline_summary_t::duplicate): Update.
17964 (inline_edge_duplication_hook): Turn to ...
17965 (ipa_call_summary_t::duplicate): ... this one.
17966 (inline_edge_removal_hook): Turn to ...
17967 (ipa_call_summary_t::remove): ... this one.
17968 (dump_inline_edge_summary): Turn to ...
17969 (dump_ipa_call_summary): ... this one.
17970 (estimate_function_body_sizes): Update.
17971 (inline_update_callee_summaries): Update.
17972 (remap_edge_change_prob): Update.
17973 (remap_edge_summaries): Update.
17974 (inline_merge_summary): Update.
17975 (do_estimate_edge_time): Update.
17976 (inline_generate_summary): Update.
17977 (inline_read_section): Update.
17978 (inline_read_summary): Update.
17979 (inline_free_summary): Update.
17980 * ipa-inline.c (can_inline_edge_p): Update.
17981 (compute_inlined_call_time): Update.
17982 (want_inline_small_function_p): Update.
17983 (edge_badness): Update.
17984 (early_inliner): Update.
17985 * ipa-inline.h (inline_edge_summary): Turn to ...
17986 (ipa_call_summary): ... this one.
17987 (ipa_call_summary_t): New class.
17988 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
17989 (ipa_call_summaries): New.
17990 (inline_edge_summary): Remove.
17991 (estimate_edge_growth): Update.
17992 * ipa-profile.c (ipa_propagate_frequency_1): Update.
17993 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
17994 * ipa-split.c (execute_split_functions): Update.
17995 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
17997 2017-05-23 Tom de Vries <tom@codesourcery.com>
17999 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
18000 attributes): Document rdrand effective target.
18002 2017-05-23 Tom de Vries <tom@codesourcery.com>
18004 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
18005 attributes): Sort alphabetically.
18007 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
18009 * config/avr/genmultilib.awk: Use gsub instead of gensub.
18011 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
18014 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
18015 V2DF/V2DI splat into two separate patterns, one that handles
18016 registers, and the other that only handles memory. Drop support
18017 for splatting from a GPR on ISA 2.07 and then splitting the
18018 splat into direct move and splat.
18019 (vsx_splat_<mode>_reg): Likewise.
18020 (vsx_splat_<mode>_mem): Likewise.
18022 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
18024 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
18026 2017-05-22 Jakub Jelinek <jakub@redhat.com>
18028 PR middle-end/80809
18029 * omp-low.c (finish_taskreg_remap): New function.
18030 (finish_taskreg_scan): If unit size of ctx->record_type
18031 is non-constant, unshare the size expression and replace
18032 decls in it with possible outer var refs.
18034 PR middle-end/80809
18035 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
18036 GOVD_SHARED rather than GOVD_PRIVATE with it.
18037 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
18038 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
18040 PR middle-end/80853
18041 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
18042 as last argument to build_outer_var_ref for pointer bases of array
18043 section reductions.
18045 2017-05-19 Martin Sebor <msebor@redhat.com>
18047 * print-tree.c (print_node): Print DECL_READ_P flag.
18049 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18051 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
18052 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
18053 * cgraph.c: Likewise.
18054 * cgraphunit.c: Likewise.
18055 * gengtype.c: Likewise.
18056 * ipa-cp.c: Likewise.
18057 * ipa-devirt.c: Likewise.
18058 * ipa-icf.c: Likewise.
18059 * ipa-predicate.c: Likewise.
18060 * ipa-profile.c: Likewise.
18061 * ipa-prop.c: Likewise.
18062 * ipa-split.c: Likewise.
18064 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
18065 edge_predicate_pool, dump_inline_hints,
18066 inline_summary::account_size_time, redirect_to_unreachable,
18067 edge_set_predicate, set_hint_predicate,
18068 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
18069 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
18070 inline_summary_t::remove, remap_hint_predicate_after_duplication,
18071 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
18072 ipa_call_summary_t::remove, initialize_growth_caches,
18073 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
18074 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
18075 mark_modified, unmodified_parm_1, unmodified_parm,
18076 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
18077 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
18078 compute_bb_predicates, will_be_nonconstant_expr_predicate,
18079 will_be_nonconstant_predicate, record_modified_bb_info,
18080 get_minimal_bb, record_modified, param_change_prob,
18081 phi_result_unknown_predicate, predicate_for_phi_result,
18082 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
18083 estimate_function_body_sizes, compute_inline_parameters,
18084 compute_inline_parameters_for_curren, pass_data_inline_parameters,
18085 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
18086 inline_update_callee_summaries, remap_edge_change_prob,
18087 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
18088 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
18089 inline_analyze_function, inline_summary_t::insert,
18090 inline_generate_summary, read_ipa_call_summary, inline_read_section,
18091 inline_read_summary, write_ipa_call_summary, inline_write_summary,
18092 inline_free_summary): Move to ipa-fnsummary.h
18093 (predicate_t): Remove.
18094 * ipa-fnsummary.c: New file.
18095 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
18096 (enum inline_hints_vals, inline_hints, agg_position_info,
18097 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
18098 inline_summaries, ipa_call_summary, ipa_call_summary_t,
18099 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
18100 dump_inline_summary, dump_inline_hints, inline_generate_summary,
18101 inline_read_summary, inline_write_summary, inline_free_summary,
18102 inline_analyze_function, initialize_inline_failed,
18103 inline_merge_summary, inline_update_overall_summary,
18104 compute_inline_parameters): Move to ipa-fnsummary.h
18105 * ipa-fnsummary.h: New file.
18106 * ipa-inline-transform.h: Include ipa-inline.h.
18107 * ipa-inline.c: LIkewise.
18109 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18111 * ipa-inline.c (edge_badness): Use inlined_time instead of
18112 inline_summaries->get.
18114 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18116 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
18118 2017-05-22 Nathan Sidwell <nathan@acm.org>
18120 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
18121 (fdump-lang): Document 'raw' option.
18122 * dumpfile.h (TDI_tu): Delete.
18123 * dumpfile.c (dump_files): Remove translation-unit.
18124 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
18126 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
18128 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
18129 command option from $(AWK) call.
18130 * config/avr/genmultilib.awk: Simplify and rewrite so that it
18131 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
18132 [FORMAT]: Remove handling of variable.
18133 * config/avr/t-multilib: Regenerate.
18135 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18137 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
18139 (dump_inline_summary): Do not print self_time.
18140 (estimate_function_body_sizes): Do not set self_time.
18141 (compute_inline_parameters): Likewise.
18142 (inline_read_section, inline_write_summary): Do not stream self_time.
18143 * ipa-inline.h (inline_summary): Drop self_time.
18145 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18147 * ipa-inline-analysis.c (account_size_time): Rename to ...
18148 (inline_summary::account_size_time): ... this one.
18149 (reset_ipa_call_summary): Turn to ...
18150 (ipa_call_summary::reset): ... this one.
18151 (reset_inline_summary): Turn to ...
18152 (inline_summary::reset): ... this one.
18153 (inline_summary_t::remove): Update.
18154 (inline_summary_t::duplicate): Update.
18155 (ipa_call_summary_t::remove): Update.
18156 (dump_inline_summary): Update.
18157 (estimate_function_body_sizes): Update.
18158 (compute_inline_parameters): Update.
18159 (estimate_node_size_and_time): Update.
18160 (inline_merge_summary): Update.
18161 (inline_update_overall_summary): Update.
18162 (inline_read_section): Update.
18163 (inline_write_summary): Update.
18164 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
18165 add account_size_time and reset member functions.
18166 (ipa_call_summary): Add reset function.
18167 * ipa-predicate.h (predicate::operator &): Constify.
18169 2017-05-22 Richard Biener <rguenther@suse.de>
18171 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
18173 2017-05-19 Jason Merrill <jason@redhat.com>
18175 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
18177 2017-05-19 Marek Polacek <polacek@redhat.com>
18180 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
18181 TYPE_OVERFLOW_WRAPS checks.
18183 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
18185 * tree-core.h (enum omp_clause_default_kind): Add
18186 "OMP_CLAUSE_DEFAULT_PRESENT".
18187 * tree-pretty-print.c (dump_omp_clause): Handle it.
18188 * gimplify.c (enum gimplify_omp_var_data): Add
18189 "GOVD_MAP_FORCE_PRESENT".
18190 (gimplify_adjust_omp_clauses_1): Map it to
18191 "GOMP_MAP_FORCE_PRESENT".
18192 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
18194 * gimplify.c (oacc_default_clause): Clarify.
18196 2017-05-19 Nathan Sidwell <nathan@acm.org>
18198 LANG_HOOK_REGISTER_DUMPS
18199 * toplev.c (general_init): Call register dump lang hook.
18200 * doc/invoke.texi: Document -fdump-lang option family.
18201 * dumpfile.c (dump_files): Remove class dump here.
18202 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
18203 * dumpfile.h (tree_dump_index): Remove TDI_class.
18204 * langhooks-def.h (lhd_register_dumps): Declare.
18205 (LANG_HOOKS_REGISTER_DUMPS): Define.
18206 (LANG_HOOKS_INITIALIZER): Add it.
18207 * langhooks.c (lhd_register_dumps): Define.
18208 * langhooks.h (struct lang_hooks): Add register_dumps.
18210 2017-05-19 Nathan Sidwell <nathan@acm.org>
18212 * context.h (context::set_passes): New.
18213 * context.c (context::context): Do not create pass manager.
18214 * toplev.c (general_init): Create pass manager here.
18216 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
18218 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
18219 use this splitter if two add or or instructions would also work for
18220 the constant we want to generate.
18222 2017-05-19 Richard Biener <rguenther@suse.de>
18225 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
18226 predicate evaluation.
18228 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
18230 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
18232 * ipa-inline.c (want_inline_small_function_p): Do not cast to
18235 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
18237 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
18238 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
18239 (inline_edge_summary_vec): Turn into ...
18240 (ipa_call_summaries): ... this one.
18241 (redirect_to_unreachable, edge_set_predicate,
18242 evaluate_properties_for_edge, inline_summary_alloc,
18243 reset_ipa_call_summary, reset_inline_summary,
18244 inline_summary_t::duplicate): Update.
18245 (inline_edge_duplication_hook): Turn to ...
18246 (ipa_call_summary_t::duplicate): ... this one.
18247 (inline_edge_removal_hook): Turn to ...
18248 (ipa_call_summary_t::remove): ... this one.
18249 (dump_inline_edge_summary): Turn to ...
18250 (dump_ipa_call_summary): ... this one.
18251 (estimate_function_body_sizes): Update.
18252 (inline_update_callee_summaries): Update.
18253 (remap_edge_change_prob): Update.
18254 (remap_edge_summaries): Update.
18255 (inline_merge_summary): Update.
18256 (do_estimate_edge_time): Update.
18257 (inline_generate_summary): Update.
18258 (inline_read_section): Update.
18259 (inline_read_summary): Update.
18260 (inline_free_summary): Update.
18261 * ipa-inline.c (can_inline_edge_p): Update.
18262 (compute_inlined_call_time): Update.
18263 (want_inline_small_function_p): Update.
18264 (edge_badness): Update.
18265 (early_inliner): Update.
18266 * ipa-inline.h (inline_edge_summary): Turn to ...
18267 (ipa_call_summary): ... this one.
18268 (ipa_call_summary_t): New class.
18269 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
18270 (ipa_call_summaries): New.
18271 (inline_edge_summary): Remove.
18272 (estimate_edge_growth): Update.
18273 * ipa-profile.c (ipa_propagate_frequency_1): Update.
18274 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
18275 * ipa-split.c (execute_split_functions): Update.
18276 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
18278 2017-05-19 Richard Biener <rguenther@suse.de>
18280 PR middle-end/80764
18281 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
18283 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
18285 * config/rs6000/rs6000.c (struct machine_function): Add field
18286 fpr_is_wrapped_separately.
18287 (rs6000_get_separate_components): Use 64 components. Handle the
18288 new FPR components.
18289 (rs6000_components_for_bb): Handle the FPR components.
18290 (rs6000_emit_prologue_components): Handle the FPR components.
18291 (rs6000_emit_epilogue_components): Handle the FPR components.
18292 (rs6000_set_handled_components): Handle the FPR components.
18293 (rs6000_emit_prologue): Don't output prologue code for those FPRs
18294 that are already separately shrink-wrapped.
18295 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
18296 that are already separately shrink-wrapped.
18298 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
18301 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
18304 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
18305 (define_peephole2 for Altivec d-form load): Add peepholes to catch
18306 cases where the register allocator uses a move and an offsettable
18307 memory operation to/from a FPR register on ISA 2.06/2.07.
18308 (define_peephole2 for Altivec d-form store): Likewise.
18310 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
18313 * config/i386/mmx.md (*mov<mode>_internal): Enable
18314 alternatives 11, 12, 13 and 14 also for 32bit targets.
18315 Remove alternatives 15, 16, 17 and 18.
18316 * config/i386/sse.md (vec_concatv2di): Change
18317 alternative (!x, *y) to (x, ?!*Yn).
18319 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
18321 * dumpfile.h (enum dump_kind): Remove stray comma.
18323 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
18325 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
18326 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
18327 predicate::num_conditions
18328 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
18329 (CHANGED): turn into predicate::changed.
18330 (agg_position_info): Move to ipa-predicate.h
18331 (add_condition, predicate::add_clause, predicate::operator &=,
18332 predicate::or_with, predicate::evaluate, predicate::probability,
18333 dump_condition, dump_clause, predicate::dump,
18334 predicate::remap_after_duplication, predicate::remap_after_inlining,
18335 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
18336 (evaluate_conditions_for_known_args): Update.
18337 (set_cond_stmt_execution_predicate): Update.
18338 * ipa-inline.h: Include ipa-predicate.h
18339 (condition, inline_param_summary, conditions, agg_position_info,
18340 predicate): Move to ipa-predicate.h
18341 * ipa-predicate.c: New file.
18342 * ipa-predicate.h: New file.
18344 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
18346 * final.c (leaf_function_p): Check we are not in a sequence.
18348 2017-05-18 Martin Liska <mliska@suse.cz>
18350 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
18351 * dumpfile.c (dump_register): Use new enum dump_kind.
18352 (get_dump_file_name): Likewise.
18353 (dump_enable_all): Likewise.
18354 (dump_switch_p_1): Likewise.
18355 (enable_rtl_dump_file): Remove usage of TDF_RTL.
18356 * dumpfile.h (enum dump_kind): New enum type.
18357 (struct dump_file_info): Create constructor and
18358 format fields and comments.
18359 * passes.c (pass_manager::register_one_dump_file):
18361 * statistics.c (statistics_early_init): Likewise.
18362 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
18363 TDF_TREE with TDF_SLIM.
18364 (gather_memory_references_ref): Likewise.
18366 2017-05-18 Martin Liska <mliska@suse.cz>
18368 * vec.h (struct vnull): Use it.
18370 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
18372 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
18373 (true_predicate, false_predicate, true_predicate_p,
18374 false_predicate_p): Remove.
18375 (single_cond_predicate, not_inlined_predicate): Turn to member function
18377 (add_condition): Update.
18378 (add_clause): Turn to...
18379 (predicate::add_clause): ... this one; update; allow passing NULL
18381 (and_predicates): Turn to ...
18382 (predicate::operator &=): ... this one.
18383 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
18384 (or_predicates): Turn to ...
18385 (predicate::or_with): ... this one.
18386 (evaluate_predicate): Turn to ...
18387 (predicate::evaluate): ... this one.
18388 (predicate_probability): Turn to ...
18389 (predicate::probability): ... this one.
18390 (dump_condition): Update.
18391 (dump_predicate): Turn to ...
18392 (predicate::dump): ... this one.
18393 (account_size_time): Update.
18394 (edge_set_predicate): Update.
18395 (set_hint_predicate): UPdate.
18396 (evaluate_conditions_for_known_args): Update.
18397 (evaluate_properties_for_edge): Update.
18398 (remap_predicate_after_duplication): Turn to...
18399 (predicate::remap_after_duplication): ... this one.
18400 (remap_hint_predicate_after_duplication): Update.
18401 (inline_summary_t::duplicate): UPdate.
18402 (dump_inline_edge_summary): Update.
18403 (dump_inline_summary): Update.
18404 (set_cond_stmt_execution_predicate): Update.
18405 (set_switch_stmt_execution_predicate): Update.
18406 (compute_bb_predicates): Update.
18407 (will_be_nonconstant_expr_predicate): Update.
18408 (will_be_nonconstant_predicate): Update.
18409 (phi_result_unknown_predicate): Update.
18410 (predicate_for_phi_result): Update.
18411 (array_index_predicate): Update.
18412 (estimate_function_body_sizes): Update.
18413 (estimate_node_size_and_time): Update.
18414 (estimate_ipcp_clone_size_and_time): Update.
18415 (remap_predicate): Rename to ...
18416 (predicate::remap_after_inlining): ... this one.
18417 (remap_hint_predicate): Update.
18418 (inline_merge_summary): Update.
18419 (inline_update_overall_summary): Update.
18420 (estimate_size_after_inlining): Update.
18421 (read_predicate): Rename to ...
18422 (predicate::stream_in): ... this one.
18423 (read_inline_edge_summary): Update.
18424 (write_predicate): Rename to ...
18425 (predicate::stream_out): ... this one.
18426 (write_inline_edge_summary): Update.
18427 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
18428 (clause_t): Turn to uint32_t
18429 (predicate): Turn to class; implement constructor and operators
18431 (size_time_entry): Update.
18432 (inline_summary): Update.
18433 (inline_edge_summary): Update.
18435 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
18437 * fold-const.c (fold_binary_loc): Move transformation...
18438 * match.pd (C - X CMP X): ... here.
18440 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
18442 * config/sparc/sparc.c (sparc_option_override): Set function
18443 alignment for -mcpu=niagara7 to 64 to match the I$ line.
18444 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
18446 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
18448 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
18450 2017-05-18 Marek Polacek <polacek@redhat.com>
18453 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
18454 (pass_ubsan::execute): Call gimple_assign_single_p instead of
18455 gimple_assign_load_p.
18457 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
18459 PR middle-end/80692
18460 * real.c (do_compare): Give decimal_do_compare preference over
18461 comparing just the signs.
18463 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
18465 * doc/md.texi (Canonicalization of Instructions): Describe the
18466 canonical form of instructions that inherently set a condition
18469 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
18471 PR middle-end/80775
18472 * tree-cfg.c: Move deletion of unreachable case statements to after
18473 the merging of consecutive case labels.
18475 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
18477 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
18478 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
18479 restoring of callee-saved registers.
18481 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
18483 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
18484 * config/visium/visium.c (single_set_and_flags): Likewise.
18485 * config/visium/visium.md (Substitutions): Likewise.
18487 2017-05-17 Martin Liska <mliska@suse.cz>
18489 * cfg.c: Introduce dump_flags_t type and
18490 use it instead of int type.
18492 * cfghooks.c: Likewise.
18493 * cfghooks.h (struct cfg_hooks): Likewise.
18494 * cfgrtl.c: Likewise.
18495 * cfgrtl.h: Likewise.
18496 * cgraph.c (cgraph_node::get_body): Likewise.
18497 * coretypes.h: Likewise.
18498 * domwalk.c: Likewise.
18499 * domwalk.h: Likewise.
18500 * dumpfile.c (struct dump_option_value_info): Likewise.
18501 (dump_enable_all): Likewise.
18502 (dump_switch_p_1): Likewise.
18503 (opt_info_switch_p): Likewise.
18504 * dumpfile.h (enum tree_dump_index): Likewise.
18505 (struct dump_file_info): Likewise.
18506 * genemit.c: Likewise.
18507 * generic-match-head.c: Likewise.
18508 * gengtype.c (open_base_files): Likewise.
18509 * gimple-pretty-print.c: Likewise.
18510 * gimple-pretty-print.h: Likewise.
18511 * graph.c (print_graph_cfg): Likewise.
18512 * graphite-scop-detection.c (dot_all_sese): Likewise.
18513 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
18514 * loop-unroll.c (report_unroll): Likewise.
18515 * passes.c (pass_manager::register_one_dump_file): Likewise.
18516 * print-tree.c: Likewise.
18517 * statistics.c: Likewise.
18518 * tree-cfg.c: Likewise.
18519 * tree-cfg.h: Likewise.
18520 * tree-dfa.c: Likewise.
18521 * tree-dfa.h: Likewise.
18522 * tree-dump.c (dump_function): Likewise.
18523 * tree-dump.h (struct dump_info): Likewise.
18524 * tree-pretty-print.c: Likewise.
18525 * tree-pretty-print.h: Likewise.
18526 * tree-ssa-live.c: Likewise.
18527 * tree-ssa-live.h: Likewise.
18528 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
18529 * tree-vect-loop.c: Likewise.
18530 * tree-vect-slp.c: Likewise.
18532 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
18533 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18535 PR tree-optimization/80457
18536 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
18537 of all arguments to a statement as scalar_to_vec operations.
18538 (vectorizable_call): Adjust call to vect_model_simple_cost for
18540 (vectorizable_conversion): Likewise.
18541 (vectorizable_assignment): Likewise.
18542 (vectorizable_shift): Likewise.
18543 (vectorizable_operation): Likewise.
18544 (vectorizable_comparison): Likewise.
18545 (vect_is_simple_cond): Record the def types for operands.
18546 (vectorizable_condition): Likewise, call vect_model_simple_cost.
18547 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
18548 for statement argument count.
18550 2017-05-16 Carl Love <cel@us.ibm.com>
18552 * config/rs6000/rs6000-c: Add support for built-in functions
18553 vector unsigned long long vec_bperm (vector unsigned long long,
18554 vector unsigned char)
18555 vector signed long long vec_mule (vector signed int,
18557 vector unsigned long long vec_mule (vector unsigned int,
18558 vector unsigned int)
18559 vector signed long long vec_mulo (vector signed int,
18561 vector unsigned long long vec_mulo (vector unsigned int,
18562 vector unsigned int)
18563 vector signed char vec_sldw (vector signed char,
18564 vector signed char,
18566 vector unsigned char vec_sldw (vector unsigned char,
18567 vector unsigned char,
18569 vector signed short vec_sldw (vector signed short,
18570 vector signed short,
18572 vector unsigned short vec_sldw (vector unsigned short,
18573 vector unsigned short,
18575 vector signed int vec_sldw (vector signed int,
18578 vector unsigned int vec_sldw (vector unsigned int,
18579 vector unsigned int,
18581 vector signed long long vec_sldw (vector signed long long,
18582 vector signed long long,
18584 vector unsigned long long vec_sldw (vector unsigned long long,
18585 vector unsigned long long,
18587 * config/rs6000/rs6000-c: Add support for built-in functions
18588 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
18589 * config/rs6000/altivec.h: Add defintion for vec_sldw.
18590 * doc/extend.texi: Update the built-in documentation for the
18591 new built-in functions.
18593 2017-05-16 Marek Polacek <polacek@redhat.com>
18597 * tree.c (save_expr): Don't fold the expression.
18599 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
18601 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
18602 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
18603 and (?*y,m). Update insn attributes.
18605 2017-05-16 Martin Liska <mliska@suse.cz>
18607 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
18608 flags argument of print_gimple_stmt, print_gimple_expr,
18609 print_generic_stmt and print_generic_expr.
18610 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
18611 * coretypes.h: Likewise.
18612 * except.c (dump_eh_tree): Likewise.
18613 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
18614 * gimple-pretty-print.h: Likewise.
18615 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
18616 (backprop::push_to_worklist): Likewise.
18617 (backprop::pop_from_worklist): Likewise.
18618 (backprop::process_use): Likewise.
18619 (backprop::intersect_uses): Likewise.
18620 (note_replacement): Likewise.
18621 * gimple-ssa-store-merging.c
18622 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
18623 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
18624 (pass_store_merging::execute): Likewise.
18625 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
18626 (ssa_base_cand_dump_callback): Likewise.
18627 (dump_incr_vec): Likewise.
18628 (replace_refs): Likewise.
18629 (replace_mult_candidate): Likewise.
18630 (create_add_on_incoming_edge): Likewise.
18631 (create_phi_basis): Likewise.
18632 (insert_initializers): Likewise.
18633 (all_phi_incrs_profitable): Likewise.
18634 (introduce_cast_before_cand): Likewise.
18635 (replace_one_candidate): Likewise.
18636 * gimplify.c (gimplify_expr): Likewise.
18637 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
18638 (set_rename): Likewise.
18639 (rename_uses): Likewise.
18640 (copy_loop_phi_nodes): Likewise.
18641 (add_close_phis_to_merge_points): Likewise.
18642 (copy_loop_close_phi_args): Likewise.
18643 (copy_cond_phi_args): Likewise.
18644 (graphite_copy_stmts_from_block): Likewise.
18645 (translate_pending_phi_nodes): Likewise.
18646 * graphite-poly.c (print_pdr): Likewise.
18647 (dump_gbb_cases): Likewise.
18648 (dump_gbb_conditions): Likewise.
18649 (print_scop_params): Likewise.
18650 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
18651 (build_cross_bb_scalars_use): Likewise.
18652 (gather_bbs::before_dom_children): Likewise.
18653 * hsa-dump.c (dump_hsa_immed): Likewise.
18654 * ipa-cp.c (print_ipcp_constant_value): Likewise.
18655 (get_replacement_map): Likewise.
18656 * ipa-inline-analysis.c (dump_condition): Likewise.
18657 (estimate_function_body_sizes): Likewise.
18658 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
18659 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
18660 * ipa-prop.c (ipa_dump_param): Likewise.
18661 (ipa_print_node_jump_functions_for_edge): Likewise.
18662 (ipa_modify_call_arguments): Likewise.
18663 (ipa_modify_expr): Likewise.
18664 (ipa_dump_param_adjustments): Likewise.
18665 (ipa_dump_agg_replacement_values): Likewise.
18666 (ipcp_modif_dom_walker::before_dom_children): Likewise.
18667 * ipa-pure-const.c (check_stmt): Likewise.
18668 (pass_nothrow::execute): Likewise.
18669 * ipa-split.c (execute_split_functions): Likewise.
18670 * omp-offload.c (dump_oacc_loop_part): Likewise.
18671 (dump_oacc_loop): Likewise.
18672 * trans-mem.c (tm_log_emit): Likewise.
18673 (tm_memopt_accumulate_memops): Likewise.
18674 (dump_tm_memopt_set): Likewise.
18675 (dump_tm_memopt_transform): Likewise.
18676 * tree-cfg.c (gimple_verify_flow_info): Likewise.
18677 (print_loop): Likewise.
18678 * tree-chkp-opt.c (chkp_print_addr): Likewise.
18679 (chkp_gather_checks_info): Likewise.
18680 (chkp_get_check_result): Likewise.
18681 (chkp_remove_check_if_pass): Likewise.
18682 (chkp_use_outer_bounds_if_possible): Likewise.
18683 (chkp_reduce_bounds_lifetime): Likewise.
18684 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
18685 (chkp_mark_completed_bounds): Likewise.
18686 (chkp_register_incomplete_bounds): Likewise.
18687 (chkp_mark_invalid_bounds): Likewise.
18688 (chkp_maybe_copy_and_register_bounds): Likewise.
18689 (chkp_build_returned_bound): Likewise.
18690 (chkp_get_bound_for_parm): Likewise.
18691 (chkp_build_bndldx): Likewise.
18692 (chkp_get_bounds_by_definition): Likewise.
18693 (chkp_generate_extern_var_bounds): Likewise.
18694 (chkp_get_bounds_for_decl_addr): Likewise.
18695 * tree-chrec.c (chrec_apply): Likewise.
18696 * tree-data-ref.c (dump_data_reference): Likewise.
18697 (dump_subscript): Likewise.
18698 (dump_data_dependence_relation): Likewise.
18699 (analyze_overlapping_iterations): Likewise.
18700 * tree-inline.c (expand_call_inline): Likewise.
18701 (tree_function_versioning): Likewise.
18702 * tree-into-ssa.c (dump_defs_stack): Likewise.
18703 (dump_currdefs): Likewise.
18704 (dump_names_replaced_by): Likewise.
18705 (dump_update_ssa): Likewise.
18706 (update_ssa): Likewise.
18707 * tree-object-size.c (pass_object_sizes::execute): Likewise.
18708 * tree-parloops.c (build_new_reduction): Likewise.
18709 (try_create_reduction_list): Likewise.
18710 (ref_conflicts_with_region): Likewise.
18711 (oacc_entry_exit_ok_1): Likewise.
18712 (oacc_entry_exit_single_gang): Likewise.
18713 * tree-pretty-print.h: Likewise.
18714 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
18715 (get_scalar_evolution): Likewise.
18716 (add_to_evolution): Likewise.
18717 (get_loop_exit_condition): Likewise.
18718 (analyze_evolution_in_loop): Likewise.
18719 (analyze_initial_condition): Likewise.
18720 (analyze_scalar_evolution): Likewise.
18721 (instantiate_scev): Likewise.
18722 (number_of_latch_executions): Likewise.
18723 (gather_chrec_stats): Likewise.
18724 (final_value_replacement_loop): Likewise.
18725 (scev_const_prop): Likewise.
18726 * tree-sra.c (dump_access): Likewise.
18727 (disqualify_candidate): Likewise.
18728 (create_access): Likewise.
18729 (reject): Likewise.
18730 (maybe_add_sra_candidate): Likewise.
18731 (create_access_replacement): Likewise.
18732 (analyze_access_subtree): Likewise.
18733 (analyze_all_variable_accesses): Likewise.
18734 (sra_modify_assign): Likewise.
18735 (initialize_constant_pool_replacements): Likewise.
18736 (find_param_candidates): Likewise.
18737 (decide_one_param_reduction): Likewise.
18738 (replace_removed_params_ssa_names): Likewise.
18739 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
18740 * tree-ssa-copy.c (dump_copy_of): Likewise.
18741 (copy_prop_visit_cond_stmt): Likewise.
18742 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
18743 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
18744 (record_equivalences_from_stmt): Likewise.
18745 * tree-ssa-dse.c (compute_trims): Likewise.
18746 (delete_dead_call): Likewise.
18747 (delete_dead_assignment): Likewise.
18748 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
18749 (forward_propagate_into_cond): Likewise.
18750 (pass_forwprop::execute): Likewise.
18751 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
18752 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
18754 (move_computations_worker): Likewise.
18755 (execute_sm): Likewise.
18756 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
18757 (remove_exits_and_undefined_stmts): Likewise.
18758 (remove_redundant_iv_tests): Likewise.
18759 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
18760 (adjust_iv_update_pos): Likewise.
18761 * tree-ssa-math-opts.c (bswap_replace): Likewise.
18762 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
18763 (value_replacement): Likewise.
18764 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
18765 * tree-ssa-pre.c (print_pre_expr): Likewise.
18766 (get_representative_for): Likewise.
18767 (create_expression_by_pieces): Likewise.
18768 (insert_into_preds_of_block): Likewise.
18769 (eliminate_insert): Likewise.
18770 (eliminate_dom_walker::before_dom_children): Likewise.
18771 (eliminate): Likewise.
18772 (remove_dead_inserted_code): Likewise.
18773 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
18774 * tree-ssa-reassoc.c (get_rank): Likewise.
18775 (eliminate_duplicate_pair): Likewise.
18776 (eliminate_plus_minus_pair): Likewise.
18777 (eliminate_not_pairs): Likewise.
18778 (undistribute_ops_list): Likewise.
18779 (eliminate_redundant_comparison): Likewise.
18780 (update_range_test): Likewise.
18781 (optimize_range_tests_var_bound): Likewise.
18782 (optimize_vec_cond_expr): Likewise.
18783 (rewrite_expr_tree): Likewise.
18784 (rewrite_expr_tree_parallel): Likewise.
18785 (linearize_expr): Likewise.
18786 (break_up_subtract): Likewise.
18787 (linearize_expr_tree): Likewise.
18788 (attempt_builtin_powi): Likewise.
18789 (attempt_builtin_copysign): Likewise.
18790 (transform_stmt_to_copy): Likewise.
18791 (transform_stmt_to_multiply): Likewise.
18792 (dump_ops_vector): Likewise.
18793 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
18794 (print_scc): Likewise.
18795 (set_ssa_val_to): Likewise.
18796 (visit_reference_op_store): Likewise.
18797 (visit_use): Likewise.
18798 (sccvn_dom_walker::before_dom_children): Likewise.
18799 (run_scc_vn): Likewise.
18800 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
18802 (expr_hash_elt::print): Likewise.
18803 (const_and_copies::pop_to_marker): Likewise.
18804 (const_and_copies::record_const_or_copy_raw): Likewise.
18805 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
18806 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
18807 (dump_predicates): Likewise.
18808 (find_uninit_use): Likewise.
18809 (warn_uninitialized_phi): Likewise.
18810 (pass_late_warn_uninitialized::execute): Likewise.
18811 * tree-ssa.c (verify_vssa): Likewise.
18812 (verify_ssa): Likewise.
18813 (maybe_optimize_var): Likewise.
18814 * tree-vrp.c (dump_value_range): Likewise.
18815 (dump_all_value_ranges): Likewise.
18816 (dump_asserts_for): Likewise.
18817 (register_edge_assert_for_2): Likewise.
18818 (vrp_visit_cond_stmt): Likewise.
18819 (vrp_visit_switch_stmt): Likewise.
18820 (vrp_visit_stmt): Likewise.
18821 (vrp_visit_phi_node): Likewise.
18822 (simplify_cond_using_ranges_1): Likewise.
18823 (fold_predicate_in): Likewise.
18824 (evrp_dom_walker::before_dom_children): Likewise.
18825 (evrp_dom_walker::push_value_range): Likewise.
18826 (evrp_dom_walker::pop_value_range): Likewise.
18827 (execute_early_vrp): Likewise.
18829 2017-05-16 Richard Biener <rguenther@suse.de>
18831 * dwarf2out.c (loc_list_from_tree_1): Do not create
18832 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
18834 2017-05-16 Richard Biener <rguenther@suse.de>
18836 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
18838 (note_variable_value_in_expr): If we resolved the decl ref
18839 do not push to the stack.
18841 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
18843 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
18844 operations in fast-math mode.
18845 (vaddq_f16): Likewise.
18846 (vmul_f16): Likewise.
18847 (vmulq_f16): Likewise.
18848 (vsub_f16): Likewise.
18849 (vsubq_f16): Likewise.
18850 * config/arm/neon.md (add<mode>3): New.
18852 (fma:<VH:mode>3): New. Also remove outdated comment.
18855 2017-05-16 Martin Liska <mliska@suse.cz>
18859 * ipa-devirt.c (warn_types_mismatch): Fix typo.
18860 (odr_types_equivalent_p): Likewise.
18862 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
18864 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
18866 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
18869 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
18870 non-interunit SSE move alternatives with '?'.
18871 (zero-extendsidi peephole2): New peephole to skip intermediate
18872 general register in SSE zero-extend sequence.
18874 2017-05-15 Jeff Law <law@redhat.com>
18876 * reorg.c (relax_delay_slots): Create a new variable to hold
18877 the temporary target rather than clobbering TARGET_LABEL.
18879 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
18880 missing argument to extract_bit_field call.
18881 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
18883 2017-05-15 Martin Liska <mliska@suse.cz>
18886 * gcc.c (process_command): Do not allow empty argument of -o option.
18888 2017-05-15 Renlin Li <renlin.li@arm.com>
18890 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
18891 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
18892 * config/aarch64/constraints.md (Usf): Add long call check.
18893 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
18894 (call_value): Likewise.
18895 (sibcall): Likewise.
18896 (sibcall_value): Likewise.
18898 (call_value_insn): New.
18899 (sibcall_insn): Update rtx pattern.
18900 (sibcall_value_insn): Likewise.
18901 (call_internal): Remove.
18902 (call_value_internal): Likewise.
18903 (sibcall_internal): Likewise.
18904 (sibcall_value_internal): Likewise.
18905 (call_reg): Likewise.
18906 (call_symbol): Likewise.
18907 (call_value_reg): Likewise.
18908 (call_value_symbol): Likewise.
18910 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
18913 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
18915 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
18917 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
18918 compatible with CCGOCmode and with CCZmode.
18920 2017-05-14 Martin Sebor <msebor@redhat.com>
18922 PR middle-end/77671
18923 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
18924 (gimple_fold_builtin_snprintf): Same.
18925 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
18926 (gimple_fold_builtin_snprintf): Same.
18927 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
18928 of character types.
18929 (is_call_safe): New function.
18930 (try_substitute_return_value): Call it.
18931 (try_simplify_call): New function.
18932 (pass_sprintf_length::handle_gimple_call): Call it.
18934 2017-05-14 Martin Sebor <msebor@redhat.com>
18936 PR middle-end/80669
18937 * builtins.c (expand_builtin_stpncpy): Simplify.
18939 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
18941 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
18942 * config/i386/i386.h
18943 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
18944 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
18945 (struct machine_function): Add new members call_ms2sysv,
18946 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
18947 (struct machine_frame_state): New fields sp_realigned and
18948 sp_realigned_offset.
18949 * config/i386/i386.c
18950 (enum xlogue_stub): New enum.
18951 (enum xlogue_stub_sets): New enum.
18952 (class xlogue_layout): New class.
18953 (struct ix86_frame): New fields stack_realign_allocate_offset,
18954 stack_realign_offset and outlined_save_offset. Modify comments to
18955 detail stack layout when using out-of-line stubs.
18956 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
18957 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
18958 -mcall-ms2sysv-xlogues.
18959 (stub_managed_regs): New static variable.
18960 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
18961 registers managed by out-of-line stub.
18962 (disable_call_ms2sysv_xlogues): New function.
18963 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
18964 m->call_ms2sysv when appropriate and compute frame layout for
18966 (sp_valid_at, fp_valid_at): New inline functions.
18967 (choose_basereg): New function.
18968 (choose_baseaddr): Add align parameter, use choose_basereg and modify
18970 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
18971 Use align parameter of choose_baseaddr to generated aligned SSE movs
18973 (pro_epilogue_adjust_stack): Modify to track
18974 machine_frame_state::sp_realigned.
18975 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
18976 (ix86_nsaved_sseregs): Likewise.
18977 (ix86_emit_save_regs): Likewise.
18978 (ix86_emit_save_regs_using_mov): Likewise.
18979 (ix86_emit_save_sse_regs_using_mov): Likewise.
18980 (get_scratch_register_on_entry): Likewise.
18981 (gen_frame_set): New function.
18982 (gen_frame_load): Likewise.
18983 (gen_frame_store): Likewise.
18984 (emit_outlined_ms2sysv_save): Likewise.
18985 (emit_outlined_ms2sysv_restore): Likewise.
18986 (ix86_expand_prologue): Modify stack re-alignment code and call
18987 emit_outlined_ms2sysv_save when appropriate.
18988 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
18989 parameter rtx_insn *insn, which allows the function to be used to only
18990 generate the notes.
18991 (ix86_expand_epilogue): Modify validity checks of frame and stack
18992 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
18993 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
18994 * config/i386/predicates.md
18995 (save_multiple): New predicate.
18996 (restore_multiple): Likewise.
18997 * config/i386/sse.md
18998 (save_multiple<mode>): New pattern.
18999 (save_multiple_realign<mode>): Likewise.
19000 (restore_multiple<mode>): Likewise.
19001 (restore_multiple_and_return<mode>): Likewise.
19002 (restore_multiple_leave_return<mode>): Likewise.
19003 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
19005 2017-05-14 Julia Koval <julia.koval@intel.com>
19007 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
19008 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
19009 (__builtin_ia32_xsetbv): New builtins.
19010 * config/i386/i386.c (ix86_expand_special_args_builtin):
19012 (ix86_expand_builtin): Special expand for new intrinsics.
19013 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
19014 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
19015 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
19017 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19019 * cfganal.c (inverted_post_order_compute): Change argument type
19021 * cfganal.h (inverted_post_order_compute): Adjust prototype.
19022 * df-core.c (rest_of_handle_df_initialize): Adjust.
19023 (rest_of_handle_df_finish): Likewise.
19024 (df_analyze_1): Likewise.
19025 (df_analyze): Likewise.
19026 (loop_inverted_post_order_compute): Change argument to be a vec *.
19027 (df_analyze_loop): Adjust.
19028 (df_get_n_blocks): Likewise.
19029 (df_get_postorder): Likewise.
19030 * df.h (struct df_d): Change field to be a vec.
19031 * lcm.c (compute_laterin): Adjust.
19032 (compute_available): Likewise.
19033 * lra-lives.c (lra_create_live_ranges_1): Likewise.
19034 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
19035 * tree-ssa-pre.c (compute_antic): Likewise.
19037 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19039 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
19040 (depth_first_search::depth_first_search): Change structure init
19041 function to this constructor.
19042 (depth_first_search::add_bb): Rename function to this member.
19043 (depth_first_search::execute): Likewise.
19044 (flow_dfs_compute_reverse_finish): Adjust.
19046 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19048 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
19049 (longest_simple_path): Likewise.
19050 * shrink-wrap.c (spread_components): Likewise.
19051 (disqualify_problematic_components): Likewise.
19052 (emit_common_heads_for_components): Likewise.
19053 (emit_common_tails_for_components): Likewise.
19054 (insert_prologue_epilogue_for_components): Likewise.
19056 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19058 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
19061 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19063 * df-core.c (df_set_blocks): Start using auto_bitmap.
19064 (df_compact_blocks): Likewise.
19065 * df-problems.c (df_rd_confluence_n): Likewise.
19066 * df-scan.c (df_insn_rescan_all): Likewise.
19067 (df_process_deferred_rescans): Likewise.
19068 (df_update_entry_block_defs): Likewise.
19069 (df_update_exit_block_uses): Likewise.
19070 (df_entry_block_bitmap_verify): Likewise.
19071 (df_exit_block_bitmap_verify): Likewise.
19072 (df_scan_verify): Likewise.
19073 * lra-constraints.c (lra_constraints): Likewise.
19074 (undo_optional_reloads): Likewise.
19075 (lra_undo_inheritance): Likewise.
19076 * lra-remat.c (calculate_gen_cands): Likewise.
19077 (do_remat): Likewise.
19078 * lra-spills.c (assign_spill_hard_regs): Likewise.
19079 (spill_pseudos): Likewise.
19080 * tree-ssa-pre.c (bitmap_set_and): Likewise.
19081 (bitmap_set_subtract_values): Likewise.
19083 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19085 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
19086 management with auto_bitmap.
19087 (fix_inter_tick): Likewise.
19088 (fix_recovery_deps): Likewise.
19089 * ira.c (add_store_equivs): Likewise.
19090 (find_moveable_pseudos): Likewise.
19091 (split_live_ranges_for_shrink_wrap): Likewise.
19092 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
19093 (rtx_reuse_manager::seen_def_p): Likewise.
19094 (rtx_reuse_manager::set_seen_def): Likewise.
19095 * print-rtl.h (class rtx_reuse_manager): Likewise.
19097 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19099 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
19101 (migrate_btr_def): Likewise.
19102 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
19103 * df-core.c (loop_post_order_compute): Likewise.
19104 (loop_inverted_post_order_compute): Likewise.
19105 * hsa-common.h: Likewise.
19106 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
19107 * init-regs.c (initialize_uninitialized_regs): Likewise.
19108 * ipa-inline.c (resolve_noninline_speculation): Likewise.
