1 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
3 * x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
5 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
7 * config/alpha/alpha.c (alpha_split_conditional_move):
8 Use std::swap instead of manually swapping.
9 (alpha_stdarg_optimize_hook): Ditto.
10 (alpha_canonicalize_comparison): Ditto.
12 2017-10-12 Bin Cheng <bin.cheng@arm.com>
14 * tree-loop-distribution.c (struct builtin_info): New struct.
15 (struct partition): Refactor fields into struct builtin_info.
16 (partition_free): Free struct builtin_info.
17 (build_size_arg_loc, build_addr_arg_loc): Delete.
18 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
19 information from struct builtin_info.
20 (find_single_drs): New function refactored from classify_partition.
21 Also moved builtin validity checks to this function.
22 (compute_access_range, alloc_builtin): New functions.
23 (classify_builtin_st, classify_builtin_ldst): New functions.
24 (classify_partition): Refactor code into functions find_single_drs,
25 classify_builtin_st and classify_builtin_ldst.
26 (distribute_loop): Don't do runtime alias check when distributing
28 (find_seed_stmts_for_distribution): New function.
29 (pass_loop_distribution::execute): Refactor code finding seed
30 stmts into above function. Support distribution for the innermost
31 two-level loop nest. Adjust dump information.
33 2017-10-12 Bin Cheng <bin.cheng@arm.com>
35 * tree-loop-distribution.c: Adjust the general comment.
36 (NUM_PARTITION_THRESHOLD): New macro.
37 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
38 (classify_partition): Skip builtin pattern of loop nest's inner loop.
39 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
40 in call to build_partition_graph.
41 (finalize_partitions): New parameter. Make loop distribution more
42 conservative by fusing more partitions.
43 (distribute_loop): Don't do runtime alias check in case of loop nest
45 (find_seed_stmts_for_distribution): New function.
46 (prepare_perfect_loop_nest): New function.
47 (pass_loop_distribution::execute): Refactor code finding seed stmts
48 and loop nest into above functions. Support loop nest distribution.
49 Adjust dump information accordingly.
51 2017-10-12 Bin Cheng <bin.cheng@arm.com>
53 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
54 and set PTYPE_SEQUENTIAL for merged partition.
56 2017-10-12 Richard Biener <rguenther@suse.de>
58 PR tree-optimization/69728
60 2017-09-19 Richard Biener <rguenther@suse.de>
62 PR tree-optimization/69728
63 * graphite-sese-to-poly.c (schedule_error): New global.
64 (add_loop_schedule): Handle empty domain by failing the
66 (build_original_schedule): Handle schedule_error.
68 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
69 domain by returning an unchanged schedule.
71 2017-10-12 Jakub Jelinek <jakub@redhat.com>
73 * genrecog.c (validate_pattern): For VEC_SELECT verify that
74 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
76 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
78 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
81 2017-10-12 Jakub Jelinek <jakub@redhat.com>
84 * expr.c (store_field): Don't optimize away bitsize == 0 store
85 from CALL_EXPR with addressable return type.
87 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
89 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
90 * config/rs6000/rs6000.md (sel): Delete mode attribute.
91 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
92 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
93 TARGET_ISEL instead of TARGET_ISEL<sel>.
95 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
97 * config/rs6000/rs6000.c
98 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
100 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
102 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
104 (reg_or_cint_operand): Fix comment.
105 (reg_or_zero_operand): New predicate.
106 * config/rs6000/rs6000-protos.h (output_isel): Delete.
107 * config/rs6000/rs6000.c (output_isel): Delete.
108 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
109 instead of reg_or_cint_operand. Output instruction directly (not via
111 (isel_unsigned_<mode>): Ditto.
112 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
113 gpc_reg_operand. Add an instruction alternative for this. Output
114 instruction directly.
115 (*isel_reversed_unsigned_<mode>): Ditto.
117 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
119 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
120 (TARGET_CANONICALIZE_COMPARISON): Define.
122 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
125 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
126 Check whether the dest is REG before adding REG_EQUIV note.
128 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
131 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
133 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
134 (make_hard_regno_born, make_hard_regno_dead): Update
135 bb_killed_pseudos and bb_gen_pseudos.
137 2017-10-11 Nathan Sidwell <nathan@acm.org>
139 * incpath.h (enum incpath_kind): Name enum, prefix values.
140 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
141 * incpath.c (heads, tails): Use INC_MAX.
142 (add_env_var_paths, add_standard_paths): Use incpath_kind.
143 (merge_include_chains, split_quote_chain,
144 register_include_chains): Update incpath_kind names.
145 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
146 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
148 (add_framework_path, darwin_register_objc_includes): Likewise.
149 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
151 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
153 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
154 Do not use float_operator operator predicate.
155 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
156 * config/i386/predicates.md (float_operator): Remove predicate.
158 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
160 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
161 (*jccxf_i387): Ditto.
162 (*jcc<mode>_i387): Ditto.
163 (*jccu<mode>_i387): Ditto.
164 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
165 (*jcc_*_i387 splitters): Remove.
166 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
167 * config/i386/i386.c (ix86_split_fp_branch): Remove.
168 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
171 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
173 * profile-count.h (slow_safe_scale_64bit): New function.
174 (safe_scale_64bit): New inline.
175 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
176 * profile-count.c: Include wide-int.h
177 (slow_safe_scale_64bit): New.
179 2017-10-11 Nathan Sidwell <nathan@acm.org>
181 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
182 HAS_DECL_ASSEMBLER_NAME_P.
183 * gimple-expr.c (gimple_decl_printable_name: Check
184 HAS_DECL_ASSEMBLER_NAME_P too.
185 * ipa-utils.h (type_in_anonymous_namespace_p): Check
186 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
187 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
188 * passes.c (rest_of_decl_compilation): Check
189 HAS_DECL_ASSEMBLER_NAME_P too.
190 * recog.c (verify_changes): Likewise.
191 * tree-pretty-print.c (dump_decl_name): Likewise.
192 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
194 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
195 (SET_DECL_ASSEMBLER_NAME): Use it.
196 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
197 (COPY_DECL_ASSEMBLER_NAME): Likewise.
198 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
200 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
202 * config.gcc (i386, x86_64): Add extra objects.
203 * i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
204 (ix86_min_insn_size): Declare.
205 (ix86_issue_rate): Declare.
206 (ix86_adjust_cost): Declare.
207 (ia32_multipass_dfa_lookahead): Declare.
208 (ix86_macro_fusion_p): Declare.
209 (ix86_macro_fusion_pair_p): Declare.
210 (ix86_bd_has_dispatch): Declare.
211 (ix86_bd_do_dispatch): Declare.
212 (ix86_core2i7_init_hooks): Declare.
213 (ix86_atom_sched_reorder): Declare.
214 * i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
215 (COSTS_N_BYTES): Move to x86-tune-costs.h.
216 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
217 (rip_relative_addr_p): Rename to ...
218 (ix86_rip_relative_addr_p): ... this one; export.
219 (memory_address_length): Update.
220 (ix86_issue_rate): Move to x86-tune-sched.c.
221 (ix86_flags_dependent): Move to x86-tune-sched.c.
222 (ix86_agi_dependent): Move to x86-tune-sched.c.
223 (exact_dependency_1): Move to x86-tune-sched.c.
224 (exact_store_load_dependency): Move to x86-tune-sched.c.
225 (ix86_adjust_cost): Move to x86-tune-sched.c.
226 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
227 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
228 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
229 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
230 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
231 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
232 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
233 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
234 (min_insn_size): Rename to ...
235 (ix86_min_insn_size): ... this one; export.
236 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
237 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
238 (core2i7_first_cycle_multipass_backtrack): Move to
239 x86-tune-sched-core.c.
240 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
241 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
242 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
243 (ix86_avoid_jump_mispredicts): Update.
244 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
245 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
246 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
247 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
248 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
249 (MAX_INSN): Move to ix86-tune-sched-bd.c.
250 (MAX_IMM): Move to ix86-tune-sched-bd.c.
251 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
252 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
253 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
254 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
255 (MAX_STORE): Move to ix86-tune-sched-bd.c.
256 (BIG): Move to ix86-tune-sched-bd.c.
257 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
258 (enum insn_path): Move to ix86-tune-sched-bd.c.
259 (get_mem_group): Move to ix86-tune-sched-bd.c.
260 (is_cmp): Move to ix86-tune-sched-bd.c.
261 (dispatch_violation): Move to ix86-tune-sched-bd.c.
262 (is_branch): Move to ix86-tune-sched-bd.c.
263 (is_prefetch): Move to ix86-tune-sched-bd.c.
264 (init_window): Move to ix86-tune-sched-bd.c.
265 (allocate_window): Move to ix86-tune-sched-bd.c.
266 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
267 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
268 (process_end_window): Move to ix86-tune-sched-bd.c.
269 (allocate_next_window): Move to ix86-tune-sched-bd.c.
270 (find_constant): Move to ix86-tune-sched-bd.c.
271 (get_num_immediates): Move to ix86-tune-sched-bd.c.
272 (has_immediate): Move to ix86-tune-sched-bd.c.
273 (get_insn_path): Move to ix86-tune-sched-bd.c.
274 (get_insn_group): Move to ix86-tune-sched-bd.c.
275 (count_num_restricted): Move to ix86-tune-sched-bd.c.
276 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
277 (add_insn_window): Move to ix86-tune-sched-bd.c.
278 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
279 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
280 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
281 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
282 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
283 (do_dispatch): Move to ix86-tune-sched-bd.c.
284 (has_dispatch): Move to ix86-tune-sched-bd.c.
285 * i386/t-i386: Add new object files.
286 * i386/x86-tune-costs.h: New file.
287 * i386/x86-tune-sched-atom.c: New file.
288 * i386/x86-tune-sched-bd.c: New file.
289 * i386/x86-tune-sched-core.c: New file.
290 * i386/x86-tune-sched.c: New file.
292 2017-10-11 Liu Hao <lh_mouse@126.com>
294 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
295 the generic version below instead.
296 (should_colorize): Recognize Windows consoles as terminals
298 * pretty-print.c [__MINGW32__] (write_all): New function.
299 [__MINGW32__] (find_esc_head): Likewise.
300 [__MINGW32__] (find_esc_terminator): Likewise.
301 [__MINGW32__] (eat_esc_sequence): Likewise.
302 [__MINGW32__] (mingw_ansi_fputs): New function that handles
304 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
307 2017-10-11 Richard Biener <rguenther@suse.de>
309 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
310 Properly call analyze_scalar_evolution with the loop of the stmt.
312 2017-10-11 Richard Biener <rguenther@suse.de>
314 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
315 * tree-core.h (tree_base): Add chrec_var union member.
316 * tree.h (CHREC_VAR): Remove.
317 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
318 * tree-chrec.h (build_polynomial_chrec): Adjust.
319 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
320 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
322 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
324 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
325 * match.pd: ... here.
326 ((T) X == (T) Y): Relax condition.
328 2017-10-11 Bin Cheng <bin.cheng@arm.com>
330 PR tree-optimization/82472
331 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
333 (break_alias_scc_partitions): Update postorder number.
335 2017-10-11 Martin Liska <mliska@suse.cz>
338 * opts.c (parse_no_sanitize_attribute): Do not use error_value
340 * opts.h (parse_no_sanitize_attribute): Remove last argument.
342 2017-10-11 Martin Liska <mliska@suse.cz>
344 * print-rtl.c (print_insn): Move declaration of idbuf
345 to same scope as name.
347 2017-10-11 Martin Liska <mliska@suse.cz>
352 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
355 2017-10-11 Martin Liska <mliska@suse.cz>
358 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
361 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
363 * config.gcc (default_gnu_indirect_function): Default to yes for
364 arm*-*-linux* with glibc.
366 2017-10-11 Richard Biener <rguenther@suse.de>
368 * tree-scalar-evolution.c (get_scalar_evolution): Handle
369 default-defs and types we do not want to analyze.
370 (interpret_loop_phi): Replace unreachable code with an assert.
371 (compute_scalar_evolution_in_loop): Remove and inline ...
372 (analyze_scalar_evolution_1): ... here, replacing condition with
373 what makes the intent clearer. Remove handling of cases
374 get_scalar_evolution now handles.
376 2017-10-10 Jim Wilson <wilson@tuliptree.org>
378 PR rtl-optimization/81434
379 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
380 comment for main loop. In sched_group_found if, also add checks for
381 pass and min_cost_group.
383 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
385 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
386 (rs6000_insn_cost): New function.
387 * config/rs6000/rs6000.md (cost): New attribute.
389 2017-10-10 Jakub Jelinek <jakub@redhat.com>
390 H.J. Lu <hongjiu.lu@intel.com>
394 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
395 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
396 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
397 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
398 to OPTION_MASK_ISA_AVX512VL - builtins that have both
399 OPTION_MASK_ISA_MMX and some other bit set require both
400 mmx and the ISAs without the mmx bit.
401 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
402 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
403 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
404 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
405 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
406 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
407 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
408 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
409 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
410 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
411 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
412 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
413 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
414 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
415 Add OPTION_MASK_ISA_MMX.
417 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
419 * config.gcc (armv7*-*-freebsd*): New target.
420 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
422 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
424 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
425 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
428 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
430 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
431 a template parameter.
432 (WIDE_INT_REF_FOR): Update accordingly.
433 * tree.h (wi::int_traits <const_tree>): Delete.
434 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
435 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
436 (wi::tree_to_wide_ref): New typedef.
437 (wi::to_wide): New function.
438 * calls.c (get_size_range): Use wi::to_wide when operating on
440 * cgraph.c (cgraph_node::create_thunk): Likewise.
441 * config/i386/i386.c (ix86_data_alignment): Likewise.
442 (ix86_local_alignment): Likewise.
443 * dbxout.c (stabstr_O): Likewise.
444 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
445 * expr.c (const_vector_from_tree): Likewise.
446 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
447 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
448 (fold_negate_expr_1, int_const_binop_1, const_binop)
449 (fold_convert_const_int_from_real, optimize_bit_field_compare)
450 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
451 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
452 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
453 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
454 (fold_not_const, round_up_loc): Likewise.
455 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
456 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
457 (alloca_call_type): Likewise.
458 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
459 * godump.c (go_output_typedef): Likewise.
460 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
461 * internal-fn.c (get_min_precision): Likewise.
462 * ipa-cp.c (ipcp_store_vr_results): Likewise.
463 * ipa-polymorphic-call.c
464 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
465 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
466 (ipa_modify_call_arguments): Likewise.
467 * match.pd: Likewise.
468 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
469 * print-tree.c (print_node_brief, print_node): Likewise.
470 * stmt.c (expand_case): Likewise.
471 * stor-layout.c (layout_type): Likewise.
472 * tree-affine.c (tree_to_aff_combination): Likewise.
473 * tree-cfg.c (group_case_labels_stmt): Likewise.
474 * tree-data-ref.c (dr_analyze_indices): Likewise.
475 (prune_runtime_alias_test_list): Likewise.
476 * tree-dump.c (dequeue_and_dump): Likewise.
477 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
478 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
479 * tree-pretty-print.c (dump_generic_node): Likewise.
480 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
481 (simple_iv_with_niters): Likewise.
482 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
483 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
484 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
485 * tree-ssa-loop-niter.c (split_to_var_and_offset)
486 (refine_value_range_using_guard, number_of_iterations_ne_max)
487 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
488 (get_cst_init_from_scev, record_nonwrapping_iv)
489 (scev_var_range_cant_overflow): Likewise.
490 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
491 * tree-ssa-pre.c (compute_avail): Likewise.
492 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
493 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
494 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
495 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
496 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
497 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
498 (dump_case_nodes, try_switch_expansion): Likewise.
499 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
500 (vect_do_peeling): Likewise.