19109 (inline_small_functions): Likewise.
19110 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
19111 * ira.c (combine_and_move_insns): Likewise.
19112 (build_insn_chain): Likewise.
19113 * loop-invariant.c (find_invariants): Likewise.
19114 * lower-subreg.c (propagate_pseudo_copies): Likewise.
19115 * predict.c (tree_predict_by_opcode): Likewise.
19116 (predict_paths_leading_to): Likewise.
19117 (predict_paths_leading_to_edge): Likewise.
19118 (estimate_loops_at_level): Likewise.
19119 (estimate_loops): Likewise.
19120 * shrink-wrap.c (try_shrink_wrapping): Likewise.
19121 (spread_components): Likewise.
19122 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
19123 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
19124 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
19125 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
19126 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
19127 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
19128 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
19129 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
19130 (mark_threaded_blocks): Likewise.
19131 (thread_through_all_blocks): Likewise.
19132 * tree-ssa.c (verify_ssa): Likewise.
19133 (execute_update_addresses_taken): Likewise.
19134 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
19136 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19138 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
19140 (post_order_compute): Likewise.
19141 (inverted_post_order_compute): Likewise.
19142 (pre_and_rev_post_order_compute_fn): Likewise.
19144 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19146 * genrecog.c (int_set::int_set): Explicitly construct our
19147 auto_vec base class.
19148 * vec.h (auto_vec::auto_vec): New constructor.
19150 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19152 * bitmap.h (class auto_bitmap): New constructor taking
19153 bitmap_obstack * argument.
19155 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19157 * bitmap.h (class auto_bitmap): Change type of m_bits to
19158 bitmap_head, and adjust ctor / dtor and member operators.
19160 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
19162 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
19163 when returned register mode doesn't match original mode.
19165 2017-05-12 Jeff Law <law@redhat.com>
19166 Jakub Jelinek <jakub@redhat.com>
19168 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
19169 we look for cc setter after the compare-elim changes.
19170 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
19171 within the vector to match what compare-elim now expects.
19172 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
19173 (xorsi3_flags, one_cmplsi2_flags): Likewise.
19175 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
19176 after the compare-elim changes.
19177 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
19178 the vector to match what compare-elim now expects.
19179 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
19180 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
19181 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
19182 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
19183 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
19185 * config/visium/visium.c (single_set_and_flags): Fix where
19186 we look for cc setter after the compare-elim changes.
19187 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
19188 with the vector to match what compare-elim now expects.
19189 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
19190 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
19191 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
19192 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
19193 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
19194 (neg<mode>2_insn_set_overflow): Likewise.
19196 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
19198 PR middle-end/79794
19199 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
19200 maybe_expand_insn call, set ops[0].target. If still set after call,
19201 set alt_rtl. Add extra arg to recursive calls.
19202 (extract_bit_field): Add alt_rtl argument. Pass to
19204 * expmed.h (extract_bit_field): Fix prototype.
19205 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
19206 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
19207 to extract_bit_field_calls.
19208 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
19209 Pass alt_rtl to extract_bit_field calls.
19210 * calls.c (store_unaligned_arguments_into_psuedos)
19211 load_register_parameters): Pass extra NULL to extract_bit_field calls.
19212 * optabs.c (maybe_legitimize_operand): Clear op->target when call
19214 * optabs.h (struct expand_operand): Add target bitfield.
19216 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
19218 * compare-elim.c (try_eliminate_compare): Canonicalize
19219 operation with embedded compare to
19220 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
19221 (set (reg) (operation)].
19223 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
19225 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
19228 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
19229 cost of adding a carry flag for ADC instruction.
19230 [case MINUS]: Ignore the cost of subtracting a carry flag
19231 for SBB instruction.
19233 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
19235 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
19237 * config/rs6000/bmiintrin.h: New file.
19238 * config/rs6000/bmi2intrin.h: New file.
19239 * config/rs6000/x86intrin.h: New file.
19241 2017-05-12 Jeff Law <law@redhat.com>
19243 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
19246 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
19248 PR middle-end/80707
19249 * tree-cfg.c: Remove cfg edges of unreachable case statements.
19251 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19253 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
19254 early expansion of vector divide builtins.
19255 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
19256 builtins identified as having unsigned arguments.
19258 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19260 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
19261 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
19262 expansion of vector logical operations (and, andc, or, xor,
19265 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19267 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
19268 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
19270 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19272 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
19273 early GIMPLE expansion of vector multiplies.
19275 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19277 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
19278 TARGET_HAVE_MOVT conditional.
19279 (movt splitter): Likewise.
19281 2017-05-12 Richard Biener <rguenther@suse.de>
19283 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
19284 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
19285 Fold all stmts not inplace.
19287 2017-05-12 Richard Biener <rguenther@suse.de>
19289 PR tree-optimization/80713
19290 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
19291 inserted_exprs bit for not removed stmts.
19293 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
19295 PR middle-end/69921
19296 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
19297 parallelized" attribute for parallelized OpenACC kernels.
19298 * omp-offload.c (execute_oacc_device_lower): Use it.
19300 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
19301 Set "oacc kernels" attribute.
19302 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
19303 parameter. Adjust all users.
19304 (oacc_fn_attrib_kernels_p): Remove function.
19305 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
19306 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
19307 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
19308 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
19309 assert "oacc kernels" attribute is set.
19311 2017-05-11 Carl Love <cel@us.ibm.com>
19313 * config/rs6000/rs6000-c: Add support for built-in functions
19314 vector unsigned char vec_popcnt (vector signed char)
19315 vector unsigned char vec_popcnt (vector unsigned char)
19316 vector unsigned short vec_popcnt (vector signed short)
19317 vector unsigned short vec_popcnt (vector unsigned short)
19318 vector unsigned int vec_popcnt (vector signed int)
19319 vector unsigned int vec_popcnt (vector unsigned int)
19320 vector unsigned long long vec_popcnt (vector signed long long)
19321 vector unsigned long long vec_popcnt (vector unsigned long long)
19322 vector signed long long vec_slo (vector signed long long,
19323 vector signed char)
19324 vector signed long long vec_slo (vector signed long long,
19325 vector unsigned char)
19326 vector unsigned long long vec_slo (vector unsigned long long,
19327 vector signed char)
19328 vector unsigned long long vec_slo (vector unsigned long long,
19329 vector unsigned char)
19330 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
19331 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
19332 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
19333 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
19334 * doc/extend.texi: Update the built-in documentation file for the
19335 new built-in functions.
19337 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
19339 * attribs.h (sorted_attr_string): Move machine independent
19340 functions for target clone support from the i386 port to common
19341 code. Rename ix86_function_versions to common_function_versions.
19342 Rename make_name to make_unique_name.
19343 (common_function_versions): Likewise.
19344 (make_unique_name): Likewise.
19345 (make_dispatcher_decl): Likewise.
19346 (is_function_default_version): Likewise.
19347 * attribs.c (attr_strcmp): Likewise.
19348 (sorted_attr_string): Likewise.
19349 (common_function_versions): Likewise.
19350 (make_unique_name): Likewise.
19351 (make_dispatcher_decl): Likewise.
19352 (is_function_default_version): Likewise.
19353 * config/i386/i386.c (attr_strcmp): Likewise.
19354 (sorted_attr_string): Likewise.
19355 (ix86_function_versions): Likewise.
19356 (make_name): Likewise.
19357 (make_dispatcher_decl): Likewise.
19358 (is_function_default_version): Likewise.
19359 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
19361 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19364 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
19365 Account for direct move costs for vec_construct of integer
19368 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
19371 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
19372 (UNSPEC_STX_ATOMIC): Ditto.
19373 (loaddi_via_sse): New insn.
19374 (storedi_via_sse): Ditto.
19375 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
19376 Update corresponding peephole2 patterns.
19377 (atomic_storedi_fpu): Ditto.
19379 2017-05-11 Julia Koval <julia.koval@intel.com>
19381 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
19382 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
19384 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
19385 (__builtin_ia32_rsqrt14ss_mask): New builtins.
19386 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
19388 2017-05-11 Nathan Sidwell <nathan@acm.org>
19390 * graphite-poly.c: Include dumpfile.h.
19392 * dumpfle.h (dump_function): Declare here ...
19393 * tree-dump.h (dump_function): ... not here.
19394 * dumpfile.c: #include tree-cfg.h.
19395 (dump_function): Move here from ...
19396 * tree-dump.c (dump_function): ... here.
19397 * gimplify.c: #include splay-tree.h, not tree-dump.h.
19398 * graphite-poly.c: Don't include tree-dump.h.
19399 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
19400 * print-tree.c: Likewise.
19401 * stor-layout.c: Likewise.
19402 * tree-nested.c: Likewise.
19404 * dumpfile.c (dump_start): Use TDF_FLAGS.
19405 (dump_enable_all): Fix TDF_KIND check thinko.
19407 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
19409 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19410 array entries to represent two legal parameterizations of the
19411 overloaded __builtin_cmpb function, as represented by the
19412 P6_OV_BUILTIN_CMPB constant.
19413 (altivec_resolve_overloaded_builtin): Add special case handling
19414 for the __builtin_cmpb function, as represented by the
19415 P6_OV_BUILTIN_CMPB constant.
19416 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
19417 (BU_P6_64BIT_2): New macro.
19418 (BU_P6_OVERLOAD_2): New macro
19419 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
19420 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
19421 (CMPB): Add overload support to represent both 32-bit and 64-bit
19422 compare-bytes function.
19423 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
19424 support for TARGET_CMPB.
19425 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
19426 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19427 documentation of the __builtin_cmpb overloaded built-in function.
19429 2017-05-11 Richard Biener <rguenther@suse.de>
19431 PR tree-optimization/80705
19432 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
19433 bases are not vectorizable.
19435 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19437 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
19438 when counting register pressure.
19440 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19442 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
19443 (iv_ca_more_deps): Renamed to ...
19444 (iv_ca_compare_deps): ... this.
19445 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
19447 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19449 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
19451 (determine_group_iv_costs): ... here.
19452 (find_inv_vars_cb): Record inv var if it's not recorded before.
19454 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19456 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
19457 (get_shiftadd_cost): Ditto.
19459 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19461 * tree-ssa-address.c: Include header file.
19462 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
19464 (add_to_parts): Refactor.
19465 (addr_to_parts): New parameter. Update use of move_hint_to_base.
19466 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
19469 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19471 PR tree-optimization/53090
19472 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
19474 (extract_cond_operands): Detect condition with IV on both sides
19475 and return COMP_IV_EXPR_2.
19476 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
19477 (rewrite_use_compare): Simplify by removing call to function
19478 extract_cond_operands.
19480 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19482 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
19483 (extract_cond_operands): Detect condition comparing against non-
19484 invariant bound and return appropriate enum value.
19485 (find_interesting_uses_cond): Update use of extract_cond_operands.
19486 Handle its return value accordingly.
19487 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
19489 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19491 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
19492 nonlinear iv_use computation in loop invariant sensitive way.
19494 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19496 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
19497 (find_iv_candidates): Call relate_compare_use_with_all_cands.
19499 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19501 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
19502 (dump_cand): Support iv_cand.inv_exprs.
19503 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
19505 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
19508 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19510 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
19512 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
19513 as local function. Include necessary header files.
19514 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
19516 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19518 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
19520 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19522 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
19523 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
19524 RSHIFT_EXPR and BIT_NOT_EXPR.
19526 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19528 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
19529 (adjust_setup_cost): New parameter supporting round up adjustment.
19530 (struct address_cost_data): Delete.
19531 (force_expr_to_var_cost): Don't bound cost with spill_cost.
19532 (split_address_cost, ptr_difference_cost): Delete.
19533 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
19534 (struct ainc_cost_data): New struct.
19535 (get_address_cost_ainc): New function.
19536 (get_address_cost, get_computation_cost): Reimplement.
19537 (determine_group_iv_cost_address): Record inv_expr for all uses of
19539 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
19540 (iv_ca_has_deps): Reimplemented to ...
19541 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
19543 (iv_ca_extend): Call iv_ca_more_deps.
19545 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19547 * tree-ssa-address.c (struct mem_address): Move to header file.
19548 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
19549 * tree-ssa-address.h (struct mem_address): Move from C file.
19550 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
19552 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19554 * tree-affine.h (aff_combination_type): New interface.
19555 (aff_combination_zero_p): Remove static.
19556 (aff_combination_const_p): New interface.
19557 (aff_combination_singleton_var_p): New interfaces.
19559 2017-05-11 Richard Biener <rguenther@suse.de>
19561 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
19562 Skip unreachable blocks and destinations.
19563 (eliminate): Move stmt removal and fixup ...
19564 (fini_eliminate): ... here. Skip inserted exprs.
19565 (pass_pre::execute): Move fini_pre after fini_eliminate.
19566 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
19567 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
19568 PRE to get rid of dead code that has invalid SSA form and
19569 split critical edges again.
19571 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19573 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19575 2017-05-11 Richard Biener <rguenther@suse.de>
19577 * passes.c (execute_function_todo): Verify loops if they are
19578 said to be up-to-date.
19579 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
19580 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
19582 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
19585 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
19586 handle calling assemble_external ourself.
19589 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
19590 modes with zero size. Enhance comment.
19592 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19594 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
19595 built-ins for vec_xl and vec_xst with short and char pointer
19598 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
19600 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
19601 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
19602 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
19603 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
19604 (_mm_maskz_min_round_ss): New intrinsics.
19605 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
19606 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
19607 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
19608 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
19609 (__builtin_ia32_minss_mask_round): New builtins.
19610 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19611 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
19612 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
19614 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
19615 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
19617 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
19620 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
19622 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
19623 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
19624 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
19625 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
19626 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
19627 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
19628 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
19629 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19630 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
19631 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
19632 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
19633 (__builtin_ia32_mulss_mask_round): New builtins.
19634 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19635 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
19636 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
19638 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
19639 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
19641 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
19644 2017-05-10 Julia Koval <julia.koval@intel.com>
19646 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
19647 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
19648 (_mm256_setr_m128i): New intrinsics.
19650 2017-05-10 Julia Koval <julia.koval@intel.com>
19652 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
19653 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
19654 (_mm_maskz_rcp14_ss): New intrinsics.
19655 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
19656 (__builtin_ia32_rcp14ss_mask): New builtins.
19657 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
19659 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
19661 PR tree-optimization/51513
19662 * tree-cfg.c (gimple_seq_unreachable_p): New function.
19663 (assert_unreachable_fallthru_edge_p): Use it.
19664 (group_case_labels_stmt): Likewise.
19665 * tree-cfg.h: Prototype it.
19666 * stmt.c: Include cfghooks.h and tree-cfg.h.
19667 (emit_case_dispatch_table) <gap_label>: New local variable.
19668 Use it to fill dispatch table gaps.
19669 Test for default_label before updating probabilities.
19670 (expand_case) <default_label>: Remove unneeded initialization.
19671 Test for unreachable default case statement and remove its edge.
19672 Set default_label accordingly.
19673 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
19675 2017-05-10 Carl Love <cel@us.ibm.com>
19677 * config/rs6000/rs6000-c: Add support for built-in functions
19678 vector signed char vec_neg (vector signed char)
19679 vector signed short int vec_neg (vector short int)
19680 vector signed int vec_neg (vector signed int)
19681 vector signed long long vec_neg (vector signed long long)
19682 vector float vec_neg (vector float)
19683 vector double vec_neg (vector double)
19684 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
19686 * config/rs6000/altivec.h: Add define for vec_neg
19687 * doc/extend.texi: Update the built-in documentation for the
19688 new built-in functions.
19690 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19692 PR tree-optimization/77644
19693 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
19695 2017-05-10 Nathan Sidwell <nathan@acm.org>
19697 * dumpfile.h (TDI_lang_all): New.
19698 (TDF_KIND): New. Renumber others
19699 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
19701 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
19703 (get_dump_file_name): Adjust suffix generation.
19704 (dump_enable_all): Use TDF_KIND.
19705 * doc/invoke.texi (-fdump-lang-all): Document.
19707 * dumpfile.h: Tabify.
19709 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
19712 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
19713 Move member access before delete.
19715 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
19717 * tree-inline.c (expand_call_inline): Split block at stmt
19720 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
19723 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
19724 are now unused after splitting mov{sf,sd}_hardfloat.
19725 (f32_lr2): Likewise.
19726 (f32_lm): Likewise.
19727 (f32_lm2): Likewise.
19728 (f32_li): Likewise.
19729 (f32_li2): Likewise.
19730 (f32_lv): Likewise.
19731 (f32_sr): Likewise.
19732 (f32_sr2): Likewise.
19733 (f32_sm): Likewise.
19734 (f32_sm2): Likewise.
19735 (f32_si): Likewise.
19736 (f32_si2): Likewise.
19737 (f32_sv): Likewise.
19738 (f32_dm): Likewise.
19739 (f32_vsx): Likewise.
19740 (f32_av): Likewise.
19741 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
19742 For movsf, order stores so the VSX stores occur before the GPR
19743 store which encourages the register allocator to use a traditional
19744 FPR instead of a GPR. For movsd, order the stores so that the GPR
19745 store comes before the VSX stores to allow the power6 to work.
19746 This is due to the power6 not having a 32-bit integer store
19747 instruction from a FPR.
19748 (movsf_hardfloat): Likewise.
19749 (movsd_hardfloat): Likewise.
19751 2017-05-09 Martin Sebor <msebor@redhat.com>
19753 PR translation/80280
19754 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
19757 PR translation/80280
19758 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
19759 data member added in r247778.
19760 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
19762 2017-05-09 Nathan Sidwell <nathan@acm.org>
19764 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
19766 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
19769 2017-05-09 Marek Polacek <polacek@redhat.com>
19771 * doc/invoke.texi: Fix typo.
19773 2017-05-09 Richard Biener <rguenther@suse.de>
19775 * tree-vrp.c (vrp_val_is_max): Adjust comment.
19776 (vrp_val_is_min): Likewise.
19777 (set_value_range_to_value): Likewise.
19778 (set_value_range_to_nonnegative): Likewise.
19779 (gimple_assign_nonzero_p): Likewise.
19780 (gimple_stmt_nonzero_p): Likewise.
19781 (vrp_int_const_binop): Likewise. Remove unreachable case.
19782 (adjust_range_with_scev): Adjust comments.
19783 (compare_range_with_value): Likewise.
19784 (extract_range_from_phi_node): Likewise.
19785 (test_for_singularity): Likewise.
19787 2017-05-09 Richard Biener <rguenther@suse.de>
19789 * tree-vrp.c (get_single_symbol): Add assert that we don't
19790 get overflowed constants as invariant part.
19791 (compare_values_warnv): Add comment before the TREE_NO_WARNING
19792 checks. Use wi::cmp instead of recursing for integer constants.
19793 (compare_values): Just ignore whether we assumed undefined
19794 overflow instead of failing the compare.
19795 (extract_range_for_var_from_comparison_expr): Add comment before the
19796 TREE_NO_WARNING sets.
19797 (test_for_singularity): Likewise.
19798 (extract_range_from_comparison): Do not disable optimization
19799 when we assumed undefined overflow.
19800 (extract_range_basic): Remove init of unused var.
19802 2017-05-09 Richard Biener <rguenther@suse.de>
19804 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
19805 (extract_range_from_multiplicative_op_1): Adjust.
19806 (extract_range_from_binary_expr_1): Use int_const_binop.
19808 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
19811 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
19812 rs6000_store_data_bypass_p in seven define_bypass directives and
19813 in several comments.
19814 * config/rs6000/rs6000-protos.h: Add prototype for
19815 rs6000_store_data_bypass_p function.
19816 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
19817 function implements slightly different (rs6000-specific) semantics
19818 than store_data_bypass_p, returning false rather than aborting
19819 with assertion error when arguments do not satisfy the
19820 requirements of store data bypass.
19821 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
19822 rs6000_store_data_bypass_p.
19824 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
19826 * config/xtensa/xtensa-protos.h
19827 (xtensa_initial_elimination_offset): New declaration.
19828 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
19829 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
19830 macro definition, add case for FRAME_POINTER_REGNUM when
19831 FRAME_GROWS_DOWNWARD.
19832 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
19833 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
19834 xtensa_initial_elimination_offset.
19836 2017-05-08 Nathan Sidwell <nathan@acm.org>
19838 * doc/invoke.texi: Alphabetize -fdump options.
19840 2017-05-08 Martin Sebor <msebor@redhat.com>
19842 PR translation/80280
19843 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
19845 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
19847 * target.def (compute_frame_layout): New optional target hook.
19848 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
19849 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
19850 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
19852 * reload1.c (verify_initial_elim_offsets): Likewise.
19853 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
19854 (use_simple_return_p): Call arm_compute_frame_layout if needed.
19855 (arm_get_frame_offsets): Split up into this ...
19856 (arm_compute_frame_layout): ... and this function.
19858 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
19860 * config/aarch64/constraints.md (Usa): New constraint.
19861 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
19863 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
19865 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
19866 with_multilib_list after it has been checked.
19868 2017-05-08 Richard Biener <rguenther@suse.de>
19870 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
19871 (bitmap_set_subtract_values): Likewise.
19873 2017-05-08 Richard Biener <rguenther@suse.de>
19875 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
19876 (gimple_assign_nonzero): ... this and remove strict_overflow_p
19878 (gimple_stmt_nonzero_warnv_p): Rename to ...
19879 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
19881 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
19882 (extract_range_basic): Adjust, do not disable propagation on
19883 strict overflow sensitive simplification.
19884 (vrp_visit_cond_stmt): Likewise.
19886 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
19888 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
19889 body size unconditionally.
19891 2017-05-07 Jeff Law <law@redhat.com>
19894 2017-05-06 Jeff Law <law@redhat.com>
19895 PR tree-optimization/78496
19896 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
19899 PR tree-optimization/78496
19900 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
19901 (simplify_stmt_using_ranges): Call it.
19902 (vrp_dom_walker::before_dom_children): Extract equivalences
19903 from an ASSERT_EXPR with an equality comparison against a
19906 2017-05-06 Jeff Law <law@redhat.com>
19908 PR tree-optimization/78496
19909 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
19912 PR tree-optimization/78496
19913 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
19914 (simplify_stmt_using_ranges): Call it.
19915 (vrp_dom_walker::before_dom_children): Extract equivalences
19916 from an ASSERT_EXPR with an equality comparison against a
19919 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
19921 * lra-constraints.c (lra_copy_reg_equiv): New function.
19922 (split_reg): Use it to copy equivalence information from the
19923 original register to the spill register.
19925 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
19927 PR rtl-optimization/75964
19928 * simplify-rtx.c (simplify_const_relational_operation): Remove
19929 invalid handling of comparisons of integer ABS.
19931 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
19933 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
19934 initialize to zero.
19935 (init_regs): Remove declaration.
19936 (function_arg_advance_32): Initialize error_p as boolean variable.
19938 2017-05-05 Nathan Sidwell <nathan@acm.org>
19940 * store-motion.c (remove_reachable_equiv_notes): Reformat long
19941 lines. Use for (;;).
19943 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19945 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
19946 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
19947 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
19948 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
19949 VF=2 that require versioning.
19951 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19953 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
19956 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19958 * diagnostic.h (diagnostic_override_option_index): Convert from
19959 macro to inline function.
19961 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19963 * diagnostic.c (last_module_changed_p): New function.
19964 (set_last_module): New function.
19965 (diagnostic_report_current_module): Convert macro usage to
19966 the above functions.
19967 * diagnostic.h (diagnostic_context::last_module): Strengthen
19968 from const line_map * to const line_map_ordinary *.
19969 (diagnostic_last_module_changed): Delete macro.
19970 (diagnostic_set_last_module): Delete macro.
19972 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19974 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
19975 with diagnostic_report_diagnostic.
19976 (diagnostic_n_impl_richloc): Likewise.
19977 * diagnostic.h (report_diagnostic): Delete macro.
19978 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
19979 with diagnostic_report_diagnostic.
19980 * substring-locations.c (format_warning_va): Likewise.
19982 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19984 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
19985 save/restor of format_spec. Move option-printing code to...
19986 (print_option_information): ...this new function, and
19987 reimplement by simply printing to the pretty_printer,
19988 rather than appending to the format string.
19990 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19992 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
19993 handling logic into...
19994 (update_effective_level_from_pragmas): ...this new function.
19996 2017-05-04 Andrew Waterman <andrew@sifive.com>
19998 * config/riscv/riscv.opt (mstrict-align): New option.
19999 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
20000 (SLOW_UNALIGNED_ACCESS): Define.
20001 (riscv_slow_unaligned_access): Declare.
20002 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
20004 (riscv_slow_unaligned_access): New variable.
20005 (rocket_tune_info): Set slow_unaligned_access to true.
20006 (optimize_size_tune_info): Set slow_unaligned_access to false.
20007 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
20008 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
20009 (riscv_option_override): Set riscv_slow_unaligned_access.
20010 * doc/invoke.texi: Add -mstrict-align to RISC-V.
20012 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
20014 * config/riscv/riscv.md: Unify indentation.
20016 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20021 * config/rs6000/rs6000.md (u code attribute): Add FIX and
20023 (extendsi<mode>2): Add support for doing sign extension via
20024 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
20025 don't have ISA 3.0 instructions.
20026 (extendsi<mode>2 splitter): Likewise.
20027 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
20028 generate the normal insns since SImode can now go in vector
20029 registers. Disallow the special UNSPECs needed for previous
20030 machines to hide SImode being used. Add new insns
20031 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
20032 (fix_trunc<mode>si2_stfiwx): Likewise.
20033 (fix_trunc<mode>si2_internal): Likewise.
20034 (fixuns_trunc<mode>si2): Likewise.
20035 (fixuns_trunc<mode>si2_stfiwx): Likewise.
20036 (fctiw<u>z_<mode>_smallint): Likewise.
20037 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
20038 of floating point to 32-bit integer from doing a direct move to
20039 the GPR registers to do a store.
20040 (fctiwz_<mode>): Break long line.
20042 2017-05-05 Bin Cheng <bin.cheng@arm.com>
20044 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
20045 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
20046 (addr_list, addr_offset_valid_p): New.
20047 (split_address_groups): Check offset validity with above function.
20048 (gt-tree-ssa-loop-ivopts.h): Include header file.
20050 2017-05-05 Nathan Sidwell <nathan@acm.org>
20052 * config.gcc (arm*-*-*): Add missing 'fi'.
20054 2017-05-05 Steve Ellcey <sellcey@cavium.com>
20056 * doc/invoke.texi (-fopt-info): Explicitly say order of options
20057 included in -fopt-info does not matter.
20058 * doc/optinfo.texi (-fopt-info): Fix description of default
20059 behavour. Explicitly say order of options included in -fopt-info
20062 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
20064 * config.gcc: Allow combinations of aprofile and rmprofile values for
20065 --with-multilib-list.
20066 * config/arm/t-multilib: New file.
20067 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
20068 variables. Remove setting of ISA and floating-point ABI in
20069 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
20070 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
20071 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
20072 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
20074 * config/arm/t-rmprofile: Likewise except for the matches changes.
20075 * doc/install.texi (--with-multilib-list): Document the combination of
20076 aprofile and rmprofile values and warn about pitfalls in doing that.
20078 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
20080 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
20081 (movdi_aarch64): Likewise.
20083 2017-05-05 Jakub Jelinek <jakub@redhat.com>
20085 PR tree-optimization/80632
20086 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
20088 (build_arrays): Initialize it for virtual phis.
20089 (fix_phi_nodes): Use it for virtual phis.
20091 PR tree-optimization/80558
20092 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
20093 [x, y] op z into [x op, y op z] for op & or | if conditions
20096 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
20097 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20100 * config/arm/arm.md (use_literal_pool): Remove.
20101 (64-bit immediate split): No longer takes cost into consideration
20102 if arm_disable_literal_pool is enabled.
20103 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
20104 used when arm_disable_literal_pool is enabled.
20105 (arm_max_const_double_inline_cost): Remove use of
20106 arm_disable_literal_pool.
20107 (push_minipool_fix): Add assert.
20108 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
20109 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
20110 (no_literal_pool_sf_immediate): New.
20112 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20114 PR tree-optimization/80613
20115 * tree-ssa-dce.c (propagate_necessity): Remove cases for
20116 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
20118 2017-05-05 Richard Biener <rguenther@suse.de>
20120 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
20122 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
20124 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
20125 of this flag from insn conditions due to removal from r247495.
20127 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
20129 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
20131 (arm_early_store_addr_dep_ptr): Likewise.
20132 * config/arm/aarch-common-protos.h
20133 (arm_early_load_addr_dep_ptr): Add prototype.
20134 (arm_early_store_addr_dep_ptr): Likewise.
20135 * config/arm/cortex-a53.md: Add new bypasses.
20137 2017-05-05 Jakub Jelinek <jakub@redhat.com>
20139 * tree.c (next_type_uid): Change type to unsigned.
20140 (type_hash_canon): Decrement back next_type_uid if
20141 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
20142 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
20145 2017-05-04 Martin Sebor <msebor@redhat.com>
20147 * builtins.c: Fix a trivial typo in a comment.
20149 PR middle-end/79234
20150 * builtins.c (check_sizes): Adjust to handle reading past the end.
20151 Avoid printing excessive upper bound of ranges. Use %E to print
20152 tree nodes instead of converting them to %wu.
20153 (expand_builtin_memchr): New function.
20154 (compute_dest_size): Rename...
20155 (compute_objsize): ...to this.
20156 (expand_builtin_memcpy): Adjust.
20157 (expand_builtin_mempcpy): Adjust.
20158 (expand_builtin_strcat): Adjust.
20159 (expand_builtin_strcpy): Adjust.
20160 (check_strncat_sizes): Adjust.
20161 (expand_builtin_strncat): Adjust.
20162 (expand_builtin_strncpy): Adjust and simplify.
20163 (expand_builtin_memset): Adjust.
20164 (expand_builtin_bzero): Adjust.
20165 (expand_builtin_memcmp): Adjust.
20166 (expand_builtin): Handle memcmp.
20167 (maybe_emit_chk_warning): Check strncat just once.
20169 2017-05-04 Martin Sebor <msebor@redhat.com>
20171 PR preprocessor/79214
20172 PR middle-end/79222
20173 PR middle-end/79223
20174 * builtins.c (check_sizes): Add inlining context and issue
20175 warnings even when -Wno-system-headers is set.
20176 (check_strncat_sizes): Same.
20177 (expand_builtin_strncat): Same.
20178 (expand_builtin_memmove): New function.
20179 (expand_builtin_stpncpy): Same.
20180 (expand_builtin): Handle memmove and stpncpy.
20182 2017-05-04 Bin Cheng <bin.cheng@arm.com>
20184 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
20185 which is not used any more.
20187 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20189 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
20191 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20193 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
20194 (cortexa53_tunings): Likewise.
20195 (cortexa57_tunings): Likewise.
20196 (cortexa72_tunings): Likewise.
20197 (cortexa73_tunings): Likewise.
20199 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20201 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
20202 Set loop alignment to 8.
20204 2017-05-04 Martin Sebor <msebor@redhat.com>
20206 PR translation/80280
20207 * builtins.c (expand_builtin_object_size): Add missing quoting to
20208 %D and like directives.
20209 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
20210 (hsa_type_for_tree_type): Same.
20211 (verify_function_arguments): Same.
20212 * symtab.c (symbol_table::change_decl_assembler_name): Same.
20213 * varasm.c (get_section): Same.
20216 2017-05-04 Martin Sebor <msebor@redhat.com>
20218 PR translation/80280
20219 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
20221 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20223 * config/aarch64/aarch64.c (generic_addrcost_table):
20224 Change HI/TI mode setting.
20226 2017-05-04 Martin Jambor <mjambor@suse.cz>
20228 PR tree-optimization/80622
20229 * tree-sra.c (comes_initialized_p): New function.
20230 (build_accesses_from_assign): Only set write lazily when
20231 comes_initialized_p is false.
20232 (analyze_access_subtree): Use comes_initialized_p.
20233 (propagate_subaccesses_across_link): Assert !comes_initialized_p
20234 instead of testing for PARM_DECL.
20236 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20238 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
20239 constraint on operand 0 to allow more general addressing modes.
20240 Adjust output template.
20241 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
20243 * config/aarch64/aarch64-protos.h
20244 (aarch64_address_valid_for_prefetch_p): Declare prototype.
20245 * config/aarch64/constraints.md (Dp): New address constraint.
20246 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
20249 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
20251 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
20252 update use of estimate_ipcp_clone_size_and_time.
20253 (estimate_local_effects): Update use of
20254 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
20255 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
20256 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
20257 Return nonspecialized time.
20259 2017-05-04 Richard Biener <rguenther@suse.de>
20261 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
20262 for the last VUSE which def dominates the PHI. Directly call
20264 (get_continuation_for_phi_1): Remove.
20266 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
20268 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
20269 to explain the use of truncating division. Cap the number of
20270 iterations to the maximum given by nb_iterations_upper_bound,
20273 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20275 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
20276 * configure: Regenerate.
20277 * config.in: Regenerate.
20278 * config/i386/driver-mingw32.c: new file.
20279 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
20280 * config.host: Link driver-mingw32.o on MinGW host.
20281 * doc/install.texi: Document new --enable-mingw-wildcard configure
20284 2017-05-04 Marek Polacek <polacek@redhat.com>
20286 PR tree-optimization/80612
20287 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
20289 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20290 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
20292 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
20293 (movt splitter): Likewise.
20294 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
20295 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
20296 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
20297 block for Thumb-1 with MOVT.
20298 (thumb2_legitimate_address_p): Move code block ...
20299 (can_avoid_literal_pool_for_label_p): ... into this new function.
20300 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
20302 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
20303 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
20304 "M-profile targets with the MOVT instruction".
20306 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20308 * config/arm/arm-builtins.c (arm_init_builtins): Rename
20309 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
20310 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
20312 2017-05-04 Martin Liska <mliska@suse.cz>
20314 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
20315 variable cond_code.
20317 2017-05-04 Richard Biener <rguenther@suse.de>
20319 * tree.c (array_at_struct_end_p): Handle arrays at struct
20320 end with flexarrays more conservatively. Refactor and treat
20321 arrays of arrays or aggregates more strict. Fix
20322 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
20323 * tree.c (array_at_struct_end_p): Adjust prototype.
20324 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
20325 * gimple-fold.c (get_range_strlen): Likewise.
20326 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
20328 2017-05-04 Richard Biener <rguenther@suse.de>
20330 PR tree-optimization/31130
20331 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
20333 (supports_overflow_infinity): Likewise.
20334 (is_negative_overflow_infinity): Likewise.
20335 (is_positive_overflow_infinity): Likewise.
20336 (is_overflow_infinity): Likewise.
20337 (stmt_overflow_infinity): Likewise.
20338 (overflow_infinity_range_p): Likewise.
20339 (usable_range_p): Remove as always returning true.
20340 (make_overflow_infinity): Remove.
20341 (negative_overflow_infinity): Likewise.
20342 (positive_overflow_infinity): Likewise.
20343 (avoid_overflow_infinity): Likewise.
20344 (set_value_range): Adjust accordingly.
20345 (set_value_range_to_nonnegative): Likewise, remove now unused
20346 overflow_infinity arg.
20347 (vrp_operand_equal_p): Adjust.
20348 (update_value_range): Likewise.
20349 (range_int_cst_singleton_p): Likewise.
20350 (operand_less_p): Likewise.
20351 (compare_values_warnv): Likewise.
20352 (extract_range_for_var_from_comparison_expr): Likewise.
20353 (vrp_int_const_binop): Likewise.
20354 (zero_nonzero_bits_from_vr): Likewise.
20355 (extract_range_from_multiplicative_op_1): Likewise.
20356 (extract_range_from_binary_expr_1): Likewise.
20357 (extract_range_from_unary_expr): Likewise.
20358 (extract_range_from_comparison): Likewise.
20359 (extract_range_basic): Likewise.
20360 (adjust_range_with_scev): Likewise.
20361 (compare_ranges): Likewise.
20362 (compare_range_with_value): Likewise.
20363 (dump_value_range): Likewise.
20364 (test_for_singularity): Likewise, remove strict_overflow_p parameter
20366 (simplify_cond_using_ranges): Adjust.
20368 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
20370 * brig-builtins.def: Added a builtin for class_f64.
20371 * builtin-types.def: Added a builtin type needed by class_f64.
20373 2017-05-03 Jason Merrill <jason@redhat.com>
20375 * timevar.def: Add TV_CONSTEXPR.
20377 2017-05-03 David Malcolm <dmalcolm@redhat.com>
20379 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
20381 2017-05-03 Martin Jambor <mjambor@suse.cz>
20383 * ipa-prop.c (ipa_update_after_lto_read): Removed.
20384 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
20385 * ipa-cp.c (ipcp_propagate_stage): Do not call
20386 ipa_update_after_lto_read.
20387 * ipa-inline.c (ipa_inline): Likewise.
20389 2017-05-03 Martin Jambor <mjambor@suse.cz>
20391 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
20392 tag. Added a default constructor and a destructor.
20393 (ipa_edge_args_sum_t): New class;
20394 (ipa_edge_args_sum): Declare.
20395 (ipa_edge_args_vector): Remove declaration.
20396 (IPA_EDGE_REF): Use ipa_edge_args_sum.
20397 (ipa_free_edge_args_substructures): Remove declaration.
20398 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
20399 (ipa_edge_args_info_available_for_edge_p): Likewise.
20400 * ipa-prop.c (ipa_edge_args_vector): Removed.
20401 (edge_removal_hook_holder): Likewise.
20402 (edge_duplication_hook_holder): Likewise.
20403 (ipa_edge_args_sum): New variable.
20404 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
20405 ipa_edge_args_vector.
20406 (ipa_free_edge_args_substructures): Likewise.
20407 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
20408 ipa_edge_args_vector.
20409 (ipa_edge_removal_hook): Turned into method
20410 ipa_edge_args_sum_t::remove.
20411 (ipa_edge_duplication_hook): Turned into method
20412 ipa_edge_args_sum_t::duplicate.
20413 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
20414 registering edge hooks.
20415 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
20416 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
20417 ipa_edge_args_sum instead of ipa_edge_args_vector.
20418 * ipa-profile.c (ipa_profile): Likewise.
20420 2017-05-03 Martin Jambor <mjambor@suse.cz>
20422 * symbol-summary.h (function_summary): New method exists.
20423 (function_summary::symtab_removal): Deallocate through release.
20424 (call_summary): New class.
20425 (gt_ggc_mx): New overload.
20426 (gt_pch_nx): Likewise.
20427 (gt_pch_nx): Likewise.
20429 2017-05-03 Jeff Law <law@redhat.com>
20431 PR tree-optimization/78496
20432 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
20433 from simplify_cond_using_ranges. Split off code to walk
20434 backwards through casts into ...
20435 (simplify_cond_using_ranges_2): New function.
20436 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
20437 (execute_vrp): After identifying jump threads, call
20438 simplify_cond_using_ranges_2.
20440 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
20443 * ipa-inline.h (inline_summary): Add ctor.
20444 (create_ggc): Do not use ggc_cleared_alloc.
20446 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
20447 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20449 * gcc.c (handle_braces): Support escaping in switch matching
20451 * doc/invoke.texi (Spec Files): Document it.
20452 Remove superfluous @code markup in items.
20454 2017-05-03 David Malcolm <dmalcolm@redhat.com>
20456 * diagnostic-show-locus.c (struct column_range): New struct.
20457 (get_affected_columns): New function.
20458 (get_printed_columns): New function.
20459 (struct correction): New struct.
20460 (correction::ensure_capacity): New function.
20461 (correction::ensure_terminated): New function.
20462 (struct line_corrections): New struct.
20463 (line_corrections::~line_corrections): New dtor.
20464 (line_corrections::add_hint): New function.
20465 (layout::print_trailing_fixits): Reimplement in terms of the new
20467 (selftest::test_overlapped_fixit_printing): New function.
20468 (selftest::diagnostic_show_locus_c_tests): Call it.
20470 2017-05-03 Nathan Sidwell <nathan@acm.org>
20472 Canonicalize canonical type hashing
20473 * tree.h (type_hash_canon_hash): Declare.
20474 * tree.c (type_hash_list, attribute_hash_list): Move into
20475 type_hash_canon_hash.
20476 (build_type_attribute_qual_variant): Break out hash code calc into
20477 type_hash_canon_hash.
20478 (type_hash_canon_hash): New. Generic type hash computation.
20479 (build_range_type_1, build_array_type_1, build_function_type,
20480 build_method_type_directly, build_offset_type, build_complex_type,
20481 make_vector_type): Call it.
20483 2017-05-03 Richard Biener <rguenther@suse.de>
20485 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
20486 When all DRs have unknown misaligned do not always peel
20487 when there is a store but apply the same costing model as if
20488 there were only loads.
20490 2017-05-03 Richard Biener <rguenther@suse.de>
20493 PR tree-optimization/80492
20494 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
20495 compare_base_decls returning dont-know properly.
20497 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
20499 * config/arm/iterators.md (CCSI): New mode iterator.
20500 (arch): New mode attribute.
20501 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
20502 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
20503 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
20504 code iterator for success result mode.
20505 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
20506 the corresponding new insn generators.
20508 2017-05-03 Bin Cheng <bin.cheng@arm.com>
20511 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20512 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
20514 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
20516 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
20517 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
20518 (DDR_A): Wrap DDR argument in brackets.
20519 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
20520 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
20521 (DDR_REVERSED_P): Likewise.
20523 2017-05-03 Jakub Jelinek <jakub@redhat.com>
20525 PR tree-optimization/79472
20526 * tree-switch-conversion.c (struct switch_conv_info): Add
20527 contiguous_range and default_case_nonstandard fields.
20528 (collect_switch_conv_info): Compute contiguous_range and
20529 default_case_nonstandard fields, don't clear final_bb if
20530 contiguous_range and only the default case doesn't have the required
20532 (check_all_empty_except_final): Set default_case_nonstandard instead
20533 of failing if contiguous_range and the default case doesn't have empty
20535 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
20536 and only the default case doesn't have the required constants. Skip
20538 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
20539 if default_case_nonstandard.
20540 (build_constructors): Build constant 1 just once. Assert that default
20541 values aren't inserted in between cases if contiguous_range. Skip
20543 (build_arrays): Skip virtual phis.
20544 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
20545 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
20546 Handle virtual phis.
20547 (gen_inbound_check): Handle default_case_nonstandard case.
20548 (process_switch): Adjust check_final_bb caller. Call
20549 gather_default_values with the first non-default case instead of
20550 default case if default_case_nonstandard.
20552 2017-05-02 Nathan Sidwell <nathan@acm.org>
20554 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
20555 check. Fix formatting.
20557 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
20559 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
20560 errors when comparing specialized and unspecialized times.
20562 2017-05-02 David Malcolm <dmalcolm@redhat.com>
20564 * diagnostic-show-locus.c
20565 (layout::should_print_annotation_line_p): Make private.
20566 (layout::print_annotation_line): Make private.
20567 (layout::annotation_line_showed_range_p): Make private.
20568 (layout::show_ruler): Make private.
20569 (layout::print_source_line): Make private. Pass in line and
20570 line_width, rather than calling location_get_source_line. Drop
20572 (layout::print_leading_fixits): New method.
20573 (layout::print_any_fixits): Rename to...
20574 (layout::print_trailing_fixits): ...this, and make private.
20575 Don't print newline fixits.
20576 (diagnostic_show_locus): Move logic for printing one row into...
20577 (layout::print_line): ...this new function. Move the
20578 location_get_source_line call and error-handling from
20579 print_source_line to here. Call print_leading_fixits, and rename
20580 print_any_fixits to print_trailing_fixits.
20581 (selftest::test_fixit_insert_containing_newline): Update now that
20582 newlines are partially supported.
20583 (selftest::test_fixit_insert_containing_newline_2): New test.
20584 (selftest::test_fixit_replace_containing_newline): Update comments.
20585 (selftest::diagnostic_show_locus_c_tests): Call the new test.
20586 * edit-context.c (class added_line): New class.
20587 (class edited_line): Describe newline handling in comment.
20588 (edited_line::actually_edited_p): New method.
20589 (edited_line::print_content): Delete redundant decl.
20590 (edited_line::m_predecessors): New field.
20591 (edited_file::print_content): Call edited_line::print_content.
20592 (edited_file::print_diff): Update to support newlines.
20593 (edited_file::print_diff_hunk): Likewise.
20594 (edited_file::print_run_of_changed_lines): New function.
20595 (edited_file::print_diff_line): Convert to...
20596 (print_diff_line): ...this.
20597 (edited_file::get_effective_line_count): New function.
20598 (edited_line::edited_line): Initialize new field m_predecessors.
20599 (edited_line::~edited_line): Clean up m_predecessors.
20600 (edited_line::apply_fixit): Handle newlines.
20601 (edited_line::get_effective_line_count): New function.
20602 (edited_line::print_content): New function.
20603 (edited_line::print_diff_lines): New function.
20604 (selftest::test_applying_fixits_insert_containing_newline): New
20606 (selftest::test_applying_fixits_replace_containing_newline): New
20608 (selftest::insert_line): New function.
20609 (selftest::test_applying_fixits_multiple_lines): Add example of
20611 (selftest::edit_context_c_tests): Call the new tests.
20613 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20615 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
20616 parameter cand. Update dump information.
20617 (get_computation_cost): Update uses.
20619 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20621 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
20622 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
20623 (get_computation_at, rewrite_use_address): Update use of
20624 get_computation_aff.
20626 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20628 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
20629 (get_computation): Delete.
20630 (get_computation_cost): Implement like get_computation_cost_at.
20631 Use get_computation_at.
20632 (get_computation_cost_at): Delete.
20633 (rewrite_use_nonlinear_expr): Use get_computation_at.
20634 (rewrite_use_compare, remove_unused_ivs): Ditto.
20636 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20638 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
20640 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20642 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
20643 (ivopts_global_cost_for_size): Rename parameter and update uses.
20644 (iv_ca_recount_cost): Update uses.
20645 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
20646 candidates seperately in n_invs and n_cands.
20647 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
20649 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20651 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
20652 (find_inv_vars_cb): New.
20653 (find_depends): Renamed to ...
20654 (find_inv_vars): ... this.
20655 (add_candidate_1, force_var_cost): Call find_inv_vars.
20656 (split_address_cost, determine_group_iv_cost_cond): Ditto.
20658 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20660 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
20661 inv_vars. Add inv_exprs.
20662 (struct iv_cand): Rename depends_on to inv_vars.
20663 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
20664 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
20665 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
20666 (dump_cand): Dump inv_vars.
20667 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
20668 (record_invariant, find_depends, add_candidate_1): Ditto.
20669 (set_group_iv_cost, force_var_cost): Ditto.
20670 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
20671 (get_computation_cost_at, get_computation_cost): Ditto.
20672 (determine_group_iv_cost_generic): Ditto.
20673 (determine_group_iv_cost_address): Ditto.
20674 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
20675 (determine_group_iv_costs): Ditto.
20676 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
20677 (iv_ca_set_remove_invariants): Renamed to ...
20678 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
20679 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
20680 (iv_ca_set_add_invariants): Renamed to ...
20681 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
20682 (iv_ca_set_cp): Use iv_ca_set_add_invs.
20683 (iv_ca_has_deps): Support inv_vars and inv_exprs.
20684 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
20685 (create_new_ivs): Remove useless dump.
20687 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20689 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
20691 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
20692 (iv_ca_set_no_cp, create_new_iv): Ditto.
20694 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20696 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
20698 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
20700 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
20701 function tree_check2.
20703 2017-05-02 Martin Liska <mliska@suse.cz>
20705 * doc/gcov.texi: Add missing preposition.
20706 * gcov.c (function_info::function_info): Properly fill up
20707 all member variables.
20709 2017-05-02 Tamar Christina <tamar.christina@arm.com>
20711 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
20713 2017-05-02 Tamar Christina <tamar.christina@arm.com>
20715 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
20717 2017-05-02 Martin Liska <mliska@suse.cz>
20720 * lto-streamer-in.c (lto_read_tree_1): Remove
20721 LTO_STREAMER_DEBUG.
20722 * lto-streamer.c (struct tree_hash_entry): Likewise.
20723 (struct tree_entry_hasher): Likewise.
20724 (tree_entry_hasher::hash): Likewise.
20725 (tree_entry_hasher::equal): Likewise.
20726 (lto_streamer_init): Likewise.
20727 (lto_orig_address_map): Likewise.
20728 (lto_orig_address_get): Likewise.
20729 (lto_orig_address_remove): Likewise.
20730 * lto-streamer.h: Likewise.
20731 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
20732 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
20734 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
20736 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
20737 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
20738 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
20739 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
20740 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
20741 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
20742 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
20743 (mm_maskz_sub_ss): New intrinsics.
20744 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20745 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
20746 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
20747 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
20748 (__builtin_ia32_subss_mask_round): New builtins.
20749 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20750 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20751 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
20753 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
20754 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
20756 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
20759 2017-05-02 Martin Jambor <mjambor@suse.cz>
20761 PR tree-optimization/78687
20762 * tree-sra.c (access): New field parent.
20763 (process_subtree_disqualification): New function.
20764 (disqualify_candidate): Call it.
20765 (build_accesses_from_assign): Reset write flag if creating an
20767 (build_access_subtree): Fill in parent field and also prpagate
20768 down grp_write flag.
20769 (create_artificial_child_access): New parameter set_grp_write, set
20770 grp_write to its value.
20771 (propagate_subaccesses_across_link): Also propagate grp_write flag
20773 (propagate_all_subaccesses): Push the closest parent back to work
20774 queue if add_access_to_work_queue returned true.
20776 2017-05-02 Richard Biener <rguenther@suse.de>
20778 * common.opt (fstrict-overflow): Alias negative to fwrapv.
20779 * doc/invoke.texi (fstrict-overflow): Remove all traces of
20780 -fstrict-overflow documentation.
20781 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
20782 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
20783 flag_strict_overflow.
20784 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
20785 * lto-opts.c (lto_write_options): Do not stream it.
20786 * lto-wrapper.c (merge_and_complain): Do not handle it.
20787 * opts.c (default_options_table): Do not set -fstrict-overflow.
20788 (finish_options): Likewise do not clear it when sanitizing.
20789 * simplify-rtx.c (simplify_const_relational_operation): Do not
20790 test flag_strict_overflow.
20792 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
20794 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
20795 using enabled attribute.
20796 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
20797 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
20798 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
20799 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
20800 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
20801 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
20802 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
20803 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
20804 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
20805 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
20807 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
20809 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
20811 2017-05-02 Richard Biener <rguenther@suse.de>
20813 PR tree-optimization/80591
20815 2017-04-10 Richard Biener <rguenther@suse.de>
20817 * tree-ssa-structalias.c (find_func_aliases): Properly handle
20820 2017-05-02 Richard Biener <rguenther@suse.de>
20822 PR tree-optimization/80549
20823 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
20824 (cleanup_tree_cfg_noloop): Create forwarders to known loop
20825 headers if they do not have a preheader.
20827 2017-05-02 Martin Liska <mliska@suse.cz>
20830 * common.opt: Fix typo.
20831 * doc/invoke.texi: Likewise.
20833 2017-05-01 Jan Beulich <jbeulich@suse.com>
20835 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
20836 swapping, add (x,x,m,x,n) alternative.
20838 2017-05-01 Nathan Sidwell <nathan@acm.org>
20840 * calls.c (combine_pending_stack_adjustment_and_call): Remove
20841 unnecessary unadjusted_alignment check.
20843 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
20846 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
20848 * gimplify.c (gimplify_cilk_detach): New function.
20849 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
20850 * tree-core.h: Document EXPR_CILK_SPAWN.
20851 * tree.h (EXPR_CILK_SPAWN): Define.
20853 2017-05-01 David Malcolm <dmalcolm@redhat.com>
20855 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
20856 to use new fixit_hint representation, using the "replace" logic.
20857 (get_line_span_for_fixit_hint): Likewise.
20858 (layout::print_any_fixits): Likewise.
20859 (selftest::test_one_liner_many_fixits): Rename to...
20860 (selftest::test_one_liner_many_fixits_1): ...this, and update
20861 comment and expected output to reflect that the multiple fix-it
20862 hints are now consolidated into one insertion.
20863 (selftest::test_one_liner_many_fixits_2): New test.
20864 (selftest::test_diagnostic_show_locus_one_liner): Update for
20866 (selftest::test_fixit_consolidation): Update for fix-it API
20868 * diagnostic.c (print_parseable_fixits): Likewise.
20869 * edit-context.c (edited_line::m_line_events): Convert from
20870 auto_vec <line_event *> to auto_vec <line_event>.
20871 (class line_event): Convert from abstract base class to a concrete
20872 class, taking over the role of replace_event.
20873 (class insert_event): Delete.
20874 (class replace_event): Rename to class line_event. Convert to
20876 (edit_context::add_fixits): Reimplement.
20877 (edit_context::apply_insert): Delete.
20878 (edit_context::apply_replace): Rename to...
20879 (edit_context::apply_fixit): ...this. Convert to half-open range.
20880 (edited_file::apply_insert): Delete.
20881 (edited_file::apply_replace): Rename to...
20882 (edited_file::apply_fixit): ...this.
20883 (edited_line::~edited_line): Drop deletion of events.
20884 (edited_line::apply_insert): Delete.
20885 (edited_line::apply_replace): Rename to...
20886 (edited_line::apply_fixit): ...this. Convert to half-open range.
20887 Update for change to type of m_line_events.
20888 * edit-context.h (edit_context::apply_insert): Delete.
20889 (edit_context::apply_replace): Rename to...
20890 (edit_context::apply_fixit): ...this.
20892 2017-05-01 Martin Sebor <msebor@redhat.com>
20894 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
20897 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
20900 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
20901 __get_cpuid_max returns 0.
20902 (__get_cpuid_count): Ditto.
20904 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
20906 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
20907 replacement expression is another instance of one of its arguments.
20909 2017-05-01 Jakub Jelinek <jakub@redhat.com>
20912 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
20913 check for stack push/pop autoinc.
20914 * config/i386/i386.c (ix86_agi_dependent): Return false
20915 if the only reason why modified_in_p returned true is that
20916 addr is SP based and set_insn is a push or pop.
20918 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
20920 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
20923 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
20926 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
20927 (account_size_time): Use two predicates - exec_pred and
20929 (evaluate_conditions_for_known_args): Compute both clause and
20931 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
20932 (inline_summary_t::duplicate): Update.
20933 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
20935 (compute_inline_parameters): Likewise.
20936 (estimate_edge_size_and_time): Update caluclation of time.
20937 (estimate_node_size_and_time): Compute both time and nonspecialized
20939 (estimate_ipcp_clone_size_and_time): Update.
20940 (inline_merge_summary): Update.
20941 (do_estimate_edge_time): Update.
20942 (do_estimate_edge_size): Update.
20943 (do_estimate_edge_hints): Update.
20944 (inline_read_section, inline_write_summary): Stream both new predicates.
20945 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
20947 (compute_inlined_call_time): Cleanup.
20948 (big_speedup_p): Update.
20949 (edge_badness): Update.
20950 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
20951 (size_time_entry): Replace predicate by exec_predicate and
20952 nonconst_predicate.
20953 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
20954 (estimate_edge_time): Return also nonspec_time.
20955 (reset_edge_growth_cache): Update.
20957 2017-04-29 Jakub Jelinek <jakub@redhat.com>
20959 PR rtl-optimization/80491
20960 * ifcvt.c (noce_process_if_block): When looking for x setter
20961 with missing else_bb, don't check only the insn right before
20962 cond_earliest, but look for the last insn that x is modified in
20963 within the same bb.
20965 PR rtl-optimization/80491
20966 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
20968 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
20970 PR tree-optimization/80487
20971 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
20973 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20975 PR tree-optimization/79697
20976 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
20977 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
20978 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
20980 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
20981 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
20983 2017-04-28 Martin Sebor <msebor@redhat.com>
20985 PR tree-optimization/80523
20986 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
20987 (init_target_to_host_charmap, target_to_host, target_strtol10): New
20989 (maybe_warn, format_directive, parse_directive): Use new functions.
20990 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
20992 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
20994 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
20996 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
20998 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
20999 target_header_dir): Set correctly.
21000 * configure: Regenerated.
21001 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21002 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21003 instead of SYSTEM_HEADER_DIR.
21005 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
21007 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
21008 (estimate_local_effects): Likewise.
21009 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
21010 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
21011 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
21012 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
21013 do_estimate_edge_time, estimate_edge_time): Likewise.
21014 * ipa-inline-analysis.c (estimate_node_size_and_time,
21015 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
21016 (estimate_time_after_inlining): Remove.
21018 2017-04-28 Martin Liska <mliska@suse.cz>
21020 * doc/gcov.texi: Enhance documentation of gcov.
21022 2017-04-28 Martin Liska <mliska@suse.cz>
21024 * doc/gcov.texi: Sort options in alphabetic order.
21025 * doc/gcov-dump.texi: Likewise.
21026 * doc/gcov-tool.texi: Likewise.
21027 * gcov.c (print_usage): Likewise.
21028 * gcov-dump.c (print_usage): Likewise.
21029 * gcov-tool.c (print_merge_usage_message): Likewise.
21030 (print_rewrite_usage_message): Likewise.
21031 (print_overlap_usage_message): Likewise.
21033 2017-04-28 Martin Liska <mliska@suse.cz>
21035 PR gcov-profile/53915
21036 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
21038 2017-04-28 Martin Liska <mliska@suse.cz>
21040 PR gcov-profile/79891
21041 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
21042 is marked by compiler as living on a line.
21043 (get_cycles_count): Remove usage of the union.
21044 (output_intermediate_file): Likewise.
21045 (find_source): Fix GNU coding style.
21046 (accumulate_line_counts): Remove old non-all block mode.
21047 (output_lines): Remove usage of the union.
21048 * profile.c (output_location): Include all BBs, even if
21049 belonging to a same line (and file) as a previous BB.
21051 2017-04-28 Martin Liska <mliska@suse.cz>
21053 * gcov.c (process_args): Handle new argument 'w'.
21054 (read_graph_file): Assign ID to BBs.
21055 (output_branch_count): Display BB # if verbose flag is set.
21056 (output_lines): Likewise for arcs.
21057 (print_usage): Add '--verbose' option help.
21058 * doc/gcov.texi: Document --verbose (-w) option.
21060 2017-04-28 Martin Liska <mliska@suse.cz>
21062 * gcov.c (struct block_location_info): New struct.
21063 (process_file): Fill up the new structure.
21064 (read_graph_file): Replace usage of encoding by the newly added
21066 (add_line_counts): Likewise.
21067 (accumulate_line_counts): Remove usage of the union.
21068 (function_info::function_info): New function.
21069 (function_info::~function_info): Likewise.
21070 (process_file): Call delete instead of release_function.
21071 (release_function): Release the function.
21072 (release_structures): Call delete instead of release_function.
21073 (solve_flow_graph): Replace usage of num_blocks.
21074 (find_exception_blocks): Likewise.
21075 (output_lines): Fix GNU coding style.
21077 2017-04-28 Martin Liska <mliska@suse.cz>
21080 * coverage.c (coverage_remove_note_file): New function.
21081 * coverage.h: Declare the function.
21082 * toplev.c (finalize): Clean if an error has been seen.
21084 2017-04-28 Martin Liska <mliska@suse.cz>
21086 PR gcov-profile/80031
21087 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
21088 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
21089 * gcov.c (read_graph_file): Read just number of blocks.
21090 * profile.c (branch_prob): Do not stream 0 flags per a basic
21093 2017-04-28 Martin Liska <mliska@suse.cz>
21095 * gcov-dump.c (tag_*): Add new argument to declarations.
21096 (dump_gcov_file): Likewise.
21097 (tag_blocks): Add and use new argument depth.
21098 (tag_arcs): Likewise.
21099 (tag_lines): Likewise.
21100 (tag_counters): Likewise.
21101 (tag_summary): Likewise.
21102 (dump_working_sets): Use depth to do a proper indentation.
21104 2017-04-28 Jakub Jelinek <jakub@redhat.com>
21107 * cgraph.h (symtab_node::debug_symtab): No longer inline.
21108 * symtab.c (symtab_node::debug_symtab): Move definition here.
21110 2017-04-28 Richard Biener <rguenther@suse.de>
21112 * lto-streamer.h (LTO_major_version): Bump to 7.
21114 2017-04-28 Richard Biener <rguenther@suse.de>
21116 * tree-vrp.c (assert_info): New struct.
21117 (add_assert_info): New helper.
21118 (register_edge_assert_for_2): Refactor to add asserts to a vector
21120 (register_edge_assert_for_1): Likewise.
21121 (register_edge_assert_for): Likewise.
21122 (finish_register_edge_assert_for): New helper actually registering
21123 asserts where live on edge.
21124 (find_conditional_asserts): Adjust.
21125 (find_switch_asserts): Likewise.
21126 (evrp_dom_walker::try_find_new_range): Generalize.
21127 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
21129 2017-04-27 Marek Polacek <polacek@redhat.com>
21132 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
21133 arg10 and arg11 to itype.
21135 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
21137 * doc/extend.texi (Object Size Checking): Improve grammar.
21139 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
21142 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
21143 that the logic for permitting reciprocal estimates matches that
21146 2017-04-27 Jakub Jelinek <jakub@redhat.com>
21149 * tree.c (type_cache_hasher::equal): Only compare
21150 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
21151 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
21152 non-aggregate element types.
21153 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
21154 about the flag on ARRAY_TYPEs in the comment, formatting fix.
21156 2017-04-27 Richard Biener <rguenther@suse.de>
21158 PR middle-end/80533
21159 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
21160 stripping ARRAY_REFs from MEM_EXPR make sure we're not
21161 keeping a reference to a trailing array.
21163 2017-04-27 Richard Biener <rguenther@suse.de>
21165 PR middle-end/80539
21166 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
21167 being in loop-closed SSA form conservatively.
21168 (chrec_fold_multiply_poly_poly): Likewise.
21170 2017-04-27 Tamar Christina <tamar.christina@arm.com>
21172 PR middle-end/79665
21173 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
21174 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
21176 2017-04-27 Jakub Jelinek <jakub@redhat.com>
21179 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
21180 (aarch64_function_arg_alignment): Return unsigned int again, but still
21181 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
21182 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
21183 Don't emit -Wpsabi note.
21184 (aarch64_function_arg_boundary): Likewise.
21185 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
21188 2017-04-26 Nathan Sidwell <nathan@acm.org>
21190 * tree.h (crc32_unsigned_n): Declare.
21191 (crc32_unsigned, crc32_unsigned): Make inline.
21192 * tree.c (crc32_unsigned_bits): Replace with ...
21193 (crc32_unsigned_n): ... this.
21194 (crc32_unsigned, crc32_byte): Remove.
21195 (crc32_string): Remove unnecessary braces.
21197 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
21199 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
21200 * ipa-inline-analysis.c (MAX_TIME): Remove.
21201 (account_size_time): Use sreal for time.
21202 (dump_inline_summary): Update.
21203 (estimate_function_body_sizes): Update.
21204 (estimate_edge_size_and_time): Update.
21205 (estimate_calls_size_and_time): Update.
21206 (estimate_node_size_and_time): Update.
21207 (inline_merge_summary): Update.
21208 (inline_update_overall_summary): Update.
21209 (estimate_time_after_inlining): Update.
21210 (inline_read_section): Update.
21211 (inline_write_summary): Update.
21212 * ipa-inline.c (compute_uninlined_call_time): Update.
21213 (compute_inlined_call_time): Update.
21214 (recursive_inlining): Update.
21215 (inline_small_functions): Update.
21216 (dump_overall_stats): Update.
21217 * ipa-inline.h: Include sreal.h.
21218 (size_time_entry): Turn time to sreal.
21219 (inline_summary): Turn self_time nad time to sreal.
21221 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
21223 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
21225 (sreal::stream_out, sreal::stream_in): New.
21226 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
21228 2017-04-25 Jakub Jelinek <jakub@redhat.com>
21230 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
21233 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
21236 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
21238 (dimode_scalar_chain::compute_convert_gain): Ditto.
21239 (dimode_scalar_chain::make_vector_copies): Ditto.
21240 (dimode_scalar_chain::convert_reg): Ditto.
21241 (dimode_scalar_chain::convert_insn): Ditto.
21242 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
21243 (VI248_AVX512BW_1): New mode iterator.
21244 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
21245 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
21248 2017-04-25 Martin Sebor <msebor@redhat.com>
21250 PR tree-optimization/80497
21251 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
21252 constants are representable in HOST_WIDE_INT.
21253 (parse_directive): Ditto.
21255 2017-04-25 Martin Sebor <msebor@redhat.com>
21258 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
21259 (new_zero_array): Adjust signature.
21260 (dom_info::dom_init): Used unsigned rather that size_t.
21261 (dom_info::dom_info): Same.
21263 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21264 Jakub Jelinek <jakub@redhat.com>
21267 * config/arm/arm.c: Include gimple.h.
21268 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
21269 returns negative, increment ncrn only if it returned positive.
21270 (arm_needs_doubleword_align): Return int instead of bool,
21271 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
21272 members, but if there is any such non-FIELD_DECL
21273 > PARM_BOUNDARY aligned decl, return -1 instead of false.
21274 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
21275 returns negative, increment nregs only if it returned positive.
21276 (arm_setup_incoming_varargs): Likewise.
21277 (arm_function_arg_boundary): Emit -Wpsabi note if
21278 arm_needs_doubleword_align returns negative, return
21279 DOUBLEWORD_ALIGNMENT only if it returned positive.
21281 2017-04-25 Marek Polacek <polacek@redhat.com>
21284 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
21285 first argument to type.
21287 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
21290 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
21291 type checks to test for compatibility instead of equality.
21293 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21294 Jakub Jelinek <jakub@redhat.com>
21297 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
21299 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
21300 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
21301 the alignment computation, but return their maximum in warn_alignment.
21302 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
21303 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
21305 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
21306 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
21309 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21311 * config/arc/simdext.md (dmpyh): Fix typo.
21313 2017-04-25 Richard Biener <rguenther@suse.de>
21315 PR tree-optimization/80492
21316 * alias.c (compare_base_decls): Handle registers with asm
21317 specification conservatively.
21318 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
21319 compare_base_decls returning dont-know properly.
21321 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21323 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
21324 (legitimate_offset_address_p): New function.
21325 (arc_legitimate_address_p): Use above function.
21327 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21329 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
21331 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21333 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
21334 ACCH registers whenever they are available.
21336 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21338 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
21339 double regs fix when not used.
21341 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21343 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
21345 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
21346 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
21348 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21350 * config/arc/arc.c (arc_output_addsi): Check for h-register class
21351 when emitting short ADD instructions.
21353 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21355 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
21357 (cmpsi_cc_c_insn): Likewise.
21358 (cbranchsi4_scratch): Compute proper instruction length using
21359 compact_hreg_operand.
21360 * config/arc/predicates.md (compact_hreg_operand): New predicate.
21362 2017-04-25 Richard Biener <rguenther@suse.de>
21364 PR middle-end/80509
21365 * passes.c (pass_manager::pass_manager): Initialize
21366 m_name_to_pass_map.
21368 2017-04-25 Richard Biener <rguenther@suse.de>
21370 PR tree-optimization/79201
21371 * tree-ssa-sink.c (statement_sink_location): Handle calls.
21373 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21376 * config/s390/vector.md: Split MEM->GPR vector moves for
21377 non-s_operand addresses.
21379 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21382 * config/s390/predicates.md (reload_const_wide_int_operand): New
21384 * config/s390/s390.md ("movti"): Remove d/P alternative.
21385 ("movti_bigconst"): New pattern definition.
21387 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
21390 * s390-protos.h (s390_expand_cs_hqi): Removed.
21391 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
21392 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
21393 modes as well as CCZ1mode and CCZmode.
21394 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
21395 signature of s390_emit_compare_and_swap.
21396 (s390_expand_cs_hqi): Likewise, make static.
21397 (s390_expand_cs_tdsi): Generate an explicit compare before trying
21398 compare-and-swap, in some cases.
21399 (s390_expand_cs): Wrapper function.
21400 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
21402 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
21403 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
21404 patterns for small and large integers. Forbid symref memory operands.
21405 Move expander to s390.c. Require cc register.
21406 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
21407 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
21408 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
21409 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
21410 symref memory operands. Remove CC mode and call s390_match_ccmode
21412 ("atomic_exchange<mode>"): Allow and implement all integer modes.
21414 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
21416 * config/s390/s390.md (define_peephole2): New peephole to help
21417 combining the load-and-test pattern with volatile memory.
21419 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
21421 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
21422 with CCZmode for TARGET_Z196.
21424 2017-04-25 Jakub Jelinek <jakub@redhat.com>
21426 PR rtl-optimization/80501
21427 * combine.c (make_compound_operation_int): Set subreg_code to SET
21428 even for AND with mask of the sign bit of mode.
21430 PR rtl-optimization/80500
21431 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
21432 sum's initial value.
21434 2017-04-25 Julian Brown <julian@codesourcery.com>
21435 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21437 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
21439 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
21441 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
21443 2017-04-25 Julian Brown <julian@codesourcery.com>
21444 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21446 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
21447 (thunderx2t99_sha): New Reservation.
21449 2017-04-25 Julian Brown <julian@codesourcery.com>
21450 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21452 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
21453 type for 1-element load.
21455 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
21457 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
21459 2017-04-24 Martin Jambor <mjambor@suse.cz>
21461 PR tree-optimization/80293
21462 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
21463 char arrays not totally scalarizable if it is false.
21464 (analyze_all_variable_accesses): Pass correct value in the new
21465 parameter. Add a statistics counter.
21467 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
21469 PR middle-end/79931
21470 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
21472 2017-04-24 Richard Biener <rguenther@suse.de>
21474 PR tree-optimization/80494
21475 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
21476 out for complex types.
21478 2017-04-24 Richard Biener <rguenther@suse.de>
21480 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
21481 * tree-ssa-sccvn.c (print_scc): Print SCC size.
21482 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
21483 (DFS): Adjust and never fail.
21484 (sccvn_dom_walker::fail): Remove.
21485 (sccvn_dom_walker::before_dom_children): Adjust.
21486 (run_scc_vn): Likewise and never fail.
21487 * tree-ssa-pre.c (pass_pre::execute): Adjust.
21488 (pass_fre::execute): Likewise.
21490 2017-04-24 Richard Biener <rguenther@suse.de>
21492 PR tree-optimization/79725
21493 * tree-ssa-sink.c (statement_sink_location): Return whether
21494 failure reason was zero uses. Move that check later.
21495 (sink_code_in_bb): Deal with zero uses by removing the stmt
21498 2017-04-24 Richard Biener <rguenther@suse.de>
21501 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
21502 pointer-based references.
21504 2017-04-24 Richard Biener <rguenther@suse.de>
21507 * pass_manager.h (pass_manager::operator new): Remove.
21508 (pass_manager::operator delete): Likewise.
21509 * passes.c (pass_manager::operator new): Remove.
21510 (pass_manager::operator delete): Likewise.
21511 (pass_manager::pass_manager): Zero individual pass members.
21513 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
21516 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
21517 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
21518 Check "XEXP (src, 1)" operand here.
21519 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
21520 Check "XEXP (src, 1)" operand here.
21521 (dimode_scalar_chain::make_vector_copies): Detect count register
21522 of a shift instruction. Zero extend count register from QImode
21523 to DImode to satisfy vector shift pattern count operand predicate.
21524 Substitute vector shift count operand with a DImode copy.
21525 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
21528 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
21530 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
21531 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
21532 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
21533 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
21534 (UNSPEC_NOREX_MEM): Remove definition.
21536 2017-04-21 Richard Biener <rguenther@suse.de>
21538 PR tree-optimization/79547
21539 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
21540 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
21541 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
21542 without any constraints.
21544 2017-04-21 Richard Biener <rguenther@suse.de>
21546 PR tree-optimization/78847
21547 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
21549 2017-04-21 Richard Biener <rguenther@suse.de>
21551 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
21552 (build_distinct_type_copy): Likewise.
21553 (build_variant_type_copy): Likewise.
21554 * tree.c (build_qualified_type): Pass down mem-stat info.
21555 (build_distinct_type_copy): Likewise.
21556 (build_variant_type_copy): Likewise.
21558 2017-04-21 Richard Biener <rguenther@suse.de>
21560 PR tree-optimization/80237
21561 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
21563 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
21564 for a simplified result.
21566 2016-04-21 Richard Biener <rguenther@suse.de>
21568 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
21569 sth as strict as a simple_iv but a chrec without symbols and an
21570 operand defined in the loop we are peeling (and not some subloop).
21571 (propagate_constants_for_unrolling): Propagate all constants.
21573 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
21576 * config/i386/i386.c (print_reg): Remove assert for disalowed
21577 regno values, call output_operand_lossage instead.
21579 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
21582 * config/i386/constraints.md (Yc): New register constraint.
21583 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
21584 Use Yc constraint for alternative 2 of operand 0. Remove
21585 preferred_for_speed attribute.
21587 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
21589 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
21590 lastprivate clauses in SIMT case.
21592 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
21594 * doc/invoke.texi (-Wextra-semi): Document new warning option.
21596 2017-04-20 Richard Biener <rguenther@suse.de>
21598 PR tree-optimization/57796
21599 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
21600 as N scalar stores.
21601 (vect_model_load_cost): Cost gathers as N scalar loads.
21603 2017-04-20 Richard Biener <rguenther@suse.de>
21605 * ggc-page.c (ggc_allocated_p): Rename to ...
21606 (safe_lookup_page_table_entry): ... this and return the lookup
21608 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
21610 2017-04-20 Richard Biener <rguenther@suse.de>
21612 PR tree-optimization/80453
21613 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
21614 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
21615 from the conditions.
21616 (vn_phi_eq): Pass them down.
21617 (vn_phi_lookup): Record them.
21618 (vn_phi_insert): Likewise.
21620 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
21622 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
21623 uninitialized variable warning to avoid buffer overrun.
21625 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
21628 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
21629 is suppressed for '{ 0 }' in C.
21631 2017-04-20 Jakub Jelinek <jakub@redhat.com>
21633 * BASE-VER: Set to 8.0.0.
21635 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
21637 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
21638 priority .init_array and .fini_array section with SECTION_NOTYPE
21641 2017-04-20 Jakub Jelinek <jakub@redhat.com>
21643 PR middle-end/80423
21644 * tree.h (build_array_type): Add typeless_storage default argument.
21645 * tree.c (type_cache_hasher::equal): Also compare
21646 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
21647 (build_array_type): Add typeless_storage argument, set
21648 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
21650 (build_nonshared_array_type): Adjust build_array_type_1 caller.
21651 (build_array_type): Likewise. Add typeless_storage argument.
21653 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
21654 Jakub Jelinek <jakub@redhat.com>
21656 PR tree-optimization/80426
21657 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
21658 operation on symbolic operands, also compute the overflow for the
21659 invariant part when the operation degenerates into a negation.
21661 2017-04-19 Jakub Jelinek <jakub@redhat.com>
21664 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
21665 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
21668 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
21670 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
21673 * config/avr/avr.c (tree.h): Include it.
21674 (cgraph.h): Include it.
21675 (avr_encode_section_info): Don't warn for uninitialized progmem
21676 variable if it's just an alias.
21678 2017-04-19 Richard Biener <rguenther@suse.de>
21681 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
21682 when needed by AutoPGO.
21684 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
21687 * doc/lto.texi: Remove an extra 'that'.
21689 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
21691 PR rtl-optimization/80429
21692 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
21693 are only used in debug insns.
21695 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
21696 Vladimir Makarov <vmakarov@redhat.com>
21698 * config/sparc/predicates.md (input_operand): Add comment. Return
21699 true for any memory operand when LRA is in progress.
21700 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
21702 2017-04-18 Jeff Law <law@redhat.com>
21705 * mips.md ({return,simple_return}_internal): Do not overwrite
21708 2017-04-18 Jakub Jelinek <jakub@redhat.com>
21710 PR tree-optimization/80443
21711 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
21712 instead of adding 1, subtract -1 and similarly instead of subtracting
21715 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
21717 PR rtl-optimization/80357
21718 * haifa-sched.c (tmp_bitmap): New variable.
21719 (model_recompute): Handle duplicate use records.
21720 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
21721 (free_global_sched_pressure_data): Free it.
21723 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
21726 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
21727 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21728 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21729 instead of SYSTEM_HEADER_DIR.
21731 2017-04-18 Jeff Law <law@redhat.com>
21733 PR middle-end/80422
21734 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
21735 predecessors after walking up the insn chain.
21737 2017-04-18 Jakub Jelinek <jakub@redhat.com>
21740 * dwarf2out.c (modified_type_die): Try harder not to emit internal
21741 sizetype type into debug info.
21743 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
21746 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
21747 unneeded test for TARGET_UPPER_REGS_SF.
21748 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
21750 2017-04-18 Jakub Jelinek <jakub@redhat.com>
21753 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
21754 instead of gsi_after_labels.
21756 2017-04-18 Jeff Law <law@redhat.com>
21758 * regcprop.c (maybe_mode_change): Avoid creating copies of the
21762 2017-04-13 Jeff Law <law@redhat.com>
21763 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
21764 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
21766 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
21769 * config/avr/avr.c (intl.h): Include it.
21770 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
21772 2017-04-18 Martin Liska <mliska@suse.cz>
21774 PR gcov-profile/78783
21775 * gcov-tool.c (gcov_output_files): Validate that destination
21776 file is either removed by the tool or by a user.
21778 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
21779 Guy Benyei <guybe@mellanox.com>
21781 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
21782 block, and do not negate it, the stored id is already negative.
21784 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
21786 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
21788 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
21791 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
21792 masks of options that should be turned off if the VSX vector
21793 options are turned off.
21794 (OTHER_P8_VECTOR_MASKS): Likewise.
21795 (OTHER_VSX_VECTOR_MASKS): Likewise.
21796 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
21797 rs6000_disable_incompatible_switches to validate no type switches
21799 (rs6000_incompatible_switch): New function to disallow turning on
21800 other vector options if -mno-vsx, -mno-power8-vector, or
21801 -mno-power9-vector are specified.
21803 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21805 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
21807 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21809 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
21810 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
21811 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
21812 (ARG_POINTER_CFA_OFFSET): Likewise.
21814 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21816 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
21817 conditions to take advantage of various optimizations.
21819 2017-04-13 Jeff Law <law@redhat.com>
21821 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
21822 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
21823 (zero_extendsidi2_dext): Likewise.
21825 2017-04-13 Jakub Jelinek <jakub@redhat.com>
21828 * fold-const.c (fold_ternary_loc): Revert
21829 use op0 instead of fold_convert_loc (loc, type, arg0) part of
21832 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
21834 PR rtl-optimization/80343
21835 * lra-remat.c (update_scratch_ops): Assign original hard reg to
21836 new scratch pseudo.
21838 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
21841 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
21842 to ubsan_encode_value.
21844 2017-04-13 Jeff Law <law@redhat.com>
21846 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
21847 appearing in DEBUG_INSNs.
21849 2017-04-13 Martin Liska <mliska@suse.cz>
21851 PR gcov-profile/80413
21852 * gcov-io.c (gcov_write_string): Copy to buffer just when
21853 allocated size is greater than zero.
21855 2017-04-13 Jakub Jelinek <jakub@redhat.com>
21858 * dwarf2out.c (decls_for_scope): Ignore declarations of
21859 current_function_decl in BLOCK_NONLOCALIZED_VARS.
21861 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
21864 * ipa-visibility.c (non_local_p): Fix typos.
21865 (localize_node): When localizing symbol in same comdat group,
21866 dissolve the group only when we know external symbols are going
21868 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
21870 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21872 PR tree-optimization/79390
21873 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
21874 order does not result in usable sequence, retry with reversed operand
21880 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
21881 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
21882 op0 instead of fold_convert_loc (loc, type, arg0).
21884 2017-04-12 Jeff Law <law@redhat.com>
21886 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
21887 has a delay slot in the generated code.
21889 * config/cris/cris.md (cris_preferred_reload_class): Return
21890 GENNONACR_REGS rather than GENERAL_REGS.
21892 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21895 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
21896 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
21897 signedness of the result type.
21899 2017-04-12 Richard Biener <rguenther@suse.de>
21900 Jeff Law <law@redhat.com>
21902 PR tree-optimization/80359
21903 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
21904 trim stores to TARGET_MEM_REFs.
21906 2017-04-12 Richard Biener <rguenther@suse.de>
21908 PR tree-optimization/79390
21909 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
21910 threading case even more.
21912 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
21915 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
21916 for quad_address_p for TImode, instead of just not indexed_address.
21918 2017-04-12 Richard Biener <rguenther@suse.de>
21919 Bernd Edlinger <bernd.edlinger@hotmail.de>
21921 PR middle-end/79671
21922 * alias.c (component_uses_parent_alias_set_from): Handle
21923 TYPE_TYPELESS_STORAGE.
21924 (get_alias_set): Likewise.
21925 * tree-core.h (tree_type_common): Add typeless_storage flag.
21926 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
21927 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
21928 for types containing members with TYPE_TYPELESS_STORAGE.
21929 (place_field): Likewise.
21930 (layout_type): Likewise for ARRAY_TYPE.
21931 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
21932 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
21933 TYPE_TYPELESS_STORAGE.
21934 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
21936 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21939 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
21940 first argument to type.
21942 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21946 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
21947 CONST0_RTX (mode) rather than const0_rtx where appropriate.
21948 (rs6000_expand_binop_builtin): Likewise.
21949 (rs6000_expand_ternop_builtin): Likewise; also add missing
21950 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
21951 vshasigma built-ins.
21952 * doc/extend.texi: Document that vec_xxpermdi's third argument
21953 must be a constant.
21955 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
21957 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
21958 Use shift_const cost parameter when calculating gain of STV shifts.
21960 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
21962 PR rtl-optimization/70478
21963 * lra-constraints.c (process_alt_operands): Check memory for
21964 disfavoring memory insn operand.
21966 2017-04-11 Jakub Jelinek <jakub@redhat.com>
21968 PR middle-end/80100
21969 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
21970 left shift in unsigned HOST_WIDE_INT type.
21972 PR rtl-optimization/80385
21973 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
21974 (not (neg X)) into (plus X -1) for complex or non-integral modes.
21977 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
21978 if they have any depend clauses.
21980 2017-04-11 Martin Liska <mliska@suse.cz>
21983 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
21984 * ipa-split.c (split_function): Create a local comdat symbol
21985 if caller is in a comdat group.
21987 2017-04-11 Martin Liska <mliska@suse.cz>
21990 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
21993 2017-04-11 Martin Sebor <msebor@redhat.com>
21995 PR middle-end/80364
21996 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
21997 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
21999 (directive::set_width, directive::set_precision, format_character):
22001 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
22004 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
22007 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
22008 conflict, set target->arch_name instead of target->cpu_name.
22010 2017-04-11 Richard Biener <rguenther@suse.de>
22012 PR tree-optimization/80374
22013 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
22014 build_zero_cst, remove fold_convertible_p check again.
22016 2017-04-11 Martin Liska <mliska@suse.cz>
22019 * ubsan.c (instrument_object_size): Do not instrument register
22022 2017-04-11 Jakub Jelinek <jakub@redhat.com>
22025 * config/i386/i386-builtin-types.def
22026 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
22027 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
22028 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
22029 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
22030 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
22031 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
22032 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
22033 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
22034 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
22035 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
22036 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
22037 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
22038 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
22039 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
22040 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
22041 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
22042 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
22043 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
22044 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
22045 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
22046 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
22047 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
22048 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
22049 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
22050 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
22051 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
22052 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
22053 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
22054 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
22055 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
22056 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
22057 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
22058 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
22059 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
22060 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
22061 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
22062 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
22063 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
22064 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
22065 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
22066 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
22067 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
22068 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
22069 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
22070 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
22071 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
22073 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
22074 flag to second_arg_count, handle 4 argument function type _COUNT
22075 aliases, handle second_arg_count on second argument rather than last.
22077 2017-04-10 Jeff Law <law@redhat.com>
22079 PR tree-optimization/80374
22080 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
22081 record anything if we can not convert integer_zero_node to the
22084 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
22087 * config/rs6000/rs6000.c (rs6000_option_override_internal):
22088 Enhance special handling given to the TARGET_P9_MINMAX option in
22089 relation to certain other options.
22091 2017-04-10 Bin Cheng <bin.cheng@arm.com>
22093 PR tree-optimization/80153
22094 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
22095 remove POINTER_PLUS_EXPR's base part directly, rather than through
22098 2017-04-10 Richard Biener <rguenther@suse.de>
22099 Bin Cheng <bin.cheng@arm.com>
22101 PR tree-optimization/80153
22102 * tree-affine.c (aff_combination_to_tree): Get base pointer from
22103 the first element of pointer type aff_tree. Build result expr in
22105 (add_elt_to_tree): Convert to type unconditionally. Remove other
22106 fold_convert calls.
22107 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
22108 (rewrite_use_nonlinear_expr): Check invariant using iv information.
22110 2017-04-10 Richard Biener <rguenther@suse.de>
22112 * tree-ssa-structalias.c (find_func_aliases): Properly handle
22115 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
22117 PR rtl-optimization/70478
22118 * lra-constraints.c (curr_small_class_check): New.
22119 (update_and_check_small_class_inputs): New.
22120 (process_alt_operands): Update curr_small_class_check. Disfavor
22121 alternative insn memory operands. Check available regs for small
22124 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
22127 * config/mips/mips.opt (-mvirt): Update description.
22128 * doc/invoke.texi (-mvirt): Likewise.
22130 2017-04-10 Richard Biener <rguenther@suse.de>
22132 PR middle-end/80362
22133 * fold-const.c (fold_binary_loc): Look at unstripped ops when
22134 looking for NEGATE_EXPR in -A / -B to A / B folding.
22136 2017-04-10 Martin Liska <mliska@suse.cz>
22138 PR gcov-profile/80224
22139 * gcov.c (print_usage): Fix usage string.
22140 (get_gcov_intermediate_filename): Remove.
22141 (output_gcov_file): Use both for normal and intermediate format.
22142 (generate_results): Do not initialize special file for
22143 intermediate format.
22145 2017-04-10 Richard Biener <rguenther@suse.de>
22147 PR tree-optimization/80304
22148 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
22151 2017-04-10 Nathan Sidwell <nathan@acm.org>
22154 * config/rs6000/rs6000.c (rs6000_vector_type): New.