501 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
502 * tree-vect-stmts.c (vectorizable_load): Likewise.
503 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
504 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
505 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
506 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
507 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
508 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
509 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
510 (evrp_dom_walker::before_dom_children): Likewise.
511 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
512 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
513 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
514 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
515 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
516 (get_range_pos_neg): Likewise.
517 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
518 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
519 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
520 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
521 * config/avr/avr.c (avr_fold_builtin): Likewise.
522 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
523 * config/msp430/msp430.c (msp430_attr): Likewise.
524 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
525 * config/powerpcspe/powerpcspe-c.c
526 (altivec_resolve_overloaded_builtin): Likewise.
527 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
528 (rs6000_expand_ternop_builtin): Likewise.
529 * config/rs6000/rs6000-c.c
530 (altivec_resolve_overloaded_builtin): Likewise.
531 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
532 (rs6000_expand_ternop_builtin): Likewise.
533 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
535 2017-10-10 Bin Cheng <bin.cheng@arm.com>
537 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
538 when copying loop nest with only one inner loop.
540 2017-10-10 Richard Biener <rguenther@suse.de>
542 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
543 blocks if SCEV is active.
544 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
546 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
547 (scev_initialize): Assert we are not yet initialized.
549 2017-10-10 Bin Cheng <bin.cheng@arm.com>
551 * tree-loop-distribution.c (generate_loops_for_partition): Remove
552 inner loop's exit stmt by making it always exit the loop, otherwise
553 we would generate an infinite empty loop.
555 2017-10-10 Bin Cheng <bin.cheng@arm.com>
557 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
558 renaming variables in new preheader if it's deleted.
560 2017-10-10 Bin Cheng <bin.cheng@arm.com>
562 * tree-loop-distribution.c (struct partition): Remove unused field
563 loops of the structure.
564 (partition_alloc, partition_free): Ditto.
565 (build_rdg_partition_for_vertex): Ditto.
567 2017-10-09 Jeff Law <law@redhat.com>
569 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
570 return type to match prototype and documentation.
572 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
574 * config/rs6000/rs6000.c (processor_costs): Move to ...
575 * config/rs6000/rs6000.h: ... here.
576 (rs6000_cost): Declare.
578 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
580 * except.c (setjmp_fn): New global variable.
581 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
582 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
583 if DONT_USE_BUILTIN_SETJMP is defined.
585 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
587 * target.def (insn_cost): New hook.
588 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
589 * doc/tm.texi: Regenerate.
590 * rtlanal.c (insn_cost): Use the new hook.
592 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
594 * combine.c (combine_validate_cost): Compute the new insn_cost,
595 not just pattern_cost.
596 (try_combine): Adjust comment.
598 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
600 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
602 * combine.c (uid_insn_cost): Adjust comment.
603 (combine_validate_cost): Adjust comment. Use pattern_cost instead
605 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
606 * dse.c (find_shift_sequence): Ditto.
607 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
608 (bb_valid_for_noce_process_p): Use pattern_cost.
609 * rtl.h (insn_rtx_cost): Delete.
610 (pattern_cost): New prototype.
611 (insn_cost): New prototype.
612 * rtlanal.c (insn_rtx_cost): Rename to...
613 (pattern_cost): ... this.
616 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
618 * config/i386/i386.md (*jcc_2): Remove insn pattern.
619 (*jcc<mode>_0_r_i387): Ditto.
620 (*jccxf_r_i387): Ditto.
621 (*jcc<mode>_r_i387): Ditto.
622 (*jccu<mode>_r_i387): Ditto.
623 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
624 (*jcc): Rename from *jcc_1.
626 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
628 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
629 deferred rescans after the lvx/stvx recombination pre-pass.
631 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
633 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
634 memory operation instruction support.
635 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
636 (rs6000-ibm-aix[789]*): Likewise.
637 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
638 Document new functions.
640 2017-10-09 Richard Biener <rguenther@suse.de>
642 PR tree-optimization/82397
643 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
644 equality only for semantically equal trees.
646 2017-10-09 Richard Biener <rguenther@suse.de>
648 PR tree-optimization/82449
649 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
650 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
651 allow constant addresses.
652 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
655 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
657 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
660 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
663 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
666 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
669 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
671 2017-10-09 Jakub Jelinek <jakub@redhat.com>
674 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
675 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
677 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
679 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
680 (WI_BINARY_PREDICATE_RESULT): Likewise.
681 (wi::binary_traits::operator_result): New type.
682 (wi::binary_traits::predicate_result): Likewise.
683 (generic_wide_int::operator~, unary generic_wide_int::operator-)
684 (generic_wide_int::operator==, generic_wide_int::operator!=)
685 (generic_wide_int::operator&, generic_wide_int::and_not)
686 (generic_wide_int::operator|, generic_wide_int::or_not)
687 (generic_wide_int::operator^, generic_wide_int::operator+
688 (binary generic_wide_int::operator-, generic_wide_int::operator*):
690 (operator~, unary operator-, operator==, operator!=, operator&)
691 (operator|, operator^, operator+, binary operator-, operator*): New
693 * expr.c (get_inner_reference): Use wi::bit_and_not.
694 * fold-const.c (fold_binary_loc): Likewise.
695 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
696 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
697 (bit_value_binop): Likewise.
698 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
699 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
700 (extract_range_from_binary_expr_1): Likewise.
701 (masked_increment): Likewise.
702 (simplify_bit_ops_using_ranges): Likewise.
704 2017-10-09 Martin Jambor <mjambor@suse.cz>
707 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
708 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
709 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
711 (hsa_fixup_mov_insn_type): New function.
712 (hsa_op_with_type::get_in_type): Use it.
713 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
714 immediates in an assert.
715 (hsa_op_with_type::extend_int_to_32bit): New method.
716 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
717 types. Convert to dest type if necessary.
718 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
719 (reg_for_gimple_ssa): Pass false as min32int to
720 hsa_type_for_scalar_tree_type.
721 (gen_hsa_addr): Fixup type when creating addresable temporary.
722 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
723 (gen_hsa_unary_operation): Extend operands and convert to dest type if
724 necessary. Call hsa_fixup_mov_insn_type.
725 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
726 extend operands and convert to dest type if necessary.
727 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
728 to dest type if necessary.
729 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
730 if conversion nt necessary and size matches.
731 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
732 to dest type if necessary.
733 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
734 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
736 (gen_hsa_clrsb): Likewise.
737 (gen_hsa_ffs): Likewise.
738 (gen_hsa_divmod): Extend operands and convert to dest type if
740 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
742 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
744 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
745 Remove empty default arguments. Use a brace block as output
747 (conditional return): Ditto.
749 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
750 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
752 (group_ending_nop): Ditto.
754 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
755 (splitters for those): Ditto.
757 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
759 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
760 a conditional jump (and the compare for it) so that pc_rtx is the
762 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
763 for the deleted and renamed ctr<mode>_internal[234] patterns.
764 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
765 Delete second conditional return pattern.
766 (ctr<mode>_internal2): Delete this second bdnz pattern.
767 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
768 (ctr<mode>_internal4): Delete this second bdz pattern.
770 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
772 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
773 (always_initialized_rtx_for_ssa_name_p): New predicate.
774 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
775 (finish_out_of_ssa): Free new field of SA.
776 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
777 * tree-ssa-coalesce.c: Include tree-ssa.h.
778 (get_parm_default_def_partitions): Remove extern keyword.
779 (get_undefined_value_partitions): New function.
780 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
781 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
783 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
785 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
787 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
789 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
791 * i386.c (ix86_expand_set_or_movmem): Disable 512bit loops for targets
794 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
796 * i386.c (has_dispatch): Disable for Ryzen.
798 2017-10-08 Olivier Hainque <hainque@adacore.com>
800 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
801 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
803 (thumb_set_return_address): Likewise.
805 2017-10-08 Olivier Hainque <hainque@adacore.com>
807 * common/config/arm/arm-common.c (arm_except_unwind_info):
808 Handle DWARF2_UNWIND_INFO.
810 2017-10-07 Michael Collison <michael.collison@arm.com>
812 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
815 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
817 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
818 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
819 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
820 defined, force the creation of a new block for a dispatch label.
822 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
824 * invoke.texi (Wsuggest-attribute=cold): Document.
825 * common.opt (Wsuggest-attribute=cold): New
826 * ipa-pure-const.c (warn_function_cold): New function.
827 * predict.c (compute_function_frequency): Use it.
828 * predict.h (warn_function_cold): Declare.
830 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
832 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
835 2017-10-06 Martin Liska <mliska@suse.cz>
837 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
838 keyword for member functions.
839 (struct sanopt_tree_couple): New struct.
840 (struct sanopt_tree_couple_hash): New function.
841 (struct sanopt_ctx): Add new hash_map.
842 (has_dominating_ubsan_ptr_check): New function.
843 (record_ubsan_ptr_check_stmt): Likewise.
844 (maybe_optimize_ubsan_ptr_ifn): Likewise.
845 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
846 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
848 2017-10-06 Sudakshina Das <sudi.das@arm.com>
851 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
852 aarch64_simd_valid_immediate on CONST_VECTORs.
853 (aarch64_reg_or_bic_imm): Likewise.
855 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
857 PR rtl-optimization/82396
858 * haifa-sched.c (ready_sort_real): Disable qsort checking.
860 2017-10-06 Sebastian Pop <sebpop@gmail.com>
862 * graphite-dependences.c (scop_get_reads): Move code to...
863 (scop_get_must_writes): Move code to...
864 (scop_get_may_writes): Move code to...
865 (scop_get_reads_and_writes): ... here.
866 (scop_get_dependences): Call scop_get_reads_and_writes.
868 2017-10-06 Jakub Jelinek <jakub@redhat.com>
870 PR tree-optimization/82434
871 * fold-const.h (can_native_encode_type_p,
872 can_native_encode_string_p): Remove.
873 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
874 don't encode anything, just return what would be otherwise returned.
875 (native_encode_fixed, native_encode_complex, native_encode_vector):
877 (native_encode_string): Likewise. Inline by hand
878 can_native_encode_string_p.
879 (can_native_encode_type_p): Remove.
880 (can_native_encode_string_p): Remove.
881 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
882 STRING_CSTs using can_native_encode_string_p, test all
883 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
884 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
885 argument from native_encode_expr.
886 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
887 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
890 2017-10-06 Richard Biener <rguenther@suse.de>
892 PR tree-optimization/82397
893 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
894 operand_equal_p but rely on data_ref_compare_tree for detecting
896 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
897 to match up with dr_group_sort_cmp.
899 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
902 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
904 * config/s390/s390-builtin-types.def: Regenerate.
906 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
909 * config/s390/s390-builtin-types.def: Regenerate.
910 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
911 Change flag from B_VXE to B_VX.
912 (s390_vec_min_dbl): Remove B_VXE flag.
914 2017-10-06 Richard Biener <rguenther@suse.de>
916 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
917 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
918 translate_isl_ast_to_gimple::is_valid_rename,
919 translate_isl_ast_to_gimple::get_rename,
920 translate_isl_ast_to_gimple::get_def_bb_for_const,
921 translate_isl_ast_to_gimple::get_new_name,
922 translate_isl_ast_to_gimple::collect_all_ssa_names,
923 translate_isl_ast_to_gimple::copy_loop_phi_args,
924 translate_isl_ast_to_gimple::collect_all_ssa_names,
925 translate_isl_ast_to_gimple::copy_loop_phi_args,
926 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
927 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
928 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
929 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
930 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
931 translate_isl_ast_to_gimple::copy_cond_phi_args,
932 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
933 translate_isl_ast_to_gimple::edge_for_new_close_phis,
934 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
935 translate_isl_ast_to_gimple::rename_uses,
936 translate_isl_ast_to_gimple::rename_all_uses): Remove.
937 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
938 (set_rename_for_each_def): Likewise.
939 (graphite_copy_stmts_from_block): Handle debug stmt resetting
940 here. Handle rewriting SCEV analyzable uses here.
941 (copy_bb_and_scalar_dependences): Generate code for PHI
943 (graphite_regenerate_ast_isl): Adjust.
944 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
945 (add_write, add_read): New functions.
946 (build_cross_bb_scalars_def): Use it and simplify.
947 (build_cross_bb_scalars_use): Likewise.
948 (graphite_find_cross_bb_scalar_vars): Inline into...
949 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
950 simulating out-of-SSA. Compute liveout and add dependencies.
951 (build_scops): Force an empty entry block.
952 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
954 (sese_build_liveouts): Declare.
955 (sese_trivially_empty_bb_p): Likewise.
956 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
957 compute liveout and debug_liveout.
958 (sese_bad_liveouts_use): Remove.
959 (sese_reset_debug_liveouts_bb): Likewise.
960 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
961 (sese_build_liveouts): Build liveout and debug_liveout and store
963 (new_sese_info): Adjust.
964 (free_sese_info): Likewise.
965 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
966 do not build liveout here.
967 (move_sese_in_condition): Adjust region entry.
968 (scev_analyzable_p): Match up with chrec_apply requirements.
969 (sese_trivially_empty_bb_p): New.
970 * tree-into-ssa.c (get_reaching_def): Properly support generating
971 default-defs for incremental rewrite of anonymous names.
973 2017-10-06 Richard Biener <rguenther@suse.de>
975 * graphite-sese-to-poly.c (extract_affine): For casts increasing
976 precision do not perform modulo reduction.
978 2017-10-06 Richard Biener <rguenther@suse.de>
980 PR tree-optimization/82436
981 * tree-vect-slp.c (vect_supported_load_permutation_p): More
982 conservatively choose the vectorization factor when checking
983 whether we can perform the required load permutation.
984 (vect_transform_slp_perm_load): Assert when we may not fail.
986 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
988 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
989 message for incompatible -msdata=* and -mcall-* options.
991 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
993 * i386.c (ia32_multipass_dfa_lookahead): Default to issue rate
994 for post-reload scheduling.
996 2017-10-05 Tamar Christina <tamar.christina@arm.com>
998 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
1000 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
1002 * i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
1003 to improve monte carlo in scimark.
1005 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
1007 * i386.c (ix86_size_cost, i386_cost, i486_cost, pentium_cost,
1008 lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
1009 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
1010 pentium4_cost, nocona_cost): Set reassociation width to 1.
1011 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
1012 width to 2 for fp operations and 1 otherwise.
1013 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
1015 (atom_cost): Set reassociation width to 2.
1016 (slm_cost, generic_cost): Set fp reassociation width
1017 to 2 and 1 otherwise.
1018 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
1019 (core_cost): Set fp reassociation width to 4 and vector to 2.
1020 (ix86_reassociation_width): Rewrite using cost table; special case
1021 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
1022 and TARGET_AVX128_OPTIMAL.
1023 * i386.h (processor_costs): Add
1024 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
1025 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
1026 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
1027 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
1028 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
1029 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
1031 2017-10-05 Nathan Sidwell <nathan@acm.org>
1033 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
1035 2017-10-05 Tamar Christina <tamar.christina@arm.com>
1037 * config/arm/arm.c (arm_test_fpu_data): New.
1038 (arm_run_selftests): Call arm_test_fpu_data.
1040 2017-10-04 Nathan Sidwell <nathan@acm.org>
1042 * toplev.c (toplev::main): Remove excess parens on pretty_printer
1044 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
1046 2017-10-04 Sudakshina Das <sudi.das@arm.com>
1048 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
1049 check type for aarch64_simd_valid_immediate.
1050 (aarch64_output_simd_mov_immediate): Update prototype.
1051 (aarch64_simd_valid_immediate): Update prototype.
1052 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
1053 support for ORR-immediate.
1054 (and<mode>3): modified pattern to add support for BIC-immediate.
1055 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
1056 now checks for valid immediate for BIC and ORR based on new enum
1058 (aarch64_output_simd_mov_immediate): Function now used to output
1059 BIC/ORR imm as well based on new enum argument.
1060 * config/aarch64/constraints.md (Do): New vector immediate constraint.