22155 (rs6000_init_builtins): Use it.
22157 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22159 * config/arm/arm.md (<mrc>): Add mode to SET source.
22160 (<mrrc>): Likewise.
22162 2017-04-10 Richard Biener <rguenther@suse.de>
22164 PR middle-end/80344
22165 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
22167 2017-04-10 Jakub Jelinek <jakub@redhat.com>
22170 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
22171 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
22172 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
22173 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
22174 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
22175 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
22176 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
22177 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
22178 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
22179 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
22180 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
22181 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
22182 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
22183 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
22184 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
22185 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
22186 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
22187 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
22188 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
22189 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
22190 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
22191 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
22192 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
22194 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
22196 PR rtl-optimization/70478
22197 * lra-constraints.c: Reverse the last patch.
22199 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
22201 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
22202 Add comment for WCHAR_T.
22204 2017-04-08 Martin Liska <mliska@suse.cz>
22207 2017-04-07 Martin Liska <mliska@suse.cz>
22210 * ipa-split.c (split_function): Add function part to a same comdat
22213 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22216 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
22218 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
22220 * rs6000/rs6000.c (vec_load_pendulum): Rename...
22221 (vec_pairing): ...to this.
22222 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
22223 (rs6000_sched_init): Adjust for name change.
22224 (struct rs6000_sched_context): Likewise.
22225 (rs6000_init_sched_context): Likewise.
22226 (rs6000_set_sched_context): Likewise.
22228 2017-04-07 Jakub Jelinek <jakub@redhat.com>
22234 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
22236 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
22237 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
22238 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
22240 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
22242 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
22244 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
22246 PR rtl-optimization/70703
22247 * ira-color.c (update_conflict_hard_regno_costs): Use
22248 int64_t instead of HOST_WIDE_INT.
22250 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
22252 PR rtl-optimization/70478
22253 * lra-constraints.c (process_alt_operands): Disfavor alternative
22254 insn memory operands.
22256 2017-04-07 Jeff Law <law@redhat.com>
22258 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
22259 CALL and NOTE_INSN_CALL_ARG_LOCATION.
22261 2017-04-07 Martin Liska <mliska@suse.cz>
22264 * config/aarch64/aarch64.c (aarch64_process_target_attr):
22265 Show error message instead of an ICE.
22267 2017-04-07 Martin Liska <mliska@suse.cz>
22270 * ipa-split.c (split_function): Add function part to a same comdat
22273 2017-04-07 Richard Biener <rguenther@suse.de>
22275 PR middle-end/80341
22276 * tree.c (get_unwidened): Also handle ! for_type case for
22278 * convert.c (do_narrow): Split out from ...
22279 (convert_to_integer_1): ... here. Do not pass final truncation
22280 type to get_unwidened for TRUNC_DIV_EXPR.
22282 2017-04-07 Richard Biener <rguenther@suse.de>
22284 * tree-affine.c (wide_int_ext_for_comb): Take type rather
22286 (aff_combination_const): Adjust.
22287 (aff_combination_scale): Likewise.
22288 (aff_combination_add_elt): Likewise.
22289 (aff_combination_add_cst): Likewise.
22290 (aff_combination_convert): Likewise.
22291 (add_elt_to_tree): Likewise. Remove unused argument.
22292 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
22294 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
22296 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
22298 * config/arm/arm.c (arm_default_short_enums): Use
22299 ARM_DEFAULT_SHORT_ENUMS.
22300 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
22302 2017-04-06 Jakub Jelinek <jakub@redhat.com>
22305 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
22306 members with redundant out-of-class redeclaration.
22308 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
22311 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
22312 * config/i386/i386.md (*zero_extendsidi2):
22313 Add (?*x,*x) and (?*v,*v) alternatives.
22315 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
22318 * config/i386/i386.c (ix86_expand_builtin)
22319 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
22320 mode from insn data. Convert operands to insn operand mode.
22321 Copy operands that don't satisfy insn predicate to a register.
22323 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
22325 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
22328 2017-04-06 Richard Biener <rguenther@suse.de>
22330 PR tree-optimization/80334
22331 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
22332 preserve alignment of accesses.
22334 2017-04-06 Richard Biener <rguenther@suse.de>
22336 PR tree-optimization/80262
22337 * tree-sra.c (build_ref_for_offset): Preserve address-space
22339 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
22340 Drop useless address-space information on MEM_REF offsets.
22342 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
22344 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
22346 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
22348 PR rtl-optimization/70703
22349 * ira-color.c (update_conflict_hard_regno_costs): Use
22350 HOST_WIDE_INT instead of long.
22352 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
22355 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
22356 not defined for x86_64 target. Add -mmmx target option when __SSE2__
22358 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
22359 for x86_64 target. Handle -m3dnowa option.
22361 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
22363 PR rtl-optimization/70703
22364 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
22365 (update_conflict_hard_regno_costs): Use long instead of unsigned
22366 arithmetic for cost calculation.
22368 2017-04-05 Jakub Jelinek <jakub@redhat.com>
22369 Bernd Edlinger <bernd.edlinger@hotmail.de>
22372 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
22375 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
22378 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
22379 ptr_mode with Pmode throughout.
22380 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
22381 into probe_stack_range and use DImode.
22383 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
22386 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
22387 call_eh_return is true.
22389 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22391 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
22392 Initialize last_match_fntype_index.
22394 2017-04-05 Jakub Jelinek <jakub@redhat.com>
22397 * tree-nvr.c: Include internal-fn.h.
22398 (pass_return_slot::execute): Ignore internal calls without
22401 2017-04-04 Jakub Jelinek <jakub@redhat.com>
22402 Richard Biener <rguenther@suse.de>
22405 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
22406 captures used multiple times, except for the last use.
22407 * generic-match-head.c: Include gimplify.h.
22409 2017-04-04 Jakub Jelinek <jakub@redhat.com>
22411 PR tree-optimization/79390
22412 * target.h (struct noce_if_info): Declare.
22413 * targhooks.h (default_noce_conversion_profitable_p): Declare.
22414 * target.def (noce_conversion_profitable_p): New target hook.
22415 * ifcvt.h (struct noce_if_info): New type, moved from ...
22416 * ifcvt.c (struct noce_if_info): ... here.
22417 (noce_conversion_profitable_p): Renamed to ...
22418 (default_noce_conversion_profitable_p): ... this. No longer
22420 (noce_try_store_flag_constants, noce_try_addcc,
22421 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
22422 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
22423 instead of noce_conversion_profitable_p.
22424 * config/i386/i386.c: Include ifcvt.h.
22425 (ix86_option_override_internal): Don't override
22426 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
22427 (ix86_noce_conversion_profitable_p): New function.
22428 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
22429 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
22430 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
22431 * doc/tm.texi: Regenerated.
22433 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22435 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
22438 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22441 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
22442 instructions for small multiply cores.
22444 2017-04-04 Jeff Law <law@redhat.com>
22446 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
22448 (mips_expand_vec_perm_const): Initialize elements in orig_perm
22449 that are not set by the loop over the elements.
22451 2017-04-04 Jakub Jelinek <jakub@redhat.com>
22454 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
22455 int mode, convert_modes it to mode as unsigned, otherwise use
22456 lowpart_subreg to mode rather than SImode.
22457 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
22458 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
22459 Use DImode instead of SImode for the shift count operand.
22460 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
22463 2017-04-04 Richard Biener <rguenther@suse.de>
22465 PR middle-end/80281
22466 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
22467 arithmetic done for the negate or the plus. Simplify.
22468 (A - (-B) -> A + B): Likewise.
22469 * fold-const.c (split_tree): Make sure to not negate pointers.
22471 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
22473 PR rtl-optimization/60818
22474 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
22475 a compare of comparisons with the thing compared if this results
22476 in a different machine mode.
22478 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
22480 * alias.c (base_alias_check): Fix typo in comment.
22481 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
22482 * cgraphunit.c (symbol_table::compile): Likewise.
22483 * collect2.c (maybe_run_lto_and_relink): Likewise.
22484 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
22485 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
22486 * config/avr/avr.c (avr_map_op_t): Likewise.
22487 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
22488 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
22489 * config/epiphany/epiphany.md (movcc): Likewise.
22490 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
22491 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
22493 * config/mips/mips.c (mips_save_restore_reg): Likewise.
22494 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
22495 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
22496 * config/sh/sh.c (sh_rtx_costs): Likewise.
22497 * fold-const.c (fold_truth_andor): Likewise.
22498 * genautomata.c (collapse_flag): Likewise.
22499 * gengtype.h (struct type::u::s): Likewise.
22500 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
22501 * input.c (FORMAT_AMOUNT): Likewise.
22502 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
22503 (known_aggs_to_agg_replacement_list): Likewise.
22504 * ipa-inline-analysis.c: Likewise.
22505 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
22506 * ipa-polymorphic-call.c
22507 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
22508 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
22509 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
22511 * modulo-sched.c (apply_reg_moves): Likewise.
22512 * omp-expand.c (build_omp_regions_1): Likewise.
22513 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
22514 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
22515 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
22516 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
22517 * value-prof.c: Likewise.
22518 * var-tracking.c (val_reset): Likewise.
22520 2017-04-03 Richard Biener <rguenther@suse.de>
22522 PR tree-optimization/80275
22523 * fold-const.c (split_address_to_core_and_offset): Handle
22526 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
22528 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
22529 descriptors is at least equal to that of functions.
22531 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
22533 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
22535 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
22538 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
22539 (mov<IMOD4:mode>): New expander.
22540 (*mov<IMOD4:mode>_internal): New insn and split pattern.
22542 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
22544 PR rtl-optimization/79405
22545 * fwprop.c (propagations_left): New variable.
22546 (forward_propagate_into): Decrement it.
22547 (fwprop_init): Initialize it.
22548 (fw_prop): If the variable has reached zero, stop propagating.
22549 (fwprop_addr): Ditto.
22551 2017-03-31 Jakub Jelinek <jakub@redhat.com>
22554 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
22555 a FUNCTION_DECL, pass it as decl instead of origin to
22558 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
22560 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
22563 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
22566 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
22567 TARGET_VSX_SMALL_INTEGER.
22569 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22571 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
22572 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
22574 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
22576 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
22577 extraction from odd-numbered MSA register.
22579 2017-03-31 Jakub Jelinek <jakub@redhat.com>
22581 PR middle-end/80173
22582 * expmed.c (store_bit_field_1): Don't attempt to create
22583 a word subreg out of hard registers wider than word if they
22584 have HARD_REGNO_NREGS of 1 for their mode.
22586 PR middle-end/80163
22587 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
22588 conversions to integer types wider than word and pointer.
22591 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
22592 (rtx_equal_for_cselib_p): Pass 0 to it.
22593 * cselib.c (cselib_hasher::equal): Likewise.
22594 (rtx_equal_for_cselib_1): Add depth argument. If depth
22595 is 128, don't look up VALUE locs and punt. Increment
22596 depth in recursive calls when walking VALUE locs.
22598 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
22600 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
22601 (make_gcov_file_name): Use the canonical path name for generating
22603 (read_line): Fix handling of files with ascii null bytes.
22605 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
22607 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
22608 to initialise a vector register instead
22609 of using a const_int.
22611 2017-03-30 Jakub Jelinek <jakub@redhat.com>
22613 PR translation/80189
22614 * gimplify.c (omp_default_clause): Use %qs instead of %s in
22615 diagnostic messages.
22617 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
22620 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
22621 (dfp_diex_<mode>): Update mode of operand 1.
22622 * doc/extend.texi (dxex, dxexq): Document change to return type.
22623 (diex, diexq): Document change to argument type.
22625 2017-03-30 Martin Jambor <mjambor@suse.cz>
22628 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
22629 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
22630 it reflects the signature changes performed at the callee side.
22631 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
22632 to cgraph_build_function_type_skip_args.
22633 (build_function_decl_skip_args): Adjust call to the above function.
22635 2017-03-30 Jakub Jelinek <jakub@redhat.com>
22638 * config/i386/sse.md
22639 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
22640 register as dest whenever it is a MEM not rtx_equal_p to the
22641 corresponding dup operand, and when forcing into reg move the
22642 reg into the memory afterwards.
22643 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
22644 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
22645 for the force_reg mode.
22646 (avx512vl_vextractf128<mode>): Use register as dest either
22647 always when a MEM, or when it is a MEM not rtx_equal_p to the
22648 corresponding dup operand, or even not when it is a CONST_VECTOR
22649 depending on the mode and lo vs. hi.
22650 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
22652 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
22653 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
22654 Likewise. Require that operands[2] is even.
22655 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
22656 Remove extraneous parens. Require that operands[2] is a multiple
22658 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
22659 operands[0] is a MEM if <mask_applied>, the predicates/constraints
22660 disallow memory then.
22662 2017-03-30 Richard Biener <rguenther@suse.de>
22664 PR tree-optimization/77498
22665 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
22666 to non-constants over backedges.
22668 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
22670 PR rtl-optimization/80233
22671 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
22672 as last_combined_insn. Do not test for BARRIER_P separately.
22674 2017-03-29 Andreas Schwab <schwab@suse.de>
22677 * calls.c (prepare_call_address): Convert funexp to Pmode before
22678 copying to temp reg.
22680 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22682 PR tree-optimization/80158
22683 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
22684 Handle possible future case of more than one alternate
22686 (replace_rhs_if_not_dup): Likewise.
22687 (replace_one_candidate): Likewise.
22689 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
22691 PR rtl-optimization/80193
22692 * ira.c (ira): Do not check allocation for LRA.
22694 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
22696 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
22697 (nvptx_output_simt_exit): Declare.
22698 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
22699 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
22700 (init_softstack_frame): Move initialization of crtl->is_leaf to...
22701 (nvptx_declare_function_name): ...here. Emit declaration of local
22702 memory space buffer for omp_simt_enter insn.
22703 (nvptx_output_unisimt_switch): New.
22704 (nvptx_output_softstack_switch): New.
22705 (nvptx_output_simt_enter): New.
22706 (nvptx_output_simt_exit): New.
22707 * config/nvptx/nvptx.h (struct machine_function): New fields
22708 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
22709 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
22710 (UNSPECV_SIMT_EXIT): Ditto.
22711 (omp_simt_enter_insn): New insn.
22712 (omp_simt_enter): New expansion.
22713 (omp_simt_exit): New insn.
22714 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
22716 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
22717 (expand_GOMP_SIMT_ENTER_ALLOC): New.
22718 (expand_GOMP_SIMT_EXIT): New.
22719 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
22720 (GOMP_SIMT_ENTER_ALLOC): Ditto.
22721 (GOMP_SIMT_EXIT): Ditto.
22722 * target-insns.def (omp_simt_enter): New insn.
22723 (omp_simt_exit): Ditto.
22724 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
22726 (lower_rec_simd_input_clauses): Implement SIMT privatization.
22727 (lower_rec_input_clauses): Likewise.
22728 (lower_lastprivate_clauses): Handle SIMT privatization.
22730 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
22731 (ompdevlow_adjust_simt_enter): New.
22732 (find_simtpriv_var_op): New.
22733 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
22734 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
22736 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
22737 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
22738 (copy_decl_for_dup_finish): Ditto.
22740 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
22742 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
22745 * config/i386/i386.c (ix86_option_override_internal): Always
22746 allow -mpreferred-stack-boundary=3 for 64-bit targets.
22748 2017-03-28 Bin Cheng <bin.cheng@arm.com>
22750 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
22752 2017-03-28 Bin Cheng <bin.cheng@arm.com>
22754 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
22755 mark new edge's irreducible flag accordign to it.
22756 (vect_do_peeling): Check loop preheader edge's irreducible flag
22757 and pass it to function slpeel_add_loop_guard.
22759 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
22761 PR tree-optimization/80218
22762 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
22763 Update block frequencies and counts.
22765 2017-03-28 Richard Biener <rguenther@suse.de>
22767 PR tree-optimization/78644
22768 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
22769 of a simplification result we may not use it at all.
22771 2017-03-28 Richard Biener <rguenther@suse.de>
22774 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
22775 without arguments, generate default definition of a SSA name.
22777 2017-03-28 Richard Biener <rguenther@suse.de>
22779 PR middle-end/80222
22780 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
22781 TYPE_REF_CAN_ALIAS_ALL references.
22782 * fold-const.c (fold_indirect_ref_1): Likewise.
22784 2017-03-28 Martin Liska <mliska@suse.cz>
22787 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
22788 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
22790 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
22791 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
22793 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
22794 (EXTRA_SPECS): Define.
22795 (SUBTARGET_EXTRA_SPECS): Likewise.
22796 (SUBTARGET_CPP_SPEC): Likewise.
22797 * config/arc/elf.h (EXTRA_SPECS): Renamed to
22798 SUBTARGET_EXTRA_SPECS.
22799 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
22801 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
22803 * config/arc/simdext.md (vst64_insn): Update pattern.
22804 (vld32wh_insn): Likewise.
22805 (vld32wl_insn): Likewise.
22806 (vld64_insn): Likewise.
22807 (vld32_insn): Likewise.
22809 2017-03-28 Marek Polacek <polacek@redhat.com>
22812 * fold-const.c (fold_comparison): Use protected_set_expr_location
22813 instead of SET_EXPR_LOCATION.
22815 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
22817 * tree.c (add_expr): Avoid name lookup warning.
22819 2017-03-27 Jeff Law <law@redhat.com>
22821 PR tree-optimization/80216
22822 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
22823 function name. Limit recursion depth.
22824 (record_temporary_equivalences): Corresponding changes.
22826 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
22828 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
22831 2017-03-27 Jakub Jelinek <jakub@redhat.com>
22834 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
22836 * cfgcleanup.c (reg_note_cfa_p): New array.
22837 (insns_have_identical_cfa_notes): New function.
22838 (old_insns_match_p): Don't cross-jump in between /f
22839 and non-/f instructions. If both i1 and i2 are frame related,
22840 verify all CFA notes, their order and content.
22842 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
22845 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
22846 HImode and SImode with zero extend to DImode to one insn.
22847 (bswap<mode>2_extenddi): Likewise.
22848 (bswapsi2_extenddi): Likewise.
22849 (bswaphi2_extendsi): Likewise.
22850 (bswaphi2): Combine bswap HImode and SImode into one insn.
22851 Separate memory insns from swapping register.
22852 (bswapsi2): Likewise.
22853 (bswap<mode>2): Likewise.
22854 (bswaphi2_internal): Delete, no longer used.
22855 (bswapsi2_internal): Likewise.
22856 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
22857 store, and gpr<-gpr swap insns.
22858 (bswap<mode>2_store): Likewise.
22859 (bswaphi2_reg): Register only splitter, combine with the splitter.
22860 (bswaphi2 splitter): Likewise.
22861 (bswapsi2_reg): Likewise.
22862 (bswapsi2 splitter): Likewise.
22863 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
22864 the insns into load, store, and register/register insns.
22865 (bswapdi2_ldbrx): Likewise.
22866 (bswapdi2_load): Likewise.
22867 (bswapdi2_store): Likewise.
22868 (bswapdi2_reg): Likewise.
22870 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
22872 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
22873 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
22875 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
22878 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
22880 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
22881 special handling for target option conflicts between dform
22882 options (-mpower9-dform, -mpower9-dform-vector,
22883 -mpower9-dform-scalar) and -mno-direct-move.
22885 2017-03-27 Richard Biener <rguenther@suse.de>
22887 PR tree-optimization/80181
22888 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
22890 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22892 * config/arc/predicates.md (move_double_src_operand): Replace the
22893 call to move_double_src_operand with a call to address_operand.
22895 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22897 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
22898 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
22899 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
22901 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22903 * config/arc/predicates.md (long_immediate_loadstore_operand):
22904 Consider scaled addresses cases.
22906 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22908 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
22909 restored when in interrupt.
22910 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
22911 doesn't have delay slot.
22913 2017-03-27 Richard Biener <rguenther@suse.de>
22916 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
22917 inlined thunk clones.
22919 2017-03-27 Jakub Jelinek <jakub@redhat.com>
22922 * asan.c (instrument_derefs): Copy over last operand from
22923 original COMPONENT_REF to the new COMPONENT_REF with
22924 DECL_BIT_FIELD_REPRESENTATIVE.
22925 * ubsan.c (instrument_object_size): Likewise.
22927 2017-03-27 Richard Biener <rguenther@suse.de>
22929 PR tree-optimization/80170
22930 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
22931 sure DR/SCEV didnt fold in constants we do not see when looking
22932 at the reference base alignment.
22934 2017-03-27 Richard Biener <rguenther@suse.de>
22936 PR middle-end/80171
22937 * gimple-fold.c (fold_ctor_reference): Properly guard against
22938 NULL return value from canonicalize_constructor_val.
22940 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
22943 * config/i386/i386.c (ix86_expand_builtin)
22944 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
22945 flags reg setting and flags reg using instructions.
22946 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
22947 clobbering instructions to zero extend op2.
22949 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
22951 * doc/install.texi (Configuration) <--with-aix-soname>:
22952 Update link to AIX ld.
22954 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
22956 PR rtl-optimization/80160
22957 PR rtl-optimization/80159
22958 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
22959 reg_alternate_class into account.
22961 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
22964 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
22965 to consider in curr_insn_transform.
22967 2017-03-24 Jakub Jelinek <jakub@redhat.com>
22969 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
22970 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
22971 and emit_mode_inner.
22973 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22975 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
22976 argument to the overloaded builtin variants. Use the new flag to
22977 deprecate certain builtin variants.
22978 * config/s390/s390-builtin-types.def: Add new builtin types.
22979 * config/s390/s390-builtins.h: Support new flags field for
22980 overloaded builtins.
22981 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
22982 (s390_macro_to_expand): Enable vector float data type.
22983 (s390_cpu_cpp_builtins_internal): Indicate support of the new
22984 builtins by incrementing the __VEC__ version number.
22985 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
22987 (s390_resolve_overloaded_builtin): Emit error messages depending
22988 on the builtin flags.
22989 * config/s390/s390.c (s390_expand_builtin): Support additional
22990 flags argument. Change error message to match the messages
22991 emitted in s390-c.c.
22992 * config/s390/s390.md: New UNSPEC_* constants.
22993 (op_type): Add new instruction types.
22994 * config/s390/vecintrin.h: Add new builtins and test data class
22996 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
22997 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
22998 (VEC_INEXACT, VEC_NOINEXACT): New constants.
22999 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
23000 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
23001 ("vec_mergel<mode>"): V_HW -> VEC_HW.
23003 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
23004 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
23005 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
23006 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
23008 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
23009 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
23010 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
23011 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
23013 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
23014 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
23015 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
23016 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
23017 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
23018 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
23019 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
23021 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
23022 ("vec_scatter_element<V_HW_4:mode>_DI")
23023 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
23024 ("vec_fpint<mode>", "vflls")
23025 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
23026 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
23027 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
23028 ("*vec_cmphe<mode>_cc"): ... these.
23030 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
23031 mode constant instead of magic value.
23033 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23035 * config/s390/s390.c (s390_expand_vec_compare): Support other
23036 vector floating point modes than just V2DF.
23037 (s390_expand_vcond): Likewise.
23038 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
23039 (s390_cannot_change_mode_class): Prevent mode changes between TF
23040 and V1TF in vector registers.
23041 * config/s390/s390.md (DF, SF): New mode attributes.
23042 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
23043 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
23044 SFmode support for VRs.
23045 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
23047 (VFT, VF_HW): New mode iterators.
23048 (vw, sdx): New mode attributes.
23049 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
23050 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
23051 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
23052 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
23053 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
23054 also the new vector floating point modes. Renaming to ...
23056 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
23057 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
23058 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
23059 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
23060 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
23061 ("vec_unordered<mode>"): ... these.
23063 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
23064 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
23065 ("*vec_extendv2df"): New insn definitions.
23067 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23069 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
23070 ("mulditi3_2", "*muldi3_sign"): New patterns.
23071 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
23072 rename the pattern definition.
23074 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23076 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
23078 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
23080 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23082 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
23083 instruction if possible.
23084 * config/s390/vector.md (vec_halfnumelts): New mode
23086 ("*vec_vllezlf<mode>"): New pattern.
23088 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23090 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
23091 ("popcountv4si2", "popcountv2di2"): Rename to ...
23092 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
23093 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
23095 ("popcount<mode>2_vxe"): New pattern.
23097 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23099 * common/config/s390/s390-common.c (processor_flags_table): Add
23101 * config.gcc: Add arch12.
23102 * config/s390/driver-native.c (s390_host_detect_local_cpu):
23103 Default to arch12 for unknown CPU model numbers.
23104 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
23105 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
23106 PROCESSOR_max sanity check.
23107 * config/s390/s390-opts.h (enum processor_type): Add
23109 * config/s390/s390.c (processor_table): Add arch12.
23110 (s390_expand_builtin): Add check for B_VXE flag.
23111 (s390_issue_rate): Add PROCESSOR_ARCH12.
23112 (s390_get_sched_attrmask): Likewise.
23113 (s390_get_unit_mask): Likewise.
23114 (s390_sched_score): Enable z13 scheduling for arch12.
23115 (s390_sched_reorder): Likewise.
23116 (s390_sched_variable_issue): Likewise.
23117 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
23119 (s390_tune_attr): Use z13 scheduling also for arch12.
23120 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
23121 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
23122 (TARGET_VXE_P): New macros.
23123 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
23124 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
23125 * config/s390/s390.opt: Add arch12 as processor_type.
23127 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23129 * config/s390/s390.md
23130 ("fixuns_truncdddi2", "fixuns_trunctddi2")
23131 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
23132 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
23134 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
23135 Rename expanders to ...
23137 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
23138 ("fixuns_truncdddi2_emu"): ... these.
23140 ("fixuns_trunc<mode>si2_emu"): New expander.
23142 ("*fixuns_truncdfdi2_z13"): Rename to ...
23143 ("*fixuns_truncdfdi2_vx"): ... this.
23145 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23147 * config/s390/2964.md: Remove the single element vector compare
23148 instructions which are no longer used.
23149 * config/s390/s390.c (s390_select_ccmode): Remove handling of
23151 (s390_canonicalize_comparison): Remove handling of DFmode
23153 (s390_expand_vec_compare_scalar): Remove function.
23154 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
23155 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
23157 ("*cmp<mode>_ccs"): Add wfcdb instruction.
23159 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23161 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
23163 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
23164 will anyway by matched by mov<mode>_64dfp.
23166 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23168 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
23169 vlef/vstef. Add missing operand to vleif.
23171 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23173 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
23174 pair for all vector types with 64 bit elements.
23175 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
23176 * config/s390/vector.md (V_HW_64): ... here.
23177 (V_128_NOSINGLE): New mode iterator.
23178 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
23179 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
23180 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
23181 ("*vec_load_pairv2di"): Change to ...
23182 ("*vec_load_pair<mode>"): ... this one.
23184 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23186 * config/s390/constraints.md: Add comments.
23187 (jKK): Reject element sizes > 8 bytes.
23188 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
23190 * config/s390/s390.md: Add the s_operand checks formerly in
23191 s390_split_ok_p to various splitters where they are still
23193 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
23194 for 128 bit vectors. Plus two splitters.
23196 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23198 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
23201 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23204 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
23205 error if the boundary argument is not constant.
23207 2017-03-24 Jakub Jelinek <jakub@redhat.com>
23209 PR rtl-optimization/80112
23210 * loop-doloop.c (doloop_condition_get): Don't check condition
23211 if cmp isn't SET with IF_THEN_ELSE src.
23213 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23215 PR tree-optimization/80158
23216 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
23217 replacing a candidate statement, also replace it for the
23218 candidate's alternate interpretation.
23219 (replace_rhs_if_not_dup): Likewise.
23220 (replace_one_candidate): Likewise.
23222 2017-03-24 Richard Biener <rguenther@suse.de>
23224 PR tree-optimization/80167
23225 * graphite-isl-ast-to-gimple.c
23226 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
23228 (translate_isl_ast_to_gimple::get_rename): Likewise.
23230 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
23232 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
23233 handling of certain combinations of target options, including the
23234 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
23235 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
23237 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23240 * config/arm/arm.md (*load_multiple): Add reload_completed to
23241 matching condition.
23243 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23244 Richard Biener <rguenth@suse.de>
23246 PR tree-optimization/79908
23247 PR tree-optimization/80136
23248 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
23249 been cast away, gimplify_and_add suffices.
23251 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
23253 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
23255 2017-03-23 Richard Biener <rguenther@suse.de>
23257 PR tree-optimization/80032
23258 * gimplify.c (gimple_push_cleanup): Forced unconditional
23259 cleanups still have to go to the conditional_cleanups
23262 2017-03-22 Jakub Jelinek <jakub@redhat.com>
23264 PR tree-optimization/80072
23265 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
23267 (next_operand_entry_id): Change type to unsigned int.
23268 (sort_by_operand_rank): Make sure to return the right return value
23269 even if unsigned fields are bigger than INT_MAX.
23270 (struct oecount): Change cnt and id type to unsigned int.
23271 (oecount_hasher::equal): Formatting fix.
23272 (oecount_cmp): Make sure to return the right return value
23273 even if unsigned fields are bigger than INT_MAX.
23274 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
23277 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
23278 TREE_READONLY on result if writing it more than once.
23281 * doc/invoke.texi (-fsanitize=thread): Document that with
23282 -fnon-call-exceptions atomics are not able to throw
23286 * tsan.c: Include tree-eh.h.
23287 (instrument_builtin_call): Call maybe_clean_eh_stmt or
23288 maybe_clean_or_replace_eh_stmt where needed.
23289 (instrument_memory_accesses): Add cfg_changed argument.
23290 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
23291 if it returned true.
23292 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
23294 PR rtl-optimization/63191
23295 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
23296 wrapper function, moved the whole old content into ...
23297 (ix86_delegitimize_address_1): ... this. New inline function.
23298 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
23299 true as last argument instead of ix86_delegitimize_address.
23301 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
23303 * config/aarch64/aarch64.c (generic_branch_cost): Copy
23304 cortexa57_branch_cost.
23306 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
23308 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
23310 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23313 * doc/md.texi (Constraints): Document wA constraint.
23314 * config/rs6000/constraints.md (wA): New.
23315 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
23316 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
23317 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
23318 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
23320 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
23323 * gimplify.c (is_oacc_declared): New function.
23324 (oacc_default_clause): Use it to set default flags for acc declared
23325 variables inside parallel regions.
23326 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
23327 declared variables.
23328 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
23329 declare attribute to any decl as necessary.
23331 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
23334 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
23335 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
23336 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
23337 (arm_arch_lpae): This.
23338 * config/arm/arm.c (arm_arch7ve): Rename into ...
23339 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
23340 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
23343 2017-03-22 Martin Liska <mliska@suse.cz>
23346 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
23347 error message instead of an ICE.
23349 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23351 * doc/extend.texi (6.11 Additional Floating Types): Revise.
23353 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
23355 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
23357 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
23360 2017-03-21 Martin Sebor <msebor@redhat.com>
23362 * doc/extend.texi: Use "cannot" instead of "can't."
23363 * doc/hostconfig.texi: Same.
23364 * doc/install.texi: Same.
23365 * doc/invoke.texi: Same.
23366 * doc/loop.texi: Same.
23367 * doc/md.texi: Same.
23368 * doc/objc.texi: Same.
23369 * doc/rtl.texi: Same.
23370 * doc/tm.texi: Same.
23371 * doc/tm.texi.in: Same.
23372 * doc/trouble.texi: Same.
23374 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
23377 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
23378 (collect_checksum_attributes): Set it.
23379 (die_checksum_ordered): Use it.
23381 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23383 PR tree-optimization/79908
23384 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
23385 change: For a VA_ARG whose LHS has been cast away, use
23386 force_gimple_operand to construct the side effects.
23388 2017-03-21 David Malcolm <dmalcolm@redhat.com>
23390 PR translation/80001
23391 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
23392 more amenable to translation.
23393 (oacc_loop_auto_partitions): Likewise.
23395 2017-03-21 Marek Polacek <polacek@redhat.com>
23396 Martin Sebor <msebor@redhat.com>
23398 PR tree-optimization/80109
23399 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
23400 on INTEGRAL_TYPE_P.
23402 2017-03-21 Jakub Jelinek <jakub@redhat.com>
23403 Segher Boessenkool <segher@kernel.crashing.org>
23406 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
23407 check reg_used_between_p between insn and one of succ or succ2
23408 depending on if succ is artificial insn not inserted into insn
23411 2017-03-21 Martin Liska <mliska@suse.cz>
23413 PR gcov-profile/80081
23414 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
23415 * doc/gcc.texi: Include gcov-dump stuff.
23416 * doc/gcov-dump.texi: New file.
23418 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
23420 PR rtl-optimization/79150
23421 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
23422 conditional jump, if the jump is the last insn of the loop.
23424 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23425 Richard Biener <rguenth@suse.de>
23427 PR tree-optimization/79908
23428 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
23429 been cast away, use force_gimple_operand to construct the side
23432 2017-03-21 Martin Liska <mliska@suse.cz>
23434 PR libfortran/79956
23435 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
23438 2017-03-21 Brad Spengler <spender@grsecurity.net>
23441 * plugin.c (htab_hash_plugin): New function.
23442 (add_new_plugin): Use it and adjust.
23443 (parse_plugin_arg_opt): Adjust.
23444 (init_one_plugin): Likewise.
23446 2017-03-21 Richard Biener <rguenther@suse.de>
23448 PR tree-optimization/80032
23449 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
23450 if set force the cleanup to happen unconditionally.
23451 (gimplify_target_expr): Push inserted clobbers with force_uncond
23452 to avoid them being removed by control-dependent DCE.
23454 2017-03-21 Richard Biener <rguenther@suse.de>
23456 PR tree-optimization/80122
23457 * tree-inline.c (copy_bb): Do not expans va-arg packs or
23458 va_arg_pack_len when the inlined call stmt requires pack
23460 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
23462 2017-03-21 Jakub Jelinek <jakub@redhat.com>
23465 * tsan.c (instrument_builtin_call): If the memory model argument
23466 is not a constant, assume it is valid.
23469 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
23472 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
23474 PR rtl-optimization/79910
23475 * combine.c (can_combine_p): Do not allow combining an I0 or I1
23476 if its dest is used by an insn before I2 (other than the combined
23477 insns themselves, which are properly handled already).
23479 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
23482 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
23484 * combine.c (record_used_regs): New static function.
23485 (try_combine): Handle situations where there is an additional
23486 instruction between I2 and I3 which needs to have a LOG_LINK
23490 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
23492 * combine.c (try_combine): Delete redundant i1 test. Call
23493 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
23495 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23498 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
23499 alternatives 13/14.
23501 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23503 PR tree-optimization/80054
23504 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
23505 the optimization if a PHI or any of its arguments is not dominated
23506 by the candidate's basis. Use gphi* rather than gimple* as
23508 (replace_profitable_candidates): Clean up a gimple* variable that
23509 should be a gphi* variable.
23511 2017-03-20 Martin Sebor <msebor@redhat.com>
23514 * doc/extend.texi (attribute constructor): Document present limitation.
23516 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
23519 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
23520 __POWER9_VECTOR__ #ifdef control, change template definition to
23521 use Power9-specific built-in function.
23522 (vec_any_eq): Likewise.
23523 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
23524 to control outcomes from this test.
23525 (vector_ae_<mode>p): For VEC_F modes, likewise.
23527 2017-03-20 Ian Lance Taylor <iant@google.com>
23529 * config/i386/i386.c (ix86_function_regparm): Save an extra
23530 register for -fsplit-stack with DECL_STATIC_CHAIN.
23532 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
23535 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
23536 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
23538 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
23540 * config/riscv/riscv.c (riscv_print_operand): Use "fence
23542 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
23545 2017-03-20 Marek Polacek <polacek@redhat.com>
23548 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
23550 2017-03-20 Richard Biener <rguenther@suse.de>
23552 PR tree-optimization/80113
23553 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
23554 allocate extra SSA name for PHI def.
23555 (add_close_phis_to_outer_loops): Likewise.
23556 (add_close_phis_to_merge_points): Likewise.
23557 (copy_loop_close_phi_args): Likewise.
23558 (copy_cond_phi_nodes): Likewise.
23560 2017-03-20 Martin Liska <mliska@suse.cz>
23562 PR middle-end/79753
23563 * tree-chkp.c (chkp_build_returned_bound): Do not build
23564 returned bounds for a LHS that's not a BOUNDED_P type.
23566 2017-03-20 Martin Liska <mliska@suse.cz>
23570 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
23571 COMPLEX_CST and VECTOR_CST.
23573 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23576 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
23577 target operand. A new splitter adds the clobber statement in case
23578 the target operand is dead anyway.
23580 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
23582 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
23583 to age-old versions of binutils and glibc.
23585 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
23587 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
23589 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
23591 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
23593 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
23595 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
23596 requirement for binutils 2.13.
23598 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
23600 * combine.c (try_combine): Delete redundant i1 test. Call
23601 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
23603 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
23605 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
23606 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
23608 <riscv64-*-elf>: Re-arrange section
23609 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
23610 <riscv32-*-linux>: Likewise.
23611 <riscv64-*-elf>: Likewise
23612 <riscv64-*-linux>: Likewise.
23614 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
23617 * aarch64.opt(verbose-cost-dump): Fix typo.
23619 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
23622 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
23623 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
23625 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
23627 * reload.c (find_reloads): When reloading a nonoffsettable address,
23628 use RELOAD_OTHER for it and its address reloads.
23630 PR rtl-optimization/79910
23631 * combine.c (record_used_regs): New static function.
23632 (try_combine): Handle situations where there is an additional
23633 instruction between I2 and I3 which needs to have a LOG_LINK
23636 2017-03-17 Jeff Law <law@redhat.com>
23638 PR tree-optimization/71437
23639 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
23640 conditional in the hash table first.
23641 (vrp_dom_walker::before_dom_children): Extract condition from
23642 ASSERT_EXPR. Record condition, its inverion and any implied
23643 conditions as well.
23645 2017-03-17 Marek Polacek <polacek@redhat.com>
23646 Markus Trippelsdorf <markus@trippelsdorf.de>
23648 PR tree-optimization/80079
23649 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
23652 2017-03-17 Richard Biener <rguenther@suse.de>
23654 PR middle-end/80075
23655 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
23656 Properly verify the LHS before the RHS possibly claims to be
23658 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
23661 2017-03-17 Martin Jambor <mjambor@suse.cz>
23663 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
23664 (List of -O2 options): Likewise.
23665 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
23668 2017-03-17 Tom de Vries <tom@codesourcery.com>
23670 * gcov-dump.c (print_usage): Print bug_report_url.
23672 2017-03-17 Richard Biener <rguenther@suse.de>
23674 PR middle-end/80050
23675 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
23676 (parser::peek): Likewise.
23678 2017-03-17 Richard Biener <rguenther@suse.de>
23680 PR tree-optimization/80048
23681 * sese.c (free_sese_info): Properly release rename_map and
23682 copied_bb_map elements.
23684 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
23686 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
23687 Add linked-list forward and backlinks. Insert on
23688 construction, remove on destruction.
23689 (class pass_store_merging): Add m_stores_head field.
23690 (pass_store_merging::terminate_and_process_all_chains):
23691 Iterate over m_stores_head list.
23692 (pass_store_merging::terminate_all_aliasing_chains):
23694 (pass_store_merging::execute): Check for debug stmts first.
23695 Push new chains onto the m_stores_head stack.
23697 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
23700 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
23701 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
23702 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
23704 2017-03-16 Jeff Law <law@redhat.com>
23706 PR tree-optimization/71437
23707 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
23708 member function. Implementation moved into after_dom_children
23709 member function and into the threader's thread_outgoing_edges
23711 (dom_opt_dom_walker::after_dom_children): Simplify by moving
23712 some code into new thread_outgoing_edges.
23713 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
23714 definition. Simplify marker handling (do it here). Assume we always
23715 have the available expression and the const/copies tables.
23716 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
23718 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
23719 * tree-vrp.c (equiv_stack): No longer file scoped.
23720 (vrp_dom_walker): New class.
23721 (vrp_dom_walker::before_dom_children): New member function.
23722 (vrp_dom_walker::after_dom_children): Likewise.
23723 (identify_jump_threads): Setup domwalker. Use it rather than
23724 walking edges in a random order by hand. Simplify setup/finalization.
23725 (finalize_jump_threads): Remove.
23726 (vrp_finalize): Do not call identify_jump_threads here.
23727 (execute_vrp): Do it here instead and call thread_through_all_blocks
23730 PR tree-optimization/71437
23731 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
23733 (simplify_stmt_for_jump_threading): Add basic_block argument. All
23735 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
23736 (dom_opt_dom_walker::thread_across_edge): Remove
23737 handle_dominating_asserts argument. All callers changed.
23738 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
23739 changes. Remove calls to lhs_of_dominating_assert. Other
23740 uses of handle_dominating_asserts turn into unconditional code
23741 (simplify_control_stmt_condition_1): Likewise.
23742 (simplify_control_stmt_condition): Likewise.
23743 (thread_through_normal_block, thread_across_edge): Likewise.
23744 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
23745 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
23746 object if it is not an SSA_NAME.
23747 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
23748 before calling into the VRP specific simplifiers.
23749 (identify_jump_threads): Remove handle_dominating_asserts
23752 2017-03-16 Jakub Jelinek <jakub@redhat.com>
23755 * tree-diagnostic.c (default_tree_printer): No longer static.
23756 * tree-diagnostic.h (default_tree_printer): New prototype.
23758 2017-03-16 Tamar Christina <tamar.christina@arm.com>
23760 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
23761 Change ins into fmov.
23763 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23765 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
23766 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
23767 Use h_con constraint for operand 1.
23768 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
23769 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
23771 2017-03-15 Jeff Law <law@redhat.com>
23773 PR tree-optimization/71437
23774 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
23775 (record_temporary_equivalences): Use it.
23777 PR tree-optimization/71437
23778 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
23779 tree-ssa-scopedtables.
23780 (lookup_avail_expr, build_and_record_new_cond): Likewise.
23781 (record_conditions, record_cond, vuse_eq): Likewise.
23782 (record_edge_info): Adjust to API tweak of record_conditions.
23783 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
23784 (record_temporary_equivalences, optimize_stmt): Likewise.
23785 (eliminate_redundant_computations): Likewise.
23786 (record_equivalences_from_stmt): Likewise.
23787 * tree-ssa-scopedtables.c: Include options.h and params.h.
23788 (vuse_eq): New function, moved from tree-ssa-dom.c
23789 (build_and_record_new_cond): Likewise.
23790 (record_conditions): Likewise. Accept vector of conditions rather
23791 than edge_equivalence structure for first argument.
23792 for the first argument.
23793 (avail_exprs_stack::lookup_avail_expr): New member function, moved
23794 from tree-ssa-dom.c.
23795 (avail_exprs_stack::record_cond): Likewise.
23796 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
23797 from tree-ssa-dom.c.
23798 (avail_exprs_stack): Add new member functions lookup_avail_expr
23800 (record_conditions): Declare.
23802 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
23805 * lra-constraints.c (process_alt_operands): Increase reject for
23806 reloading an input/output operand.
23808 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
23811 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
23812 insns to convert from signed/unsigned char/short to IEEE 128-bit
23814 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
23816 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
23819 * config/i386/i386.c (ix86_vector_duplicate_value): Create
23820 subreg of inner mode for values already in registers.
23822 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
23824 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
23825 iteration reg is used after the loop.
23827 2017-03-14 Martin Sebor <msebor@redhat.com>
23829 PR tree-optimization/79800
23830 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
23831 precision in negative-positive range.
23832 (format_floating): Call non-const overload with adjusted precision.
23834 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
23837 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
23840 2017-03-14 Martin Sebor <msebor@redhat.com>
23842 PR middle-end/80020
23843 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
23844 * builtins.def (aligned_alloc): Use it.
23847 * Makefile.in (GTFILES): Add calls.c.
23848 * calls.c: Include "gt-calls.h".
23850 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
23852 PR rtl-optimization/79728
23853 * regs.h (struct target_regs): New field
23854 x_contains_allocatable_regs_of_mode.
23855 (contains_allocatable_regs_of_mode): New macro.
23856 * reginfo.c (init_reg_sets_1): Initialize it, and change
23857 contains_reg_of_mode so it includes global regs as well.
23858 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
23859 rather than contains_regs_of_mode.
23861 2017-03-14 Martin Liska <mliska@suse.cz>
23863 * doc/invoke.texi: Document options that can't be combined with
23864 -fcheck-pointer-bounds.
23866 2017-03-14 Martin Liska <mliska@suse.cz>
23868 PR middle-end/79831
23869 * doc/invoke.texi (-Wchkp): Document the option.
23871 2017-03-14 Martin Liska <mliska@suse.cz>
23873 * Makefile.in: Install gcov-dump.
23875 2017-03-14 Martin Liska <mliska@suse.cz>
23877 * multiple_target.c (expand_target_clones): Bail out for
23878 an invalid attribute.
23880 2017-03-14 Richard Biener <rguenther@suse.de>
23882 * alias.c (struct alias_set_entry): Pack properly.
23883 * cfgloop.h (struct loop): Likewise.
23884 * cse.c (struct set): Likewise.
23885 * ipa-utils.c (struct searchc_env): Likewise.
23886 * loop-invariant.c (struct invariant): Likewise.
23887 * lra-remat.c (struct cand): Likewise.
23888 * recog.c (struct change_t): Likewise.
23889 * rtl.h (struct address_info): Likewise.
23890 * symbol-summary.h (function_summary): Likewise.
23891 * tree-loop-distribution.c (struct partition): Likewise.
23892 * tree-object-size.c (struct object_size_info): Likewise.
23893 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
23894 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
23895 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
23896 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
23897 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
23898 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
23899 (struct _stmt_vec_info): Likewise.
23901 2017-03-14 Martin Liska <mliska@suse.cz>
23904 * multiple_target.c (create_dispatcher_calls): Check that
23905 a target can create a function dispatcher.
23907 2017-03-14 Martin Liska <mliska@suse.cz>
23910 * multiple_target.c (expand_target_clones): Drop local.local
23911 flag for default implementation.
23913 2017-03-14 Richard Biener <rguenther@suse.de>
23915 PR tree-optimization/80030
23916 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
23918 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
23920 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
23921 gcc_fallthrough() instead of __attribute__((fallthrough));
23923 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
23925 * doc/gcc.texi: Remove "up" link to (DIR).
23926 * doc/gccint.texi: Ditto.
23928 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
23930 * doc/install.texi (Specific) <avr>: Remove reference to
23933 2017-03-13 Jeff Law <law@redhat.com>
23935 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
23936 attribute rather than comments.
23938 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
23939 match_scratch operand is highest.
23941 2017-03-13 Martin Liska <mliska@suse.cz>
23943 PR middle-end/78339
23944 * ipa-pure-const.c (warn_function_noreturn): If the declarations
23945 is a CHKP clone, use original declaration.
23947 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23949 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
23950 (arc_conditional_register_usage): Use a different allocation order
23951 when optimizing for size.
23952 * common/config/arc/arc-common.c (arc_option_optimization_table):
23953 Section anchors default on when optimizing for size.
23955 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23957 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
23959 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23961 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
23962 * config/arc/arc.md (cpu_facility): Add cd variant.
23963 (*movqi_insn): Add code density variant.
23964 (*movhi_insn): Likewise.
23965 (*movqi_insn): Likewise.
23966 (*addsi3_mixed): Likewise.
23967 (subsi3_insn): Likewise.
23969 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23971 * config/arc/arc.md (movsi_cond_exec): Update constraint.
23973 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23975 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
23976 expressions with MINUS and UNARY ops.
23978 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23981 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
23983 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
23984 between vec_select and vector argument.
23985 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
23986 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
23987 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
23988 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
23989 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
23990 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
23992 2017-03-13 Richard Biener <rguenther@suse.de>
23995 * params.def (vect-max-peeling-for-alignment): Fix typo.
23997 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23999 * doc/install.texi (Specific) <mips-*-*>: Remove description of
24000 issue that only occurred with binutils below 2.18.
24002 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
24004 * doc/install.texi (Specific) <cris-axis-elf>: No longer
24005 refer to binutils 2.11/2.12 minimum.
24007 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
24009 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
24010 ftp.kernel.org and simplify binutils requirement.
24012 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
24014 * doc/invoke.texi (Warning Options): Fix spelling of link-time
24016 (Optimize Options): Ditto. Also remove redundancy.
24018 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24020 PR translation/79848
24021 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
24023 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
24024 to G_ to avoid double translation.
24026 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24028 PR translation/79923
24029 * auto-profile.c (get_combined_location): Convert leading
24030 character of diagnostics to lower case and remove trailing period.
24031 (read_profile): Likewise for various diagnostics.
24032 * config/arm/arm.c (arm_option_override): Remove trailing period
24033 from various diagnostics.
24034 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
24035 (msp430_expand_delay_cycles): Likewise.
24037 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24040 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
24041 full command-line argument, rather than just "str".
24042 (aarch64_validate_march): Likewise.
24043 (aarch64_validate_mtune): Likewise.
24045 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
24047 PR rtl-optimization/78911
24048 * lra-assigns.c (must_not_spill_p): New function.
24049 (spill_for): Use it.
24051 2017-03-10 Jakub Jelinek <jakub@redhat.com>
24053 PR tree-optimization/79981
24054 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
24055 ATOMIC_COMPARE_EXCHANGE ifn result.
24056 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
24057 IFN_ATOMIC_COMPARE_EXCHANGE.
24059 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24062 * opts.c (parse_sanitizer_options): Add missing question mark to
24063 "did you mean" message.
24065 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24067 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
24069 (VMULEUH_UNS): Likewise.
24070 (VMULOUB_UNS): Likewise.
24071 (VMULOUH_UNS): Likewise.
24072 * config/rs6000/rs6000.c (builtin_function_type): Remove
24073 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
24075 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24078 * read-rtl-function.c (function_reader::read_rtx_operand): Update
24079 x with result of extra_parsing_for_operand_code_0.
24080 (function_reader::extra_parsing_for_operand_code_0): Convert
24081 return type from void to rtx, returning x. When reading
24082 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
24083 larger size containing struct block_symbol.
24085 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
24087 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
24088 -mfloat128-hardware without -m64.
24090 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
24093 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
24094 entries to the case statement that marks unsigned arguments to
24095 overloaded functions.
24097 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
24099 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
24100 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
24102 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
24105 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
24106 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
24108 2017-03-10 Martin Liska <mliska@suse.cz>
24112 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
24113 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
24116 2017-03-10 Olivier Hainque <hainque@adacore.com>
24118 * tree-switch-conversion (array_value_type): Start by resetting
24119 candidate type to it's main variant.
24121 2017-03-10 Jakub Jelinek <jakub@redhat.com>
24123 PR rtl-optimization/79909
24124 * combine.c (try_combine): Use simplify_replace_rtx on individual
24125 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
24126 of the whole CALL_INSN_FUNCTION_USAGE.
24128 PR tree-optimization/79972
24129 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
24130 get_range_info on SSA_NAMEs. Formatting fixes.
24132 2017-03-10 Richard Biener <rguenther@suse.de>
24133 Jakub Jelinek <jakub@redhat.com>
24135 PR tree-optimization/77975
24136 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
24137 edge to be constant.
24138 (get_val_for): For constant x return it. Formatting fix.
24139 (loop_niter_by_eval): Avoid pointless looping if the next iteration
24140 would use the same bases as the current one.
24142 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24144 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
24145 instead of vec_select for V1TImode.
24146 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
24148 (VSX_LE_128): Add V1TI to this mode iterator.
24149 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
24150 (*vsx_le_perm_store_<mode>): Likewise.
24151 (pre-reload splitter for VSX stores): Likewise.
24152 (post-reload splitter for VSX stores): Likewise.
24153 (*vsx_xxpermdi2_le_<mode>): Likewise.
24154 (*vsx_lxvd2x2_le_<mode>): Likewise.
24155 (*vsx_stxvd2x2_le_<mode>): Likewise.
24157 2017-03-09 Michael Eager <eager@eagercon.com>
24159 Correct failures with --enable-checking=yes,rtl.
24161 * config/microblaze/microblaze.c (microblaze_expand_shift):
24162 Replace GET_CODE test with CONST_INT_P and INTVAL test with
24163 test for const0_rtx.
24164 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
24165 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
24167 2017-03-09 Richard Biener <rguenther@suse.de>
24169 PR tree-optimization/79977
24170 * graphite-scop-detection.c (scop_detection::merge_sese):
24171 Handle the case of extra exits to blocks dominating the entry.
24173 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
24175 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
24178 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
24180 PR rtl-optimization/79949
24181 * lra-constraints.c (process_alt_operands): Check memory when
24182 trying to predict a cycle. Print about the overall increase.
24184 2017-03-09 Richard Biener <rguenther@suse.de>
24186 PR middle-end/79971
24187 * gimple-expr.c (useless_type_conversion_p): Preserve
24188 TYPE_SATURATING for fixed-point types.
24190 2017-03-09 Richard Biener <rguenther@suse.de>
24193 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
24194 alignment of BLKmode params.
24196 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24199 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
24200 (VALL_NO_V2Q): Likewise.
24202 * config/aarch64/aarch64-simd.md
24203 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
24205 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
24206 VALL_NO_V2Q mode iterator.
24207 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
24209 2017-03-09 Martin Liska <mliska@suse.cz>
24211 PR tree-optimization/79631
24212 * tree-chkp-opt.c (chkp_is_constant_addr): Call
24213 tree_int_cst_sign_bit just for INTEGER constants.
24215 2017-03-09 Martin Liska <mliska@suse.cz>
24219 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
24222 2017-03-09 Marek Polacek <polacek@redhat.com>
24225 * tree.c (inchash::add_expr): Handle TREE_VEC.
24227 2017-03-09 Martin Liska <mliska@suse.cz>
24230 (chkp_narrow_size_and_offset): New function.
24231 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
24232 (void chkp_parse_bit_field_ref): New function.
24233 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
24234 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
24236 2017-03-09 Martin Liska <mliska@suse.cz>
24239 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
24240 (chkp_find_bounds_1): Remove gcc_unreachable.
24242 2017-03-09 Jakub Jelinek <jakub@redhat.com>
24245 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
24246 BUILT_IN_SYNC*, determine the access type from the size suffix and
24247 always build a MEM_REF with that type. Handle forgotten
24248 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
24251 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
24252 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
24253 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
24254 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
24255 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
24256 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
24257 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
24258 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
24259 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
24260 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
24261 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
24262 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
24263 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
24264 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
24265 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
24266 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
24267 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
24268 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
24269 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
24270 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
24271 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
24272 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
24273 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
24274 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
24275 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
24276 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
24277 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
24278 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
24279 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
24280 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
24281 definitions outside of __OPTIMIZE__ guarded section.
24284 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
24285 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
24286 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
24287 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
24290 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24292 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
24293 ("vfenez<mode>"): Add missing constraints.
24295 2017-03-08 Martin Sebor <msebor@redhat.com>
24298 * config/nds32/nds32.c (nds32_option_override):
24299 Fix misspelled diagnostic.
24301 2017-03-08 Jakub Jelinek <jakub@redhat.com>
24304 * gimplify.c (gimplify_omp_for): Replace index var in outer
24305 taskloop statement with an artificial variable and add
24306 OMP_CLAUSE_PRIVATE clause for it.
24308 2017-03-08 Richard Biener <rguenther@suse.de>
24310 PR tree-optimization/79955
24311 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
24312 for accesses that are completely outside of the variable.
24314 2017-03-08 Andrew Haley <aph@redhat.com>
24316 PR tree-optimization/79943
24317 * tree-ssa-loop-split.c (compute_new_first_bound): When
24318 calculating the new upper bound, (END-BEG) should be added, not
24321 2017-03-08 Jakub Jelinek <jakub@redhat.com>
24323 * config/avr/avr.md (setmemhi): Make sure match_dup
24324 operand number comes before match_scratch.
24326 2017-03-08 Richard Biener <rguenther@suse.de>
24328 PR tree-optimization/79920
24329 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
24330 with ncopies == 1 to ...
24331 (vect_transform_slp_perm_load): ... here. Properly compute
24332 all element loads by iterating VF times over the group. Do
24333 not handle ncopies (computed in a broken way) in
24334 vect_create_mask_and_perm.
24336 2017-03-08 Jakub Jelinek <jakub@redhat.com>
24339 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
24340 is a uniform vector, use uniform_vector_p return value instead of
24341 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
24343 2017-03-07 Marek Polacek <polacek@redhat.com>
24345 PR middle-end/79809
24346 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
24347 (alloca_call_type): Likewise.
24349 2017-03-07 Martin Liska <mliska@suse.cz>
24351 * gcov.c (process_args): Put comment to correct location.
24353 2017-03-07 Martin Liska <mliska@suse.cz>
24355 PR middle-end/68270
24356 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
24357 Use array_at_struct_end_p instead of DECL_CHAIN (field).
24358 (chkp_narrow_bounds_for_field): Likewise.
24359 (chkp_parse_array_and_component_ref): Pass one more argument to
24362 2017-03-07 Richard Biener <rguenther@suse.de>
24364 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
24367 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
24369 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
24370 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
24372 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24375 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
24376 to end of description.
24377 (PARAM_MAX_STORES_TO_MERGE): Likewise.
24379 2017-03-07 Jakub Jelinek <jakub@redhat.com>
24381 PR rtl-optimization/79901
24382 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
24384 (*avx512f_<code><mode>3<mask_name>): ... this.
24385 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
24386 iterator instead of VI8_AVX2_AVX512BW.
24388 PR rtl-optimization/79901
24389 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
24390 min/max expander, expand it using expand_vec_cond_expr.
24393 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
24396 2017-03-06 Jakub Jelinek <jakub@redhat.com>
24399 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
24400 to void * for PCH reasons.
24401 * dwarf2out.c (output_loc_operands, output_die): Cast
24402 v.val_vec.array to unsigned char *.
24404 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
24407 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
24410 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
24412 PR rtl-optimization/79571
24413 * lra-constraints.c (process_alt_operands): Calculate static
24414 reject and subtract it from overall when only addresses will be
24417 2017-03-06 Julia Koval <julia.koval@intel.com>
24420 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
24421 incoming stack boundary to 128 for 64-bit targets.
24423 2017-03-06 Richard Biener <rguenther@suse.de>
24425 PR tree-optimization/79894
24426 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
24427 to NULL after folding it.
24429 2017-03-06 Richard Biener <rguenther@suse.de>
24431 PR tree-optimization/79824
24432 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
24433 check disabling peeling for gaps.
24435 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
24437 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
24438 attributes): Document gettimeofday.
24440 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
24442 * config/s390/s390.c (s390_option_override_internal): Set
24443 PARAM_MIN_VECT_LOOP_BOUND
24445 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
24447 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
24448 * config/s390/s390.md: Likewise.
24450 2017-03-06 Jakub Jelinek <jakub@redhat.com>
24453 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
24454 (<avx2_avx512>_perm<mode>): Rename to ...
24455 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
24457 (<avx512>_perm<mode>_mask): Rename to ...
24458 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
24460 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
24461 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
24462 instead of VI8F_256_512.
24463 (avx512f_perm<mode>): New define_expand.
24464 (avx512f_perm<mode>_mask): Likewise.
24465 (avx512f_perm<mode>_1<mask_name>): New define_insn.
24466 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
24468 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
24470 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
24471 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
24473 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
24475 2017-03-06 Martin Liska <mliska@suse.cz>
24478 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
24479 when having a SSA NAME w/o VAR_DECL assigned to it.
24481 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
24483 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
24484 msa_dpsub_<su>_d): Fix MODE for vec_select.
24486 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
24488 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
24490 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
24492 2017-03-06 Richard Biener <rguenther@suse.de>
24494 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
24495 * plugin.c (register_plugin_info): Likewise.
24496 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
24498 2017-03-05 Jakub Jelinek <jakub@redhat.com>
24500 * config/i386/sse.md (sse_storehps, sse_storelps,
24501 avx_<castmode><avxsizesuffix>_<castmode>,
24502 avx512f_<castmode><avxsizesuffix>_<castmode>,
24503 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
24504 in condition that at least one operand is not a MEM.
24506 2017-03-03 Jakub Jelinek <jakub@redhat.com>
24508 PR middle-end/79805
24509 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
24510 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
24512 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
24513 gimple_call_nothrow_p flag based on whether original builtin can throw.
24514 If it can, emit following stmts on the fallthrough edge.
24515 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
24516 don't create new bb if inserting just debug stmts on the edge, try to
24517 insert them on the fallthru bb or just reset debug stmts.
24519 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
24522 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
24523 restore recog_data (including the operand rtxes inside it) around
24524 the call to get_insn_template.
24526 2017-03-03 Martin Sebor <msebor@redhat.com>
24528 PR tree-optimization/79699
24529 * context.c (context::~context): Free MPFR caches to avoid
24530 a memory leak on program exit.
24532 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24534 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
24535 Use wide_int::ulow () instead of .elt (0).
24537 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
24539 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
24540 (*pushxf): Limit oF constraint to 32bit targets and add oC
24541 constraint for 64bit targets.
24542 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
24543 (*pushdf): Change rmF constraint to rmC.
24545 2017-03-03 Martin Liska <mliska@suse.cz>
24547 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
24548 Remove unused variable.
24550 2017-03-03 Jakub Jelinek <jakub@redhat.com>
24553 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
24554 is a memory operand, increase num_memory.
24555 (ix86_expand_args_builtin): Likewise.
24557 2017-03-03 Jan Hubicka <jh@suse.cz>
24560 * ipa-devirt.c (maybe_record_node): Properly handle
24561 __cxa_pure_virtual visibility.
24563 2017-03-03 Martin Liska <mliska@suse.cz>
24565 PR tree-optimization/79803
24566 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
24568 (pass_loop_prefetch::execute): Disabled optimization if an
24569 assumption about L1 cache size is not met.
24571 2017-03-03 Martin Liska <mliska@suse.cz>
24573 PR rtl-optimization/79574
24574 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
24575 (hash_scan_set): Likewise.
24576 (dump_hash_table): Likewise.
24577 (hoist_code): Likewise.
24579 2017-03-03 Richard Biener <rguenther@suse.de>
24581 * fixed-value.c (fixed_from_string): Restore use of elt (1)
24582 in place of uhigh ().
24583 (fixed_convert_from_real): Likewise.
24585 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
24588 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
24590 2017-03-03 Richard Biener <rguenther@suse.de>
24592 PR middle-end/79818
24593 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
24594 TYPE_OVERFLOW_UNDEFINED check.
24596 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24598 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
24600 (vector_ae_<mode>_p): Likewise.
24601 (vector_nez_<mode>_p): Likewise.
24602 (vector_ne_v2di_p): Likewise.
24603 (vector_ae_v2di_p): Likewise.
24604 (vector_ne_<mode>_p): Likewise.
24605 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
24607 (vsx_tsqrt<mode>2_fe): Likewise.
24609 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
24612 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
24614 2017-03-02 Jakub Jelinek <jakub@redhat.com>
24616 PR rtl-optimization/79780
24617 * cprop.c (one_cprop_pass): When second and further conditional trap
24618 in a single basic block is turned into an unconditional trap, turn it
24619 into a deleted note to avoid RTL verification failures.
24621 2017-03-02 Richard Biener <rguenther@suse.de>
24623 * fold-const.c (const_binop): Use ulow () instead of elt (0).
24625 2017-03-02 Richard Biener <rguenther@suse.de>
24627 PR tree-optimization/79345
24629 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
24630 param and abort the walk, returning -1 if it is hit.
24631 (walk_aliased_vdefs): Take a limit param and pass it on.
24632 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
24633 defaulting to 0 and return a signed int.
24634 * tree-ssa-uninit.c (struct check_defs_data): New struct.
24635 (check_defs): New helper.
24636 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
24637 about uninitialized memory.
24638 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
24639 bogus uninitialized warning.
24640 (fixed_convert_from_real): Likewise.
24642 2017-03-02 Bin Cheng <bin.cheng@arm.com>
24644 PR tree-optimization/66768
24645 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
24646 iv_use if base object can't be determined.
24648 2017-03-02 Jakub Jelinek <jakub@redhat.com>
24650 PR tree-optimization/79345
24651 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
24652 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
24653 (get_pattern_stats): Initialize it.
24654 * genemit.c (gen_expand): Verify match_scratch numbers come after
24655 match_operand/match_dup numbers.
24656 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
24657 match_scratch numbers.
24658 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
24660 * config/s390/s390.md (trunctdsd2): Likewise.
24662 2017-03-02 Richard Biener <rguenther@suse.de>
24664 * wide-int.h (wide_int_storage::operator=): Implement in terms
24667 2017-03-02 Richard Biener <rguenther@suse.de>
24669 PR tree-optimization/79777
24670 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
24671 the to insert expression to sth existing.
24673 2017-03-01 Martin Sebor <msebor@redhat.com>
24675 PR middle-end/79692
24676 * gimple-ssa-sprintf.c
24677 (directive::known_width_and_precision): New function.
24678 (format_integer): Use it.
24679 (get_mpfr_format_length): Consider the full range of precision
24680 when computing %g output with the # flag. Set the likely byte
24681 count to 3 rather than 1 when precision is indeterminate.
24682 (format_floating): Correct the lower bound of precision.
24684 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24686 * doc/invoke.texi: Document default code model for 64-bit Linux.
24688 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24691 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
24692 udiv rather than div since input pattern is unsigned.
24694 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
24696 * config/i386/i386.c (print_reg): Warn for values of
24697 unsupported size in integer register.
24699 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
24702 * config/rs6000/predicates.md (current_file_function_operand): Do
24703 not allow self calls to be local if the function is replaceable.
24705 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
24708 * config/rs6000/altivec.h (vec_ctz and others): Change the
24709 preprocessor macro that controls conditional compilation from
24710 _ARCH_PWR9 to __POWER9_VECTOR__.
24711 (vec_all_ne): Change parameterization of __altivec_scalar_pred
24712 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
24713 control (instead of _ARCH_PWR9 control) so that template
24714 definition uses power9-specific function.
24715 (vec_any_eq): Likewise.
24716 (vec_all_ne): Change macro definition to use a power9-specific
24717 expansion under #ifdef __POWER9_VECTOR__ control (instead of
24718 _ARCH_PWR9 control).
24719 (vec_any_eq) Likewise.
24720 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
24721 expansion for CMPNEF to remove support for xvcmpnesp instruction.
24722 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
24723 support for xvcmpnedp instruction.
24724 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
24725 macro expansion so that Power9 implementation of vec_all_ne does
24726 not use the AltiVec predicate framework.
24727 (VCMPNEH_P): Likewise.
24728 (VCMPNEW_P): Likewise.
24729 (VCMPNED_P): Likewise.
24730 (VCMPNEFP_P): Likewise.
24731 (VCMPNEDP_P): Likewise.
24732 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
24733 implementation of vec_any_eq to not use AltiVec predicate
24735 (VCMPAEH_P): Likewise.
24736 (VCMPAEW_P): Likewise.
24737 (VCMPAED_P): Likewise.
24738 (VCMPAEFP_P): Likewise.
24739 (VCMPAEDP_P): Likewise.
24740 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
24741 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
24742 not use the AltiVec predicate framework.
24743 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
24744 of vec_any_eq to not use AltiVec predicate framework.
24745 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
24746 support for predefined __POWER9_VECTOR__ macro to indicate that
24747 Power9 instruction selection is enabled.
24748 (altivec_overloaded_builtins): Remove extraneous
24749 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
24750 function argument types RS6000_BTI_bool_V16QI and
24751 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
24752 entry for overloaded function argument types RS6000_BTI_bool_V4SI
24753 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
24754 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
24755 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
24756 Power9 for implementations of vec_cmpne. Change the signature for
24757 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
24758 (representing vec_all_ne) to remove the previously described first
24759 argument of type RS6000_BTI_INTSI, as this was an artifact of
24760 reliance on the AltiVec predicate framework, which is no longer
24761 used in the implementation of these functions. Add
24762 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
24763 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
24764 since, unlike the AltiVec predicate framework implementation, we
24765 do not share function descriptors between vec_alle and vec_anyeq.
24766 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
24767 set of modes that receive special treatment even when
24768 TARGET_P9_VECTOR is true. The special treatment emits code that
24769 does not depend on Power9 instructions.
24770 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
24771 define_expand to not rely on AltiVec predicate framework.
24772 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
24774 (vector_ne_v2di_p): Change this define_expand to not rely on
24775 AltiVec predicate framework.
24776 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
24778 (vector_ne_<mode>_p): Change this define_expand to not rely on
24779 AltiVec predicate framework.
24780 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
24782 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
24783 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
24784 define_insn pattern.
24785 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
24786 define_insn pattern because the xvcmpne<VSs>. instruction is not
24788 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
24789 instruction is not supported.
24791 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24793 * config/nvptx/nvptx.c: Include intl.h.
24795 2017-03-01 Martin Jambor <mjambor@suse.cz>
24798 * ipa-prop.h (ipa_bits): Removed field known.
24799 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
24800 to pointers. Adjusted their comments to warn about their sharing.
24801 (ipcp_transformation_summary): Change bits to a vector of pointers.
24802 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
24803 (ipa_get_ipa_bits_for_value): Declare.
24804 * tree-vrp.h (value_range): Mark as GTY((for_user)).
24805 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
24806 (ipa_bits_hash_table): Likewise.
24807 (ipa_vr_ggc_hash_traits): Likewise.
24808 (ipa_vr_hash_table): Likewise.
24809 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
24810 being pointers and vr_known being removed.
24811 (ipa_set_jf_unknown): Likewise.
24812 (ipa_get_ipa_bits_for_value): New function.
24813 (ipa_set_jfunc_bits): Likewise.
24814 (ipa_get_value_range): New overloaded functions.
24815 (ipa_set_jfunc_vr): Likewise.
24816 (ipa_compute_jump_functions_for_edge): Use the above functions to
24817 construct bits and vr parts of jump functions.
24818 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
24819 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
24821 (ipcp_grow_transformations_if_necessary): Also allocate
24822 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
24824 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
24825 them. Fix too long lines.
24826 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
24827 vr_known being removed.
24828 (ipa_read_jump_function): Use new setter functions to construct bits
24829 and vr parts of jump functions or set them to NULL.
24830 (write_ipcp_transformation_info): Adjust for bits being pointers.
24831 (read_ipcp_transformation_info): Likewise.
24832 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
24834 Include gt-ipa-prop.h.
24835 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
24837 (ipcp_store_bits_results): Likewise.
24838 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
24839 Do not write to existing jump functions but use a temporary instead.
24841 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24844 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
24845 attempt to use its first operand as BIT_FIELD_REF base.
24847 2017-03-01 Richard Biener <rguenther@suse.de>
24849 PR middle-end/79721
24850 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
24851 interpolating formula in wrapping arithmetic.
24852 (chrec_apply): Convert chrec_evaluate return value to wanted type.
24854 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24856 PR tree-optimization/79734
24857 * tree-vect-generic.c (expand_vector_condition): Optimize
24858 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
24859 Handle VEC_COND_EXPR where comparison has different inner width from
24860 type's inner width.
24862 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
24864 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
24865 markup, and similar issues. Remove @opindex entries for things
24866 that aren't options. Add missing -mmpy-option entries.
24868 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24870 PR tree-optimization/79737
24871 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
24872 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
24873 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
24874 instead of byte_size. Formatting fix.
24875 (shift_bytes_in_array_right): Formatting fix.
24877 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
24880 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
24881 condition on optimize for the leaf function test.
24883 2017-02-28 Martin Liska <mliska@suse.cz>
24886 * read-rtl-function.c (function_reader::handle_unknown_directive):
24887 Bail out when one uses -flto.
24889 2017-02-28 Martin Liska <mliska@suse.cz>
24891 * common.opt: Replace space with tabular for options of <number>
24893 * config/i386/i386.opt: Show <number> value for
24894 -mlarge-data-threshold.
24895 * opts.c (print_filtered_help): Do not display number in hexadecimal
24898 2017-02-28 Martin Liska <mliska@suse.cz>
24900 * common.opt: Fix --help=option -Q for options which are of
24903 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
24905 * config/i386/i386.c (print_reg): Error out for values
24906 of 8-bit size in invalid integer register.
24908 2017-02-28 Martin Sebor <msebor@redhat.com>
24910 PR tree-optimization/79691
24911 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
24913 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24916 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
24917 gcc_unreachable with output_operand_lossage.
24919 2017-02-28 Richard Biener <rguenther@suse.de>
24921 PR tree-optimization/79740
24922 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
24924 (visit_nary_op): Insert the nary into the hashtable if we
24925 pattern-matched sth.
24926 * tree-ssa-pre.c (eliminate_insert): Robustify.
24928 2017-02-28 Richard Biener <rguenther@suse.de>
24930 PR middle-end/79731
24931 * fold-const.c (decode_field_reference): Reject out-of-bound
24934 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24936 * config/i386/i386.c: Include intl.h.
24937 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
24938 instead of just cond ? "..." : "...".
24939 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
24940 * coverage.c (read_counts_file): Likewise.
24941 * omp-offload.c: Include intl.h.
24942 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
24943 of just cond ? "..." : "...".
24944 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
24945 of just cond ? "..." : "...".
24947 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
24950 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
24952 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
24953 'tune for' CPU name.
24954 * config/arm/arm-cpu-data.h: Regenerated.
24956 2017-02-28 Richard Biener <rguenther@suse.de>
24958 PR tree-optimization/79732
24959 * tree-inline.c (expand_call_inline): Do not shadow var.
24961 2017-02-28 Richard Biener <rguenther@suse.de>
24963 PR tree-optimization/79723
24964 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
24965 address-space properly.
24967 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
24969 * doc/optinfo.texi (Optimization groups): Fix option used for
24971 * doc/invoke.texi (-fopt-info): Document "omp".
24972 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
24973 (OPTGROUP_ALL): Add OPTGROUP_OMP.
24974 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
24975 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
24976 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
24978 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
24980 * dumpfile.c (optgroup_options): Instead of "openmp", associate
24981 OPTGROUP_OMP with "omp".
24983 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
24986 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
24987 for arithmetic shift of unsigned V2DI.
24989 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
24991 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
24992 arc/linux.h headers.
24993 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
24994 (LINK_SPEC): Likewise.
24995 (ARC_TLS_EXTRA_START_SPEC): Likewise.
24996 (EXTRA_SPECS): Likewise.
24997 (STARTFILE_SPEC): Likewise.
24998 (ENDFILE_SPEC): Likewise.
24999 (LIB_SPEC): Likewise.
25000 (TARGET_SDATA_DEFAULT): Likewise.
25001 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
25002 (MULTILIB_DEFAULTS): Likewise.
25003 (DWARF2_UNWIND_INFO): Likewise.
25004 * config/arc/big.h: New file.
25005 * config/arc/elf.h: Likewise.
25006 * config/arc/linux.h: Likewise.
25007 * config/arc/t-uClibc: Remove.
25009 2017-02-27 Bin Cheng <bin.cheng@arm.com>
25011 PR tree-optimization/77536
25012 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
25013 (tree_transform_and_unroll_loop): Use above function to compute the
25014 estimated niter of unrolled loop and use it when scaling profile.
25015 Also use count info rather than frequency if it's non-zero.
25016 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
25017 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
25018 (vect_transform_loop): Call above function.
25020 2017-02-27 Richard Biener <rguenther@suse.de>
25022 PR tree-optimization/45397
25023 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
25024 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
25025 (visit_nary_op): Add pattern matching for CSEing sign-changed
25026 or truncated operations with wider ones.
25028 2017-02-27 Richard Biener <rguenther@suse.de>
25030 PR tree-optimization/79690
25031 * tree-vect-stmts.c (vectorizable_store): Use vector type
25032 built from the DR with address-space.
25034 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
25036 * doc/invoke.texi (Optimize Options): Refine the description
25037 of asan-use-after-return.
25039 2017-02-25 Alan Modra <amodra@gmail.com>
25041 PR rtl-optimization/79584
25042 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
25043 base, not ad->base_term, the reg within base. Remove assertion
25044 that ad->base == ad->base_term. Replace gen_int_mode using
25045 bogus mode with const0_rtx.
25047 2017-02-25 Jakub Jelinek <jakub@redhat.com>
25049 PR middle-end/79396
25050 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
25051 FMA_EXPR like tcc_binary or tcc_unary.
25053 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
25056 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
25058 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
25059 (output_loc_operands): Handle DW_OP_call_ref and
25060 DW_OP_GNU_variable_value.
25061 (struct variable_value_struct): New type.
25062 (struct variable_value_hasher): Likewise.
25063 (variable_value_hash): New variable.
25064 (string_types): Remove.
25065 (copy_loc_descr): New function.
25066 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
25067 (prepend_loc_descr_to_each): New function.
25068 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
25069 instead of add_loc_descr_to_each if the first argument is single
25070 location list and the second has multiple.
25071 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
25072 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
25073 when looking for variable value which doesn't have other location info.
25074 (loc_list_from_tree): Formatting fix.
25075 (gen_array_type_die): Simplify DW_AT_string_length handling.
25076 (adjust_string_types): Remove.
25077 (gen_subprogram_die): Don't call adjust_string_types nor test/set
25078 string_types. Call resolve_variable_values.
25079 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
25080 (resolve_addr_in_expr): Likewise. Add A argument.
25081 (copy_deref_exprloc): Remove deref argument. Adjust for the
25082 original expression being DW_OP_GNU_variable_value with optionally
25083 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
25084 optionally after it.
25085 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
25086 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
25087 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
25088 (variable_value_hasher::hash, variable_value_hasher::equal): New
25090 (resolve_variable_value_in_expr, resolve_variable_value,
25091 resolve_variable_values, note_variable_value_in_expr,
25092 note_variable_value): New functions.
25093 (dwarf2out_early_finish): Call note_variable_value on all toplevel
25096 2017-02-24 Jakub Jelinek <jakub@redhat.com>
25099 * opts.h (handle_generated_option): Add GENERATED_P argument.
25100 * opts-common.c (handle_option): Adjust function comment.
25101 (handle_generated_option): Add GENERATED_P argument, pass it to
25103 (control_warning_option): Pass false to handle_generated_option
25105 * opts.c (maybe_default_option): Pass true to handle_generated_option
25107 * optc-gen.awk: Likewise.
25109 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25111 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
25112 a REG, look at the REG it is a SUBREG of.
25113 (splitter for cmpeqsi_t): Ditto.
25115 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25117 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
25118 the special USEs with the pattern of the insn, not the insn itself.
25120 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
25123 * doc/invoke.texi: Document -mload-store-pairs.
25125 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25126 Sandra Loosemore <sandra@codesourcery.com>
25128 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
25129 argument isn't a CONST_INT.
25130 (nios2_alternate_compare_const): Assert op is a CONST_INT.
25131 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
25132 (nios2_validate_compare): Bypass alternate compare logic if *op2
25133 is not a CONST_INT.
25134 (ldstwm_operation_p): Return false if first_base is not a REG or
25135 if first_offset is not a CONST_INT.
25137 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25139 * config/cris/cris.md: Use correct operand in a define_peephole2.
25141 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25143 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
25145 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25147 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
25148 this_insn if it is an INSN or JUMP_INSN.
25149 (force_offsettable): Look at base, not at addr.
25150 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
25151 on things that aren't necessarily CONST_INTs.
25153 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
25155 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
25156 -mfpmath=sse is the default also for x86-32 targets with SSE2
25157 instruction set when @option{-ffast-math} is enabled
25159 2017-02-24 Jeff Law <law@redhat.com>
25161 PR rtl-optimizatoin/79286
25162 * ira.c (update_equiv_regs): Drop may_trap_p exception to
25165 2017-02-24 Richard Biener <rguenther@suse.de>
25167 PR tree-optimization/79389
25168 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
25171 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
25173 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
25174 function comment to reflect reality.
25175 (loop_exits_before_overflow): Fix typo in function description.
25177 2017-02-24 Richard Biener <rguenther@suse.de>
25179 PR tree-optimization/79389
25180 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
25181 properly that a threading opportunity exists. Detect conditional
25182 copy/constant propagation opportunities.
25184 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
25186 * config/visium/visium.md (type): Add trap.
25187 (b): New mode attribute.
25188 (*btst): Rename into...
25189 (*btst<mode>): ...this and adjust.
25190 (*cbranchsi4_btst_insn): Rename into...
25191 (*cbranch<mode>4_btst_insn): ...this and adjust.
25192 (trap): New define_insn.
25194 2017-02-23 Jakub Jelinek <jakub@redhat.com>
25196 PR tree-optimization/79389
25197 * ifcvt.c (struct noce_if_info): Add rev_cond field.
25198 (noce_reversed_cond_code): New function.
25199 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
25200 reversed_comparison_code. Formatting fix.
25201 (noce_try_store_flag): Test rev_cond != NULL in addition to
25202 reversed_comparison_code.
25203 (noce_try_store_flag_constants): Likewise.
25204 (noce_try_store_flag_mask): Likewise.
25205 (noce_try_addcc): Use rev_cond if non-NULL instead of
25206 reversed_comparison_code.
25207 (noce_try_cmove_arith): Likewise. Formatting fixes.
25208 (noce_try_minmax, noce_try_abs): Clear rev_cond.
25209 (noce_find_if_block): Initialize rev_cond.
25210 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
25211 instead of false as last argument never attempt to reverse it
25214 2017-02-23 Bin Cheng <bin.cheng@arm.com>
25216 PR tree-optimization/79663
25217 * tree-predcom.c (combine_chains): Process refs in reverse order
25218 only for ZERO length chains, and add explaining comment.
25220 2017-02-23 Jeff Law <law@redhat.com>
25222 PR tree-optimization/79578
25223 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
25224 in call to operand_equal_p.
25226 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
25229 * config/i386/cpuid.h: Fix another undefined behavior.
25231 2017-02-23 Richard Biener <rguenther@suse.de>
25233 PR tree-optimization/79683
25234 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
25235 vector types for data-refs.
25237 2017-02-23 Martin Liska <mliska@suse.cz>
25239 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
25241 2017-02-23 Jakub Jelinek <jakub@redhat.com>
25243 PR middle-end/79665
25244 * internal-fn.c (get_range_pos_neg): Moved to ...
25245 * tree.c (get_range_pos_neg): ... here. No longer static.
25246 * tree.h (get_range_pos_neg): New prototype.
25247 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
25248 are known to be in between 0 and signed maximum inclusive, try to
25249 expand both unsigned and signed divmod and use the cheaper one from
25252 2017-02-22 Jeff Law <law@redhat.com>
25254 PR tree-optimization/79578
25255 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
25256 to compare base operands.
25258 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
25261 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
25262 gpc_reg_operand instead of fpr_reg_operand.
25264 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
25266 * config/mips/mips.c (mips_return_in_memory): Force FP
25267 vector types to be returned in memory for o32 ABI.
25269 2017-02-22 Jakub Jelinek <jakub@redhat.com>
25271 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
25272 instead of DW_TAG_member for static data member declarations and don't
25273 set no_linkage_name for static inline data members.
25274 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
25277 2017-02-22 Martin Liska <mliska@suse.cz>
25279 * doc/invoke.texi: Replace inequality signs with square brackets
25282 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25284 PR tree-optimization/68644
25285 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
25287 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
25290 * lra-constraints.c (simplify_operand_subreg): Handle
25291 WORD_REGISTER_OPERATIONS targets.
25293 2017-02-22 Jakub Jelinek <jakub@redhat.com>
25296 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
25297 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
25298 elimination by swapping fld*.
25300 2017-02-22 Richard Biener <rguenther@suse.de>
25302 PR tree-optimization/79673
25303 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
25304 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
25305 irrelevant address-space qualifiers and avoiding a
25306 ADDR_SPACE_CONVERT_EXPR from fold_convert.
25308 2017-02-22 Richard Biener <rguenther@suse.de>
25310 PR tree-optimization/79666
25311 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
25312 to not symbolically negate if that may introduce undefined
25315 2017-02-22 Martin Liska <mliska@suse.cz>
25318 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
25319 * data-streamer-out.c (streamer_write_gcov_count_stream):
25321 * value-prof.c (stream_out_histogram_value): Make assert more
25322 precise based on type of counter.
25324 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
25327 * config/i386/i386.md (standard_x87sse_constant_load splitter):
25328 Use nonimmediate_operand instead of memory_operand for operand 1.
25329 (float-extend standard_x87sse_constant_load splitter): Ditto.
25331 2017-02-21 Jeff Law <law@redhat.com>
25333 PR tree-optimization/79621
25334 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
25335 blocks with edges to themselves.
25337 2017-02-21 Jakub Jelinek <jakub@redhat.com>
25340 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
25341 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
25342 Use gimple_call_builtin_p.
25345 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
25346 on temporarily removed DEBUG_INSNs.
25348 PR tree-optimization/79649
25349 * tree-loop-distribution.c (classify_partition): Give up on
25350 non-generic address space loads/stores.
25352 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
25354 * doc/loop.texi (Loop manipulation): Remove nonexistent
25355 tree_ssa_loop_version from the documentation.
25356 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
25358 2017-02-21 Jakub Jelinek <jakub@redhat.com>
25361 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
25362 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
25363 * config/rs6000/rs6000.c: Include except.h.
25364 (rs6000_expand_split_stack_prologue): Call
25365 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
25367 2017-02-21 Martin Jambor <mjambor@suse.cz>
25370 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
25371 have been analyzed.
25373 2017-02-21 Martin Jambor <mjambor@suse.cz>
25375 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
25376 for backward compatibility only.
25377 * doc/invoke.texi (Option Summary): Remove all references to
25378 -fipa-cp-alignment.
25380 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
25384 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25386 * lra-constraints.c (curr_insn_transform): Handle
25387 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
25389 2017-02-21 Martin Liska <mliska@suse.cz>
25391 * config/i386/i386.opt: Replace -masm-dialect with -masm.
25393 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
25395 PR translation/79638
25396 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
25398 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
25401 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
25402 (arm_function_ok_for_sibcall): Return false for an indirect call by
25403 descriptor if all the argument registers are used.
25404 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
25405 alignment of the function.
25407 2017-02-21 Jakub Jelinek <jakub@redhat.com>
25409 PR tree-optimization/61441
25410 * simplify-rtx.c (simplify_const_unary_operation): For
25411 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
25412 the sNaN unmodified.