1062 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
1063 (aarch64_reg_or_bic_imm): Likewise.
1065 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1067 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
1068 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
1070 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
1074 PR rtl-optimization/82396
1075 * haifa-sched.c (autopref_multipass_init): Simplify
1077 (autopref_rank_data): Simplify sort order.
1078 * sched-int.h (autopref_multipass_data_): Remove
1079 multi_mem_insn_p, min_offset and max_offset.
1081 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1083 * doc/sourcebuild.texi: Document vect_peeling_profitable.
1085 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1087 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
1090 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1092 * doc/sourcebuild.texi: Document vect_long_mult.
1094 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
1096 PR tree-optimization/82413
1097 * fold-const.c (build_range_check): Use widest_int when comparing
1098 the maximum ETYPE value with HIGH.
1100 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
1102 PR rtl-optimization/82396
1103 * haifa-sched.c (autopref_multipass_init): Simplify
1105 (autopref_rank_data): Simplify sort order.
1106 * sched-int.h (autopref_multipass_data_): Remove
1107 multi_mem_insn_p, min_offset and max_offset.
1109 2017-10-04 Jakub Jelinek <jakub@redhat.com>
1111 PR tree-optimization/82381
1112 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
1113 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
1116 PR tree-optimization/82374
1117 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
1118 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
1119 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
1120 current_function_decl to the new decl.
1122 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
1124 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
1125 helper macro for IEEE float128 hardware built-in functions.
1126 (SQRTF128_ODD): Add built-in functions with the round-to-odd
1128 (TRUNCF128_ODD): Likewise.
1129 (ADDF128_ODD): Likewise.
1130 (SUBF128_ODD): Likewise.
1131 (MULF128_ODD): Likewise.
1132 (DIVF128_ODD): Likewise.
1133 (FMAF128_ODD): Likewise.
1134 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
1135 UNSPEC_TRUNC_ROUND_TO_ODD.
1136 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
1137 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
1138 floating point round to odd instructions.
1139 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
1140 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
1141 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
1142 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
1143 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
1144 (trunc<mode>sf2_hw): Change the truncate with round to odd
1145 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
1146 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
1147 to odd hardware instructions.
1148 (sub<mode>3_odd): Likewise.
1149 (mul<mode>3_odd): Likewise.
1150 (div<mode>3_odd): Likewise.
1151 (sqrt<mode>2_odd): Likewise.
1152 (fma<mode>4_odd): Likewise.
1153 (fms<mode>4_odd): Likewise.
1154 (nfma<mode>4_odd): Likewise.
1155 (nfms<mode>4_odd): Likewise.
1156 (trunc<mode>df2_odd): Change the truncate with round to odd
1157 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
1159 * doc/extend.texi (PowerPC built-in functions): Update documentation
1160 for existing IEEE float128-bit built-in functions. Add built-in
1161 functions that generate the IEEE 128-bit floating point round to
1164 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
1166 PR rtl-optimization/77729
1167 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
1168 to (X&(C1&~C2))|C2 transformations.
1170 2017-10-03 Martin Jambor <mjambor@suse.cz>
1172 PR tree-optimization/82363
1173 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
1174 mismatch, mark lacc written regardless of racc.
1176 2017-10-03 Jakub Jelinek <jakub@redhat.com>
1178 PR tree-optimization/82381
1179 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
1180 stmt_to_insert nor wheather SSA_NAMEs are default defs.
1181 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
1182 fallthrough into reassoc_stmt_dominates_stmt_p.
1185 * combine.c (combine_instructions): Don't combine in unreachable
1188 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
1191 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
1192 function to not use the have_cpu variable. Do not set cpu_index,
1193 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
1195 (rs6000_valid_attribute_p): Remove duplicate initializations of
1196 old_optimize and func_optimize.
1197 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
1198 (rs6000_activate_target_options): Make global.
1199 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
1202 2017-10-02 Jakub Jelinek <jakub@redhat.com>
1204 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
1205 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
1206 Set *poffset to 0 and *psize and *pmax_size to -1 if
1207 *poffset + *psize overflows in HOST_WIDE_INT.
1209 PR tree-optimization/82387
1210 PR tree-optimization/82388
1211 PR tree-optimization/82389
1212 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
1213 instead of live_bytes non-NULL.
1215 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
1218 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
1221 2017-10-02 Richard Biener <rguenther@suse.de>
1223 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
1224 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
1225 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
1227 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
1229 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
1230 requested precision matches the type's.
1231 * calls.c (alloc_max_size): Calculate the new candidate size as
1232 a widest_int and use wi::to_widest when comparing it with the
1233 current candidate size.
1234 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
1235 zero rather than integer_zero_node.
1236 * match.pd: Check for a no-op conversion before using wi::add
1237 rather than after. Use tree_to_uhwi when summing small shift
1238 counts into an unsigned int.
1240 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
1241 Alan Hayward <alan.hayward@arm.com>
1242 David Sherwood <david.sherwood@arm.com>
1245 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
1246 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
1247 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
1248 POINTER_AND_FP_REGS.
1250 2017-10-02 Richard Biener <rguenther@suse.de>
1252 PR tree-optimization/82355
1253 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
1254 a mapping for the enclosing loop but avoid generating one for
1256 (copy_bb_and_scalar_dependences): Remove premature codegen
1257 error on PHIs in blocks duplicated into multiple places.
1258 * graphite-scop-detection.c
1259 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
1260 in the region use it as loop and nest to analyze the DR in.
1261 (try_generate_gimple_bb): Likewise.
1262 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
1263 (add_loop_constraints): For blocks in a loop not in the region
1264 create a dimension with a single iteration.
1265 * sese.h (gbb_loop_at_index): Remove assert.
1267 2017-10-01 Kevin Buettner <kevinb@redhat.com>
1269 * omp-expand.c (adjust_context_scope): New function.
1270 (expand_parallel_call): Call adjust_context_scope.
1272 2017-10-01 Jeff Law <law@redhat.com>
1274 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
1275 dom_opt_dom_walker class with direct access to private members.
1276 Add comments. Call test_for_singularity.
1277 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
1278 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
1279 m_dummy_cond anymore.
1280 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
1282 (pass_dominator:execute): Build the dummy_cond here and pass it
1283 to the dom_opt_dom_walker ctor.
1284 (test_for_singularity): New function.
1286 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
1287 Maya Rashish <coypu@sdf.org>
1289 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
1290 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
1291 (alpha*-*-netbsd*) Use nbsd_tm_file.
1292 (arm*-*-netbsdelf*) Likewise.
1293 (i[34567]86-*-netbsdelf*) Likewise.
1294 (x86_64-*-netbsd*) Likewise.
1295 (mips*-*-netbsd*) Likewise.
1296 (powerpc-*-netbsd*) Likewise.
1297 (sh*-*-netbsd*) Likewise.
1298 (sparc-*-netbsdelf*) Likewise.
1299 (sparc64-*-netbsd*) Likewise.
1300 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
1302 (vax-*-netbsdelf*) Likewise.
1303 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
1304 (UINT_FAST8_TYPE) Likewise.
1305 (INT_FAST16_TYPE) Check CHAR_FAST16.
1306 (UINT_FAST16_TYPE) Likewise.
1308 2017-09-30 Jakub Jelinek <jakub@redhat.com>
1311 * config/i386/i386.md
1312 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
1313 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
1314 *divmodsi4_zext_2): New define_insn_and_split.
1315 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
1316 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
1317 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
1318 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
1319 New define_insn_and_split.
1320 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
1321 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
1322 operands[1] having DImode when mode is SImode.
1324 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
1325 always SImode for DIV and MOD in REG_EQUAL notes.
1327 2017-09-29 Yury Gribov <tetra2005@gmail.com>
1330 * match.pd: Fix handling of NaNs in pattern.
1332 2017-09-29 Jeff Law <law@redhat.com>
1334 * sbitmap.c (bitmap_bit_in_range_p): New function.
1335 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
1336 * tree-ssa-dse.c (live_bytes_read): New function.
1337 (dse_classify_store): Ignore reads of dead bytes.
1339 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
1340 typos and whitespace errors.
1341 * config/i386/predicates.md (address_no_seg_operand): Likewise.
1342 * config/s390/s390.c (s390_emit_prologue): Likewise.
1344 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
1347 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
1348 with a symbol for LRA.
1350 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
1352 PR rtl-optimization/82338
1353 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
1355 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
1357 * genmodes.c (calc_wider_mode): Suppress qsort macro.
1358 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
1359 (qsort_chk): Declare.
1360 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
1361 (qsort_chk): New function.
1363 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1365 PR tree-optimization/82337
1366 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
1367 phi definition if the PHI result appears in an abnormal PHI.
1368 (find_basis_for_base_expr): Don't record a basis if the LHS of the
1369 basis appears in an abnormal PHI.
1371 2017-09-29 Richard Biener <rguenther@suse.de>
1373 * graphite-isl-ast-to-gimple.c
1374 (translate_isl_ast_to_gimple::set_codegen_error): New function.
1375 (binary_op_to_tree): Use it.
1376 (get_rename_from_scev): Likewise.
1377 (copy_loop_phi_nodes): Likewise.
1378 (copy_bb_and_scalar_dependences): Likewise.
1379 (translate_pending_phi_nodes): Likewise.
1381 2017-09-29 Jakub Jelinek <jakub@redhat.com>
1384 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
1385 for movabsq $(i32 << shift), r64.
1387 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
1389 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
1390 index when encoding %esp as %rsp to avoid 0x67 prefix.
1392 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
1394 * config/i386/i386.md (*movsf_internal, *movdf_internal):
1395 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
1397 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
1399 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
1400 Extensions with more than 16 double VFP registers.
1401 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
1402 to_clear_mask and all code related to it. Replace the remaining
1403 entry by a sbitmap and adapt code accordingly.
1405 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
1407 * brig-builtins.def: Change pure attributes to const.
1409 2017-09-28 Joseph Myers <joseph@codesourcery.com>
1411 * config.gcc (default_gnu_indirect_function): Default to yes for
1412 sparc*-*-linux* with glibc.
1414 2017-09-28 Joseph Myers <joseph@codesourcery.com>
1416 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
1417 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
1418 when creating .init_array and .fini_array sections with priority
1421 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
1424 * config/aarch64/aarch64.c
1425 (aarch64_builtin_support_vector_misalignment): Always return false
1426 when misalignment is unknown.
1428 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
1430 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
1431 this function to return false if the definition used by the swap
1432 instruction is artificial, or if the memory address from which the
1433 constant value is loaded is not represented by a base address held
1434 in a register or if the base address register is a frame or stack
1435 pointer. Additionally, return false if the base address of the
1436 loaded constant is a SYMBOL_REF but is not considered to be a
1438 (replace_swapped_load_constant): New function.
1439 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
1440 loaded constant vector with a load of a swapped constant vector.
1442 2017-09-27 Carl Love <cel@us.ibm.com>
1444 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
1445 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
1446 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
1449 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
1451 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
1452 first, always call autopref_rank_data otherwise.
1454 2017-09-27 Richard Biener <rguenther@suse.de>
1456 * graphite-scop-detection.c (find_scop_parameters): Move
1457 loop bound handling ...
1458 (gather_bbs::before_dom_children): ... here, avoiding the need
1459 to build scop_info->loop_nest.
1460 (record_loop_in_sese): Remove.
1461 * sese.h (sese_info_t::loop_nest): Remove.
1462 * sese.c (new_sese_info): Do not allocate loop_nest.
1463 (free_sese_info): Do not free loop_nest.
1465 2017-09-27 Jakub Jelinek <jakub@redhat.com>
1468 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
1469 lhs from calls if the lhs has addressable type.
1471 2017-09-27 Richard Biener <rguenther@suse.de>
1473 * graphite.h (scop::max_alias_set): New member.
1474 * graphite-scop-detection.c: Remove references to non-existing
1475 --param in comments.
1476 (build_alias_sets): Record the maximum alias set used for drs.
1477 (build_scops): Support zero as unlimited for
1478 --param graphite-max-arrays-per-scop.
1479 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
1481 (build_poly_sr_1): ... here. Compute alias set based on the
1482 maximum alias set used for drs rather than
1483 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
1485 2017-09-27 Richard Biener <rguenther@suse.de>
1487 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
1488 --param loop-block-tile-size=0 to disable tiling.
1490 2017-09-27 Richard Biener <rguenther@suse.de>
1492 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
1493 (graphite-max-nb-scop-params): Document special value zero.
1494 * domwalk.h (dom_walker::STOP): New symbolical constant.
1495 (dom_walker::dom_walker): Add optional parameter for bb to
1497 (dom_walker::~dom_walker): Declare.
1498 (dom_walker::before_dom_children): Document STOP return value.
1499 (dom_walker::m_user_bb_to_rpo): New member.
1500 (dom_walker::m_bb_to_rpo): Likewise.
1501 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
1502 mapping here if not provided by the user.
1503 (dom_walker::~dom_walker): Free bb to RPO mapping if not
1504 provided by the user.
1505 (dom_walker::STOP): Define.
1506 (dom_walker::walk): Do not compute bb to RPO mapping here.
1507 Support STOP return value from before_dom_children to stop
1509 * graphite-optimize-isl.c (optimize_isl): If the schedule
1510 is the same still generate code if -fgraphite-identity
1511 or -floop-parallelize-all are given.
1512 * graphite-scop-detection.c: Include cfganal.h.
1513 (gather_bbs::gather_bbs): Get and pass through bb to RPO
1515 (gather_bbs::before_dom_children): Return STOP for BBs
1517 (build_scops): Compute bb to RPO mapping and pass it to
1518 the domwalk. Treat --param graphite-max-nb-scop-params=0
1519 as not limiting the number of params.
1520 * graphite.c (graphite_initialize): Remove limit on the
1521 number of basic-blocks in a function.
1522 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
1523 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
1524 default value of 10.
1526 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
1528 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
1529 Adjust code to eliminate needing to do the shift right 32-bits
1530 operation after XSCVDPSPN.
1532 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1534 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
1535 ((X / Y) != 0 -> X >= Y): Likewise.
1537 2017-09-26 Carl Love <cel@us.ibm.com>
1539 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
1540 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
1541 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
1542 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
1543 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
1544 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
1545 definitions and overloading.
1546 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
1547 statement for P9V_BUILTIN_XST_LEN_R.
1548 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
1549 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
1550 define_expand and define_insn for the instructions and builtins.
1551 * doc/extend.texi: Update the built-in documentation file for the new
1553 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
1554 define_insn for the instructions
1556 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
1559 * gcc/config/netbsd-protos.h: New file.
1560 * gcc/config/netbsd.c: New file.
1561 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
1562 * gcc/config/t-netbsd: New file.
1563 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
1564 (tmake_file) Add t-netbsd.
1565 (extra_objs) Add netbsd.o.
1567 2017-09-26 Janus Weil <janus@gcc.gnu.org>
1571 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
1573 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
1575 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
1576 sign extension from a vector register to a GPR by doing a 32-bit
1577 direct move and then an EXTSW.
1578 (extendsi<mode>2 splitter): Likewise.
1579 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
1580 right or vector extract after doing XSCVDPSPN. Use
1581 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
1583 (movdi_from_sf_zero_ext): Likewise.
1584 (reload_gpr_from_vsxsf): Likewise.
1585 (p8_mfvsrd_4_disf): Delete, no longer used.
1586 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
1587 then needing to move the SFmode to a GPR to use the XSCVDPSP
1588 instruction instead of FRSP and XSCVDPSPN.
1589 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
1590 it is adjacent to the other XSCVSPDP insns.
1591 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
1592 SFmode to be in traditional Altivec registers.
1593 (vsx_xscvdpspn): Eliminate useless alternative constraint.
1594 (vsx_xscvspdpn): Likewise.
1595 (vsx_xscvspdpn_scalar): Likewise.
1597 2017-09-26 Martin Jambor <mjambor@suse.cz>
1599 * tree-sra.c (compare_access_positions): Put integral types first,
1600 stabilize sorting of integral types, remove conditions putting
1601 non-full-precision integers last.
1602 (sort_and_splice_var_accesses): Disable scalarization if a
1603 non-integert would be represented by a non-full-precision integer.
1605 2017-09-26 Joseph Myers <joseph@codesourcery.com>
1607 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
1608 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
1609 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
1610 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
1611 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
1612 conditionals inside the function instead of around it. Call
1613 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
1614 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
1616 2017-09-26 Richard Biener <rguenther@suse.de>
1618 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
1620 (scop_detection::build_scop_breadth): ... this. Removed.
1621 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
1622 (scop_detection::harmful_stmt_in_bb): Likewise.
1623 (scop_detection::graphite_can_represent_stmt): Likewise.
1624 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
1625 (scop_detection::can_represent_loop): Remove recursion, fold in ...
1626 (scop_detection::can_represent_loop_1): ... this. Removed.
1627 (scop_detection::harmful_loop_in_region): Simplify after inlining
1628 the above and remove more quadraticness.
1629 (build_scops): Adjust.
1630 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
1633 2017-09-26 Jakub Jelinek <jakub@redhat.com>
1636 * config/i386/i386.c (ix86_print_operand_address_as): Only test
1637 REGNO (base) == SP_REG if base is a REG.
1640 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
1641 if it is different SSA_NAME.
1642 (optimize_range_tests_cmp_bitwise): New function.
1643 (optimize_range_tests): Call it.
1645 2017-09-26 Richard Biener <rguenther@suse.de>
1647 PR tree-optimization/82321
1648 * graphite.c (canonicalize_loop_closed_ssa): Properly check
1649 for the def being inside the loop.
1651 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1653 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
1655 * config/s390/s390-builtins.def: Fix constraint on op4.
1657 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1659 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
1660 independent expanders.
1661 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
1662 ("vec_ordered", "vec_unordered"): New expanders.
1664 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1666 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
1669 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1671 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
1672 vec_unpacks_lo_v16qi.
1673 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
1675 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1677 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
1678 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
1679 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
1681 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1683 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
1685 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
1687 ("*vec_slb<mode>"): New insn pattern.
1688 ("vec_shr_<mode>"): New expander.
1689 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
1690 and force the shift count operand to V16QImode.
1691 ("vec_srb<mode>"): Set shift count mode to V16QI.
1693 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1695 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
1696 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
1697 ("vec_widen_smult_hi_<mode>"): New expander definitions.
1699 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
1702 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
1704 2017-09-26 Richard Biener <rguenther@suse.de>
1706 PR tree-optimization/82320
1707 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
1710 2017-09-25 Jeff Law <law@redhat.com>
1712 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
1713 prototype for new argument.
1714 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
1715 mostly extracted from rs6000_emit_allocate_stack.
1716 (rs6000_emit_probe_stack_range_stack_clash): New function.
1717 (rs6000_emit_allocate_stack): Call
1718 rs6000_emit_probe_stack_range_stack_clash as needed.
1719 (rs6000_emit_probe_stack_range): Add additional argument
1720 to call to gen_probe_stack_range{si,di}.
1721 (output_probe_stack_range): New.
1722 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
1723 (output_probe_stack_range_stack_clash): New.
1724 (rs6000_emit_prologue): Emit notes into dump file as requested.
1725 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
1726 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
1727 Add additional operand and pass it to output_probe_stack_range.
1729 2017-09-25 Bin Cheng <bin.cheng@arm.com>
1731 PR tree-optimization/82163
1732 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
1733 (checking_verify_loop_closed_ssa): New parameter.
1734 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
1735 (check_loop_closed_ssa_stmt): Delete.
1736 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
1737 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
1738 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
1741 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
1743 * brig-builtins.def: Treat HSAIL barrier builtins as
1744 setjmp/longjump style functions.
1746 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
1748 * target.def (constant_alignment): New hook.
1749 * defaults.h (CONSTANT_ALIGNMENT): Delete.
1750 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
1751 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
1752 * doc/tm.texi: Regenerate.
1753 * targhooks.h (default_constant_alignment): Declare.
1754 (constant_alignment_word_strings): Likewise.
1755 * targhooks.c (default_constant_alignment): New function.
1756 (constant_alignment_word_strings): Likewise.
1757 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
1758 instead of CONSTANT_ALIGNMENT.
1759 * varasm.c (align_variable, get_variable_align, build_constant_desc)
1760 (force_const_mem): Likewise.
1761 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
1762 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
1763 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
1764 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1765 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
1767 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
1768 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1769 constant_alignment_word_strings.
1770 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
1771 (CONSTANT_ALIGNMENT): Likewise.
1772 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1773 (arm_constant_alignment): New function.
1774 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
1775 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1776 constant_alignment_word_strings.
1777 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
1778 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1779 constant_alignment_word_strings.
1780 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
1781 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1782 (cris_constant_alignment): New function.
1783 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
1784 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1785 (epiphany_constant_alignment): New function.
1786 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
1787 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1788 constant_alignment_word_strings.
1789 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
1790 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1791 constant_alignment_word_strings.
1792 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
1793 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1794 constant_alignment_word_strings.
1795 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
1796 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
1797 * config/i386/i386.c (ix86_constant_alignment): Make static.
1798 Use the same interface as the target hook.
1799 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1800 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
1801 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1802 constant_alignment_word_strings.
1803 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
1804 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
1805 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1806 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
1807 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1808 constant_alignment_word_strings.
1809 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
1810 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1811 constant_alignment_word_strings.
1812 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
1813 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1814 constant_alignment_word_strings.
1815 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
1816 * config/microblaze/microblaze.c (microblaze_constant_alignment):
1818 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1819 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
1820 * config/mips/mips.c (mips_constant_alignment): New function.
1821 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1822 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
1823 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
1824 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1825 (mmix_constant_alignment): Make static. Use the same interface
1827 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
1828 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1829 constant_alignment_word_strings.
1830 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
1831 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1832 constant_alignment_word_strings.
1833 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
1834 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1835 constant_alignment_word_strings.
1836 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
1837 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1838 (rs6000_constant_alignment): New function.
1839 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
1840 * config/riscv/riscv.c (riscv_constant_alignment): New function.
1841 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1842 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
1843 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1844 (rs6000_constant_alignment): New function.
1845 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
1846 * config/s390/s390.c (s390_constant_alignment): New function.
1847 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1848 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
1849 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1850 constant_alignment_word_strings.
1851 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
1852 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1853 (sparc_constant_alignment): New function.
1854 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
1855 * config/spu/spu.c (spu_constant_alignment): New function.
1856 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1857 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
1858 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1859 constant_alignment_word_strings.
1860 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
1861 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1862 constant_alignment_word_strings.
1863 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
1864 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1865 constant_alignment_word_strings.
1866 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
1867 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1868 (visium_constant_alignment): New function.
1869 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
1870 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1871 (xtensa_constant_alignment): New function.
1872 * system.h (CONSTANT_ALIGNMENT): Poison.
1874 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
1876 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
1877 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
1878 (rs6000_builtin_valid_without_lhs): New helper function.
1879 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1880 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
1882 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
1884 * target.h (vec_perm_indices): Use unsigned short rather than
1886 (auto_vec_perm_indices): Likewise.
1887 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
1888 Use unsigned int rather than unsigned char.
1889 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
1891 2017-09-25 Richard Biener <rguenther@suse.de>
1893 * cfgloop.h (sort_sibling_loops): Declare.
1894 * cfgloop.c (sort_sibling_loops_cmp): New helper.
1895 (sort_sibling_loops): New function sorting the sibling loop list
1897 * graphite.c (graphite_transform_loops): Sort sibling loops.
1899 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
1901 * target.def (vec_perm_const_ok): Change sel parameter to
1903 * optabs-query.c (can_vec_perm_p): Update accordingly.
1904 * doc/tm.texi: Regenerate.
1905 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
1906 auto_vec_perm_indices and remove separate nelt field.
1907 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
1908 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
1909 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
1910 (aarch64_expand_vec_perm_const): Update accordingly.
1911 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
1912 to vec_perm_indices.
1913 * config/arm/arm.c (expand_vec_perm_d): Change perm to
1914 auto_vec_perm_indices and remove separate nelt field.
1915 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
1916 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
1917 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
1919 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
1920 to vec_perm_indices.
1921 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
1922 sel to vec_perm_indices.
1923 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
1924 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
1925 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
1927 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
1930 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
1933 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
1934 on the FUNCTION_DECL function context if it has a DIE that is a
1937 2017-09-25 Richard Biener <rguenther@suse.de>
1939 PR tree-optimization/82285
1940 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
1943 2017-09-25 Tom de Vries <tom@codesourcery.com>
1947 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
1950 2017-09-25 Richard Biener <rguenther@suse.de>
1952 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
1953 ISL errors other than isl_error_quota happen. Dump if the
1954 schedule is the same.
1955 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
1956 errors instead of aborting inside ISL.
1958 2017-09-25 Iain Sandoe <iain@codesourcery.com>
1961 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
1962 of libgcc_eh for m64.
1963 * config/i386/darwin64.h: Likewise.
1965 2017-09-25 Richard Biener <rguenther@suse.de>
1968 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
1969 attribute for incomplete types nor twice for complete ones.
1971 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
1974 * config/i386/i386.c (ix86_print_operand_address_as): Encode
1975 %esp as %rsp to avoid 0x67 prefix if there is no index or base
1978 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
1981 * config/i386/i386.opt (mprefer-avx256): Use
1982 ix86_target_flags variable.
1983 * config/i386/i386.c (ix86_target_string): Move
1984 -mprefer-avx256 to flag2_opts.
1986 2017-09-22 Jakub Jelinek <jakub@redhat.com>
1989 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
1990 and x != -1 | y != -1 into (x & y) != -1.
1992 2017-09-22 Steve Ellcey <sellcey@cavium.com>
1994 * config.gcc: Add new case statement to set
1995 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
1996 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
1997 s390x-*-linux* case statements. Added aarch64 to the list of
1998 supported architectures.
2000 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2002 PR tree-optimization/82289
2003 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
2004 STMT_VINFO_RELEVANT_P.
2006 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2007 Alan Hayward <alan.hayward@arm.com>
2008 David Sherwood <david.sherwood@arm.com>
2010 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
2011 for VR_RANGE only; don't allow VR_ANTI_RANGE.
2012 (extract_range_from_binary_expr_1): Don't call
2013 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
2015 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2016 Alan Hayward <alan.hayward@arm.com>
2017 David Sherwood <david.sherwood@arm.com>
2019 * target.def (preferred_vector_alignment): New hook.
2020 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
2022 * doc/tm.texi: Regenerate.
2023 * targhooks.h (default_preferred_vector_alignment): Declare.
2024 * targhooks.c (default_preferred_vector_alignment): New function.
2025 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
2027 (DR_TARGET_ALIGNMENT): New macro.
2028 (aligned_access_p): Update commentary.
2029 (vect_known_alignment_in_bytes): New function.
2030 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
2032 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
2033 Calculate the misalignment based on the target alignment rather than
2035 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
2036 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
2037 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
2038 the target alignment, rather than masking the element misalignment
2039 with the number of elements in a vector. Also use the target
2040 alignment when calculating the maximum number of peels.
2041 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
2042 instead of TYPE_ALIGN_UNIT.
2043 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
2044 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
2045 (vect_create_addr_base_for_vector_ref): Update call accordingly.
2046 (vect_create_data_ref_ptr): Likewise.
2047 (vect_setup_realignment): Realign by ANDing with
2048 -DR_TARGET_MISALIGNMENT.
2049 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
2050 the number of peels based on DR_TARGET_ALIGNMENT.
2051 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
2052 with the guaranteed alignment boundary when deciding whether
2054 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
2055 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
2056 (ensure_base_align): Remove stmt_info parameter. Get the
2057 target base alignment from DR_TARGET_ALIGNMENT.
2058 (vectorizable_store): Update call accordingly. Interpret
2059 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
2061 (vectorizable_load): Likewise.
2063 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2064 Alan Hayward <alan.hayward@arm.com>
2065 David Sherwood <david.sherwood@arm.com>
2067 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
2068 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
2069 (vect_enhance_data_refs_alignment): Likewise.
2071 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
2073 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
2074 error. Only quit immediately if parsing is complete.
2075 (BEGIN): Initialize fatal_err and parse_done.
2076 (begin fpu, end fpu): Check number of arguments.
2077 (begin arch, end arch): Likewise.
2078 (begin cpu, end cpu): Likewise.
2079 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
2080 (optalias): Likewise.
2082 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
2084 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
2085 * config/arm/arm-isa.h: Delete. Move definitions to ...
2086 * arm-cpus.in: ... here. Use new feature and fgroup values.
2087 * config/arm/arm.c (arm_option_override): Use lower case for feature
2089 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
2090 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
2091 * config/arm/parsecpu.awk (END): Add new command 'isa'.
2093 (print_isa_bits_for): New function.
2094 (gen_isa): New function.
2095 (gen_comm_data): Use print_isa_bits_for.
2096 (define feature): New keyword.
2097 (define fgroup): New keyword.
2098 * config/arm/t-arm (TM_H): Remove.
2099 (GTM_H): Add arm-isa.h.
2100 (arm-isa.h): Add rule to generate file.
2101 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
2102 case for feature bit names.
2104 2017-09-22 Richard Biener <rguenther@suse.de>
2106 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
2108 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
2109 print of no dependency loops ...
2110 * graphite.c (graphite_transform_loops): ... here.
2111 (canonicalize_loop_closed_ssa_form): Work from inner to outer
2113 (same_close_phi_node, remove_duplicate_close_phi,
2114 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
2115 (canonicalize_loop_closed_ssa): ... here and simplify.
2116 * graphite-optimize-isl.c: Include tree-vectorizer.h.
2117 (optimize_isl): Use dump_printf_loc to tell when we stopped
2118 optimizing because of an ISL timeout.
2120 2017-09-22 Richard Biener <rguenther@suse.de>
2122 PR tree-optimization/82291
2123 * tree-if-conv.c (predicate_mem_writes): Make sure to
2124 remove writes in blocks predicated with false.
2126 2017-09-22 Richard Biener <rguenther@suse.de>
2128 * sese.c: Include cfganal.h.
2129 (if_region_set_false_region): Remove.
2130 (create_if_region_on_edge): Likewise.
2131 (move_sese_in_condition): Re-implement without destroying
2134 2017-09-22 Richard Biener <rguenther@suse.de>
2136 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
2137 Verify both BBs contain loop PHI nodes before dispatching to
2139 (graphite_regenerate_ast_isl): Do not recompute dominators,
2140 do not verify three times. Restructure for clarity.
2141 * graphite-scop-detection.c (same_close_phi_node,
2142 remove_duplicate_close_phi, make_close_phi_nodes_unique,
2143 defined_in_loop_p, canonicalize_loop_closed_ssa,
2144 canonicalize_loop_closed_ssa_form): Simplify, remove excess
2145 checking and SSA rewrite, move to ...
2146 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
2147 (graphite_initialize): Do not pass in ctx, do not reset the
2148 SCEV cache, compute only dominators.
2149 (graphite_transform_loops): Allocate ISL ctx after
2150 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
2151 Maintain post-dominators only around build_scops.
2152 * sese.c (if_region_set_false_region): Make static. Free
2153 and recompute dominators.
2154 (move_sese_in_condition): Assert we don't get called with
2155 post-dominators computed.
2156 * sese.h (if_region_set_false_region): Remove.