25414 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
25416 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25417 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25418 instead of SYSTEM_HEADER_DIR.
25420 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
25421 Martin Liška <mliska@suse.cz>
25423 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
25424 Fix typos and grammar, use active voice, and clarify.
25426 2017-02-20 Marek Polacek <polacek@redhat.com>
25428 PR middle-end/79537
25429 * gimplify.c (gimplify_expr): Handle unused *&&L;.
25432 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
25434 2017-02-20 Jakub Jelinek <jakub@redhat.com>
25437 * config/i386/i386.c (ix86_expand_builtin): Handle
25438 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
25439 ix86_builtins_isa[fcode].isa as a requirement of those
25440 flags and any other flag in the bitmask.
25441 (ix86_init_mmx_sse_builtins): Use 0 instead of
25442 ~OPTION_MASK_ISA_64BIT as mask.
25443 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
25444 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
25445 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
25446 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
25448 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25451 * lra-constraints.c (split_reg): Check requested split mode
25452 is supported by the register.
25454 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25456 * lra-constraints.c (simplify_operand_subreg): Remove early
25459 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25462 * lra-constraints.c (curr_insn_transform): Tighten condition
25463 for converting SUBREG reloads from OP_OUT to OP_INOUT.
25465 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25468 * lra-constraints.c (curr_insn_transform): Handle
25469 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
25471 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
25474 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
25476 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
25478 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
25481 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
25484 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25486 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
25489 2017-02-18 Jakub Jelinek <jakub@redhat.com>
25491 * final.c (last_columnnum, override_columnnum): New variables.
25492 (final_start_function): Set last_columnnum, pass it to begin_prologue
25493 hook and pass 0 to dwarf2out_begin_prologue.
25494 (final_scan_insn): Update override_columnnum. Pass last_columnnum
25495 to source_line debug hook.
25496 (notice_source_line): Compute last_columnnum and for debug_column_info
25497 return true on column changes.
25498 * debug.h (struct gcc_debug_hooks): Add column argument to
25499 source_line and begin_prologue hooks.
25500 (debug_nothing_int_charstar_int_bool): Remove prototype.
25501 (debug_nothing_int_int_charstar,
25502 debug_nothing_int_int_charstar_int_bool): New prototypes.
25503 (dwarf2out_begin_prologue): Add column argument.
25504 * debug.c (do_nothing_debug_hooks): Adjust source_line and
25505 begin_prologue hooks.
25506 (debug_nothing_int_charstar_int_bool): Remove.
25507 (debug_nothing_int_int_charstar,
25508 debug_nothing_int_int_charstar_int_bool): New functions.
25509 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
25510 through to dwarf2out_source_line.
25511 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
25512 (dwarf2out_source_line): Add column argument, emit it if requested.
25513 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
25515 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
25516 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
25517 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
25518 through to dwarf2out_begin_prologue.
25519 (vmsdbgout_source_line): Add column argument, pass it through to
25520 dwarf2out_source_line.
25521 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
25522 dbxout_source_line caller.
25523 (dbxout_source_line): Add column argument.
25525 * common.opt (gno-column-info, gcolumn-info): New options.
25526 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
25527 (check_die): Also test for multiple DW_AT_decl_column attributes.
25528 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
25529 DW_AT_decl_column if requested.
25530 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
25532 (gen_variable_die): Likewise.
25533 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
25534 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
25537 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
25538 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
25539 (ix86_handle_option): Handle OPT_m3dnowa.
25540 * doc/invoke.texi (-m3dnowa): Document.
25541 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
25542 -m3dnowa instead of -m3dnow -march=athlon.
25545 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
25546 instead of gcc_assert for K, r and R code checks. Formatting fixes.
25548 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25551 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
25552 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
25553 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
25554 generator for vsx_xxpermdi_<mode>_be.
25555 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
25556 force big-endian semantics.
25557 (vsx_xxpermdi_<mode>_be): New define_expand with same
25558 implementation as previous version of vsx_xxpermdi_<mode>.
25560 2017-02-17 Jakub Jelinek <jakub@redhat.com>
25562 PR tree-optimization/79327
25563 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
25564 variable, its initialization and use.
25566 2017-02-17 Julia Koval <julia.koval@intel.com>
25568 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
25569 (OPTION_MASK_ISA_PKU_UNSET): New.
25570 (ix86_handle_option): Handle -mrdpid.
25571 * config/i386/cpuid.h (bit_RDPID): New.
25572 * config/i386/driver-i386.c (host_detect_local_cpu):
25573 Detect RDPID feature.
25574 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
25575 * config/i386/i386-c.c (ix86_target_macros_internal):
25577 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
25578 (ix86_valid_target_attribute_inner_p): Add "rdpid".
25579 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
25580 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
25581 * config/i386/i386.md (define_insn "rdpid"): New.
25582 * config/i386/i386.opt Add -mrdpid.
25583 * config/i386/immintrin.h (_rdpid_u32): New.
25585 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
25587 PR rtl-optimization/79541
25588 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
25589 instead of transforming it into USE.
25591 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
25593 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
25594 If HONOR_SNANS (SFmode) force the input to a register.
25595 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
25596 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
25597 an frsp or similar insn.
25599 2017-02-17 Martin Liska <mliska@suse.cz>
25601 PR rtl-optimization/79577
25602 * params.def (selsched-max-sched-times): Increase minimum to 1.
25604 2017-02-17 Martin Liska <mliska@suse.cz>
25606 PR rtl-optimization/79574
25607 * gcse.c (want_to_gcse_p): Prevent integer overflow.
25609 2017-02-17 Martin Liska <mliska@suse.cz>
25611 PR tree-optimization/79529
25612 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
25613 ssa_defined_default_def_p to handle cases which are implicitly
25615 * tree-ssa.c (ssa_defined_default_def_p): New function.
25616 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
25617 which are implicitly defined.
25618 * tree-ssa.h (ssa_defined_default_def_p): Declare.
25620 2017-02-17 Richard Biener <rguenther@suse.de>
25622 PR middle-end/79576
25623 * params.def (max-ssa-name-query-depth): Limit to 10.
25625 2017-02-17 Richard Biener <rguenther@suse.de>
25627 PR tree-optimization/79552
25628 * tree-ssa-structalias.c (visit_loadstore): Properly verify
25631 2017-02-17 Richard Biener <rguenther@suse.de>
25634 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
25636 2017-02-17 Marek Polacek <polacek@redhat.com>
25638 PR middle-end/79536
25639 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
25640 (fold_negate_expr): New wrapper.
25642 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
25644 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
25645 Correct terminology and de-emphasize pre-standard behavior.
25647 2017-02-16 Alan Modra <amodra@gmail.com>
25649 PR rtl-optimization/79286
25650 * ira.c (def_dominates_uses): New function.
25651 (update_equiv_regs): Don't create an equivalence for insns that
25652 may trap where the register def does not dominate the use.
25654 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
25656 PR rtl-optimization/78127
25657 * lra.c (lra): Call lra_eliminate before finish the loop after
25660 2017-02-16 Richard Biener <rguenther@suse.de>
25662 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
25664 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
25665 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
25666 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
25667 (isl_val_int_from_wi): New function.
25668 (extract_affine_gmp): Rename to ...
25669 (extract_affine_wi): ... this, take a widest_int.
25670 (extract_affine_int): Just wrap extract_affine_wi.
25671 (add_param_constraints): Use isl_val_int_from_wi.
25672 (add_loop_constraints): Likewise, and extract_affine_wi.
25674 2017-02-15 Jeff Law <law@redhat.com>
25676 PR middle-end/79521
25677 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
25678 ira_init_register_move_cost_if_necessary.
25680 2017-02-15 Martin Sebor <msebor@redhat.com>
25682 PR middle-end/32003
25683 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
25684 removed in a prior commit.
25686 2017-02-15 Bin Cheng <bin.cheng@arm.com>
25688 PR tree-optimization/79347
25689 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
25690 counters during peeling.
25692 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
25694 * Makefile.in (site.exp): Remove "set ISLVER".
25696 2017-02-15 Jakub Jelinek <jakub@redhat.com>
25699 * real.c (real_from_integer): Call real_convert even for decimal.
25701 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
25704 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
25706 2017-02-14 Andrew Pinski <apinski@cavium.com>
25708 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
25709 cores and change the partno/implementer to be correct.
25710 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
25711 the 'B" as the implementer.
25712 * config/aarch64/aarch64-tune.md: Regenerate.
25714 2017-02-14 Carl Love <cel@us.ibm.com>
25716 * config/rs6000/rs6000.c: Add case statement entry to make the
25717 xvcvuxdsp built-in argument unsigned.
25718 * config/rs6000/vsx.md: Fix the source and return operand types so they
25719 match the instruction definitions from the ISA document. Fix typo
25720 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
25723 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
25726 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
25727 member early_clobber_alts.
25728 * lra-lives.c (reg_early_clobber_p): New.
25729 (process_bb_lives): Use it.
25730 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
25731 (debug_operand_data): Initialize early_clobber_alts.
25732 (setup_operand_alternative): Set up early_clobber_alts.
25733 (collect_non_operand_hard_regs): Ditto. Pass early clobber
25734 alternatives to new_insn_reg.
25735 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
25737 (lra_update_insn_regno_info): Pass the new arg.
25739 2017-02-14 Jakub Jelinek <jakub@redhat.com>
25741 PR middle-end/79505
25742 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
25743 (new_oacc_loop_raw): Don't clear already cleared fields.
25746 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
25747 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
25748 _mm512_prefetch_i64gather_ps): New inline functions and macros.
25750 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
25753 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
25755 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
25758 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
25759 the extra instruction to the right place to store 128-bit constant
25762 2017-02-14 Martin Sebor <msebor@redhat.com>
25764 PR middle-end/79448
25765 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
25766 warning for strings of unknown length.
25768 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
25770 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
25772 2017-02-14 Jeff Law <law@redhat.com>
25775 * ira-costs.c (scan_one_insn): Initialize register move costs
25776 for pseudos seen in USE/CLOBBER insns.
25778 PR tree-optimization/79095
25779 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
25780 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
25781 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
25782 if the operands are known to be not equal, then the resulting range
25784 (intersect_ranges): If the new range is ~[0,0] and the old range is
25785 wide, then prefer ~[0,0].
25786 * tree-vrp.c (overflow_comparison_p_1): New function.
25787 (overflow_comparison_p): New function.
25788 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
25789 if NAME is used in an overflow test.
25790 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
25791 overflow check that can be expressed as an equality test, then adjust
25792 ops to be that equality test.
25794 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25796 * config/s390/s390-builtin-types.def: Remove flags argument.
25797 * config/s390/s390.c (s390_init_builtins): Likewise.
25799 2017-02-14 Martin Liska <mliska@suse.cz>
25801 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
25802 vector. Fix trailing white spaces.
25804 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
25806 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
25809 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25811 PR rtl-optimization/68664
25812 * config/arm/arm.c (arm_sched_can_speculate_insn):
25813 New function. Declare prototype.
25814 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
25816 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25818 PR rtl-optimization/68664
25819 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
25821 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
25823 2017-02-14 Amit Pawar <amit.pawar@amd.com>
25825 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
25826 max skip bytes for function, loop and jump.
25828 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25830 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
25831 ABS_EXPR for gimple dump.
25833 2017-02-14 Jakub Jelinek <jakub@redhat.com>
25836 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
25838 PR tree-optimization/79408
25839 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
25840 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
25841 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
25842 also if rhs1 is INTEGER_CST.
25844 2017-02-14 Richard Biener <rguenther@suse.de>
25846 PR middle-end/79432
25847 * tree-into-ssa.c (insert_phi_nodes): When the function can
25848 have abnormal edges rewrite SSA names with broken use-def
25849 dominance out of SSA and register them for PHI insertion.
25851 2017-02-13 Martin Sebor <msebor@redhat.com>
25853 PR middle-end/79496
25854 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
25855 clearing info.nowrite flag when snprintf size argument is a range.
25857 2017-02-13 Jakub Jelinek <jakub@redhat.com>
25859 * cprop.c (cprop_jump): Add missing space in string literal.
25860 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
25861 (get_constraint_for_component_ref): Likewise.
25862 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
25863 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
25864 * lra-constraints.c (process_alt_operands): Likewise.
25865 * ipa-inline.c (inline_small_functions): Likewise.
25866 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
25867 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
25868 * trans-mem.c (diagnose_tm_1_op): Likewise.
25869 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
25870 (grid_parallel_clauses_gridifiable): Likewise.
25872 * config/nvptx/mkoffload.c (process): Add space in between
25875 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
25876 "MOD4_SSE_REGS" and "ALL_REGS".
25878 * spellcheck.c (test_data): Add , in between "foo" and "food".
25880 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25883 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
25884 boundary crossing check and subsequent code generation agree.
25886 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25888 * config/aarch64/aarch64.c (has_memory_op): Delete.
25889 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
25892 2017-02-13 Jakub Jelinek <jakub@redhat.com>
25894 PR rtl-optimization/79388
25895 PR rtl-optimization/79450
25896 * combine.c (distribute_notes): When removing TEM_INSN for which
25897 corresponding dest has last value recorded, invalidate that last
25900 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25902 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
25903 of explicit '@'. Add missing assembly comment marker on branch costs
25906 2017-02-13 Nathan Sidwell <nathan@acm.org>
25908 * gengtype-lex.l (<in_struct>): Add '/'.
25910 2017-02-13 Martin Liska <mliska@suse.cz>
25913 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
25915 2017-02-13 Richard Biener <rguenther@suse.de>
25917 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
25919 * configure: Re-generate.
25920 * config.in: Likewise.
25921 * graphite-dependences.c: Simplify as if
25922 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
25923 * graphite-isl-ast-to-gimple.c: Likewise.
25924 * graphite-optimize-isl.c: Likewise.
25925 * graphite-poly.c: Likewise.
25926 * graphite-sese-to-poly.c: Likewise.
25927 * graphite.h: Likewise.
25928 * toplev.c: Include isl/version.h and use isl_version () for
25929 printing the ISL version.
25930 * doc/install.texi: Update ISL requirement.
25932 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
25934 * doc/standards.texi (Standards): Update reference to
25937 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
25939 * doc/extend.texi (Named Address Spaces): sourceware.org now
25941 * doc/install.texi (Binaries): Ditto.
25944 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
25946 * doc/cpp.texi: Replace "stringify"/"stringification" with C
25947 standard terminology "stringize"/"stringizing" throughout.
25948 * doc/cppinternals.texi: Likewise.
25950 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
25952 * doc/extend.texi: Fix some spelling mistakes and typos.
25953 * doc/invoke.texi: Likewise.
25955 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
25958 * params.def (inline-min-speedup) Change from 10 to 8.
25960 2017-02-11 Jakub Jelinek <jakub@redhat.com>
25962 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
25965 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
25968 * ipa-inline-analysis.c (get_minimal_bb): New function.
25969 (record_modified): Use it.
25970 (remap_edge_change_prob): Handle also ancestor functions.
25972 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
25974 * doc/contrib.texi (Contributors): Remove broken link into
25975 the Mauve CVS repository.
25977 2017-02-11 Jakub Jelinek <jakub@redhat.com>
25979 PR middle-end/79454
25980 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
25981 result computation whenever lhs doesn't have vector mode, not
25982 just when it has BLKmode.
25984 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
25986 * doc/makefile.texi (profiledbootstrap): Refer to the
25987 installation instructions only in textual form.
25989 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25992 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
25994 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
25996 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
25997 (Specific): Update mingw-w64 reference.
25999 (Specific): Remove broken link to Renesas RX processor.
26001 2017-02-10 Richard Biener <rguenther@suse.de>
26003 * toplev.c (process_options): Do not mention obsolete graphite
26004 options when printing sorry message about missing graphite support.
26005 Mention -floop-nest-optimize.
26007 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
26009 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
26010 (vtst_p16): Likewise.
26011 (vtstq_p8): Likewise.
26012 (vtstq_p16): Likewise.
26014 (vtstq_p64): Likewise.
26015 * config/arm/arm_neon.h (vgetq_lane_p64): New.
26016 (vset_lane_p64): New.
26017 (vsetq_lane_p64): New.
26019 2017-02-10 Jakub Jelinek <jakub@redhat.com>
26021 PR tree-optimization/79411
26022 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
26023 stmt operands are SSA_NAMEs used in abnormal phis.
26024 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
26027 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
26030 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
26033 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
26035 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
26037 2017-02-09 Jakub Jelinek <jakub@redhat.com>
26039 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
26043 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
26044 not arbitrary TREE_CONSTANT.
26047 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
26048 "omp declare target link" attribute unless is_global_var.
26049 * omp-offload.c (find_link_var_op): Likewise.
26051 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
26052 Chung-Lin Tang <cltang@codesourcery.com>
26054 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
26056 (gimplify_adjust_omp_clauses): Don't delete TILE.
26057 (gimplify_omp_for): Deal with TILE.
26058 * internal-fn.c (expand_GOACC_TILE): New function.
26059 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
26061 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
26062 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
26064 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
26065 avoid DIV for outermost collapse var.
26066 (expand_oacc_for): Insert tile element loop as needed. Adjust.
26067 Remove out of date comments, fix whitespace.
26068 * omp-general.c (omp_extract_for_data): Deal with tiling.
26069 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
26070 adjust OLF_DIM_BASE value.
26071 (struct omp_for_data): Add tiling field.
26072 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
26073 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
26074 for auto loops. Remove default auto determining, moved to
26075 oacc_loop_fixed_partitions.
26076 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
26077 stmts, add e_mask field.
26078 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
26079 (oacc_thread_numbers): Use oacc_dim_call.
26080 (oacc_xform_tile): New.
26081 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
26082 (finish_oacc_loop): Adjust for ifns vector.
26083 (oacc_loop_discover_walk): Append loop abstraction sites to list,
26084 add case for GOACC_TILE fns.
26085 (oacc_loop_xform_loop): Delete.
26086 (oacc_loop_process): Iterate over call list directly, and add
26087 handling for GOACC_TILE fns.
26088 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
26090 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
26091 vector partitioning to outer loops. Assign 2 partitions to loops
26092 when available. Add TILE handling.
26093 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
26094 (execite_oacc_device_lower): Process GOACC_TILE fns,
26095 ignore unknown specs.
26096 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
26097 * tree.c (omp_clause_num_ops): Adjust TILE ops.
26098 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
26100 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
26102 * configure.ac (ACX_BUGURL): Update.
26103 * configure: Regenerate.
26105 2017-02-09 Richard Biener <rguenther@suse.de>
26107 PR tree-optimization/69823
26108 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
26109 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
26111 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
26113 * config/arc/arc-c.def: Add __NPS400__ definition.
26114 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
26115 (TARGET_NPS400): Define.
26117 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
26119 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
26121 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
26122 pointer, arch_info.
26123 (arc_cpu_types): Fill the arch_info field with a pointer into the
26124 arc_arch_types table.
26125 (arc_selected_cpu): Declare.
26126 * config/arc/arc.c (arc_selected_cpu): Make global.
26127 (arc_selected_arch): Delete.
26128 (arc_base_cpu): Delete.
26129 (arc_override_options): Remove references to deleted variables,
26130 update access to arch information.
26131 (ARC_OPT): Update access to arch information.
26132 (ARC_OPTX): Likewise.
26133 * config/arc/arc.h (arc_base_cpu): Remove declaration.
26134 (TARGET_ARC600): Update access to arch information.
26135 (TARGET_ARC601): Likewise.
26136 (TARGET_ARC700): Likewise.
26137 (TARGET_EM): Likewise.
26138 (TARGET_HS): Likewise.
26139 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
26142 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
26145 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
26146 condition/operands for integer GE/LE/GEU/LEU operations.
26148 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
26150 PR translation/79397
26151 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
26154 2017-02-08 Martin Jambor <mjambor@suse.cz>
26157 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
26158 whether allocation happened.
26159 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
26160 nothing was allocated.
26162 2017-02-08 Jakub Jelinek <jakub@redhat.com>
26164 PR tree-optimization/79408
26165 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
26166 constant, but SSA_NAME with a known integer range, use the minimum
26167 of that range instead of op1 to determine if modulo can be replaced
26168 with its first operand.
26170 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26172 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
26174 2017-02-08 Richard Biener <rguenther@suse.de>
26176 PR tree-optimization/71824
26177 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
26178 Check all loops contained in the merged region.
26180 2017-02-07 Andrew Pinski <apinski@cavium.com>
26182 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
26184 2017-02-07 Andrew Pinski <apinski@cavium.com>
26186 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
26187 (thunderxt88): Likewise.
26188 (thunderxt81): Disable LSE and change v8.1 to v8.
26189 (thunderxt83): Likewise.
26191 2017-02-07 Jakub Jelinek <jakub@redhat.com>
26192 Richard Biener <rguenther@suse.de>
26194 PR middle-end/79399
26195 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
26196 type from int to size_t.
26197 * ira-costs.c (struct_costs_size): Change type from int to size_t.
26199 2017-02-07 Jakub Jelinek <jakub@redhat.com>
26201 PR rtl-optimization/79386
26202 * cprop.c (bypass_conditional_jumps): Initialize
26203 bypass_last_basic_block already before splitting bbs after
26204 unconditional traps...
26205 (bypass_conditional_jumps): ... rather than here.
26208 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
26209 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
26210 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
26211 fix -masm=intel patterns.
26213 2017-02-07 Richard Biener <rguenther@suse.de>
26215 PR tree-optimization/79256
26216 PR middle-end/79278
26217 * builtins.c (get_object_alignment_2): Use min_align_of_type
26218 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
26219 and ADJUST_FIELD_ALIGN.
26221 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
26223 * doc/tm.texi: Regenerate.
26224 * stor-layout.c (layout_decl): Adjust.
26225 (update_alignment_for_field): Likewise.
26226 (place_field): Likewise.
26227 (min_align_of_type): Likewise.
26228 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
26229 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
26230 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
26231 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
26232 * config/frv/frv.c (frv_adjust_field_align): Likewise.
26233 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
26234 * config/i386/i386.c (x86_field_alignment): Likewise.
26235 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
26236 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
26237 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
26238 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
26239 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
26240 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
26244 2017-01-30 Richard Biener <rguenther@suse.de>
26246 PR tree-optimization/79256
26247 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
26248 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
26251 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
26253 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
26254 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
26255 builtins to SImode and emit a zero-extend, if necessary.
26257 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
26259 * docs/invoke.texi (RISC-V Options): Alphabetize.
26261 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
26263 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
26266 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
26268 * config/riscv/riscv.c: New file.
26269 * common/config/riscv/riscv-common.c: Likewise.
26270 * config.gcc: Likewise.
26271 * config/riscv/constraints.md: Likewise.
26272 * config/riscv/elf.h: Likewise.
26273 * config/riscv/generic.md: Likewise.
26274 * config/riscv/linux.h: Likewise.
26275 * config/riscv/multilib-generator: Likewise.
26276 * config/riscv/peephole.md: Likewise.
26277 * config/riscv/pic.md: Likewise.
26278 * config/riscv/predicates.md: Likewise.
26279 * config/riscv/riscv-builtins.c: Likewise.
26280 * config/riscv/riscv-c.c: Likewise.
26281 * config/riscv/riscv-ftypes.def: Likewise.
26282 * config/riscv/riscv-modes.def: Likewise.
26283 * config/riscv/riscv-opts.h: Likewise.
26284 * config/riscv/riscv-protos.h: Likewise.
26285 * config/riscv/riscv.h: Likewise.
26286 * config/riscv/riscv.md: Likewise.
26287 * config/riscv/riscv.opt: Likewise.
26288 * config/riscv/sync.md: Likewise.
26289 * config/riscv/t-elf-multilib: Likewise.
26290 * config/riscv/t-linux: Likewise.
26291 * config/riscv/t-linux-multilib: Likewise.
26292 * config/riscv/t-riscv: Likewise.
26293 * configure.ac: Likewise.
26294 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
26295 Waterman as RISC-V maintainers.
26296 * doc/install.texi: Add RISC-V entries.
26297 * doc/invoke.texi: Add RISC-V options section.
26298 * doc/md.texi: Add RISC-V constraints section.
26299 * configure: Regenerated.
26301 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
26304 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
26305 false values to be constant vectors with all 0 or all 1 bits set.
26306 (vcondu<mode><mode>): Likewise.
26307 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
26309 (fpmask_comparison_operator): Update comment.
26310 (vecint_comparison_operator): New predicate.
26311 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
26312 vector conditionals when the true and false values are constant
26313 vectors with all 0 bits or all 1 bits set.
26315 2017-02-06 Martin Sebor <msebor@redhat.com>
26317 PR tree-optimization/79376
26318 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
26320 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
26322 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
26323 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
26324 to simplify split condition.
26326 2017-02-06 Jakub Jelinek <jakub@redhat.com>
26328 * omp-expand.c (oxpand_omp_atomic_fetch_op,
26329 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
26332 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
26334 PR rtl-optimization/68664
26335 * target.def (can_speculate_insn): New hook.
26336 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
26337 * doc/tm.texi: Regenerate.
26338 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
26339 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
26340 (rs6000_sched_can_speculate_insn): New function.
26342 2017-02-06 Jakub Jelinek <jakub@redhat.com>
26344 PR tree-optimization/79284
26345 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
26346 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
26347 vectorizable_mask_load_store, vectorizable_operation,
26348 vect_is_simple_cond, get_same_sized_vectype): Use it instead
26349 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
26350 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
26351 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
26352 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
26353 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
26354 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
26355 is_gimple_assign (stmt). Replace another such test with
26356 is_gimple_assign (stmt).
26358 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
26361 * config/avr/avr.c (rtl-iter.h): Include it.
26362 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
26363 (avr_legitimate_combined_insn): ...and implementation.
26365 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
26367 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
26368 * config/s390/s390.c (s390_const_operand_ok)
26369 (s390_canonicalize_comparison, s390_extract_part)
26370 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
26371 (s390_contiguous_bitmask_p, s390_rtx_costs)
26372 (legitimize_pic_address): Likewise.
26373 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
26374 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
26375 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
26376 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
26377 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
26379 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
26381 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
26382 REGNO($0) == REGNO($1).
26384 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26386 * config/s390/linux.h(SIZE_TYPE): Add comment.
26388 2017-02-06 Julian Brown <julian@codesourcery.com>
26389 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26390 Virendra Pathak <virendra.pathak@broadcom.com>
26392 * config/aarch64/aarch64-cores.def: Change the scheduler
26394 * config/aarch64/aarch64.md: Include thunderx2t99.md.
26395 * config/aarch64/thunderx2t99.md: New file.
26397 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
26399 * doc/standards.texi (Go Language): Update link to language
26402 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
26404 * tree-eh.c (lower_resx): Sanitize profile.
26405 (cleanup_empty_eh_move_lp): Likewise.
26407 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
26410 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
26412 * cfgloopmanip.h (loop_version): Update prototype.
26413 * modulo-sched.c (sms_schedule): Update call of loop_version.
26414 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
26415 * tree-parloops.c (gen_parallel_loop): Likewise.
26416 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
26417 * tree-ssa-loop-split.c (split_loop): Likewise.
26418 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
26419 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
26421 2017-02-05 Martin Liska <mliska@suse.cz>
26424 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
26426 (print_operand_address): Initialize a struct to zero.
26428 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
26430 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
26431 garbage collector only in textual form.
26433 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
26435 * doc/extend.texi (x86 specific memory model extensions for
26436 transactional memory): Simplify a phrase.
26438 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
26441 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
26442 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
26443 (atomic_storedi_1): Likewise.
26445 2017-02-04 Jakub Jelinek <jakub@redhat.com>
26447 PR tree-optimization/79338
26448 * tree-parloops.c (gather_scalar_reductions): Don't call
26449 vect_analyze_loop_form for loop->inner before destroying loop's
26452 2017-02-03 Martin Sebor <msebor@redhat.com>
26454 PR tree-optimization/79327
26455 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
26456 when precision has resulted in leading zeros.
26457 (format_integer): Adjust the likely counter to assume an unknown
26458 argument that may be zero is non-zero.
26460 2017-02-03 Jason Merrill <jason@redhat.com>
26463 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
26464 avoid copying non-taken branch.
26466 2017-02-03 Jakub Jelinek <jakub@redhat.com>
26468 PR tree-optimization/79340
26469 * tree-vect-loop.c (vectorizable_reduction): Release
26470 vec_defs elements after safe_splicing them into other vectors.
26473 PR tree-optimization/79327
26474 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
26475 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
26477 (format_integer): Use wide_int_to_tree instead of build_int_cst
26478 + to_?hwi. If argmin is NULL, just set argmin and argmax to
26479 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
26480 of shortest and longest sequence.
26482 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
26484 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
26485 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
26487 2017-02-03 Walter Lee <walt@tilera.com>
26490 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
26491 after initial stackframe link reg save.
26492 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
26494 2017-02-03 Jakub Jelinek <jakub@redhat.com>
26497 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
26498 wu for stxssp alternative.
26500 2017-02-03 Martin Sebor <msebor@redhat.com>
26502 PR tree-optimization/79352
26503 * gimple-fold.c (get_range_strlen): Add argument.
26504 (get_range_strlen): Change return type to bool.
26505 (get_maxval_strlen): Pass in a dummy argument.
26506 * gimple-fold.h (get_range_strlen): Change return type to bool.
26507 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
26508 * tree.h (array_at_struct_end_p): Add argument.
26509 * tree.c (array_at_struct_end_p): Handle it.
26511 2017-02-03 Martin Liska <mliska@suse.cz>
26514 * multiple_target.c (create_dispatcher_calls): Redirect edge
26515 from a caller of a dispatcher.
26516 (expand_target_clones): Make the clones local.
26517 (ipa_target_clone): Do both target clones and resolvers.
26518 (ipa_dispatcher_calls): Remove the pass.
26519 (pass_dispatcher_calls::gate): Likewise.
26520 (make_pass_dispatcher_calls): Likewise.
26521 * passes.def (pass_target_clone): Put as very first IPA early
26524 2017-02-03 Martin Liska <mliska@suse.cz>
26526 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
26527 in case of a function with ifunc attribute.
26529 2017-02-03 Martin Liska <mliska@suse.cz>
26531 * cgraph.c (cgraph_node::dump): Dump function version info.
26532 * symtab.c (symtab_node::dump_base): Add missing new line.
26534 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
26536 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
26537 (ifcombine_ifandif): Use it.
26539 2017-02-03 Martin Liska <mliska@suse.cz>
26541 * doc/invoke.texi: Document default value for
26542 use-after-scope-direct-emission-threshold.
26544 2017-02-03 Martin Liska <mliska@suse.cz>
26546 PR tree-optimization/79339
26547 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
26548 (format_floating): Likewise.
26550 2017-02-03 Martin Liska <mliska@suse.cz>
26553 * ipa-prop.c (ipa_node_params_t::insert): Remove current
26555 (ipa_node_params_t::remove): Likewise.
26556 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
26557 initialization from removed ipa_node_params_t::insert.
26558 (ipa_node_params::~ipa_node_params): Move from removed
26559 ipa_node_params_t::release.
26560 * symbol-summary.h (symbol_summary::m_released): New member.
26561 Do not release a summary twice. Do not allow to call finalizer
26562 for types of a summary that live in GGC memory.
26564 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26566 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
26569 2017-02-02 Martin Sebor <msebor@redhat.com>
26571 PR middle-end/79275
26572 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
26573 (format_string): Tighten up the range of output for non-constant
26574 strings and correct the expected range for wide non-constant strings.
26576 2017-02-02 Martin Sebor <msebor@redhat.com>
26578 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
26580 PR middle-end/32003
26581 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
26583 (-fdump-tree-@var): Add to index and document how to come up
26584 with pass-specific option and dump file names.
26585 (-fdump-passes): Clarify where to look for output.
26587 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
26589 PR middle-end/77445
26590 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
26591 statistics of the analyzed path; allow threading for speed when
26592 any of BBs along the path are optimized for speed.
26594 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
26596 PR middle-end/78468
26597 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
26598 settings of the virtual registers.
26601 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
26603 * explow.c (get_dynamic_stack_size): Take known alignment of stack
26604 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
26607 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26609 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
26610 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
26612 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26614 * config/s390/s390.md: Add missing comments with the expanded
26616 * config/s390/vector.md: Likewise.
26617 * config/s390/vx-builtins.md: Likewise.
26619 2017-02-02 Jakub Jelinek <jakub@redhat.com>
26622 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
26623 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
26624 conditions on a single line.
26626 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26628 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
26629 __S390_VX__ to __VX__.
26631 2017-02-01 Andrew Pinski <apinski@cavium.com>
26633 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
26634 stmt_info to record_stmt_cost.
26635 (vect_get_known_peeling_cost): Pass stmt_info if known to
26637 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
26638 cpu_vector_cost field into
26639 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
26640 field into vec_int_stmt_cost and vec_fp_stmt_cost.
26641 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
26642 splitting of scalar_stmt_cost and vec_stmt_cost.
26643 (thunderx_vector_cost): Likewise.
26644 (cortexa57_vector_cost): LIkewise.
26645 (exynosm1_vector_cost): Likewise.
26646 (xgene1_vector_cost): Likewise.
26647 (thunderx2t99_vector_cost): Improve after the splitting of the two
26649 (aarch64_builtin_vectorization_cost): Update for the splitting of
26650 scalar_stmt_cost and vec_stmt_cost.
26652 2017-02-01 Torvald Riegel <triegel@redhat.com>
26653 Richard Henderson <rth@redhat.com>
26655 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
26656 conditional on existance of a fast atomic load.
26657 * optabs-query.c (can_atomic_load_p): New function.
26658 * optabs-query.h (can_atomic_load_p): Declare it.
26659 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
26660 no fast atomic load is available for the particular size of access.
26661 (expand_atomic_compare_and_swap): Likewise.
26662 (expand_atomic_load): Likewise.
26663 (expand_atomic_store): Likewise.
26664 (expand_atomic_fetch_op): Likewise.
26665 * testsuite/lib/target-supports.exp
26666 (check_effective_target_sync_int_128): Remove x86 because it provides
26667 no fast atomic load.
26668 (check_effective_target_sync_int_128_runtime): Likewise.
26670 2017-02-01 Richard Biener <rguenther@suse.de>
26672 * graphite.c: Include tree-vectorizer.h for find_loop_location.
26673 (graphite_transform_loops): Provide opt-info for optimized nests.
26674 * tree-parloop.c (parallelize_loops): Provide opt-info for
26675 parallelized loops.
26677 2017-02-01 Richard Biener <rguenther@suse.de>
26679 PR middle-end/79315
26680 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
26681 was not set before.
26683 2017-02-01 Richard Biener <rguenther@suse.de>
26685 PR tree-optimization/71824
26686 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
26687 Verify the loops are valid in the merged SESE region.
26688 (scop_detection::can_represent_loop_1): Check analyzing the
26689 evolution of the number of iterations in the region succeeds.
26691 2017-01-31 Ian Lance Taylor <iant@golang.org>
26693 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
26694 REG_ARGS_SIZE note to 32-bit push insns and call insn.
26696 2017-01-31 David Malcolm <dmalcolm@redhat.com>
26698 PR preprocessor/79210
26699 * input.c (get_substring_ranges_for_loc): Replace line_width
26700 assertion with error-handling.
26702 2017-01-31 Richard Biener <rguenther@suse.de>
26704 PR tree-optimization/77318
26705 * graphite-sese-to-poly.c (extract_affine): Fix assert.
26706 (create_pw_aff_from_tree): Take loop parameter.
26707 (add_condition_to_pbb): Pass loop of the condition to
26708 create_pw_aff_from_tree.
26710 2017-01-31 Jakub Jelinek <jakub@redhat.com>
26712 * config/s390/s390.c (s390_asan_shadow_offset): New function.
26713 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
26715 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
26719 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
26721 (convert_int_to_float128): Likewise.
26722 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
26723 (convert_int_to_float128): Likewise.
26724 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
26725 (UNSPEC_IEEE128_CONVERT): Likewise.
26726 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
26727 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
26728 Use local variables for IBM extended format.
26729 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
26730 (fix_trunc<mode>si2_fprs): Likewise.
26731 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
26732 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
26733 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
26734 to know that we can now have integers of all sizes in vector
26736 (fix<uns>_<mode>di2_hw): Likewise.
26737 (float<uns>_<mode>si2_hw): Likewise.
26738 (fix_<mode>si2_hw): Likewise.
26739 (fixuns_<mode>si2_hw): Likewise.
26740 (float<uns>_<mode>di2_hw): Likewise.
26741 (float_<mode>di2_hw): Likewise.
26742 (float_<mode>si2_hw): Likewise.
26743 (floatuns_<mode>di2_hw): Likewise.
26744 (floatuns_<mode>si2_hw): Likewise.
26745 (xscvqp<su>wz_<mode>): Delete, no longer used.
26746 (xscvqp<su>dz_<mode>): Likewise.
26747 (xscv<su>dqp_<mode>): Likewise.
26748 (ieee128_mfvsrd_64bit): Likewise.
26749 (ieee128_mfvsrd_32bit): Likewise.
26750 (ieee128_mfvsrwz): Likewise.
26751 (ieee128_mtvsrw): Likewise.
26752 (ieee128_mtvsrd_64bit): Likewise.
26753 (ieee128_mtvsrd_32bit): Likewise.
26755 2017-01-31 Martin Liska <mliska@suse.cz>
26758 * ipa-prop.c (ipa_free_all_node_params): Call release method
26759 instead of ~sumbol_summary to not to trigger double times
26762 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
26764 PR tree-optimization/71691
26765 * bitmap.h (class auto_bitmap): New.
26766 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
26767 is_maybe_undefined instead of ssa_undefined_value_p.
26769 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26771 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
26772 __S390_ARCH_LEVEL__ to __ARCH__.
26774 2017-01-31 Jakub Jelinek <jakub@redhat.com>
26776 PR tree-optimization/79267
26777 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
26778 if should_remove_lhs_p is true.
26780 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
26783 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
26784 (add_alignment_attribute): New.
26785 (base_type_die): Add alignment attribute.
26786 (subrange_type_die): Likewise.
26787 (modified_type_die): Likewise.
26788 (gen_array_type_die): Likewise.
26789 (gen_descr_array_type_die: Likewise.
26790 (gen_enumeration_type_die): Likewise.
26791 (gen_subprogram_die): Likewise.
26792 (gen_variable_die): Likewise.
26793 (gen_field_die): Likewise.
26794 (gen_ptr_to_mbr_type_die): Likewise.
26795 (gen_struct_or_union_type_die): Likewise.
26796 (gen_subroutine_type_die): Likewise.
26797 (gen_typedef_die): Likewise.
26798 (base_type_cmp): Compare alignment attribute.
26800 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26803 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
26804 (setb_unsigned) New pattern for setb with CCUNS.
26805 * config/rs6000/rs6000.c (expand_block_compare): Use a different
26806 subfc./subfe sequence to avoid overflow problems. Generate a
26807 shorter sequence with cmpld/setb for power9.
26808 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
26809 for generating subfc. instruction.
26810 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
26811 now uses this instruction.
26813 2017-01-30 Ian Lance Taylor <iant@google.com>
26816 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
26817 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
26819 2017-01-30 Martin Sebor <msebor@redhat.com>
26821 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
26822 Move constant to the right of a relational operator.
26823 (get_mpfr_format_length, format_character, format_string): Ditto.
26824 (should_warn_p, maybe_warn): Same.
26826 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
26828 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
26831 * asan.c (get_translation_unit_decl): Remove function.
26832 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
26834 2017-01-30 Martin Liska <mliska@suse.cz>
26836 PR gcov-profile/79259
26837 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
26838 -fprofile-generate.
26840 2017-01-30 Martin Liska <mliska@suse.cz>
26843 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
26844 Initialize variables with NULL value.
26846 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
26849 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
26851 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
26853 2017-01-30 Richard Biener <rguenther@suse.de>
26855 PR tree-optimization/79276
26856 * tree-vrp.c (process_assert_insertions): Properly adjust common
26857 when removing a duplicate.
26859 * gcc.dg/torture/pr79276.c: New testcase.
26861 2017-01-30 Richard Biener <rguenther@suse.de>
26863 PR tree-optimization/79256
26864 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
26865 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
26867 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
26869 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
26872 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
26873 ("*r<noxa>sbg_<mode>_sll_bitmask")
26874 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
26875 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
26876 Use contiguous_bitmask_nowrap_operand.
26878 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26881 * config/rs6000/altivec.h (vec_xl): Revise #define.
26882 (vec_xst): Likewise.
26884 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
26886 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
26888 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
26890 PR rtl-optimization/79194
26891 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
26892 traps before call to bypass_conditional_jumps.
26894 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
26896 PR tree-optimization/71374
26897 * lra-constraints.c (check_conflict_input_operands): New.
26898 (match_reload): Use it.
26900 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
26903 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
26904 account to calculate conflict_set.
26906 2017-01-27 Bin Cheng <bin.cheng@arm.com>
26908 PR rtl-optimization/78559
26909 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
26910 other_insn in combine.
26912 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26914 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
26915 uint16_type_node for BT_UINT16.
26917 2017-01-27 David Malcolm <dmalcolm@redhat.com>
26919 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
26920 "RTL Tests" to menu.
26921 (GIMPLE Tests): New node.
26922 (RTL Tests): New node.
26924 2017-01-27 Richard Biener <rguenther@suse.de>
26926 PR tree-optimization/79245
26927 * tree-loop-distribution.c (distribute_loop): Apply cost
26928 modeling also to detected patterns.
26930 2017-01-27 Richard Biener <rguenther@suse.de>
26932 PR tree-optimization/71433
26933 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
26934 (compare_assert_loc): New function.
26935 (process_assert_insertions): Sort and optimize assert locations
26936 to remove duplicates and push down identical assertions on
26937 edges to their destination block.
26939 2017-01-27 Richard Biener <rguenther@suse.de>
26941 PR tree-optimization/79244
26942 * tree-vrp.c (remove_range_assertions): Forcefully propagate
26943 out SSA names even if abnormal.
26945 2017-01-27 Jakub Jelinek <jakub@redhat.com>
26947 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
26948 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
26949 instead of MPFR_RNDN.
26951 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
26954 * arm.c (arm_option_override): Don't call build_target_option_node
26955 until after doing all option overrides.
26956 (arm_valid_target_attribute_tree): Likewise.
26958 2017-01-27 Martin Liska <mliska@suse.cz>
26960 * doc/invoke.texi (-fprofile-arcs): Document profiling support
26961 for {cd}tors and C++ {cd}tors.
26963 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
26965 * config/s390/s390.md ("*setmem_long_and")
26966 ("*setmem_long_and_31z"): Use zero_extend instead of and.
26968 2017-01-26 Martin Sebor <msebor@redhat.com>
26970 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
26973 2017-01-26 Martin Sebor <msebor@redhat.com>
26975 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
26976 HAVE_DFmode before using XFmode or DFmode.
26977 (parse_directive): Avoid using the z length modifier to avoid
26978 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
26980 PR middle-end/78703
26981 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
26982 to accept adjustment as an array.
26983 (get_int_range): New function.
26984 (struct directive): Make width and prec arrays.
26985 (directive::set_width, directive::set_precision): Call get_int_range.
26986 (format_integer, format_floating): Handle width and precision ranges.
26987 (format_string, parse_directive): Same.