2158 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
2160 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
2161 mode attribute for TARGET_AVX512VL.
2163 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
2165 * config/i386/i386.opt (mprefer-avx256): New option.
2166 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
2168 (ix86_preferred_simd_mode): Return 256-bit AVX modes
2169 for TARGET_PREFER_AVX256.
2170 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
2172 2017-09-21 Jeff Law <law@redhat.com>
2174 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
2175 Fix dump output if the only stack space is for pushed registers.
2177 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2179 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
2182 2017-09-21 Martin Sebor <msebor@redhat.com>
2185 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
2186 code (in C++) or code that triggers warnings.
2188 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
2190 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
2192 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
2194 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
2195 * sched-rgn.c: Ditto.
2196 * sel-sched-ir.c: Ditto.
2198 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
2200 * toplev.h (set_random_seed): Adjust return type.
2201 * toplev.c (init_local_tick): Move eager initialization of random_seed
2202 to get_random_seed. Adjust comment.
2203 (init_random_seed): Inline to get_random_seed, delete.
2204 (get_random_seed): Initialize random_seed lazily.
2205 (set_random_seed): Do not return previous value.
2206 (print_switch_value): Do not call get_random_seed.
2208 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
2210 * cgraph.c (delete_function_version): New, broken out from...
2211 (cgraph_node::delete_function_version): ...here. Rename to
2212 cgraph_node::delete_function_version_by_decl. Update all uses.
2213 (cgraph_node::remove): Call delete_function_version.
2215 2017-09-21 Jakub Jelinek <jakub@redhat.com>
2218 * tree-inline.c (expand_call_inline): Emit clobber stmts for
2219 VAR_DECLs to which addressable non-volatile parameters are mapped
2220 and for id->retvar after the return value assignment. Clear
2221 id->retval and id->retbnd after inlining.
2223 2017-09-21 Richard Biener <rguenther@suse.de>
2225 PR tree-optimization/82276
2226 PR tree-optimization/82244
2227 * tree-vrp.c (build_assert_expr_for): Set
2228 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
2230 (remove_range_assertions): Revert earlier change.
2232 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
2235 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
2237 2017-09-21 Richard Biener <rguenther@suse.de>
2239 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
2240 Restore valid IL after code generation errors.
2241 * graphite.c (graphite_transform_loops): Diagnose code
2242 generation issues as MSG_MISSED_OPTIMIZATION and continue
2243 with processing SCOPs.
2245 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2246 Alan Hayward <alan.hayward@arm.com>
2247 David Sherwood <david.sherwood@arm.com>
2249 * calls.c (compute_argument_addresses): Use simplify_gen_binary
2250 rather than choosing between plus_constant and gen_rtx_<CODE>.
2251 * expr.c (emit_push_insn): Likewise.
2252 (expand_expr_real_2): Likewise.
2254 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2255 Alan Hayward <alan.hayward@arm.com>
2256 David Sherwood <david.sherwood@arm.com>
2258 * loop-unroll.c (split_iv): Call copy_rtx on the step.
2260 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2261 Alan Hayward <alan.hayward@arm.com>
2262 David Sherwood <david.sherwood@arm.com>
2264 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
2265 calling tree_to_uhwi.
2267 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2268 Alan Hayward <alan.hayward@arm.com>
2269 David Sherwood <david.sherwood@arm.com>
2271 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
2272 INTEGER_CST rather than a negative test for ADDR_EXPR.
2274 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2275 Alan Hayward <alan.hayward@arm.com>
2276 David Sherwood <david.sherwood@arm.com>
2278 * tree-vrp.c (extract_range_from_binary_expr_1): Check
2279 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
2281 2017-09-21 Richard Biener <rguenther@suse.de>
2283 PR tree-optimization/71351
2284 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
2285 graphite_create_new_loop_guard): Remove, fold remaining parts
2287 (translate_isl_ast_node_for): ... here and simplify.
2289 2017-09-21 Jakub Jelinek <jakub@redhat.com>
2292 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
2293 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
2294 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
2295 alternative always use QI mode, for -Os imov (=R,R) alternative
2296 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
2299 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2300 Jeff Law <law@redhat.com>
2302 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
2303 (allocate_stack_space): New function, partially extracted from
2305 (s390_emit_prologue): Track offset to most recent stack probe.
2306 Code to allocate space moved into allocate_stack_space.
2307 Dump actions when no stack is allocated.
2308 (s390_prologue_plus_offset): New function.
2309 (s390_emit_stack_probe): Likewise.
2311 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
2313 * common.opt (Wa, Wl, Wp, g, gz=): Add
2315 (gno-column-info): Remove.
2316 (gcolumn-info): Drop RejectNegative.
2318 (gno-record-gcc-switches): Remove.
2319 (grecord-gcc-switches): Drop RejectNegative.
2320 (gno-split-dwarf): Remove.
2321 (gsplit-dwarf): Drop RejectNegative.
2322 (gno-strict-dwarf): Remove.
2323 (gstrict-dwarf): Drop RejectNegative.
2324 * config/darwin.opt (gfull, gused): Add RejectNegative.
2325 * dwarf2out.c (gen_producer_string): Drop
2326 gno-record-gcc-switches handler.
2327 * optc-gen.awk: Add g to prefixes with negative forms.
2328 * opts-common.c (remapping_prefix_p): New.
2329 (find_opt): Check it.
2330 (generate_canonical_option): Test g prefix.
2331 (option_map): Add -gno- mapping.
2332 (add_misspelling_candidates): Check remapping_prefix_p.
2334 2017-09-20 Jeff Law <law@redhat.com>
2336 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
2337 thinko in stack clash protection support.
2339 * explow.c (compute_stack_clash_protection_loop_data): Use
2340 CONST_INT_P instead of explicit test. Verify object is a
2341 CONST_INT_P before looking at INTVAL.
2342 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
2343 instead of explicit test.
2345 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
2348 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
2349 address instead of to r1 and r11.
2351 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
2353 * config.gcc: Support "knm".
2354 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
2355 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2357 * config/i386/i386.c (m_KNM): Define.
2358 (processor_target_table): Add "knm".
2360 (ix86_option_override_internal): Add "knm".
2361 (ix86_issue_rate): Add PROCESSOR_KNM.
2362 (ix86_adjust_cost): Ditto.
2363 (ia32_multipass_dfa_lookahead): Ditto.
2364 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
2365 (fold_builtin_cpu): Add M_INTEL_KNM.
2366 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
2367 (processor_type): Add PROCESSOR_KNM.
2368 * config/i386/x86-tune.def: Add m_KNM.
2369 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
2371 2017-09-20 Richard Biener <rguenther@suse.de>
2373 PR tree-optimization/80213
2374 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
2375 are allowed in empty BBs as well.
2376 (canonicalize_loop_closed_ssa): Also look for other complex
2378 (scop_detection::get_sese): Include the loop-closed PHI block
2380 (scop_detection::merge_sese): Remove code adding extra blocks.
2381 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
2382 (build_scops): Assert the final returned scop is invalid.
2384 2017-09-20 Richard Biener <rguenther@suse.de>
2386 PR tree-optimization/82264
2387 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
2389 (vn_phi_lookup): Likewise.
2390 (vn_phi_insert): Likewise.
2392 2017-09-20 Jakub Jelinek <jakub@redhat.com>
2394 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
2395 that fits into uhwi or shwi, add DW_AT_const_value regardless
2396 of early_dwarf without going through RTL, using add_AT_unsigned
2399 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
2400 (DEBUG_LTO_ABBREV_SECTION): Likewise.
2401 (DEBUG_LTO_MACINFO_SECTION): Likewise.
2402 (DEBUG_MACRO_SECTION): Likewise.
2403 (DEBUG_LTO_MACRO_SECTION): Likewise.
2404 (DEBUG_STR_DWO_SECTION): Likewise.
2405 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
2406 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
2407 (DEBUG_LTO_DWO_LINE_SECTION): Define.
2408 (DEBUG_LTO_LINE_STR_SECTION): Define.
2409 (init_sections_and_labels): Initialize debug_line_str_section
2410 variable. Initialize debug_loc_section for -gdwarf-5 to
2411 DEBUG_LOCLISTS_SECTION. Formatting fixes.
2413 2017-09-20 Richard Biener <rguenther@suse.de>
2415 * graphite-sese-to-poly.c (extract_affine): Properly handle
2416 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
2418 2017-09-20 Richard Biener <rguenther@suse.de>
2420 PR tree-optimization/81373
2421 * graphite-scop-detection.c (build_cross_bb_scalars_def):
2422 Force SESE live-out defs to be handled even if they are
2425 2017-09-19 Jeff Law <law@redhat.com>
2427 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
2428 nothing for stack adjustments with REG_STACK_CHECK.
2429 * sched-deps.c (parse_add_or_inc): Reject insns with
2430 REG_STACK_CHECK from dependency breaking.
2431 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
2432 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
2433 * reg-notes.def (STACK_CHECK): New note.
2435 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
2436 (ix86_expand_prologue): Dump stack clash info as needed.
2437 Call ix86_adjust_stack_and_probe_stack_clash as needed.
2439 * function.c (dump_stack_clash_frame_info): New function.
2440 * function.h (dump_stack_clash_frame_info): Prototype.
2441 (enum stack_clash_probes): New enum.
2443 * config/alpha/alpha.c (alpha_expand_prologue): Also check
2444 flag_stack_clash_protection.
2445 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
2446 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
2447 (arm_frame_pointer_required): Likewise.
2448 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
2449 (ia64_expand_prologue): Likewise.
2450 * config/mips/mips.c (mips_expand_prologue): Likewise.
2451 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
2452 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
2453 (sparc_flat_expand_prologue): Likewise.
2454 * config/spu/spu.c (spu_expand_prologue): Likewise.
2456 * explow.c: Include "params.h".
2457 (anti_adjust_stack_and_probe_stack_clash): New function.
2458 (get_stack_check_protect): Likewise.
2459 (compute_stack_clash_protection_loop_data): Likewise.
2460 (emit_stack_clash_protection_loop_start): Likewise.
2461 (emit_stack_clash_protection_loop_end): Likewise.
2462 (allocate_dynamic_stack_space): Use get_stack_check_protect.
2463 Use anti_adjust_stack_and_probe_stack_clash.
2464 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
2465 (emit_stack_clash_protection_loop_start): Likewise.
2466 (emit_stack_clash_protection_loop_end): Likewise.
2467 * rtl.h (get_stack_check_protect): Prototype.
2468 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
2469 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
2470 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
2472 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
2474 * doc/tm.texi: Rebuilt.
2475 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
2476 get_stack_check_protect.
2477 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
2478 * config/arm/arm.c (arm_expand_prologue): Likewise.
2479 (arm_frame_pointer_required): Likewise.
2480 * config/i386/i386.c (ix86_expand_prologue): Likewise.
2481 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
2482 * config/mips/mips.c (mips_expand_prologue): Likewise.
2483 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
2484 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
2485 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
2486 (sparc_flat_expand_prologue): Likewise.
2488 * common.opt (-fstack-clash-protection): New option.
2489 * flag-types.h (enum stack_check_type): Note difference between
2490 -fstack-check= and -fstack-clash-protection.
2491 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
2492 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
2493 * toplev.c (process_options): Issue warnings/errors for cases
2494 not handled with -fstack-clash-protection.
2495 * doc/invoke.texi (-fstack-clash-protection): Document new option.
2496 (-fstack-check): Note additional problem with -fstack-check=generic.
2497 Note that -fstack-check is primarily for Ada and refer users
2498 to -fstack-clash-protection for stack-clash-protection.
2499 Document new params for stack clash protection.
2501 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
2503 * config/i386/i386.c (ix86_split_long_move): Do not handle
2504 address used for LEA in a special way.
2506 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
2508 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
2510 2017-09-19 Martin Sebor <msebor@redhat.com>
2513 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
2514 of incompatible types.
2516 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
2518 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
2519 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
2520 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2521 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
2523 2017-09-19 Richard Biener <rguenther@suse.de>
2525 PR tree-optimization/82244
2526 * tree-vrp.c (remove_range_assertions): Do not propagate
2527 a constant to abnormals but replace the assert with a copy.
2529 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
2531 PR rtl-optimization/57878
2532 PR rtl-optimization/68988
2533 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
2534 avoidance test involving non_reload_pseudos. Move frequency test
2535 below the general fragmentation avoidance test.
2537 2017-09-19 Richard Biener <rguenther@suse.de>
2539 PR tree-optimization/69728
2540 * graphite-sese-to-poly.c (schedule_error): New global.
2541 (add_loop_schedule): Handle empty domain by failing the
2543 (build_original_schedule): Handle schedule_error.
2545 2017-09-19 Richard Biener <rguenther@suse.de>
2547 * graphite-scop-detection.c (scop_detection::can_represent_loop):
2548 Do not iterate to sibling loops but only to siblings of inner
2551 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
2554 * config/m68k/m68k.md (moveq feeding equality comparison): Check
2555 that the registers are different.
2557 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
2559 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
2560 to processor_model and "amdfam17h" to arch_names_table.
2561 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
2563 2017-09-18 Jakub Jelinek <jakub@redhat.com>
2566 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
2568 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2569 Alan Hayward <alan.hayward@arm.com>
2570 David Sherwood <david.sherwood@arm.com>
2572 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
2574 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
2576 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
2577 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
2578 vect_schedule_slp_instance.
2579 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
2580 Update call to vect_slp_analyze_node_operations. Simplify return
2582 (vect_slp_analyze_bb_1): Update call accordingly.
2583 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
2584 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
2585 (vect_schedule_slp): Update call accordingly.
2587 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2588 Alan Hayward <alan.hayward@arm.com>
2589 David Sherwood <david.sherwood@arm.com>
2591 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
2592 with types that aren't in fact scalar.
2594 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2596 * tree-vect-slp.c (vect_record_max_nunits): New function,
2598 (vect_build_slp_tree_1): ...here.
2599 (vect_build_slp_tree_2): Call it for phis too.
2601 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2603 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
2604 to vect_get_vec_def_for_operand when getting the mask operand.
2606 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2607 Alan Hayward <alan.hayward@arm.com>
2608 David Sherwood <david.sherwood@arm.com>
2610 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
2613 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2614 Alan Hayward <alan.hayward@arm.com>
2615 David Sherwood <david.sherwood@arm.com>
2617 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
2618 calculation for vector booleans.
2620 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2621 Alan Hayward <alan.hayward@arm.com>
2622 David Sherwood <david.sherwood@arm.com>
2624 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
2626 (vect_transform_stmt): ...here.
2627 (vect_analyze_stmt): Use it instead of calling
2628 vectorizable_live_operation directly.
2630 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
2632 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
2633 non-SIMT targets in acc vector loops.
2635 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
2637 * configure.ac: Add arc and check if assembler supports gdwarf2.
2638 * configure: Regenerate.
2640 2017-09-18 Richard Biener <rguenther@suse.de>
2642 PR tree-optimization/82220
2643 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
2644 epilogue niters from the min_profitable_iters compute.
2646 2017-09-18 Jakub Jelinek <jakub@redhat.com>
2649 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
2650 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
2651 (ix86_init_pic_reg): Revert 2017-09-01 changes.
2653 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
2656 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
2657 switching to a new text section.
2659 2017-09-18 Richard Biener <rguenther@suse.de>
2661 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
2663 (build_alias_set): Reject aliases with no access function.
2665 2017-09-18 Richard Biener <rguenther@suse.de>
2667 PR tree-optimization/79622
2668 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
2670 (build_cross_bb_scalars_use): Likewise.
2672 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
2674 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
2676 2017-09-18 Alan Modra <amodra@gmail.com>
2679 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
2680 stack_pointer_rtx for count 0. Update comments. Break up
2681 large rtl expression.
2683 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
2685 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
2686 Increase to 20 bytes.
2687 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
2688 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
2689 or avx version of the stub.
2691 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
2694 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
2695 compute the minimum stack alignment. Also update preferred stack
2696 boundary for leaf functions.
2698 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
2700 PR tree-optimization/82228
2701 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
2704 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
2706 * common/config/nds32/nds32-common.c
2707 (nds32_option_optimization_table): Refine formatting.
2708 (nds32_option_optimization_table): Use -fsched-pressure and
2709 -fomit-frame-pointer for specific optimization level.
2711 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
2713 * config/nds32/nds32.c: Refine formatting and comments.
2714 * config/nds32/nds32.h: Likewise.
2715 * config/nds32/nds32.md: Likewise.
2716 * config/nds32/nds32-cost.c: Likewise.
2717 * config/nds32/nds32-isr.c: Likewise.
2718 * config/nds32/nds32-md-auxiliary.c: Likewise.
2719 * config/nds32/nds32-multiple.md: Likewise.
2720 * config/nds32/nds32-predicates.c: Likewise.
2722 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
2723 Jakub Jelinek <jakub@redhat.com>
2725 Add support for -std=c++2a.
2726 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
2728 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
2730 2017-09-15 Steve Ellcey <sellcey@cavium.com>
2733 * doc/extend.texi (Common Function Attributes): Add
2734 references to ARM, AArch64, and S/390 specific attributes.
2735 (Function Specific Option Pragmas): Add AArch64 and S/390
2736 to list of back ends that support the target pragma.
2738 2017-09-15 Nathan Sidwell <nathan@acm.org>
2740 * doc/standards.texi: Fix C++17 description. Update URLs for
2743 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
2745 * common.opt (Wcast-align=strict): New warning option.
2746 * doc/invoke.texi: Document -Wcast-align=strict.
2748 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
2750 * cgraph.h (cgraph_thunk_info): Add comments.
2751 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
2752 assert for VIRTUAL_* arguments stricter.
2754 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
2756 PR tree-optimization/71026
2757 * match.pd: Move RDIV patterns from fold-const.c
2758 * fold-const.c (distribute_real_division): Removed.
2759 (fold_binary_loc): Remove calls to distribute_real_divison.
2761 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2763 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
2764 c++1z and gnu++1z as deprecated. Change other references to
2765 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
2766 Change -Wc++1z-compat to -Wc++17-compat.
2767 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
2768 * dwarf2out.c (highest_c_language): Handle C++17.
2769 (gen_compile_unit_die): Likewise.
2771 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2773 PR rtl-optimization/82192
2774 * combine.c (make_extraction): Don't look through non-paradoxical
2775 SUBREGs or TRUNCATE if pos + len is or might be bigger than
2778 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2779 Alan Hayward <alan.hayward@arm.com>
2780 David Sherwood <david.sherwood@arm.com>
2782 * target.def (function_arg_offset): New hook.
2783 * targhooks.h (default_function_arg_offset): Declare.
2784 * targhooks.c (default_function_arg_offset): New function.
2785 * function.c (locate_and_pad_parm): Use
2786 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
2787 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
2788 (TARGET_FUNCTION_ARG_OFFSET): ...this.
2789 * doc/tm.texi: Regenerate.
2790 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
2791 * config/spu/spu.c (spu_function_arg_offset): New function.
2792 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
2793 * system.h (FUNCTION_ARG_OFFSET): Poison.
2795 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2796 Alan Hayard <alan.hayward@arm.com>
2797 David Sherwood <david.sherwood@arm.com>
2799 * target.def (truly_noop_truncation): New hook.
2800 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
2801 than TRULY_NOOP_TRUNCATION.
2802 * hooks.h (hook_bool_uint_uint_true): Declare.
2803 * hooks.c (hook_bool_uint_uint_true): New function.
2804 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
2805 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
2806 * doc/tm.texi: Regenerate.
2807 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
2808 rather than TRULY_NOOP_TRUNCATION in comments.
2809 (simplify_comparison): Likewise.
2810 (record_truncated_value): Likewise.
2811 * expmed.c (extract_bit_field_1): Likewise.
2812 (extract_split_bit_field): Likewise.
2813 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
2814 instead of TRULY_NOOP_TRUNCATION.
2815 * function.c (assign_parm_setup_block): Likewise.
2816 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
2817 * rtlhooks.c: Include target.h.
2818 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
2819 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
2820 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
2821 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
2822 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
2823 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
2824 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
2825 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
2826 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
2827 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
2828 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
2829 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
2830 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
2831 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
2832 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
2833 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
2834 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
2835 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
2836 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
2837 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
2838 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
2839 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
2840 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
2841 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
2842 * config/mips/mips.c (mips_truly_noop_truncation): New function.
2843 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2844 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2845 rather than TRULY_NOOP_TRUNCATION in comments.
2846 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
2847 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
2848 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
2849 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
2850 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
2851 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
2852 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
2853 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
2854 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
2855 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
2856 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
2857 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2858 rather than TRULY_NOOP_TRUNCATION in comments.
2859 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
2860 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
2861 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
2862 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
2863 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
2864 TRULY_NOOP_TRUNCATION condition.
2865 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
2866 (TRULY_NOOP_TRUNCATION): Delete.
2867 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
2868 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
2869 * config/spu/spu.c (spu_truly_noop_truncation): New function.
2870 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2871 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
2872 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
2873 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
2874 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2875 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2876 rather than TRULY_NOOP_TRUNCATION in comments.
2877 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
2878 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
2879 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
2880 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
2881 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
2882 * system.h (TRULY_NOOP_TRUNCATION): Poison.
2884 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
2887 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
2888 (*cmp_ior): Likewise.
2889 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
2890 (*ior_scc_scc_cmp): Likewise.
2891 (*and_scc_scc): Likewise.
2892 (*and_scc_scc_cmp): Likewise.
2894 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2895 Alan Hayard <alan.hayward@arm.com>
2896 David Sherwood <david.sherwood@arm.com>
2898 * target.def (can_change_mode_class): New hook.
2899 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
2900 (hard_regno_nregs): Likewise.
2901 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
2902 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
2903 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
2904 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
2905 (LOAD_EXTEND_OP): Update accordingly.
2906 * doc/tm.texi: Regenerate.
2907 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
2908 CANNOT_CHANGE_MODE_CLASS.
2909 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
2910 (REG_CAN_CHANGE_MODE_P): ...this new macro.
2911 * combine.c (simplify_set): Update accordingly.
2912 * emit-rtl.c (validate_subreg): Likewise.
2913 * recog.c (general_operand): Likewise.
2914 * regcprop.c (mode_change_ok): Likewise.
2915 * reload1.c (choose_reload_regs): Likewise.
2916 (inherit_piecemeal_p): Likewise.
2917 * rtlanal.c (simplify_subreg_regno): Likewise.
2918 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
2919 instead of CANNOT_CHANGE_MODE_CLASS.
2920 (reload_cse_simplify_operands): Likewise.
2921 * reload.c (push_reload): Use targetm.can_change_mode_class
2922 instead of CANNOT_CHANGE_MODE_CLASS.
2923 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
2924 REG_CANNOT_CHANGE_MODE_P.
2925 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2926 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
2927 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2928 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2929 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2930 (arm_can_change_mode_class): New function.
2931 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
2932 than CANNOT_CHANGE_MODE_CLASS in comments.
2933 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2934 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
2935 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
2936 (ix86_can_change_mode_class): ...this new function, inverting the
2937 sense of the return value.
2938 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2939 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2940 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2941 (ia64_can_change_mode_class): New function.
2942 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2943 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
2944 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
2945 (m32c_can_change_mode_class): ...this new function, inverting the
2946 sense of the return value.
2947 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2948 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2949 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
2950 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
2951 (mips_can_change_mode_class): ...this new function, inverting the
2952 sense of the return value.
2953 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2954 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2955 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2956 (msp430_can_change_mode_class): New function.
2957 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2958 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
2959 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2960 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2961 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2962 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
2963 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2964 (pa_cannot_change_mode_class): Replace with...
2965 (pa_can_change_mode_class): ...this new function, inverting the
2966 sense of the return value.
2967 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
2968 than CANNOT_CHANGE_MODE_CLASS in comments.
2969 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2970 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
2971 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2972 (pdp11_cannot_change_mode_class): Replace with...
2973 (pdp11_can_change_mode_class): ...this new function, inverting the
2974 sense of the return value.
2975 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2976 * config/powerpcspe/powerpcspe-protos.h
2977 (rs6000_cannot_change_mode_class_ptr): Delete.
2978 * config/powerpcspe/powerpcspe.c
2979 (rs6000_cannot_change_mode_class_ptr): Delete.
2980 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2981 (rs6000_option_override_internal): Assign to
2982 targetm.can_change_mode_class instead of
2983 rs6000_cannot_change_mode_class_ptr.
2984 (rs6000_cannot_change_mode_class): Replace with...
2985 (rs6000_can_change_mode_class): ...this new function, inverting the
2986 sense of the return value.
2987 (rs6000_debug_cannot_change_mode_class): Replace with...
2988 (rs6000_debug_can_change_mode_class): ...this new function.
2989 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2990 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
2991 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2992 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2993 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
2995 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
2996 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2997 (rs6000_option_override_internal): Assign to
2998 targetm.can_change_mode_class instead of
2999 rs6000_cannot_change_mode_class_ptr.
3000 (rs6000_cannot_change_mode_class): Replace with...
3001 (rs6000_can_change_mode_class): ...this new function, inverting the
3002 sense of the return value.
3003 (rs6000_debug_cannot_change_mode_class): Replace with...
3004 (rs6000_debug_can_change_mode_class): ...this new function.
3005 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3006 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
3007 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
3008 (s390_can_change_mode_class): ...this new function, inverting the
3009 sense of the return value.
3010 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3011 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3012 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
3013 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3014 (sh_cannot_change_mode_class): Replace with...
3015 (sh_can_change_mode_class): ...this new function, inverting the
3016 sense of the return value.
3017 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3018 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3019 (sparc_can_change_mode_class): New function.
3020 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3021 * config/spu/spu.c (spu_can_change_mode_class): New function.
3022 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3023 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3024 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3025 (visium_can_change_mode_class): New function.
3026 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
3028 2017-09-15 Richard Biener <rguenther@suse.de>
3030 PR tree-optimization/82217
3031 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
3032 but not undefined case.
3034 2017-09-15 Jakub Jelinek <jakub@redhat.com>
3037 * postreload.c (reload_cse_simplify_operands): Skip
3038 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
3040 2017-09-15 Richard Biener <rguenther@suse.de>
3042 PR tree-optimization/68823
3043 * graphite-scop-detection.c (build_alias_set): If we have a
3044 possible dependence check whether we can handle them by just
3045 looking at the DRs DR_ACCESS_FNs.
3046 (build_scops): If build_alias_set fails, fail the SCOP.
3048 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
3050 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
3051 to support float128 built-in functions that require the ISA 3.0
3053 (BU_FLOAT128_3_HW): Likewise.
3054 (SQRTF128): Add support for the IEEE 128-bit square root and fma
3056 (FMAF128): Likewise.
3058 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
3059 support for built-in functions that need the ISA 3.0 IEEE 128-bit
3060 floating point instructions.
3061 (rs6000_invalid_builtin): Likewise.
3062 (rs6000_builtin_mask_names): Likewise.
3063 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
3064 (RS6000_BTM_FLOAT128_HW): Likewise.
3065 (RS6000_BTM_COMMON): Likewise.
3066 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
3068 * doc/extend.texi (RS/6000 built-in functions): Document the
3069 IEEE 128-bit floating point square root and fused multiply-add
3072 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
3074 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
3075 reg (r2) isn't in the set of registers defined in the prologue.
3077 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3078 Alan Hayward <alan.hayward@arm.com>
3079 David Sherwood <david.sherwood@arm.com>
3081 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
3082 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
3083 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
3084 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
3085 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
3087 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
3088 max_vectorization_factor.
3089 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
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_min_worthwhile_factor): Delete.
3096 (vect_worthwhile_without_simd_p): Declare.
3097 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
3098 (vectorizable_reduction): Use it.
3099 * tree-vect-stmts.c (vectorizable_shift): Likewise.
3100 (vectorizable_operation): Likewise.
3102 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3103 Alan Hayward <alan.hayward@arm.com>
3104 David Sherwood <david.sherwood@arm.com>
3106 * tree-vectorizer.h (vect_get_num_copies): New function.
3107 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
3108 * tree-vect-loop.c (vectorizable_reduction): Likewise.
3109 (vectorizable_induction): Likewise.
3110 (vectorizable_live_operation): Likewise.
3111 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
3112 (vectorizable_bswap): Likewise.
3113 (vectorizable_call): Likewise.
3114 (vectorizable_conversion): Likewise.
3115 (vectorizable_assignment): Likewise.
3116 (vectorizable_shift): Likewise.
3117 (vectorizable_operation): Likewise.
3118 (vectorizable_store): Likewise.
3119 (vectorizable_load): Likewise.
3120 (vectorizable_condition): Likewise.
3121 (vectorizable_comparison): Likewise.
3122 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
3124 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3125 Alan Hayward <alan.hayward@arm.com>
3126 David Sherwood <david.sherwood@arm.com>
3128 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
3129 of vect_init_vector.
3131 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3132 Alan Hayward <alan.hayward@arm.com>
3133 David Sherwood <david.sherwood@arm.com>
3135 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
3136 an inline wrapper that provides a location.
3137 (gimple_build_vector): Likewise.
3138 * gimple-fold.c (gimple_build_vector_from_val): New function.
3139 (gimple_build_vector): Likewise.
3140 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
3141 functions to build the initial value. Always return a gimple value.
3142 (get_initial_defs_for_reduction): Likewise. Only compute
3144 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
3145 vect_init_vector on the results from get_initial_def(s)_for_reduction.
3146 (vectorizable_induction): Use gimple_build_vector rather than
3149 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3150 Alan Hayward <alan.hayward@arm.com>
3151 David Sherwood <david.sherwood@arm.com>
3153 * target.h (vec_perm_indices): New typedef.
3154 (auto_vec_perm_indices): Likewise.
3155 * optabs-query.h: Include target.h
3156 (can_vec_perm_p): Take a vec_perm_indices *.
3157 * optabs-query.c (can_vec_perm_p): Likewise.
3158 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
3159 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
3160 * tree-vect-generic.c (lower_vec_perm): Likewise.
3161 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
3162 (vect_grouped_load_supported): Likewise.
3163 (vect_shift_permute_load_chain): Likewise.
3164 (vect_permute_store_chain): Use auto_vec_perm_indices.
3165 (vect_permute_load_chain): Likewise.
3166 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
3167 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
3168 Update uses of can_vec_perm_p.
3169 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
3170 mode with a number of elements. Take a vec_perm_indices *.
3171 (vect_create_epilog_for_reduction): Update accordingly.
3172 Use auto_vec_perm_indices.
3173 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
3174 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
3175 (vect_transform_slp_perm_load): Likewise.
3176 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
3177 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
3178 (vect_gen_perm_mask_checked): Likewise.
3179 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
3180 (vect_gen_perm_mask_checked): Likewise.
3181 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
3182 (vectorizable_store): Likewise.
3183 (vectorizable_load): Likewise.
3184 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
3185 (vectorizable_bswap): Likewise.
3187 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3188 Alan Hayward <alan.hayward@arm.com>
3189 David Sherwood <david.sherwood@arm.com>
3191 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
3192 * tree.c (build_vector): Likewise.
3193 (build_vector_from_ctor): Update accordingly.
3194 (build_vector_from_val): Likewise.
3195 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
3196 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
3197 * tree-vect-generic.c (add_rshift): Likewise.
3198 (expand_vector_divmod): Likewise.