26989 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26992 * dwarf2out.c (generate_skeleton_bottom_up): For children with
26993 comdat_type_p set, just clone them, but keep the children in the
26997 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
26998 which have direct callers with -fvar-tracking-assignments enabled
27000 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
27001 inside of type units.
27003 2017-01-26 Martin Sebor <msebor@redhat.com>
27005 PR middle-end/78703
27006 * gimple-ssa-sprintf.c (struct result_range): Add likely and
27008 (struct format_result): Replace number_chars, number_chars_min,
27009 and number_chars_max with a single member of struct result_range.
27011 (format_result::operator+=): Adjust.
27012 (struct fmtresult): Remove bounded. Handle likely and unlikely
27014 (fmtresult::adjust_for_width_or_precision): New function.
27015 (fmtresult:type_max_digits): New function.
27016 (bytes_remaining): Handle likely and unlikely counters.
27017 (min_bytes_remaining): Remove.
27018 (format_percent): Simplify.
27019 (format_integer, format_floating): Set likely and unlikely counters.
27020 (get_string_length, format_character, format_string): Same.
27021 (format_plain, should_warn_p): New function.
27022 (maybe_warn): Call should_warn_p. Update diagnostic messages
27023 and handle those for all directives, including plain strings.
27024 (format_directive): Handle likely and unlikely counters.
27025 Remove unnecessary quoting from diagnostics. Add an informational
27027 (add_bytes): Remove.
27028 (pass_sprintf_length::compute_format_length): Simplify.
27029 (try_substitute_return_value): Handle likely and unlikely counters.
27031 2017-01-26 Carl Love <cel@us.ibm.com>
27033 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
27034 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
27036 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
27039 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
27040 endianess for subregs into account.
27041 * lra-constraints.c (lra_constraints): Do risky transformations
27042 always on the first iteration.
27043 * lra-lives.c (check_pseudos_live_through_calls): Add arg
27044 last_call_used_reg_set.
27045 (process_bb_lives): Define and use last_call_used_reg_set.
27046 * lra.c (lra): Always continue after lra_constraints on the first
27049 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
27051 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
27053 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
27055 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27057 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
27058 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
27059 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
27060 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
27061 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
27062 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
27063 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
27064 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
27065 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
27067 2017-01-26 Marek Polacek <polacek@redhat.com>
27070 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
27071 for the third operand.
27073 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27075 PR middle-end/79236
27076 * omp-low.c (struct omp_context): Add simt_stmt field.
27077 (scan_omp_for): Return omp_context *.
27078 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
27079 context to the _simt_ SIMD stmt.
27080 (lower_omp_for): For combined SIMD with sibling _simt_
27081 SIMD, make sure to use the same decls in _looptemp_
27082 clauses as in the sibling.
27084 2017-01-26 David Sherwood <david.sherwood@arm.com>
27086 PR middle-end/79212
27087 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
27090 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27093 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
27094 emit fld b; fld a; if possible.
27096 * brig-builtins.def: Update copyright years.
27097 * config/arm/arm_acle_builtins.def: Update copyright years.
27099 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
27102 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
27103 constraint instead of o for the stxsd instruction.
27105 2017-01-25 Carl Love <cel@us.ibm.com>
27107 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
27108 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
27110 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
27112 * doc/invoke.texi (C++ Dialect Options): Fix typo.
27114 2017-01-25 Richard Biener <rguenther@suse.de>
27116 PR tree-optimization/69264
27117 * target.def (vector_alignment_reachable): Improve documentation.
27118 * doc/tm.texi: Regenerate.
27119 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
27121 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
27122 earlier changes with respect to TYPE_USER_ALIGN.
27123 (vector_alignment_reachable_p): Likewise. Improve dumping.
27125 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27128 * config/arm/arm.md (xordi3): Force constant operand into a register
27131 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27133 * doc/invoke.texi (-fstore-merging): Correct default optimization
27134 levels at which it is enabled.
27135 (-O): Move -fstore-merging from list to...
27138 2017-01-25 Richard Biener <rguenther@suse.de>
27141 * omp-expand.c: Include debug.h.
27142 (expand_omp_taskreg): Make sure to generate early debug before
27143 outlining anything from a function.
27144 (expand_omp_target): Likewise.
27145 (grid_expand_target_grid_body): Likewise.
27147 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
27150 * asan.c (get_translation_unit_decl): New function.
27151 (asan_add_global): Extract modules file name from globals
27152 TRANSLATION_UNIT_DECL name.
27154 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
27157 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
27158 for long calls with APCS frame and VFP.
27160 2017-01-24 David Malcolm <dmalcolm@redhat.com>
27162 * cfg.c (original_copy_tables_initialized_p): New function.
27163 * cfg.h (original_copy_tables_initialized_p): New decl.
27164 * cfgrtl.c (relink_block_chain): Guard the call to
27165 free_original_copy_tables with a call to
27166 original_copy_tables_initialized_p.
27167 * cgraph.h (symtab_node::native_rtl_p): New decl.
27168 * cgraphunit.c (symtab_node::native_rtl_p): New function.
27169 (symtab_node::needed_p): Don't assert for early assembly output
27170 for __RTL functions.
27171 (cgraph_node::finalize_function): Set "force_output" for __RTL
27173 (cgraph_node::analyze): Bail out early for __RTL functions.
27174 (analyze_functions): Update assertion to support __RTL functions.
27175 (cgraph_node::expand): Bail out early for __RTL functions.
27176 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
27178 * function.h (struct function): Update comment for field
27180 * gimple-expr.c: Include "tree-pass.h".
27181 (gimple_has_body_p): Return false for __RTL functions.
27182 * Makefile.in (OBJS): Add run-rtl-passes.o.
27183 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
27185 (gcc::pass_manager::get_clean_slate): New accessor.
27186 * passes.c: Include "insn-addr.h".
27187 (should_skip_pass_p): Add logging. Update logic for running
27188 "expand" to be compatible with both __GIMPLE and __RTL. Guard
27189 property-provider override so it is only done for gimple passes.
27191 (skip_pass): New function.
27192 (execute_one_pass): Call skip_pass when skipping passes.
27193 * read-md.c (md_reader::read_char): Support filtering
27194 the input to a subset of line numbers.
27195 (md_reader::md_reader): Initialize fields
27196 m_first_line and m_last_line.
27197 (md_reader::read_file_fragment): New function.
27198 * read-md.h (md_reader::read_file_fragment): New decl.
27199 (md_reader::m_first_line): New field.
27200 (md_reader::m_last_line): New field.
27201 * read-rtl-function.c (function_reader::create_function): Only
27202 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
27203 curr_properties. Set DECL_INITIAL to a dummy block.
27204 (read_rtl_function_body_from_file_range): New function.
27205 * read-rtl-function.h (read_rtl_function_body_from_file_range):
27207 * run-rtl-passes.c: New file.
27208 * run-rtl-passes.h: New file.
27210 2017-01-24 Jeff Law <law@redhat.com>
27212 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
27215 2017-01-24 Bin Cheng <bin.cheng@arm.com>
27217 PR tree-optimization/79159
27218 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
27219 (record_nonwrapping_iv): Improve boundary using above function if no
27220 value range information.
27222 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
27223 Martin Jambor <mjambor@suse.cz>
27225 * brig-builtins.def: New file.
27226 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
27227 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
27228 (DEF_HSAIL_SAT_BUILTIN): Likewise.
27229 (DEF_HSAIL_INTR_BUILTIN): Likewise.
27230 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
27231 * builtin-types.def (BT_INT8): New.
27232 (BT_INT16): Likewise.
27233 (BT_UINT8): Likewise.
27234 (BT_UINT16): Likewise.
27235 (BT_FN_ULONG): Likewise.
27236 (BT_FN_UINT_INT): Likewise.
27237 (BT_FN_UINT_ULONG): Likewise.
27238 (BT_FN_UINT_LONG): Likewise.
27239 (BT_FN_UINT_PTR): Likewise.
27240 (BT_FN_ULONG_PTR): Likewise.
27241 (BT_FN_INT8_FLOAT): Likewise.
27242 (BT_FN_INT16_FLOAT): Likewise.
27243 (BT_FN_UINT32_FLOAT): Likewise.
27244 (BT_FN_UINT16_FLOAT): Likewise.
27245 (BT_FN_UINT8_FLOAT): Likewise.
27246 (BT_FN_UINT64_FLOAT): Likewise.
27247 (BT_FN_UINT16_UINT32): Likewise.
27248 (BT_FN_UINT32_UINT16): Likewise.
27249 (BT_FN_UINT16_UINT16_UINT16): Likewise.
27250 (BT_FN_INT_PTR_INT): Likewise.
27251 (BT_FN_UINT_PTR_UINT): Likewise.
27252 (BT_FN_LONG_PTR_LONG): Likewise.
27253 (BT_FN_ULONG_PTR_ULONG): Likewise.
27254 (BT_FN_VOID_UINT64_UINT64): Likewise.
27255 (BT_FN_UINT8_UINT8_UINT8): Likewise.
27256 (BT_FN_INT8_INT8_INT8): Likewise.
27257 (BT_FN_INT16_INT16_INT16): Likewise.
27258 (BT_FN_INT_INT_INT): Likewise.
27259 (BT_FN_UINT_FLOAT_UINT): Likewise.
27260 (BT_FN_FLOAT_UINT_UINT): Likewise.
27261 (BT_FN_ULONG_UINT_UINT): Likewise.
27262 (BT_FN_ULONG_UINT_PTR): Likewise.
27263 (BT_FN_ULONG_ULONG_ULONG): Likewise.
27264 (BT_FN_UINT_UINT_UINT): Likewise.
27265 (BT_FN_VOID_UINT_PTR): Likewise.
27266 (BT_FN_UINT_UINT_PTR: Likewise.
27267 (BT_FN_UINT32_UINT64_PTR): Likewise.
27268 (BT_FN_INT_INT_UINT_UINT): Likewise.
27269 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
27270 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
27271 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
27272 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
27273 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
27274 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
27275 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
27276 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
27277 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
27278 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
27279 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
27280 * doc/frontends.texi: List BRIG FE.
27281 * doc/install.texi (Testing): Add BRIG tesring requirements.
27282 * doc/invoke.texi (Overall Options): Mention BRIG.
27283 * doc/standards.texi (Standards): Doucment BRIG HSA version.
27285 2017-01-24 Richard Biener <rguenther@suse.de>
27287 PR translation/79208
27288 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
27290 2017-01-24 Martin Jambor <mjambor@suse.cz>
27293 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
27294 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
27295 and known_contexts.
27297 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
27299 PR middle-end/79123
27300 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
27301 casts from signed to unsigned really don't have a range.
27303 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
27305 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
27306 GMP_RNDx for compatiblity.
27308 2017-01-24 Martin Liska <mliska@suse.cz>
27311 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
27312 that would prevent us to call alloca with -1 as argument.
27314 2017-01-24 Jakub Jelinek <jakub@redhat.com>
27316 * dwarf2out.c (output_compilation_unit_header, output_file_names):
27317 Avoid -Wformat-security warning.
27319 2017-01-23 Andrew Pinski <apinski@cavium.com>
27321 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
27324 2017-01-23 Martin Sebor <msebor@redhat.com>
27326 PR middle-end/78703
27327 * gimple-ssa-sprintf.c (warn_level): New global.
27328 (format_integer): Use it here and throughout the rest of the file.
27329 Use the same switch to compute sign as base.
27330 (maybe_warn): New function.
27331 (format_directive): Factor out warnings into maybe_warn.
27332 Add debugging output. Use warn_level.
27333 (add_bytes): Use warn_level.
27334 (pass_sprintf_length::compute_format_length): Add debugging output.
27335 (try_substitute_return_value): Same.
27336 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
27338 PR middle-end/78703
27339 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
27340 (struct fmtresult, format_integer, format_floating): Adjust.
27341 (fmtresult::fmtresult): Set max correctly in two argument ctor.
27342 (get_string_length, format_string,format_directive): Same.
27343 (pass_sprintf_length::compute_format_length): Same.
27344 (try_substitute_return_value): Simplify slightly.
27346 PR middle-end/78703
27347 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
27348 (fmtresult::operator+=): Outlined.
27349 (struct fmtresult): Add ctors.
27350 (struct conversion_spec): Rename...
27351 (struct directive): ...to this. Add and remove data members.
27352 (directive::set_width, directive::set_precision): New functions.
27353 (format_percent): Use fmtresult ctor.
27354 (get_width_and_precision): Remove.
27355 (format_integer): Make naming changes. Avoid computing width and
27357 (format_floating): Same. Adjust indentation.
27358 (format_character, format_none): New functions.
27359 (format_string): Moved character handling to format_character.
27360 (format_directive): Remove arguments, change return type.
27361 (parse_directive): New function.
27362 (pass_sprintf_length::compute_format_length): Move directive
27363 parsing to parse_directive.
27365 2017-01-23 Jakub Jelinek <jakub@redhat.com>
27367 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
27368 (assign_assembler_name_if_needed): ... this.
27369 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
27370 (assign_assembler_name_if_needed): ... this.
27371 (free_lang_data_in_cgraph): Adjust callers.
27372 * cgraphunit.c (cgraph_node::analyze): Likewise.
27373 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
27376 2017-01-23 Richard Biener <rguenther@suse.de>
27378 PR tree-optimization/79088
27379 PR tree-optimization/79188
27380 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
27381 resetting loop bounds after last path deletion. Reset loop
27382 bounds of the target loop, make code match the comments.
27383 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
27384 Make sure loops need no fixups.
27386 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
27388 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
27389 exponent support with double type for first argument.
27390 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
27391 type returned by __builtin_vec_extract_sig,
27392 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
27393 functions from "vector int" to "vector unsigned int" or from
27394 "vector long long int" to "vector unsigned long long int".
27395 Changed type returned by __builtin_vec_extract_exp,
27396 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
27397 functions from "vector int" to "vector unsigned int" or from
27398 "vector long long int" to "vector unsigned long long int".
27399 Changed return type of __builtin_vec_test_data_class,
27400 __builtin_vec_test_data_class_sp, and
27401 __builtin_vec_test_data_class_dp from "vector int" to
27402 "vector bool int" or from "vector long long int" to "vector bool
27403 long long int" and changed second argument type from "unsigned
27404 int" to "int". Added new overloaded function forms "vector float
27405 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
27406 "vector float __builtin_vec_insert_exp_sp (vector float, vector
27407 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
27408 double, vector unsigned long long int)" and "vector double
27409 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
27410 long int)". Changed return type of
27411 __builtin_scalar_test_data_class and
27412 __builtin_scalar_test_data_class_sp and
27413 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
27414 int" and changed second argument from "unsigned int" to "int".
27415 Changed type returned by __builtin_scalar_test_neg,
27416 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
27417 from "int" to "bool int". Added new overloaded function form
27418 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
27419 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
27420 exponent double-precision with floating point first argument.
27421 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
27422 documentation of scalar_test_data_class, scalar_test_neg,
27423 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
27424 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
27425 vec_test_data_class built-in functions to reflect refinements in
27426 their type signatures.
27428 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
27430 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
27432 (aarch64_elf_asm_destructor): Likewise.
27434 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
27436 PR rtl-optimization/78634
27437 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
27438 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
27439 * ifcvt.c (noce_try_cmove): Add missing cost check.
27441 PR rtl-optimization/71724
27442 * combine.c (if_then_else_cond): Look for situations where it is
27443 beneficial to undo the work of one of the recursive calls.
27445 2017-01-23 Bin Cheng <bin.cheng@arm.com>
27447 PR tree-optimization/70754
27448 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
27449 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
27450 combined stmt before it if not NULL.
27451 (combine_chains): Process refs reversely and compute dominance point
27454 2017-01-23 Martin Liska <mliska@suse.cz>
27456 PR tree-optimization/79196
27457 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
27458 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
27460 (strlen_optimize_stmt): Call the renamed function.
27462 2017-01-23 Michael Matz <matz@suse.de>
27464 PR tree-optimization/78384
27465 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
27467 2017-01-23 Richard Biener <rguenther@suse.de>
27469 PR tree-optimization/79186
27470 * tree-vrp.c (register_new_assert_for): Make sure we've seen
27471 both incoming edges before moving an assert.
27473 2017-01-23 Martin Jambor <mjambor@suse.cz>
27475 * ipa-prop.c (load_from_param_1): Removed.
27476 (load_from_unmodified_param): Bits from load_from_param_1 put back
27478 (load_from_param): Removed.
27479 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
27480 with stmt. Reverted back to use of load_from_unmodified_param.
27482 2017-01-23 Martin Jambor <mjambor@suse.cz>
27485 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
27486 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
27487 field a pointer to garbage collected vector, mark lattices and
27488 ipcp_orig_node with GTY((skip)).
27489 (ipa_get_param_count): Adjust to descriptors being a pointer.
27490 (ipa_get_param): Likewise.
27491 (ipa_get_type): Likewise.
27492 (ipa_get_param_move_cost): Likewise.
27493 (ipa_set_param_used): Likewise.
27494 (ipa_get_controlled_uses): Likewise.
27495 (ipa_set_controlled_uses): Likewise.
27496 (ipa_is_param_used): Likewise.
27497 (ipa_node_params_t): Move into garbage collector. New methods insert
27499 (ipa_node_params_sum): Annotate wth GTY(()).
27500 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
27502 (ipa_load_from_parm_agg): Adjust declaration.
27503 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
27504 * ipa-profile.c (ipa_profile): Likewise.
27505 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
27506 (ipa_populate_param_decls): Make descriptors parameter garbage
27508 (ipa_dump_param): Adjust to descriptors being a pointer.
27509 (ipa_alloc_node_params): Likewise.
27510 (ipa_initialize_node_params): Likewise.
27511 (load_from_param_1): Make descriptors parameter garbage collected.
27512 (load_from_unmodified_param): Likewise.
27513 (load_from_param): Likewise.
27514 (ipa_load_from_parm_agg): Likewise.
27515 (ipa_node_params::~ipa_node_params): Removed.
27516 (ipa_free_all_node_params): Remove call to delete operator.
27517 (ipa_node_params_t::insert): New.
27518 (ipa_node_params_t::remove): Likewise.
27519 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
27520 copy known_csts and known_contexts vectors.
27521 (ipa_read_node_info): Adjust to descriptors being a pointer.
27522 (ipcp_modif_dom_walker): Make m_descriptors field garbage
27524 (ipcp_transform_function): Make descriptors variable garbage
27527 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
27529 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
27530 * config/i386/avx512dqintrin.h: Ditto.
27531 * config/i386/avx512fintrin.h: Ditto.
27532 * config/i386/i386.c: Handle new builtins.
27533 * config/i386/i386-builtin.def: Add new builtins.
27534 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
27535 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
27537 2017-01-23 Jakub Jelinek <jakub@redhat.com>
27538 Martin Liska <mliska@suse.cz>
27540 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
27541 * asan.c (asan_expand_poison_ifn): Support stores and use
27542 appropriate ASAN report function.
27543 * internal-fn.c (expand_ASAN_POISON_USE): New function.
27544 * internal-fn.def (ASAN_POISON_USE): Declare.
27545 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
27546 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
27547 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
27548 ASAN_POISON calls w/o LHS.
27549 * tree-ssa.c (execute_update_addresses_taken): Create clobber
27550 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
27551 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
27552 * gimplify.c (asan_poison_variables): Add attribute
27553 use_after_scope_memory to variables that really needs to live
27555 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
27556 having the attribute.
27558 2017-01-23 Martin Liska <mliska@suse.cz>
27560 * asan.c (create_asan_shadow_var): New function.
27561 (asan_expand_poison_ifn): Likewise.
27562 * asan.h (asan_expand_poison_ifn): New declaration.
27563 * internal-fn.c (expand_ASAN_POISON): Likewise.
27564 * internal-fn.def (ASAN_POISON): New builtin.
27565 * sanopt.c (pass_sanopt::execute): Expand
27566 asan_expand_poison_ifn.
27567 * tree-inline.c (copy_decl_for_dup_finish): Make function
27569 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
27570 * tree-ssa.c (is_asan_mark_p): New function.
27571 (execute_update_addresses_taken): Rewrite local variables
27572 (identified just by use-after-scope as addressable) into SSA.
27574 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
27576 * doc/install.texi (Specific): opensource.apple.com uses https
27577 now. Remove trailing slash.
27579 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
27581 * README.Portability: Remove note on an Irix compatibility issue.
27583 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
27585 * gcov.c (INCLUDE_ALGORITHM): Define.
27586 (INCLUDE_VECTOR): Define.
27587 No longer include <vector> and <algorithm> directly.
27589 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
27591 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
27593 * doc/invoke.texi (Code Gen Options): Ditto.
27595 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
27598 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
27600 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
27602 rtl-optimization/79125
27603 * cprop.c (local_cprop_pass): Handle cases where we make an
27604 unconditional trap.
27606 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
27610 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
27611 read from, for big endian.
27613 2017-01-20 Jiong Wang <jiong.wang@arm.com>
27615 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
27616 register pauth builtins for LP64 only.
27618 2017-01-20 Marek Polacek <polacek@redhat.com>
27621 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
27624 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
27626 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
27628 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
27629 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
27630 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
27632 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27635 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
27636 in big-endian mode when they are not a single duplicated value.
27638 2017-01-20 Richard Biener <rguenther@suse.de>
27640 * BASE-VER: Bump to 7.0.1.
27642 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
27644 * omp-low.c (omplow_simd_context): New struct. Use it...
27645 (lower_rec_simd_input_clauses): ...here and...
27646 (lower_rec_input_clauses): ...here to hold common data. Adjust all
27647 references to idx, lane, max_vf, is_simt.
27649 2017-01-20 Graham Markall <graham.markall@embecosm.com>
27651 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
27654 2017-01-20 Martin Jambor <mjambor@suse.cz>
27656 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
27657 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
27659 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
27660 (GTFILES): Rename hsa.c to hsa-common.c.
27661 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
27662 * hsa-dump.c: Likewise.
27663 * hsa-gen.c: Likewise.
27664 * hsa-regalloc.c: Likewise.
27665 * ipa-hsa.c: Likewise.
27666 * omp-expand.c: Likewise.
27667 * omp-low.c: Likewise.
27668 * toplev.c: Likewise.
27670 2017-01-20 Marek Polacek <polacek@redhat.com>
27673 * doc/invoke.texi: Document -Wduplicated-branches.
27674 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
27675 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
27676 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
27677 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
27678 return 0 only when not OEP_LEXICOGRAPHIC.
27679 (fold_build_cleanup_point_expr): Use the expression
27680 location when building CLEANUP_POINT_EXPR.
27681 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
27682 * tree.c (add_expr): Handle error_mark_node.
27684 2017-01-20 Martin Liska <mliska@suse.cz>
27687 * tree-profile.c (init_ic_make_global_vars): Do not call
27689 (gimple_init_gcov_profiler): Likewise.
27691 2017-01-20 Martin Liska <mliska@suse.cz>
27694 * cgraph.h (maybe_create_reference): Remove argument and
27696 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
27698 * ipa-cp.c (create_specialized_node): Likewise.
27699 * symtab.c (symtab_node::maybe_create_reference): Handle
27700 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
27702 2017-01-20 Martin Liska <mliska@suse.cz>
27704 * read-rtl-function.c (function_reader::create_function): Use
27705 build_decl instread of build_decl_stat.
27707 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
27709 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
27710 * config/i386/avx512dqintrin.h: Ditto.
27711 * config/i386/avx512fintrin.h: Ditto.
27712 * config/i386/i386-builtin-types.def: Add new types.
27713 * config/i386/i386.c: Handle new types.
27714 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
27715 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
27716 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
27717 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
27718 (__builtin_ia32_kshiftridi): New.
27719 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
27721 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
27725 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
27726 define to rs6000_init_stack_protect_guard.
27727 (rs6000_init_stack_protect_guard): New function.
27729 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
27730 Yunqiang Su <yunqiang.su@imgtec.com>
27732 * config.gcc (supported_defaults): Add madd4.
27733 (with_madd4): Add validation.
27734 (all_defaults): Add madd4.
27735 * config/mips/mips.opt (mmadd4): New option.
27736 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
27738 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
27740 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
27741 (ISA_HAS_FUSED_MADD4): Likewise.
27742 * doc/invoke.texi (-mmadd4): Document the new option.
27743 * doc/install.texi (--with-madd4): Document the new option.
27745 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27747 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
27748 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
27749 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
27750 (aarch64_init_pauth_hint_builtins): New.
27751 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
27752 (aarch64_expand_builtin): Expand new builtins.
27754 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27756 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
27757 * combine-stack-adj.c (no_unhandled_cfa): Handle
27758 REG_CFA_TOGGLE_RA_MANGLE.
27759 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
27760 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
27761 info for return address signing.
27762 (aarch64_expand_epilogue): Likewise.
27764 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27766 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
27767 * config/aarch64/aarch64-protos.h
27768 (aarch64_return_address_signing_enabled): New declaration.
27769 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
27771 (aarch64_expand_prologue): Sign return address before it's pushed onto
27773 (aarch64_expand_epilogue): Authenticate return address fetched from
27775 (aarch64_override_options): Sanity check for ILP32 and ISA level.
27776 (aarch64_attributes): New function attributes for "sign-return-address".
27777 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
27778 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
27779 ("*do_return"): Generate combined instructions according to key index.
27780 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
27781 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
27783 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
27784 * config/aarch64/aarch64.opt (msign-return-address=): New.
27785 * doc/extend.texi (AArch64 Function Attributes): Documents
27786 "sign-return-address=".
27787 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
27789 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
27791 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
27792 overall option summary.
27794 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27796 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
27797 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
27798 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
27799 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
27801 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
27803 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
27804 -mpower9-minmax by default for -mcpu=power9.
27805 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
27806 128-bit floating point.
27808 2017-01-20 Alan Modra <amodra@gmail.com>
27810 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
27811 optimizing for size.
27813 2017-01-20 Alan Modra <amodra@gmail.com>
27816 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
27817 for strcmp and strncmp from corresponding builtin decl.
27819 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
27821 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
27822 instead of i386/rtems-64.h.
27823 * config/i386/rtems-64.h: Remove.
27825 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
27829 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
27831 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
27833 2017-01-19 Tamar Christina <tamar.christina@arm.com>
27835 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
27836 Change int to HOST_WIDE_INT.
27837 * config/aarch64/aarch64-protos.h
27838 (aarch64_simd_gen_const_vector_dup): Likewise.
27839 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
27841 2017-01-19 David Malcolm <dmalcolm@redhat.com>
27843 * langhooks-def.h (lhd_type_for_size): New decl.
27844 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
27845 * langhooks.c (lhd_type_for_size): New function, taken from
27848 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
27850 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
27851 define_bypass for CR latency.
27852 (power9-cracked-alu): Update bypass latency and remove power9-branch.
27853 (power9-alu2): Add define_bypass for CR latency.
27855 (power9-mul): Update insn latency.
27856 (power9-mul-compare): Update insn latency, bypass latency and remove
27859 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27861 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
27863 * config/aarch64/aarch64.md
27864 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
27865 aarch64_nopcrelative_literal_loads.
27866 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
27868 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
27870 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
27871 TARGET_LOONGSON_3A.
27872 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
27874 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
27877 * config.gcc (supported_defaults): Add lxc1-sxc1.
27878 (with_lxc1_sxc1): Add validation.
27879 (all_defaults): Add lxc1-sxc1.
27880 * config/mips/mips.opt (mlxc1-sxc1): New option.
27881 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
27883 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
27884 __mips_no_lxc1_sxc1.
27885 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
27886 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
27887 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
27889 2017-01-19 Richard Biener <rguenther@suse.de>
27891 PR tree-optimization/72488
27892 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
27893 sure to restore SSA info.
27894 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
27896 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
27898 PR rtl-optimization/79121
27899 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
27900 of the inner type when shifting an extended value.
27902 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
27905 * symtab.c (symtab_node::equal_address_to): Fix comparing of
27906 interposable aliases.
27908 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
27911 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
27912 Use the evmergelohi instruction.
27913 (mov_si<mode>_e500_subreg4_2_le): Likewise.
27914 (mov_sitf_e500_subreg8_2_be): Likewise.
27915 (mov_sitf_e500_subreg12_2_le): Likewise.
27916 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
27917 (mov_si<mode>_e500_subreg4_2_be): Likewise.
27918 (mov_sitf_e500_subreg8_2_le): Likewise.
27919 (mov_sitf_e500_subreg12_2_be): Likewise.
27921 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27923 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
27924 attribute from vecsimple to vecperm.
27925 (altivec_vbpermq2): Likewise.
27927 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27930 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
27932 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27933 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
27934 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
27935 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
27936 case where N arg is SIZE_MAX.
27937 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
27938 (cmpstrsi): Add pattern.
27940 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27942 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27943 __builtin_vec_revb builtins.
27944 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
27945 built-in functions to support generation of the ISA 3.0 XXBR<x>
27946 vector byte reverse instructions.
27947 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
27948 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
27949 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
27950 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
27951 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
27952 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
27953 (P9V_BUILTIN_VEC_REVB): Likewise.
27954 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
27955 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
27956 (p9_xxbrq_v16qi): Likewise.
27957 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
27958 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
27959 (p9_xxbrh_v8hi): Likewise.
27960 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
27961 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
27962 vec_revb built-in functions.
27964 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
27966 PR rtl-optimization/78952
27967 * config/i386/i386.md (any_extract): New code iterator.
27968 (*insvqi_2): Use any_extract for source operand.
27969 (*insvqi_3): Use any_shiftrt for source operand.
27971 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
27973 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
27975 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
27977 2017-01-18 Matthias Klose <doko@ubuntu.com>
27979 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
27981 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27983 * config/rs6000/altivec.h (vec_bperm): Change #define.
27984 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
27985 (altivec_vbpermq2): New define_insn.
27986 (altivec_vbpermd): Likewise.
27987 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
27988 function interface.
27989 (VBPERMD): Likewise.
27990 (VBPERM): New polymorphic function interface.
27991 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
27992 Add entries for P9V_BUILTIN_VEC_VBPERM.
27993 * doc/extend.texi: Add interfaces for vec_bperm.
27995 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27997 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
27998 first letter of error messages.
27999 (s390_resolve_overloaded_builtin): Likewise.
28000 * config/s390/s390.c (s390_expand_builtin): Likewise.
28001 (s390_invalid_arg_for_unprototyped_fn): Likewise.
28002 (s390_valid_target_attribute_inner_p): Likewise.
28003 * config/s390/s390.md ("tabort"): Likewise.
28005 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
28007 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
28008 (ISA_AVOID_DIV_HILO): New macro.
28009 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
28010 (ISA_HAS_DDIV): Likewise.
28012 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28014 * doc/invoke.texi (fabi-version): Correct number of occurrences.
28016 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28018 * doc/invoke.texi (fabi-version): Spelling fix.
28020 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28023 * doc/invoke.texi (fabi-version): Mention mangling fix for
28026 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28029 * doc/invoke.texi (fabi-version): Document discriminator mangling.
28031 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
28034 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
28035 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
28037 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
28039 (stack_protect_test): Ditto.
28040 * config/rs6000/rs6000.opt (mstack-protector-guard=,
28041 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
28043 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
28044 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
28045 -mstack-protector-guard-offset=.
28046 (RS/6000 and PowerPC Options): Ditto.
28048 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
28050 * config/i386/i386.h (MASK_CLASS_P): New define.
28051 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
28052 there are no registers from different register sets also when
28053 mask registers are used. Update function comment.
28054 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
28055 to (*k/*r) and (*k/*km) alternatives.
28057 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
28059 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
28060 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
28061 (EH_RETURN_HANDLER_RTX): New define.
28062 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
28063 Force frame pointer in EH return functions.
28064 (aarch64_expand_epilogue): Add barrier for eh_return.
28065 (aarch64_final_eh_return_addr): Remove.
28066 (aarch64_eh_return_handler_rtx): New function.
28067 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
28069 (aarch64_eh_return_handler_rtx): New prototype.
28071 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28073 * config/rs6000/altivec.h (vec_rlmi): New #define.
28074 (vec_vrlnm): Likewise.
28075 (vec_rlnm): Likewise.
28076 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
28077 (UNSPEC_VRLNM): Likewise.
28078 (VIlong): New mode iterator.
28079 (altivec_vrl<VI_char>mi): New define_insn.
28080 (altivec_vrl<VI_char>nm): Likewise.
28081 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
28083 (VRLDNM): Likewise.
28084 (RLNM): New polymorphic function entry.
28085 (VRLWMI): New monomorphic function entry.
28086 (VRLDMI): Likewise.
28087 (RLMI): New polymorphic function entry.
28088 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
28089 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
28090 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
28093 2017-01-17 Jakub Jelinek <jakub@redhat.com>
28096 * dwarf2out.c (field_byte_offset): Restore the
28097 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
28098 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
28099 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
28102 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
28105 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
28107 2017-01-17 Jakub Jelinek <jakub@redhat.com>
28110 * dwarf2out.c (add_data_member_location_attribute): For constant
28111 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
28112 instead of DW_AT_data_member_location, DW_AT_bit_offset and
28113 DW_AT_byte_size attributes.
28115 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
28117 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
28118 after forcing to constant memory when the code model is medium.
28120 2017-01-17 Julia Koval <julia.koval@intel.com>
28123 * config/i386/avx512fintrin.h
28124 (_mm512_i32gather_ps): Change __addr type to void const*.
28125 (_mm512_mask_i32gather_ps): Ditto.
28126 (_mm512_i32gather_pd): Ditto.
28127 (_mm512_mask_i32gather_pd): Ditto.
28128 (_mm512_i64gather_ps): Ditto.
28129 (_mm512_mask_i64gather_ps): Ditto.
28130 (_mm512_i64gather_pd): Ditto.
28131 (_mm512_mask_i64gather_pd): Ditto.
28132 (_mm512_i32gather_epi32): Ditto.
28133 (_mm512_mask_i32gather_epi32): Ditto.
28134 (_mm512_i32gather_epi64): Ditto.
28135 (_mm512_mask_i32gather_epi64): Ditto.
28136 (_mm512_i64gather_epi32): Ditto.
28137 (_mm512_mask_i64gather_epi32): Ditto.
28138 (_mm512_i64gather_epi64): Ditto.
28139 (_mm512_mask_i64gather_epi64): Ditto.
28140 (_mm512_i32scatter_ps): Change __addr type to void*.
28141 (_mm512_mask_i32scatter_ps): Ditto.
28142 (_mm512_i32scatter_pd): Ditto.
28143 (_mm512_mask_i32scatter_pd): Ditto.
28144 (_mm512_i64scatter_ps): Ditto.
28145 (_mm512_mask_i64scatter_ps): Ditto.
28146 (_mm512_i64scatter_pd): Ditto.
28147 (_mm512_mask_i64scatter_pd): Ditto.
28148 (_mm512_i32scatter_epi32): Ditto.
28149 (_mm512_mask_i32scatter_epi32): Ditto.
28150 (_mm512_i32scatter_epi64): Ditto.
28151 (_mm512_mask_i32scatter_epi64): Ditto.
28152 (_mm512_i64scatter_epi32): Ditto.
28153 (_mm512_mask_i64scatter_epi32): Ditto.
28154 (_mm512_i64scatter_epi64): Ditto.
28155 (_mm512_mask_i64scatter_epi64): Ditto.
28156 * config/i386/avx512pfintrin.h
28157 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
28158 (_mm512_mask_prefetch_i32gather_ps): Ditto.
28159 (_mm512_mask_prefetch_i64gather_pd): Ditto.
28160 (_mm512_mask_prefetch_i64gather_ps): Ditto.
28161 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
28162 (_mm512_prefetch_i32scatter_ps): Ditto.
28163 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
28164 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
28165 (_mm512_prefetch_i64scatter_pd): Ditto.
28166 (_mm512_prefetch_i64scatter_ps): Ditto.
28167 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
28168 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
28169 * config/i386/avx512vlintrin.h
28170 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
28171 (_mm_mmask_i32gather_ps): Ditto.
28172 (_mm256_mmask_i32gather_pd): Ditto.
28173 (_mm_mmask_i32gather_pd): Ditto.
28174 (_mm256_mmask_i64gather_ps): Ditto.
28175 (_mm_mmask_i64gather_ps): Ditto.
28176 (_mm256_mmask_i64gather_pd): Ditto.
28177 (_mm_mmask_i64gather_pd): Ditto.
28178 (_mm256_mmask_i32gather_epi32): Ditto.
28179 (_mm_mmask_i32gather_epi32): Ditto.
28180 (_mm256_mmask_i32gather_epi64): Ditto.
28181 (_mm_mmask_i32gather_epi64): Ditto.
28182 (_mm256_mmask_i64gather_epi32): Ditto.
28183 (_mm_mmask_i64gather_epi32): Ditto.
28184 (_mm256_mmask_i64gather_epi64): Ditto.
28185 (_mm_mmask_i64gather_epi64): Ditto.
28186 (_mm256_i32scatter_ps): Change __addr type to void*.
28187 (_mm256_mask_i32scatter_ps): Ditto.
28188 (_mm_i32scatter_ps): Ditto.
28189 (_mm_mask_i32scatter_ps): Ditto.
28190 (_mm256_i32scatter_pd): Ditto.
28191 (_mm256_mask_i32scatter_pd): Ditto.
28192 (_mm_i32scatter_pd): Ditto.
28193 (_mm_mask_i32scatter_pd): Ditto.
28194 (_mm256_i64scatter_ps): Ditto.
28195 (_mm256_mask_i64scatter_ps): Ditto.
28196 (_mm_i64scatter_ps): Ditto.
28197 (_mm_mask_i64scatter_ps): Ditto.
28198 (_mm256_i64scatter_pd): Ditto.
28199 (_mm256_mask_i64scatter_pd): Ditto.
28200 (_mm_i64scatter_pd): Ditto.
28201 (_mm_mask_i64scatter_pd): Ditto.
28202 (_mm256_i32scatter_epi32): Ditto.
28203 (_mm256_mask_i32scatter_epi32): Ditto.
28204 (_mm_i32scatter_epi32): Ditto.
28205 (_mm_mask_i32scatter_epi32): Ditto.
28206 (_mm256_i32scatter_epi64): Ditto.
28207 (_mm256_mask_i32scatter_epi64): Ditto.
28208 (_mm_i32scatter_epi64): Ditto.
28209 (_mm_mask_i32scatter_epi64): Ditto.
28210 (_mm256_i64scatter_epi32): Ditto.
28211 (_mm256_mask_i64scatter_epi32): Ditto.
28212 (_mm_i64scatter_epi32): Ditto.
28213 (_mm_mask_i64scatter_epi32): Ditto.
28214 (_mm256_i64scatter_epi64): Ditto.
28215 (_mm256_mask_i64scatter_epi64): Ditto.
28216 (_mm_i64scatter_epi64): Ditto.
28217 (_mm_mask_i64scatter_epi64): Ditto.
28218 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
28219 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
28220 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
28221 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
28222 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
28223 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
28224 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
28225 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
28226 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
28227 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
28228 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
28229 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
28230 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
28231 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
28232 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
28233 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
28234 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
28235 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
28236 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
28237 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
28238 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
28239 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
28240 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
28241 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
28242 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
28243 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
28244 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
28245 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
28246 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
28247 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
28248 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
28249 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
28250 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
28251 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
28252 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
28253 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
28254 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
28255 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
28256 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
28257 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
28258 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
28259 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
28260 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
28261 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
28262 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
28263 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
28264 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
28265 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
28266 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
28267 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
28268 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
28269 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
28270 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
28271 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
28272 definitions accordingly.
28274 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
28275 Kuan-Lin Chen <kuanlinchentw@gmail.com>
28278 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
28281 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
28284 * ira-conflicts.c (ira_build_conflicts): Update total conflict
28285 hard regs for inner regno.
28287 2017-01-17 Martin Liska <mliska@suse.cz>
28290 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
28291 assumption and add comment.
28293 2017-01-17 Nathan Sidwell <nathan@acm.org>
28295 * ipa-visibility.c (localize_node): New function, broken out of ...
28296 (function_and_variable_visibility): ... here. Call it.
28298 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
28300 PR middle-end/77445
28301 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
28302 correctly set frequency of oudgoing edge.
28303 (duplicate_thread_path): Fix profile updating.
28305 2017-01-17 Jakub Jelinek <jakub@redhat.com>
28308 * configure.ac: Add GCC_BASE_VER.
28309 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
28310 version from BASE-VER file.
28311 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
28312 (gcc.o): Depend on $(BASEVER).
28313 * common.opt (dumpfullversion): New option.
28314 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
28315 * doc/invoke.texi: Document -dumpfullversion.
28316 * doc/install.texi: Document --with-gcc-major-version-only.
28317 * configure: Regenerated.
28319 2017-01-17 Richard Biener <rguenther@suse.de>
28321 PR tree-optimization/71433
28322 * tree-vrp.c (register_new_assert_for): Merge same asserts
28323 on all incoming edges.
28324 (process_assert_insertions_for): Handle insertions at the
28327 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
28329 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
28330 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
28332 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
28335 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
28338 2017-01-17 Alan Modra <amodra@gmail.com>
28341 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
28342 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
28343 symbolic stack limit when pic.
28345 2017-01-16 Martin Sebor <msebor@redhat.com>
28347 PR tree-optimization/78608
28348 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
28350 2017-01-16 Jeff Law <law@redhat.com>
28353 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
28354 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
28355 for several include directories that may be relative to sysroot.
28356 * config/i386/x-mingw32 (gplus_includedir): Define.
28357 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
28358 (native_system_includedir): Likewise.
28359 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
28360 override if TARGET_SYSTEM_ROOT is defined.
28361 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
28363 PR tree-optimization/79090
28364 PR tree-optimization/33562
28365 PR tree-optimization/61912
28366 PR tree-optimization/77485
28367 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
28368 and computed trims into the dump file.
28370 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
28372 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
28374 2017-01-16 Jakub Jelinek <jakub@redhat.com>
28377 * gimplify.c (gimplify_init_constructor): If want_value and
28378 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
28382 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
28383 sequence. Formatting fixes.
28384 (doloop_optimize): Formatting fixes.
28387 * gcc.c (debug_level_greater_than_spec_func): New function.
28388 (static_spec_functions): Add debug-level-gt spec function.
28389 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
28391 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
28392 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
28393 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
28394 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
28395 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
28396 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
28398 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
28400 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
28401 QImode fixups to general and mask registers only.
28403 2017-01-16 Carl Love <cel@us.ibm.com>
28405 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
28406 for built-in functions
28407 vector signed char vec_nabs (vector signed char)
28408 vector signed short vec_nabs (vector signed short)
28409 vector signed int vec_nabs (vector signed int)
28410 vector signed long long vec_nabs (vector signed long long)
28411 vector float vec_nabs (vector float)
28412 vector double vec_nabs (vector double)
28413 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
28415 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
28416 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
28417 * doc/extend.texi: Update the documentation file for the new built-in
28420 2017-01-16 Martin Sebor <msebor@redhat.com>
28422 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
28425 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28427 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
28428 UNSPEC_VSX__XXSPLTD to require special splat handling.
28430 2017-01-16 David Malcolm <dmalcolm@redhat.com>
28433 * system.h: Poison strndup.
28435 2017-01-16 Alan Modra <amodra@gmail.com>
28438 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
28441 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
28443 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
28445 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
28447 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
28448 call recog here. Assert that INSN_CODE (insn) is non-negative.