3199 (optimize_vector_constructor): Likewise.
3200 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
3201 (vect_transform_slp_perm_load): Likewise.
3202 (vect_schedule_slp_instance): Likewise.
3203 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
3204 (vectorizable_call): Likewise.
3205 (vect_gen_perm_mask_any): Likewise. Add elements in order.
3206 * expmed.c (make_tree): Likewise.
3207 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
3208 a vector passed to build_vector.
3209 (fold_convert_const): Likewise.
3210 (exact_inverse): Likewise.
3211 (fold_ternary_loc): Likewise.
3212 (fold_relational_const): Likewise.
3213 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
3214 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
3215 (const_unop): Likewise. Store the reduction accumulator in a
3216 variable rather than an array.
3217 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
3218 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
3219 the new vector, rather than constructing it after the input arrays.
3220 (native_interpret_vector): Use auto_vec<tree> when building
3221 a vector passed to build_vector. Add elements in order.
3222 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
3223 auto_vec<tree> when building a vector passed to build_vector.
3224 (vect_create_epilog_for_reduction): Likewise.
3225 (vectorizable_induction): Likewise.
3226 (get_initial_def_for_reduction): Likewise. Fix indentation of
3228 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
3230 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
3231 passed to build_vector.
3233 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3234 Alan Hayward <alan.hayward@arm.com>
3235 David Sherwood <david.sherwood@arm.com>
3237 * tree-core.h (tree_base::u): Add an "nelts" field.
3238 (tree_vector): Use VECTOR_CST_NELTS as the length.
3239 * tree.c (tree_size): Likewise.
3240 (make_vector): Initialize VECTOR_CST_NELTS.
3241 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
3242 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
3243 TYPE_VECTOR_SUBPARTS.
3244 * expr.c (const_vector_mask_from_tree): Consistently use "units"
3245 as the number of units, setting it from VECTOR_CST_NELTS.
3246 (const_vector_from_tree): Likewise.
3247 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
3248 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
3249 (fold_negate_expr_1): Likewise.
3250 (fold_convert_const): Likewise.
3251 (const_binop): Likewise. Differentiate the number of output and
3253 (const_unop): Likewise.
3254 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
3255 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
3256 in cases that did the opposite.
3258 2017-09-14 Richard Biener <rguenther@suse.de>
3260 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
3263 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
3265 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
3267 2017-09-14 Jakub Jelinek <jakub@redhat.com>
3270 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
3271 if and where to split a bb, except for splitting before debug insn
3272 sequences followed by non-label real insn. Delete debug insns
3273 in between basic blocks.
3275 * combine.c (make_compound_operation_int): Formatting fixes.
3277 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
3278 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3279 * config/netbsd.h (LINK_EH_SPEC): Likewise.
3280 * config/sol2.h (LINK_EH_SPEC): Likewise.
3281 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3282 * config/s390/linux.h (LINK_SPEC): Likewise.
3283 * config/freebsd.h (LINK_EH_SPEC): Likewise.
3284 * config/openbsd.h (LINK_EH_SPEC): Likewise.
3285 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3286 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
3287 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
3288 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3289 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
3290 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
3292 2017-09-13 Jakub Jelinek <jakub@redhat.com>
3294 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
3296 (ENDFILE_LINUX_SPEC): Likewise.
3297 (LINK_EH_SPEC): Likewise.
3298 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
3299 (LINK_OS_LINUX_SPEC32): Likewise.
3300 (LINK_OS_LINUX_SPEC64): Likewise.
3301 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
3302 (LINK_OS_LINUX_SPEC): Likewise.
3304 2017-09-13 Martin Liska <mliska@suse.cz>
3307 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
3308 CASE_HIGH is NULL_TREE.
3310 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
3311 Alan Hayward <alan.hayward@arm.com>
3312 David Sherwood <david.sherwood@arm.com>
3314 * target.def (secondary_memory_needed): New hook.
3315 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
3316 instead of SECONDARY_MEMORY_NEEDED.
3317 (secondary_memory_needed_mode): Likewise.
3318 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
3319 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
3320 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
3321 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
3322 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
3323 * doc/tm.texi: Regenerate.
3324 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
3325 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
3326 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3327 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
3328 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
3329 * config/i386/i386.c (inline_secondary_memory_needed): Put the
3330 mode argument first and change the reg_class arguments to reg_class_t.
3331 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
3332 Make static. Update the call to inline_secondary_memory_needed.
3333 (ix86_register_move_cost): Update the call to
3334 inline_secondary_memory_needed.
3335 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3336 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
3338 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
3339 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
3341 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
3342 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
3343 * config/mips/mips.c (mips_secondary_memory_needed): Make static
3344 and match hook interface. Add comment from mips.h.
3345 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3346 * config/mmix/mmix.md (truncdfsf2): Refer to
3347 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
3349 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
3350 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
3351 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3352 (pa_secondary_memory_needed): New function.
3353 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
3354 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
3355 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3356 (pdp11_secondary_memory_needed): Make static and match hook interface.
3357 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
3358 * config/powerpcspe/powerpcspe-protos.h
3359 (rs6000_secondary_memory_needed_ptr): Delete.
3360 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
3362 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3363 (rs6000_option_override_internal): Assign to
3364 targetm.secondary_memory_needed rather than
3365 rs6000_secondary_memory_needed_ptr.
3366 (rs6000_secondary_memory_needed): Match hook interface.
3367 (rs6000_debug_secondary_memory_needed): Likewise.
3368 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
3369 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
3370 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
3371 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3372 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
3373 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
3375 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
3376 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3377 (rs6000_option_override_internal): Assign to
3378 targetm.secondary_memory_needed rather than
3379 rs6000_secondary_memory_needed_ptr.
3380 (rs6000_secondary_memory_needed): Match hook interface.
3381 (rs6000_debug_secondary_memory_needed): Likewise.
3382 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
3383 * config/s390/s390.c (s390_secondary_memory_needed): New function.
3384 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3385 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
3386 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3387 (sparc_secondary_memory_needed): New function.
3388 * lra-constraints.c (check_and_process_move): Refer to
3389 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
3391 (curr_insn_transform): Likewise.
3392 (process_alt_operands): Use targetm.secondary_memory_needed
3393 instead of TARGET_SECONDARY_MEMORY_NEEDED.
3394 (check_secondary_memory_needed_p): Likewise.
3395 (choose_split_class): Likewise.
3396 * reload.c: Unconditionally include code that was previously
3397 conditional on SECONDARY_MEMORY_NEEDED.
3398 (push_secondary_reload): Use targetm.secondary_memory_needed
3399 instead of TARGET_SECONDARY_MEMORY_NEEDED.
3400 (push_reload): Likewise.
3401 * reload1.c: Unconditionally include code that was previously
3402 conditional on SECONDARY_MEMORY_NEEDED.
3403 (choose_reload_regs): Use targetm.secondary_memory_needed
3404 instead of TARGET_SECONDARY_MEMORY_NEEDED.
3405 (gen_reload): Likewise.
3406 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
3408 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
3409 Alan Hayward <alan.hayward@arm.com>
3410 David Sherwood <david.sherwood@arm.com>
3412 * target.def (secondary_memory_needed_mode): New hook:
3413 * targhooks.c (default_secondary_memory_needed_mode): Declare.
3414 * targhooks.h (default_secondary_memory_needed_mode): New function.
3415 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
3416 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
3417 * doc/tm.texi: Regenerate.
3418 * lra-constraints.c (check_and_process_move): Use
3419 targetm.secondary_memory_needed_mode instead of
3420 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
3421 (curr_insn_transform): Likewise.
3422 * reload.c (get_secondary_mem): Likewise.
3423 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3424 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
3426 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3427 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3428 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
3429 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3430 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
3432 * config/powerpcspe/powerpcspe-protos.h
3433 (rs6000_secondary_memory_needed_mode): Delete.
3434 * config/powerpcspe/powerpcspe.c
3435 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3436 (rs6000_secondary_memory_needed_mode): Make static.
3437 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3438 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
3440 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
3442 (rs6000_secondary_memory_needed_mode): Make static.
3443 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3444 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
3445 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3446 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3447 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
3449 (sparc_secondary_memory_needed_mode): New function.
3450 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
3452 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
3454 * config/aarch64/constraints.md (Umq): New constraint.
3455 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
3457 (mov<mode>): Update condition.
3459 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3461 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
3462 when bitposition is the same.
3464 2017-09-13 Richard Biener <rguenther@suse.de>
3466 * dwarf2out.c (output_die_symbol): Remove.
3467 (output_die): Do not output a DIEs symbol.
3469 2017-09-13 Richard Biener <rguenther@suse.de>
3472 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
3473 default-def to avoid breaking iterator update with the weird
3474 interaction with cgraph_update_edges_for_call_stmt_node.
3476 2017-09-13 Richard Biener <rguenther@suse.de>
3478 * tree-cfg.c (verify_gimple_assign_binary): Add verification
3479 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
3480 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
3481 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
3483 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
3485 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
3486 Disable pc relative literal load irrespective of
3487 TARGET_FIX_ERR_A53_84341 for default.
3489 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
3491 * config/sparc/sparc.c (output_return): Output the source location of
3492 the insn in the delay slot, if any.
3493 (output_sibcall): Likewise.
3495 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
3498 * common.opt (-static-pie): New alias.
3499 (shared): Negate static-pie.
3500 (-no-pie): Update help text.
3502 (static-pie): New option.
3503 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
3504 -static-pie support.
3505 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
3506 (LINK_EH_SPEC): Likewise.
3507 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3508 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
3509 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
3510 * gcc.c (LINK_COMMAND_SPEC): Likewise.
3511 (init_gcc_specs): Likewise.
3512 (init_spec): Likewise.
3513 (display_help): Update help message for -pie.
3514 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
3517 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
3519 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
3520 (movdi_aarch64): Likewise.
3521 (movti_aarch64): Likewise.
3523 2017-09-12 Simon Wright <simon@pushface.org>
3526 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
3527 calculation of the minor version, always output as 0.
3529 2017-09-12 Jakub Jelinek <jakub@redhat.com>
3532 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
3533 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
3534 on it early, rather than manual conversion late. For
3535 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
3536 instead of performing manual conversion.
3538 2017-09-12 Carl Love <cel@us.ibm.com>
3540 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
3541 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
3543 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
3544 VMULOSW): Add definitions.
3545 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
3546 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
3547 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
3548 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
3549 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
3551 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
3553 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
3555 (movti_aarch64): Likewise.
3556 (movdf_aarch64): Likewise.
3557 (movtf_aarch64): Likewise.
3558 (load_pairdi): Likewise.
3559 (store_pairdi): Likewise.
3560 (load_pairdf): Likewise.
3561 (store_pairdf): Likewise.
3562 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
3563 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
3564 (ldr_got_small_<mode>): Likewise.
3565 (ldr_got_small_28k_<mode>): Likewise.
3566 (ldr_got_tiny): Likewise.
3567 * config/aarch64/iterators.md (ldst_sz): New.
3568 (ldpstp_sz): Likewise.
3569 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
3571 (thunderx_load): Split load_8 to load_16.
3572 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
3574 (thunderx2t99_storepair_basic): Split store_8 to store_16.
3575 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
3576 (xgene1_store_pair): Split store_8 to store_16.
3577 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
3578 (falkor_st_0_st_sd): Split store_8 to store_16.
3580 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
3582 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
3583 and store1/2/3/4 to store_4/8/12/16.
3584 * config/aarch64/aarch64.md: Update for rename.
3585 * config/arm/arm.md: Likewise.: Likewise.
3586 * config/arm/arm.c: Likewise.
3587 * config/arm/thumb1.md: Likewise.
3588 * config/arm/thumb2.md: Likewise.
3589 * config/arm/vfp.md: Likewise.
3590 * config/arm/arm-generic.md: Likewise.
3591 * config/arm/arm1020e.md: Likewise.
3592 * config/arm/arm1026ejs.md: Likewise.
3593 * config/arm/arm1136jfs.md: Likewise.
3594 * config/arm/arm926ejs.md: Likewise.
3595 * config/arm/cortex-a15.md: Likewise.
3596 * config/arm/cortex-a17.md: Likewise.
3597 * config/arm/cortex-a5.md: Likewise.
3598 * config/arm/cortex-a53.md: Likewise.
3599 * config/arm/cortex-a57.md: Likewise.
3600 * config/arm/cortex-a7.md: Likewise.
3601 * config/arm/cortex-a8.md: Likewise.
3602 * config/arm/cortex-a9.md: Likewise.
3603 * config/arm/cortex-m4.md: Likewise.
3604 * config/arm/cortex-m7.md: Likewise.
3605 * config/arm/cortex-r4.md: Likewise.
3606 * config/arm/exynos-m1.md: Likewise.
3607 * config/arm/fa526.md: Likewise.
3608 * config/arm/fa606te.md: Likewise.
3609 * config/arm/fa626te.md: Likewise.
3610 * config/arm/fa726te.md: Likewise.
3611 * config/arm/fmp626.md: Likewise.
3612 * config/arm/iwmmxt.md: Likewise.
3613 * config/arm/ldmstm.md: Likewise.
3614 * config/arm/marvell-pj4.md: Likewise.
3615 * config/arm/xgene1.md: Likewise.
3616 * config/aarch64/thunderx.md: Likewise.
3617 * config/aarch64/thunderx2t99.md: Likewise.
3618 * config/aarch64/falkor.md: Likewise.
3620 2017-09-12 Martin Liska <mliska@suse.cz>
3622 * attribs.c (private_lookup_attribute): New function.
3623 * attribs.h (private_lookup_attribute): Declared here.
3624 (lookup_attribute): Called from this place.
3626 2017-09-12 Richard Biener <rguenther@suse.de>
3628 PR tree-optimization/82157
3629 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
3630 stmts with side-effects.
3632 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3633 Alan Hayward <alan.hayward@arm.com>
3634 David Sherwood <david.sherwood@arm.com>
3636 * target.def (hard_regno_nregs): New hook.
3637 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
3638 * targhooks.h (default_hard_regno_nregs): Declare.
3639 * targhooks.c (default_hard_regno_nregs): New function.
3640 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
3641 (TARGET_HARD_REGNO_NREGS): ...this hook.
3642 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
3643 (CLASS_MAX_NREGS): Likewise.
3644 * doc/tm.texi: Regenerate.
3645 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
3646 instead of HARD_REGNO_NREGS.
3647 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
3648 HARD_REGNO_NREGS in the comment.
3649 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
3650 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
3651 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
3652 Return an unsigned int.
3653 (TARGET_HARD_REGNO_NREGS): Redefine.
3654 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
3655 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
3656 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
3657 (arc_hard_regno_nregs): New function.
3658 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
3659 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
3660 (arm_hard_regno_nregs): New function.
3661 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
3662 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
3663 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
3664 (TARGET_HARD_REGNO_NREGS): Redefine.
3665 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
3666 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
3667 (HARD_REGNO_NREGS): Delete.
3668 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
3669 (cr16_hard_regno_nregs): New function.
3670 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
3671 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
3672 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
3673 (cris_hard_regno_nregs): New function.
3674 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
3675 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
3676 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
3677 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
3678 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
3679 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
3680 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
3681 (frv_hard_regno_nregs): Make static. Take and return an
3683 (frv_class_max_nregs): Remove outdated copy of documentation.
3684 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
3685 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
3686 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
3687 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
3688 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
3689 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
3690 (TARGET_HARD_REGNO_NREGS): Redefine.
3691 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
3692 (CLASS_MAX_NREGS): Update comment.
3693 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
3694 (ia64_hard_regno_nregs): New function.
3695 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
3696 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
3697 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
3698 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
3699 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
3701 (m32c_hard_regno_nregs): Likewise. Make static.
3702 (TARGET_HARD_REGNO_NREGS): Redefine.