28450 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
28453 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
28455 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
28456 in the currently scheduled RTL fragment.
28458 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
28460 PR rtl-optimization/78751
28461 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
28464 2017-01-14 Jeff Law <law@redhat.com>
28466 PR tree-optimization/79090
28467 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
28468 variable length stores.
28469 (compute_trims): Delete dead assignment to *trim_tail.
28470 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
28473 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
28475 PR rtl-optimization/78626
28476 PR rtl-optimization/78727
28477 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
28478 of a block, and split such blocks after everything else is finished.
28480 2017-01-14 Alan Modra <amodra@gmail.com>
28483 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
28484 target legitimate_combined_insn.
28485 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
28486 (rs6000_legitimate_combined_insn): New function.
28487 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
28489 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
28490 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
28491 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
28493 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
28495 * doc/frontends.texi (G++ and GCC): Remove references to Java.
28497 2017-01-13 Jeff Law <law@redhat.com>
28499 PR tree-optimization/33562
28500 PR tree-optimization/61912
28501 PR tree-optimization/77485
28502 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
28504 (delete_dead_assignment): Likewise.
28505 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
28506 statement to delete_dead_call and delete_dead_assignment.
28508 2017-01-13 David Malcolm <dmalcolm@redhat.com>
28511 * substring-locations.c (format_warning_va): Strengthen case 1 so
28512 that both endpoints of the substring must be within the format
28513 range for just the substring to be printed.
28515 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
28517 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
28518 * config/i386/i386.c (ix86_target_string): Add missing options
28519 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
28520 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
28521 flags_other and ix86_target_other to flags2_other. Display unknown
28523 (ix86_valid_target_attribute_inner_p): Add missing options and
28524 reorder options by implied ISAs, as in ix86_target_string.
28526 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
28528 * hash-table.h (hash_table::too_empty_p): New function.
28529 (hash_table::expand): Use it.
28530 (hash_table::traverse): Likewise.
28531 (hash_table::empty_slot): Use sizeof (value_type) instead of
28532 sizeof (PTR) to convert bytes to elements. Shrink the table
28533 if the current size is excessive for the current number of
28536 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
28538 * ira-costs.c (record_reg_classes): Break from the inner loop
28539 early once alt_fail is known to be true. Update outer loop
28540 handling accordingly.
28542 2017-01-13 Jeff Law <law@redhat.com>
28544 * tree-ssa-dse.c (decrement_count): New function.
28545 (increment_start_addr, maybe_trim_memstar_call): Likewise.
28546 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
28547 when we know the partially dead statement is a mem* function.
28549 PR tree-optimization/61912
28550 PR tree-optimization/77485
28551 * tree-ssa-dse.c: Include expr.h.
28552 (maybe_trim_constructor_store): New function.
28553 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
28555 PR tree-optimization/33562
28556 PR tree-optimization/61912
28557 PR tree-optimization/77485
28558 * doc/invoke.texi: Document new dse-max-object-size param.
28559 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
28560 * tree-ssa-dse.c: Include params.h.
28561 (dse_store_status): New enum.
28562 (initialize_ao_ref_for_dse): New, partially extracted from
28564 (valid_ao_ref_for_dse, normalize_ref): New.
28565 (setup_live_bytes_from_ref, compute_trims): Likewise.
28566 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
28567 (maybe_trim_partially_dead_store): Likewise.
28568 (maybe_trim_complex_store): Likewise.
28569 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
28570 Track what bytes live from the original store. Return tri-state
28571 for dead, partially dead or live.
28572 (dse_dom_walker): Add constructor, destructor and new private members.
28573 (delete_dead_call, delete_dead_assignment): New extracted from
28575 (dse_optimize_stmt): Make a member of dse_dom_walker.
28576 Use initialize_ao_ref_for_dse.
28578 PR tree-optimization/33562
28579 PR tree-optimization/61912
28580 PR tree-optimization/77485
28581 * sbitmap.h (bitmap_count_bits): Prototype.
28582 (bitmap_clear_range, bitmap_set_range): Likewise.
28583 * sbitmap.c (bitmap_clear_range): New function.
28584 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
28586 2017-01-13 Martin Liska <mliska@suse.cz>
28589 * function.c (set_cfun): Add new argument force.
28590 * function.h (set_cfun): Likewise.
28591 * ipa-inline-transform.c (inline_call): Use the function when
28592 strict alising from is dropped for function we inline to.
28594 2017-01-13 Richard Biener <rguenther@suse.de>
28596 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
28597 for dumping GIMPLE INTEGER_CSTs.
28599 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28601 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
28602 to 201112L since C++17.
28604 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
28607 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
28608 if -fsanitize=kernel-address is present.
28610 2017-01-13 Richard Biener <rguenther@suse.de>
28612 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
28613 as _Literal ( type ) number in case usual suffixes do not
28614 preserve all information.
28616 2017-01-13 Richard Biener <rguenther@suse.de>
28618 PR tree-optimization/77283
28619 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
28620 and ssa-iterators.h.
28621 (is_feasible_trace): Implement a cost model based on joiner
28624 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
28627 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
28628 char or short to __float128/_Float128 directly.
28630 2017-01-12 Martin Sebor <msebor@redhat.com>
28632 to -Wformat-overflow.
28633 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
28634 (min_bytes_remaining): Same.
28635 (get_string_length): Same.
28636 (format_string): Same.
28637 (format_directive): Same.
28639 (pass_sprintf_length::handle_gimple_call): Same.
28641 2017-01-12 Jakub Jelinek <jakub@redhat.com>
28643 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
28644 info.nowrite calls with no lhs that can't throw. Return bool
28645 whether gsi_remove has been called or not.
28646 (pass_sprintf_length::handle_gimple_call): Return bool whether
28647 try_substitute_return_value called gsi_remove. Formatting fix.
28648 (pass_sprintf_length::execute): Don't use gsi_remove if
28649 handle_gimple_call returned true.
28652 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
28653 be removed due to side-effects, don't remove following barrier nor
28654 turn the successor edge into fallthru edge.
28656 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28659 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
28660 element-reversing loads and stores as not swappable.
28662 2017-01-12 Nathan Sidwell <nathan@acm.org>
28663 Nicolai Stange <nicstange@gmail.com>
28665 * combine.c (try_combine): Don't ignore result of overlap checking
28666 loop. Combine overlap & asm check into single loop.
28668 2017-01-12 Richard Biener <rguenther@suse.de>
28670 * tree-pretty-print.c (dump_generic_node): Provide -gimple
28671 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
28673 2017-01-12 Richard Biener <rguenther@suse.de>
28675 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
28676 and TS_TARGET_OPTION directly derive from TS_BASE.
28677 * tree-core.h (tree_optimization_option): Derive from tree_base.
28678 (tree_target_option): Likewise.
28680 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
28682 * config/i386/i386.c (memory_address_length): Increase len
28683 only when rip_relative_addr_p returns false.
28685 2017-01-11 Julia Koval <julia.koval@intel.com>
28687 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
28688 (OPTION_MASK_ISA_SGX_SET): New.
28689 (ix86_handle_option): Handle OPT_msgx.
28690 * config.gcc: Added sgxintrin.h.
28691 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
28692 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
28693 * config/i386/i386.c (ix86_target_string): Add -msgx.
28695 (ix86_option_override_internal): Handle new options.
28696 (ix86_valid_target_attribute_inner_p): Add sgx.
28697 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
28698 * config/i386/i386.opt: Add msgx.
28699 * config/i386/sgxintrin.h: New file.
28700 * config/i386/x86intrin.h: Add sgxintrin.h.
28702 2017-01-11 Jakub Jelinek <jakub@redhat.com>
28705 * fold-const.c (maybe_nonzero_address): Return 1 for function
28707 (tree_single_nonzero_warnv_p): Don't handle function local objects
28711 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
28714 2017-01-11 David Malcolm <dmalcolm@redhat.com>
28717 * opts.c: Include "spellcheck.h"
28718 (struct string_fragment): New struct.
28719 (struct edit_distance_traits<const string_fragment &>): New
28721 (get_closest_sanitizer_option): New function.
28722 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
28724 2017-01-11 Jakub Jelinek <jakub@redhat.com>
28726 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
28728 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
28729 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
28730 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
28731 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
28732 for initial die_offset if dwarf_split_debug_info.
28733 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
28734 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
28736 (output_skeleton_debug_sections): Formatting fix. Use
28737 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
28738 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
28740 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
28742 * config/arm/cortex-a53.md: Add bypasses for
28743 cortex_a53_r2f_cvt.
28744 (cortex_a53_r2f): Only use for transfers.
28745 (cortex_a53_f2r): Likewise.
28746 (cortex_a53_r2f_cvt): Add reservation for conversions.
28747 (cortex_a53_f2r_cvt): Likewise.
28749 2017-01-11 Tamar Christina <tamar.christina@arm.com>
28751 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
28752 to all inlined functions, change static to extern.
28754 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
28757 * config/arm/arm.c (legitimize_pic_address): Handle reference to
28759 (arm_assemble_integer): Likewise.
28761 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
28763 * config.gcc: Use new awk script to check CPU, FPU and architecture
28764 parameters for --with-... options.
28765 * config/arm/parsecpu.awk: New file
28766 * config/arm/arm-cpus.in: New file.
28767 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
28769 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
28771 * config/arm/t-arm: Update dependency rules.
28772 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
28773 of processing .def files.
28774 * config/arm/genopt.sh: Deleted.
28775 * config/arm/gentune.sh: Deleted.
28776 * config/arm/arm-cores.def: Deleted.
28777 * config/arm/arm-arches.def: Deleted.
28778 * config/arm/arm-fpus.def: Deleted.
28779 * config/arm/arm-tune.md: Regenerated.
28780 * config/arm/arm-tables.opt: Regenerated.
28781 * config/arm/arm-cpu.h: New generated file.
28782 * config/arm/arm-cpu-data.h: New generated file.
28783 * config/arm/arm-cpu-cdata.h: New generated file.
28785 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
28788 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
28790 (input_varpool_node): Unpack dynamically_initialized bit.
28792 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
28794 PR rtl-optimization/79032
28795 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
28796 the alignment of the adjusted memory reference against that of MODE,
28797 instead of the alignment of the original memory reference.
28799 2017-01-11 Martin Jambor <mjambor@suse.cz>
28801 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
28803 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
28804 decorated functions.
28806 2017-01-11 Richard Biener <rguenther@suse.de>
28808 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
28809 set range/nonnull info for PHI results. Do not set it on
28810 stmts marked for removal.
28812 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
28814 * expr.c (store_field): In the bitfield case, fetch the return value
28815 from the registers before applying a single big-endian adjustment.
28816 Always do a final load for a BLKmode value not larger than a word.
28818 2017-01-10 David Malcolm <dmalcolm@redhat.com>
28821 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
28822 that we correctly handle column numbers greater than
28823 LINE_MAP_MAX_COLUMN_NUMBER.
28825 2017-01-10 Martin Sebor <msebor@redhat.com>
28827 PR middle-end/78245
28828 * gimple-ssa-sprintf.c (get_destination_size): Call
28829 {init,fini}object_sizes.
28830 * tree-object-size.c (addr_object_size): Adjust.
28831 (pass_through_call): Adjust.
28832 (pass_object_sizes::execute): Adjust.
28833 * tree-object-size.h (fini_object_sizes): Declare.
28835 2017-01-10 Martin Sebor <msebor@redhat.com>
28837 PR tree-optimization/78775
28838 * builtins.c (get_size_range): Move...
28839 * calls.c: ...to here.
28840 (alloc_max_size): Accept zero argument.
28841 (operand_signed_p): Remove.
28842 (maybe_warn_alloc_args_overflow): Call get_size_range.
28843 * calls.h (get_size_range): Declare.
28845 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
28847 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
28848 from TI's devices.csv file as of September 2016.
28849 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
28851 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
28853 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
28854 * doc/invoke.texi: Likewise.
28855 * doc/md.texi: Likewise.
28856 * doc/objc.texi: Likewise.
28858 2017-01-10 Joshua Conner <joshconner@google.com>
28860 * config/arm/fuchsia-elf.h: New file.
28861 * config/fuchsia.h: New file.
28862 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
28863 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
28865 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
28867 2016-01-10 Richard Biener <rguenther@suse.de>
28869 PR tree-optimization/79034
28870 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28871 Propagate out degenerate PHIs in the joiner.
28873 2017-01-10 Martin Liska <mliska@suse.cz>
28875 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
28876 (sort_congruence_classes_by_decl_uid): Likewise.
28877 (sort_congruence_class_groups_by_decl_uid): Likewise.
28878 (sem_item_optimizer::merge_classes): Sort class, groups in these
28879 classes and members in the groups by DECL_UID of declarations.
28880 This would make merge operations stable.
28882 2017-01-10 Martin Liska <mliska@suse.cz>
28884 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
28885 usage of m_classes_vec.
28886 (sem_item_optimizer::~sem_item_optimizer): Likewise.
28887 (sem_item_optimizer::get_group_by_hash): Likewise.
28888 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
28889 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
28890 (sem_item_optimizer::verify_classes): Likewise.
28891 (sem_item_optimizer::process_cong_reduction): Likewise.
28892 (sem_item_optimizer::dump_cong_classes): Likewise.
28893 (sem_item_optimizer::merge_classes): Likewise.
28894 * ipa-icf.h (congruence_class_hash): Rename from
28895 congruence_class_group_hash. Remove declaration of m_classes_vec.
28897 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
28899 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
28900 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
28901 * config.gcc: Add avx512vpopcntdqintrin.h.
28902 * config/i386/avx512vpopcntdqintrin.h: New.
28903 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
28904 * config/i386/i386-builtin-types.def: Add new types.
28905 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
28906 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
28907 __builtin_ia32_vpopcountq_v8di_mask): New.
28908 * config/i386/i386-c.c (ix86_target_macros_internal): Define
28909 __AVX512VPOPCNTDQ__.
28910 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
28911 (PTA_AVX512VPOPCNTDQ): Define.
28912 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
28913 TARGET_AVX512VPOPCNTDQ_P): Define.
28914 * config/i386/i386.opt: Add mavx512vpopcntdq.
28915 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
28916 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
28918 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28920 PR middle-end/77484
28921 * predict.def (PRED_CALL): Set to 67.
28923 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
28925 * expr.c (store_field): In the bitfield case, if the value comes from
28926 a function call and is of an aggregate type returned in registers, do
28927 not modify the field mode; extract the value in all cases if the mode
28928 is BLKmode and the size is not larger than a word.
28930 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
28933 * config/i386/cpuid.h: Fix undefined behavior.
28935 2017-01-04 Jeff Law <law@redhat.com>
28937 PR tree-optimization/79007
28938 PR tree-optimization/67955
28939 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
28940 conservative for pt.null when flag_non_call_exceptions is on.
28942 2017-01-09 Jakub Jelinek <jakub@redhat.com>
28944 PR translation/79019
28945 PR translation/79020
28946 * params.def (PARAM_INLINE_MIN_SPEEDUP,
28947 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
28948 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
28950 * config/avr/avr.opt (maccumulate-args): Likewise.
28951 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
28952 * common.opt (freport-bug): Likewise.
28953 * cif-code.def (CIF_FINAL_ERROR): Likewise.
28954 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
28955 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
28956 translatable string.
28957 * config/i386/i386.c (function_value_32): Likewise.
28958 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
28959 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
28961 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
28962 * common/config/msp430/msp430-common.c (msp430_handle_option):
28964 * symtab.c (symtab_node::verify_base): Likewise.
28965 * opts.c (set_debug_level): Likewise.
28966 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
28967 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
28968 missing whitespace to translatable strings.
28969 * config/avr/avr.md (bswapsi2): Fix typo in comment.
28970 * config/sh/superh.h: Likewise.
28971 * config/i386/xopintrin.h: Likewise.
28972 * config/i386/znver1.md: Likewise.
28973 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
28974 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
28975 * double-int.h (struct double_int): Likewise.
28976 * double-int.c (div_and_round_double): Likewise.
28977 * wide-int.cc: Likewise.
28978 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
28979 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
28980 * cfgcleanup.c (crossjumps_occured): Renamed to ...
28981 (crossjumps_occurred): ... this.
28982 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
28985 PR tree-optimization/78899
28986 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
28987 returning bool return struct loop *, NULL for failure and the new
28989 (versionable_outer_loop_p): Don't version outer loop if it has
28990 dont_vectorized bit set.
28991 (tree_if_conversion): When versioning outer loop, ensure
28992 tree_if_conversion is performed also on the inner loop of the
28993 non-vectorizable outer loop copy.
28994 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
28995 LOOP_VECTORIZED in inner loop of the scalar outer loop and
28996 prevent vectorization of it.
28997 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
28998 the outer loop vectorization of the non-scalar version is attempted
28999 before vectorization of the inner loop in scalar version. If
29000 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
29001 vectorization of its inner loop.
29002 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
29003 has 2 inner loops, rename also on edges from bb whose single pred
29004 is outer_loop->header. Fix typo in function comment.
29006 2017-01-09 Martin Sebor <msebor@redhat.com>
29009 * asan.c (asan_emit_stack_protection): Increase local buffer size
29010 to avoid snprintf truncation warning.
29012 2017-01-09 Andrew Pinski <apinski@cavium.com>
29014 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
29015 to reference thunderx2t99 for the tuning structure
29016 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
29018 (thunderx2t99_extra_costs): This.
29019 * config/aarch64/aarch64-tune.md: Regenerate.
29020 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
29021 (vulcan_addrcost_table): This.
29022 (vulcan_regmove_cost): Rename to ...
29023 (thunderx2t99_regmove_cost): This.
29024 (vulcan_vector_cost): Rename to ...
29025 (thunderx2t99_vector_cost): this.
29026 (vulcan_branch_cost): Rename to ...
29027 (thunderx2t99_branch_cost): This.
29028 (vulcan_tunings): Rename to ...
29029 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
29030 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
29032 2017-01-09 Martin Jambor <mjambor@suse.cz>
29036 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
29037 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
29038 (propagate_vr_accross_jump_function): Use the above function for all
29039 value range computations for pass-through jump functions and type
29040 converasion from explicit value range values.
29041 (ipcp_propagate_stage): Do not attempt to deduce types of formal
29042 parameters from TYPE_ARG_TYPES.
29043 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
29044 (ipa_write_node_info): Stream type of the actual argument.
29045 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
29047 2017-01-09 Martin Liska <mliska@suse.cz>
29050 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
29051 (lookup_compiler): Do not show error message with have_E.
29053 2017-01-09 Jakub Jelinek <jakub@redhat.com>
29055 PR tree-optimization/78938
29056 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
29057 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
29058 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
29059 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
29062 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29064 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
29067 2017-01-09 Richard Biener <rguenther@suse.de>
29069 PR tree-optimization/78997
29070 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
29071 name condition properly.
29073 2017-01-09 Richard Biener <rguenther@suse.de>
29076 * dwarf2out.c (is_cxx): New overload with context.
29077 (is_naming_typedef_decl): Use it.
29079 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
29081 * invoke.texi (Option Summary): Correct spacing in option lists
29082 and add line breaks to fix over-long lines.
29084 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
29086 PR middle-end/17660
29088 * extend.texi (Common Variable Attributes): Add xref to GCC
29089 Internals manual to explain mode attribute keywords.
29091 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
29094 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
29095 and Preprocessor Options.
29096 (Options for Linking): Document -pthread here....
29097 (RS/6000 and PowerPC Options): ...not here.
29098 (Solaris 2 Options): ...or here.
29099 * doc/cppopts.texi: Document -pthread.
29101 2017-01-08 Martin Sebor <msebor@redhat.com>
29103 PR middle-end/77708
29104 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
29105 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
29106 New member functions.
29107 (format_directive): Used them.
29109 (pass_sprintf_length::handle_gimple_call): Same.
29110 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
29111 to avoid truncation for any argument.
29112 (extract_affine_mul): Same.
29113 * tree.c (get_file_function_name): Same.
29115 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29117 PR middle-end/77484
29118 * predict.def (PRED_INDIR_CALL): Set to 86.
29120 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
29122 PR preprocessor/54124
29123 * doc/cppopts.texi: Reformat -d subtable to list the full name
29124 of the options. Add cross-reference to the docs for the general
29125 compiler -d options.
29126 * doc/invoke.texi (Developer Options): Add cross-reference to the
29127 preprocessor-specific -d option documentation.
29129 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
29131 PR preprocessor/13498
29132 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
29133 redudant material, and reflect new command-line options.
29134 (System Headers): Likewise.
29136 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
29138 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
29139 -isystem, and -idirafter. Copy-edit.
29140 * doc/cppopts.texi: Copy-edit. Remove contradiction about
29141 default for -ftrack-macro-expansion. Delete obsolete and
29142 badly-formatted implementation details about -fdebug-cpp output.
29143 * doc/cppwarnopts.texi: Copy-edit.
29145 2017-01-07 David Malcolm <dmalcolm@redhat.com>
29148 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
29149 that the transition from a max line width >= 1<<10 to narrower
29150 lines works correctly.
29152 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
29154 * doc/options.texi (PerFunction): New.
29155 * opt-functions.awk (switch_flags): Map both Optimization and
29156 PerFunction to CL_OPTIMIZATION.
29157 * opth-gen.awk: Test for PerFunction flag along with
29159 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
29160 it only when the latter is present. Skip those that don't in
29161 the hash function generator.
29162 * common.opt (fvar-tracking): Mark as PerFunction instead of
29164 (fvar-tracking-assignments): Likewise.
29165 (fvar-tracking-assignments-toggle): Likewise.
29166 (fvar-tracking-uninit): Likewise.
29168 2017-01-07 Jakub Jelinek <jakub@redhat.com>
29170 PR translation/79018
29171 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
29174 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
29177 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
29178 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
29179 TARGET_LONG_JUMP_TABLE_OFFSETS.
29180 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
29181 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
29182 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
29183 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
29184 * config/m68k/m68k.md (tablejump expander): Likewise.
29185 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
29186 TARGET_LONG_JUMP_TABLE_OFFSETS.
29187 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
29188 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
29190 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
29191 David Holsgrove <david.holsgrove@xilinx.com>
29193 * common/config/microblaze/microblaze-common.c
29194 (TARGET_EXCEPT_UNWIND_INFO): Remove.
29195 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
29197 * config/microblaze/microblaze.c (microblaze_must_save_register)
29198 (microblaze_expand_epilogue, microblaze_return_addr): Handle
29200 (microblaze_eh_return): New function.
29201 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
29202 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
29203 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
29204 * config/microblaze/microblaze.md (eh_return): New pattern.
29206 2017-01-06 Jakub Jelinek <jakub@redhat.com>
29208 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
29209 GCC_DIAGNOSTIC_STRINGIFY): Define.
29211 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
29213 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29215 * config/arm/arm.md (<mcrr>): New.
29217 * config/arm/arm.c (arm_arch5te): New.
29218 (arm_option_override): Set arm_arch5te.
29219 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
29221 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
29222 (arm_mcrr_qualifiers): ... this. New.
29223 (MRRC_QUALIFIERS): Define to...
29224 (arm_mrrc_qualifiers): ... this. New.
29225 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
29227 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
29228 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
29229 (MRRCI, mrrc, MRRC): New.
29230 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
29231 VUNSPEC_MRRC2): New.
29233 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29235 * config/arm/arm.md (<mcr>): New.
29237 * config/arm/arm.c (arm_coproc_builtin_available): Add
29238 support for mcr, mrc, mcr2 and mrc2.
29239 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
29240 (arm_mcr_qualifiers): ... this. New.
29241 (MRC_QUALIFIERS): Define to ...
29242 (arm_mrc_qualifiers): ... this. New.
29243 (MCR_QUALIFIERS): Define to ...
29244 (arm_mcr_qualifiers): ... this. New.
29245 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
29247 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
29248 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
29249 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
29250 VUNSPEC_MRC2): New.
29252 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29254 * config/arm/arm.md (*ldc): New.
29258 * config/arm/arm.c (arm_coproc_builtin_available): Add
29259 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
29260 (arm_coproc_ldc_stc_legitimate_address): New.
29261 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
29262 'qualifier_const_pointer'.
29263 (LDC_QUALIFIERS): Define to...
29264 (arm_ldc_qualifiers): ... this. New.
29265 (STC_QUALIFIERS): Define to...
29266 (arm_stc_qualifiers): ... this. New.
29267 * config/arm/arm-protos.h
29268 (arm_coproc_ldc_stc_legitimate_address): New.
29269 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
29270 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
29271 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
29272 stc2, stcl, stc2l): New.
29273 * config/arm/constraints.md (Uz): New.
29274 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
29275 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
29276 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
29277 VUNSPEC_STC2L): New.
29279 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29281 * config/arm/arm.md (<cdp>): New.
29282 * config/arm/arm.c (neon_const_bounds): Rename this ...
29283 (arm_const_bounds): ... this.
29284 (arm_coproc_builtin_available): New.
29285 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
29286 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
29287 (CDP_QUALIFIERS): Define to...
29288 (arm_cdp_qualifiers): ... this. New.
29290 (arm_expand_builtin_args): Add case for 6 arguments.
29291 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
29292 (arm_const_bounds): ... this.
29293 (arm_coproc_builtin_available): New.
29294 * config/arm/arm_acle.h (__arm_cdp): New.
29296 * config/arm/arm_acle_builtins.def (cdp): New.
29298 * config/arm/iterators.md (CDPI,CDP,cdp): New.
29299 * config/arm/neon.md: Rename all 'neon_const_bounds' to
29300 'arm_const_bounds'.
29301 * config/arm/types.md (coproc): New.
29302 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
29303 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
29304 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
29305 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
29307 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29309 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
29310 (UBINOP_QUALIFIERS): New.
29312 (acle_builtin_data): New. Change comment.
29313 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
29314 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
29315 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
29316 arm_acle_builtins.def.
29317 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
29318 (arm_init_acle_builtins): New.
29319 (CRC32_BUILTIN): Remove.
29320 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
29321 crc32cb, crc32ch and crc32cw.
29322 (arm_init_crc32_builtins): Remove.
29323 (arm_init_builtins): Use arm_init_acle_builtins rather
29324 than arm_init_crc32_builtins.
29325 (arm_expand_acle_builtin): New.
29326 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
29327 * config/arm/arm_acle_builtins.def: New.
29329 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29331 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
29332 (arm_builtin_datum): ... this.
29333 (arm_init_neon_builtin): Rename to ...
29334 (arm_init_builtin): ... this. Add a new parameters PREFIX
29335 and USE_SIG_IN_NAME.
29336 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
29337 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
29338 'arm_builtin_datum'.
29339 (arm_init_vfp_builtins): Likewise.
29340 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
29341 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
29342 (arm_expand_neon_args): Rename to ...
29343 (arm_expand_builtin_args): ... this. Rename builtin_arg
29344 enum values and differentiate between ARG_BUILTIN_MEMORY
29345 and ARG_BUILTIN_NEON_MEMORY.
29346 (arm_expand_neon_builtin_1): Rename to ...
29347 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
29348 values, arm_expand_builtin_args and add bool parameter NEON.
29349 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
29350 (arm_expand_vfp_builtin): Likewise.
29351 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
29353 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29355 PR middle-end/77484
29356 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
29357 * predict.c (tree_estimate_probability_bb): Reverse direction of
29358 polymorphic call predictor.
29360 2017-01-06 David Malcolm <dmalcolm@redhat.com>
29362 * passes.c (execute_one_pass): Split out pass-skipping logic into...
29363 (determine_pass_name_match): ...this new function and...
29364 (should_skip_pass_p): ...this new function.
29366 2017-01-06 Nathan Sidwell <nathan@acm.org>
29368 * ipa-visibility.c (function_and_variable_visibility): Reformat
29369 comments and long lines. Remove extrneous if.
29370 * symtab.c (symtab_node::make_decl_local): Fix code format.
29371 (symtab_node::set_section_for_node): Fix comment typo.
29373 2017-01-06 Martin Liska <mliska@suse.cz>
29376 * lra-constraints.c: Rename invariant to lra_invariant.
29377 * predict.c (set_even_probabilities): Initialize e to NULL.
29379 2017-01-05 Martin Sebor <msebor@redhat.com>
29381 PR tree-optimization/78910
29382 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
29383 (format_integer): Correct off-by-one error in the handling
29384 of precision with negative numbers in signed conversions..
29386 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
29388 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
29390 2017-01-05 Jakub Jelinek <jakub@redhat.com>
29392 PR tree-optimization/71016
29393 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
29394 factor_out_conditional_conversion. Formatting fix.
29395 (factor_out_conditional_conversion): Add cond_stmt argument.
29396 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
29397 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
29400 2017-01-05 David Malcolm <dmalcolm@redhat.com>
29402 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
29403 read-rtl-function.o, and selftest-rtl.o.
29404 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
29405 (selftest::aarch64_test_loading_full_dump): New function.
29406 (selftest::aarch64_run_selftests): New function.
29407 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
29408 selftest::aarch64_run_selftests.
29409 * config/i386/i386.c
29410 (selftest::ix86_test_loading_dump_fragment_1): New function.
29411 (selftest::ix86_test_loading_call_insn): New function.
29412 (selftest::ix86_test_loading_full_dump): New function.
29413 (selftest::ix86_test_loading_unspec): New function.
29414 (selftest::ix86_run_selftests): Call the new functions.
29415 * emit-rtl.c (maybe_set_max_label_num): New function.
29416 * emit-rtl.h (maybe_set_max_label_num): New decl.
29417 * function.c (instantiate_decls): Guard call to
29418 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
29419 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
29421 * gensupport.c (gen_reader::gen_reader): Pass "false"
29422 for new "compact" param of rtx_reader.
29423 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
29424 rather than an empty string for NULL strings.
29425 * read-md.c: Potentially include config.h rather than bconfig.h.
29426 Wrap include of errors.h with #ifdef GENERATOR_FILE.
29427 (have_error): New global, copied from errors.c.
29428 (md_reader::read_name): Rename to...
29429 (md_reader::read_name_1): ...this, adding "out_loc" param,
29430 and converting "missing name or number" to returning false, rather
29432 (md_reader::read_name): Reimplement in terms of read_name_1.
29433 (md_reader::read_name_or_nil): New function.
29434 (md_reader::read_string): Handle "(nil)" by returning NULL.
29435 (md_reader::md_reader): Add new param "compact".
29436 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
29437 (md_reader::read_file): New method.
29438 * read-md.h (md_reader::md_reader): Add new param "compact".
29439 (md_reader::read_file): New method.
29440 (md_reader::is_compact): New accessor.
29441 (md_reader::read_name): Convert return type from void to file_location.
29442 (md_reader::read_name_or_nil): New decl.
29443 (md_reader::read_name_1): New decl.
29444 (md_reader::m_compact): New field.
29445 (noop_reader::noop_reader): Pass "false" for new "compact" param
29447 (rtx_reader::rtx_reader): Add new "compact" param.
29448 (rtx_reader::read_rtx_operand): Make virtual and convert return
29449 type from void to rtx.
29450 (rtx_reader::read_until): New decl.
29451 (rtx_reader::handle_any_trailing_information): New virtual function.
29452 (rtx_reader::postprocess): New virtual function.
29453 (rtx_reader::finalize_string): New virtual function.
29454 (rtx_reader::m_in_call_function_usage): New field.
29455 (rtx_reader::m_reuse_rtx_by_id): New field.
29456 * read-rtl-function.c: New file.
29457 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
29458 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
29459 (selftest::verify_three_block_rtl_cfg): New decl.
29460 * read-rtl-function.h: New file.
29461 * read-rtl.c: Potentially include config.h rather than bconfig.h.
29462 For host, include function.h, memmodel.h, and emit-rtl.h.
29463 (one_time_initialization): New function.
29464 (struct compact_insn_name): New struct.
29465 (compact_insn_names): New array.
29466 (find_code): Handle insn codes in compact dumps.
29467 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
29468 (bind_subst_iter_and_attr): Likewise.
29469 (add_condition_to_string): Likewise.
29470 (add_condition_to_rtx): Likewise.
29471 (apply_attribute_uses): Likewise.
29472 (add_current_iterators): Likewise.
29473 (apply_iterators): Likewise.
29474 (initialize_iterators): Guard usage of apply_subst_iterator with
29475 #ifdef GENERATOR_FILE.
29476 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
29477 (md_reader::read_mapping): Likewise.
29478 (add_define_attr_for_define_subst): Likewise.
29479 (add_define_subst_attr): Likewise.
29480 (read_subst_mapping): Likewise.
29481 (check_code_iterator): Likewise.
29482 (rtx_reader::read_rtx): Likewise. Move one-time initialization
29484 (one_time_initialization): New function.
29485 (rtx_reader::read_until): New method.
29486 (read_flags): New function.
29487 (parse_reg_note_name): New function.
29488 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
29489 Handle reuse_rtx ids.
29490 Wrap iterator lookup within #ifdef GENERATOR_FILE.
29491 Add parsing support for RTL dumps, mirroring the special-cases in
29492 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
29493 values, and calling handle_any_trailing_information.
29494 (rtx_reader::read_rtx_operand): Convert return type from void
29495 to rtx, returning return_rtx. Handle case 'e'. Call
29496 finalize_string on XSTR and XTMPL fields.
29497 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
29498 "(nil)" values were omitted. Call the postprocess vfunc on the
29500 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
29501 class ctor. Initialize m_in_call_function_usage. Call
29502 one_time_initialization.
29503 * rtl-tests.c (selftest::test_uncond_jump): Call
29504 set_new_first_and_last_insn.
29505 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
29506 * selftest-rtl.c: New file.
29507 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
29508 (selftest::get_insn_by_uid): New decl.
29509 * selftest-run-tests.c (selftest::run_tests): Call
29510 read_rtl_function_c_tests.
29511 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
29512 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
29515 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
29517 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
29518 operands in a special way. Assert that pos+len <= mode precision.
29520 2017-01-05 Jakub Jelinek <jakub@redhat.com>
29522 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
29523 3 argument Alias with unlimited for the negative form.
29524 (fno-vect-cost-model): Removed.
29526 2017-01-05 Martin Liska <mliska@suse.cz>
29528 * hsa-gen.c (gen_hsa_divmod): New function.
29529 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
29531 2017-01-05 Martin Liska <mliska@suse.cz>
29534 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
29537 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29539 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
29540 small constant length operands.
29542 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29544 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
29545 between loop iterations.
29547 2017-01-05 Martin Liska <mliska@suse.cz>
29550 * gimplify.c (gimplify_decl_expr): Compare to
29551 asan_poisoned_variables instread of checking flags.
29552 (gimplify_target_expr): Likewise.
29553 (gimplify_expr): Likewise.
29554 (gimplify_function_tree): Conditionally initialize
29555 asan_poisoned_variables.
29557 2017-01-04 Jeff Law <law@redhat.com>
29559 PR tree-optimizatin/78812
29560 * rtl.h (contains_mem_rtx_p): Prototype.
29561 * ifcvt.c (containts_mem_rtx_p): Move from here to...
29562 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
29563 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
29564 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
29565 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
29567 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29569 * input.c (assert_char_at_range): Default-initialize actual_range.
29571 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29573 * df-scan.c (df_ref_create_structure): Make regno unsigned,
29574 to match the caller.
29576 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29578 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
29579 insns after final jump in test to emit dummy move.
29581 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29583 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
29584 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
29586 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29588 * multiple_target.c (create_dispatcher_calls): Init e_next.
29589 * tree-ssa-loop-split.c (split_loop): Init border.
29590 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
29593 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
29598 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
29599 (altivec_register_operand): Do not return true if the operand
29600 contains a SUBREG mixing SImode and SFmode.
29601 (vsx_register_operand): Likewise.
29602 (vsx_reg_sfsubreg_ok): New predicate.
29603 (vfloat_operand): Do not return true if the operand contains a
29604 SUBREG mixing SImode and SFmode.
29605 (vint_operand): Likewise.
29606 (vlogical_operand): Likewise.
29607 (gpc_reg_operand): Likewise.
29608 (int_reg_operand): Likewise.
29609 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
29610 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
29611 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
29613 (rs6000_emit_move_si_sf_subreg): New helper function.
29614 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
29615 fixup SUBREGs involving SImode and SFmode.
29616 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
29617 numbers for the new peephole2 optimization.
29618 (peephole2 for SFmode unions): New peephole2 to optimize cases in
29619 the GLIBC math library that do AND/IOR/XOR operations on single
29620 precision floating point.
29621 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
29622 target macros to say whether we need to avoid SUBREGs mixing
29624 (TARGET_ALLOW_SF_SUBREG): Likewise.
29625 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
29626 (UNSPEC_SI_FROM_SF): Likewise.
29627 (iorxor): Change spacing.
29628 (and_ior_xor): New iterator for AND, IOR, and XOR.
29629 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
29630 (movdi_from_sf_zero_ext): Likewise.
29631 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
29632 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
29633 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
29634 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
29635 (fms<mode>4): Likewise.
29636 (fnma<mode>4): Likewise.
29637 (fnms<mode>4): Likewise.
29638 (nfma<mode>4): Likewise.
29639 (nfms<mode>4): Likewise.
29641 2017-01-04 Marek Polacek <polacek@redhat.com>
29644 * doc/invoke.texi: Document -Wpointer-compare.
29646 2017-01-04 Jakub Jelinek <jakub@redhat.com>
29648 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
29651 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
29652 descriptions for -gdwarf-5 and emit them as uleb128 instead of
29655 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
29658 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
29659 documentation of the powerpc_popcntb_ok attribute.
29660 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29661 code to issue warning messages if a requested CPU configuration is
29662 not supported by the binary (assembler and loader) toolchain.
29663 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
29664 made to define a built-in function that has been disabled.
29665 (paired_init_builtins): Add assertion to prevent ICE if attempt is
29666 made to define a built-in function that has been disabled.
29667 (altivec_init_builtins): Add comment explaining why definition
29668 of the DST built-in functions is not preceded by an assertion
29669 check. Add assertions to prevent ICE if attempts are made to
29670 define an altivec predicate or an abs* built-in function that has
29672 (htm_init_builtins): Add comment explaining why definition of the
29673 htm built-in functions is not preceded by an assertion check.
29675 2017-01-04 Jeff Law <law@redhat.com>
29677 PR tree-optimizatin/67955
29678 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
29679 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
29680 the points-to solution does not include pt_null. Use DECL_PT_UID
29683 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
29685 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
29686 Use gen_int_mode instead of gen_lopwart for const_int operands.
29688 2017-01-04 Jakub Jelinek <jakub@redhat.com>
29690 PR tree-optimization/71563
29691 * match.pd: Simplify X << Y into X if Y is known to be 0 or
29692 out of range value - has low bits known to be zero.
29694 2017-01-04 Alan Modra <amodra@gmail.com>
29696 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
29697 * configure: Regenerate.
29698 * config.in: Regenerate.
29700 2017-01-04 Jakub Jelinek <jakub@redhat.com>
29703 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
29704 a substring of the message, but strcmp with the whole message. Ifdef
29705 ENABLE_NLS, translate the message first using dgettext.
29707 2017-01-03 Jeff Law <law@redhat.com>
29709 PR tree-optimizatin/78856
29710 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
29711 (mark_threaded_blocks): Remove code to truncate thread paths that
29712 cross multiple loop headers. Instead invalidate the cached loop
29713 iteration information and handle case of a thread path walking
29714 into an irreducible region.
29716 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
29719 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
29720 assertions. Add support for doing the signbit if the IEEE 128-bit
29721 floating point value is in a GPR.
29722 * config/rs6000/rs6000.md (Fsignbit): Delete.
29723 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
29724 Update the length attribute if the value is in a GPR.
29725 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
29726 the sign or zero extension instruction, since the value is always 0/1.
29727 (signbit<mode>2_dm2): Delete using <Fsignbit>.
29730 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
29731 extracting SImode to a GPR register so that we can generate a
29732 store, limit the vector to be in a traditional Altivec register
29733 for the vextuwrx instruction.
29735 2017-01-03 Ian Lance Taylor <iant@google.com>
29737 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
29739 2017-01-03 Martin Sebor <msebor@redhat.com>
29741 PR tree-optimization/78696
29742 * gimple-ssa-sprintf.c (format_floating): Correct handling of
29743 precision. Use MPFR for %f for greater fidelity. Correct handling
29745 (pass_sprintf_length::compute_format_length): Set width and precision
29746 specified by asrerisk to void_node for vararg functions.
29747 (try_substitute_return_value): Adjust dump output.
29749 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
29751 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
29753 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
29755 * doc/invoke.texi (SPARC options): Document -mlra as the default.
29756 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
29757 -mlra/-mno-lra was passed to the compiler.
29759 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
29761 PR rtl-optimization/65618
29762 * emit-rtl.c (try_split): Move initialization of "before" and
29763 "after" to just before the call to emit_insn_after_setloc.
29765 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
29767 * doc/md.texi (Standard Names): Remove reference to Java frontend.
29769 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
29771 * dwarf2out.c (gen_enumeration_type_die): When
29772 -gno-strict-dwarf, add a DW_AT_encoding attribute.
29774 2017-01-03 Jakub Jelinek <jakub@redhat.com>
29776 PR tree-optimization/78965
29777 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
29778 Change first argument from const call_info & to call_info &. For %n
29779 set info.nowrite to false.
29781 PR middle-end/78901
29782 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
29783 possibly throwing calls.
29785 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
29786 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
29787 and fns handling, rather than in a separate case SSA_NAME.
29789 2017-01-02 Jeff Law <law@redhat.com>
29791 * config/darwin-driver.c (darwin_driver_init): Const-correctness
29792 fixes for first_period and second_period variables.
29794 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
29797 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
29798 (*insvqi_1): New insn pattern.
29799 (*insvqi_1_mem_rex64): Ditto.
29800 (*insvqi_2): Ditto.
29801 (*insvqi_3): Rename from *insvqi.
29803 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
29805 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
29807 * doc/cfg.texi (Edges): Remove reference to Java.
29808 (Maintaining the CFG): Ditto.
29810 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29812 PR middle-end/77674
29813 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
29814 transparent aliases.
29816 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29818 PR middle-end/77484
29819 * predict.def (PRED_CALL): Update hitrate.
29820 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
29821 * predict.c (tree_estimate_probability_bb): Split CALL predictor
29822 into direct/indirect/polymorphic variants.
29824 2017-01-01 Jakub Jelinek <jakub@redhat.com>
29826 Update copyright years.
29828 * gcc.c (process_command): Update copyright notice dates.
29829 * gcov-dump.c (print_version): Ditto.
29830 * gcov.c (print_version): Ditto.
29831 * gcov-tool.c (print_version): Ditto.
29832 * gengtype.c (create_file): Ditto.
29833 * doc/cpp.texi: Bump @copying's copyright year.
29834 * doc/cppinternals.texi: Ditto.
29835 * doc/gcc.texi: Ditto.
29836 * doc/gccint.texi: Ditto.
29837 * doc/gcov.texi: Ditto.
29838 * doc/install.texi: Ditto.
29839 * doc/invoke.texi: Ditto.
29841 Copyright (C) 2017 Free Software Foundation, Inc.
29843 Copying and distribution of this file, with or without modification,
29844 are permitted in any medium without royalty provided the copyright
29845 notice and this notice are preserved.