3703 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
3704 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
3705 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
3706 (m68k_hard_regno_nregs): New function.
3707 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
3708 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
3709 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
3710 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
3711 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
3712 Take and return an unsigned int.
3713 (TARGET_HARD_REGNO_NREGS): Redefine.
3714 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
3715 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
3716 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
3717 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
3718 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
3719 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
3720 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
3721 (msp430_hard_regno_nregs): Make static. Take and return an
3723 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
3724 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
3725 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
3726 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
3727 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
3728 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
3729 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
3730 (TARGET_HARD_REGNO_NREGS): Redefine.
3731 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
3732 (PA_HARD_REGNO_NREGS): ...this.
3733 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
3734 (PA_HARD_REGNO_NREGS): ...this.
3735 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
3736 (pa_hard_regno_nregs): New function.
3737 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
3738 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
3739 (pdp11_hard_regno_nregs): New function.
3740 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
3741 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
3742 (rs6000_hard_regno_nregs_hook): New function.
3743 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
3744 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
3745 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
3746 Take and return an unsigned int. Move earlier in file.
3747 (TARGET_HARD_REGNO_NREGS): Redefine.
3748 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
3749 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
3750 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
3751 (rl78_hard_regno_nregs): Make static. Take and return an
3753 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
3754 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
3755 (rs6000_hard_regno_nregs_hook): New function.
3756 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
3757 * config/rx/rx.c (rx_hard_regno_nregs): New function.
3758 (TARGET_HARD_REGNO_NREGS): Redefine.
3759 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
3760 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
3761 instead of HARD_REGNO_NREGS.
3762 (s390_hard_regno_nregs): New function.
3763 (s390_hard_regno_mode_ok): Add comment from s390.h.
3764 (TARGET_HARD_REGNO_NREGS): Redefine.
3765 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
3766 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
3767 (sh_hard_regno_nregs): New function.
3768 (sh_pass_in_reg_p): Use it.
3769 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
3770 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
3771 (sparc_hard_regno_nregs): New function.
3772 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
3773 * config/spu/spu.c (spu_hard_regno_nregs): New function.
3774 (spu_function_arg_advance): Use it, supplying a valid register number.
3775 (TARGET_HARD_REGNO_NREGS): Redefine.
3776 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
3777 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
3778 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
3779 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
3780 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
3781 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
3782 (CLASS_MAX_NREGS): Remove copy of old documentation.
3783 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
3784 (visium_hard_regno_nregs): New function.
3785 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
3786 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
3787 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
3788 xtensa_hard_regno_nregs): New function.
3789 * system.h (HARD_REGNO_NREGS): Poison.
3791 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3793 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
3794 hard_regno_nregs instead of HARD_REGNO_NREGS.
3795 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
3796 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
3797 (c6x_expand_epilogue): Likewise.
3798 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
3799 (frv_read_iacc_argument): Likewise.
3800 * config/sh/sh.c: Include regs.h.
3801 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
3802 (regs_used): Likewise.
3803 (output_stack_adjust): Likewise.
3804 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
3805 * expmed.c: Include regs.h.
3806 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
3807 * ree.c: Include regs.h.
3808 (combine_reaching_defs): Use hard_regno_nregs instead of
3810 (add_removable_extension): Likewise.
3812 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3814 * regs.h (hard_regno_nregs): Turn into a function.
3815 (end_hard_regno): Update accordingly.
3816 * caller-save.c (setup_save_areas): Likewise.
3817 (save_call_clobbered_regs): Likewise.
3818 (replace_reg_with_saved_mem): Likewise.
3819 (insert_restore): Likewise.
3820 (insert_save): Likewise.
3821 * combine.c (can_change_dest_mode): Likewise.
3822 (move_deaths): Likewise.
3823 (distribute_notes): Likewise.
3824 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
3825 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
3826 (rs6000_split_multireg_move): Likewise.
3827 (rs6000_register_move_cost): Likewise.
3828 (rs6000_memory_move_cost): Likewise.
3829 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
3830 (rs6000_split_multireg_move): Likewise.
3831 (rs6000_register_move_cost): Likewise.
3832 (rs6000_memory_move_cost): Likewise.
3833 * cselib.c (cselib_reset_table): Likewise.
3834 (cselib_lookup_1): Likewise.
3835 * emit-rtl.c (set_mode_and_regno): Likewise.
3836 * function.c (aggregate_value_p): Likewise.
3837 * ira-color.c (setup_profitable_hard_regs): Likewise.
3838 (check_hard_reg_p): Likewise.
3839 (calculate_saved_nregs): Likewise.
3840 (assign_hard_reg): Likewise.
3841 (improve_allocation): Likewise.
3842 (calculate_spill_cost): Likewise.
3843 * ira-emit.c (modify_move_list): Likewise.
3844 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
3845 (ira_hard_reg_in_set_p): Likewise.
3846 * ira.c (setup_reg_mode_hard_regset): Likewise.
3847 (clarify_prohibited_class_mode_regs): Likewise.
3848 (check_allocation): Likewise.
3849 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3850 (lra_setup_reg_renumber): Likewise.
3851 (setup_try_hard_regno_pseudos): Likewise.
3852 (spill_for): Likewise.
3853 (assign_hard_regno): Likewise.
3854 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
3855 * lra-constraints.c (in_class_p): Likewise.
3856 (lra_constraint_offset): Likewise.
3857 (simplify_operand_subreg): Likewise.
3858 (lra_constraints): Likewise.
3859 (split_reg): Likewise.
3860 (split_if_necessary): Likewise.
3861 (invariant_p): Likewise.
3862 (inherit_in_ebb): Likewise.
3863 * lra-lives.c (process_bb_lives): Likewise.
3864 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
3865 (get_hard_regs): Likewise.
3866 (do_remat): Likewise.
3867 * lra-spills.c (assign_spill_hard_regs): Likewise.
3868 * mode-switching.c (create_pre_exit): Likewise.
3869 * postreload.c (reload_combine_recognize_pattern): Likewise.
3870 * recog.c (peep2_find_free_register): Likewise.
3871 * regcprop.c (kill_value_regno): Likewise.
3872 (set_value_regno): Likewise.
3873 (copy_value): Likewise.
3874 (maybe_mode_change): Likewise.
3875 (find_oldest_value_reg): Likewise.
3876 (copyprop_hardreg_forward_1): Likewise.
3877 * regrename.c (check_new_reg_p): Likewise.
3878 (regrename_do_replace): Likewise.
3879 * reload.c (push_reload): Likewise.
3880 (combine_reloads): Likewise.
3881 (find_dummy_reload): Likewise.
3882 (operands_match_p): Likewise.
3883 (find_reloads): Likewise.
3884 (find_equiv_reg): Likewise.
3885 (reload_adjust_reg_for_mode): Likewise.
3886 * reload1.c (count_pseudo): Likewise.
3887 (count_spilled_pseudo): Likewise.
3888 (find_reg): Likewise.
3889 (clear_reload_reg_in_use): Likewise.
3890 (free_for_value_p): Likewise.
3891 (allocate_reload_reg): Likewise.
3892 (choose_reload_regs): Likewise.
3893 (reload_adjust_reg_for_temp): Likewise.
3894 (emit_reload_insns): Likewise.
3895 (delete_output_reload): Likewise.
3896 * rtlanal.c (subreg_get_info): Likewise.
3897 * sched-deps.c (sched_analyze_reg): Likewise.
3898 * sel-sched.c (init_regs_for_mode): Likewise.
3899 (mark_unavailable_hard_regs): Likewise.
3900 (choose_best_reg_1): Likewise.
3901 (verify_target_availability): Likewise.
3902 * valtrack.c (dead_debug_insert_temp): Likewise.
3903 * var-tracking.c (track_loc_p): Likewise.
3904 (emit_note_insn_var_location): Likewise.
3905 * varasm.c (make_decl_rtl): Likewise.
3906 * reginfo.c (choose_hard_reg_mode): Likewise.
3907 (init_reg_modes_target): Refer directly to
3908 this_target_regs->x_hard_regno_nregs.
3910 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3912 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
3913 instead of hard_regno_nregs.
3915 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3917 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
3918 end_hard_regno instead of hard_regno_nregs.
3919 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
3920 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
3921 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
3922 * ira-color.c (improve_allocation): Likewise.
3923 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3924 * lra-lives.c (mark_regno_live): Likewise.
3925 (mark_regno_dead): Likewise.
3926 * lra-remat.c (operand_to_remat): Likewise.
3927 * lra.c (collect_non_operand_hard_regs): Likewise.
3928 * postreload.c (reload_combine_note_store): Likewise.
3929 (move2add_valid_value_p): Likewise.
3930 * reload.c (regno_clobbered_p): Likewise.
3932 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3934 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
3936 * config/v850/v850.c (v850_reorg): Likewise.
3937 * reload.c (refers_to_regno_for_reload_p): Likewise.
3938 (find_equiv_reg): Likewise.
3939 * reload1.c (reload_reg_reaches_end_p): Likewise.
3941 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3943 * caller-save.c (add_used_regs): Use REG_NREGS instead of
3945 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
3946 * config/arm/arm.c (output_move_neon): Likewise.
3947 (arm_attr_length_move_neon): Likewise.
3948 (neon_split_vcombine): Likewise.
3949 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
3950 (c6x_mark_reg_written): Likewise.
3951 (c6x_dwarf_register_span): Likewise.
3952 * config/i386/i386.c (ix86_save_reg): Likewise.
3953 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
3954 (rws_access_reg): Likewise.
3955 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
3956 * mode-switching.c (create_pre_exit): Likewise.
3957 * ree.c (combine_reaching_defs): Likewise.
3958 (add_removable_extension): Likewise.
3959 * regcprop.c (find_oldest_value_reg): Likewise.
3960 (copyprop_hardreg_forward_1): Likewise.
3961 * reload.c (reload_inner_reg_of_subreg): Likewise.
3962 (push_reload): Likewise.
3963 (combine_reloads): Likewise.
3964 (find_dummy_reload): Likewise.
3965 (reload_adjust_reg_for_mode): Likewise.
3966 * reload1.c (find_reload_regs): Likewise.
3967 (forget_old_reloads_1): Likewise.
3968 (reload_reg_free_for_value_p): Likewise.
3969 (reload_adjust_reg_for_temp): Likewise.
3970 (emit_reload_insns): Likewise.
3971 (delete_output_reload): Likewise.
3972 * sel-sched.c (choose_best_reg_1): Likewise.
3973 (choose_best_pseudo_reg): Likewise.
3975 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3976 Alan Hayward <alan.hayward@arm.com>
3977 David Sherwood <david.sherwood@arm.com>
3979 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
3980 * target.def (slow_unaligned_access): New hook.
3981 * targhooks.h (default_slow_unaligned_access): Declare.
3982 * targhooks.c (default_slow_unaligned_access): New function.
3983 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
3984 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
3985 * doc/tm.texi: Regenerate.
3986 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
3987 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
3988 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
3990 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
3991 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
3993 (rs6000_slow_unaligned_access): New function.
3994 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
3995 (expand_block_compare): Likewise.
3996 (expand_strn_compare): Likewise.
3997 (rs6000_rtx_costs): Likewise.
3998 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
3999 (riscv_slow_unaligned_access): Likewise.
4000 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
4001 (riscv_slow_unaligned_access_p): ...this and make static.
4002 (riscv_option_override): Update accordingly.
4003 (riscv_slow_unaligned_access): New function.
4004 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
4005 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
4006 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
4007 (rs6000_slow_unaligned_access): New function.
4008 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
4009 (rs6000_rtx_costs): Likewise.
4010 * config/rs6000/rs6000-string.c (expand_block_compare)
4011 (expand_strn_compare): Use targetm.slow_unaligned_access instead
4012 of SLOW_UNALIGNED_ACCESS.
4013 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
4014 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
4015 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
4016 of SLOW_UNALIGNED_ACCESS.
4017 * expmed.c (simple_mem_bitfield_p): Likewise.
4018 * expr.c (alignment_for_piecewise_move): Likewise.
4019 (emit_group_load_1): Likewise.
4020 (emit_group_store): Likewise.
4021 (copy_blkmode_from_reg): Likewise.
4022 (emit_push_insn): Likewise.
4023 (expand_assignment): Likewise.
4024 (store_field): Likewise.
4025 (expand_expr_real_1): Likewise.
4026 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
4027 * lra-constraints.c (simplify_operand_subreg): Likewise.
4028 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
4029 * gimple-ssa-store-merging.c: Likewise in block comment at start
4031 * tree-ssa-strlen.c: Include target.h.
4032 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
4033 of SLOW_UNALIGNED_ACCESS.
4034 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
4036 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4038 PR rtl-optimization/82185
4039 * expmed.c (emit_store_flag_int): Only test tem if it has been
4042 2017-09-12 Richard Biener <rguenther@suse.de>
4045 * match.pd ((FTYPE) N CMP CST): Fix typo.
4047 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
4049 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
4051 (mips_near_type_p): Add 'short_call' attribute as a synonym
4053 * doc/extend.texi (short_call): Document new function attribute.
4055 2017-09-12 Jakub Jelinek <jakub@redhat.com>
4058 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
4059 assertion check that in the condition.
4060 (get_atomic_generic_size): Likewise. Before testing if parameter
4061 has pointer type, if it has array type, call for C++
4062 default_conversion to perform array-to-pointer conversion.
4064 2017-09-12 Richard Biener <rguenther@suse.de>
4066 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
4067 for operations we cannot scalarize.
4069 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
4071 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
4072 vectors heap vectors. Clean up comments.
4073 Make visited_bbs a reference.
4074 (profitable_jump_thread_path): Make GC
4075 vectors heap vectors. Clean up comments.
4077 (convert_and_register_jump_thread_path): Make GC vectors heap
4079 (check_subpath_and_update_thread_path): Same. Clean up comments.
4080 Make visited_bbs a reference.
4081 (handle_phi): Abstract common code to to
4082 register_jump_thread_path_if_profitable.
4083 Rename VAR_BB to DEF_BB.
4085 Make GC vectors heap vectors.
4086 Make visited_bbs a reference.
4087 (handle_assignment): Same.
4088 (register_jump_thread_path_if_profitable): New.
4089 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
4091 Make GC vectors heap vectors. Clean up comments.
4092 Make visited_bbs a reference.
4093 (find_jump_threads_backwards): Make visited_bbs live in the stack.
4094 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
4097 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
4100 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
4101 words of E_DImode object are reachable by xtensa_uimm8x4 access.
4103 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
4105 Revert r251800 and r251799.
4107 2017-09-11 Martin Jambor <mjambor@suse.cz>
4110 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
4111 arguments in advance.
4112 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
4113 use it to find predecessor edges.
4114 (naive_outof_ssa): Collect vector of predecessors.
4116 2017-09-08 Jason Merrill <jason@redhat.com>
4118 PR c++/70029 - ICE with ref-qualifier and -flto
4119 * langhooks.h (struct lang_hooks_for_types): Add
4120 copy_lang_qualifiers.
4121 * attribs.c (build_type_attribute_qual_variant): Use it.
4122 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
4124 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
4125 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
4127 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
4130 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
4131 (*mulsi3_sp64): New instruction.
4132 (mulsi3): New expander.
4134 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
4136 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
4138 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4140 * sancov.c: Include memmodel.h.
4142 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
4145 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
4148 2017-09-07 Carl Love <cel@us.ibm.com>
4150 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
4151 the sldi instruction.
4153 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
4155 * sancov.c: Include tm_p.h.
4157 2017-09-07 Jakub Jelinek <jakub@redhat.com>
4160 * output.h (switch_to_other_text_partition): New declaration.
4161 * varasm.c (switch_to_other_text_partition): New function.
4162 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
4163 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
4164 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
4165 to the other text partition before emitting LCL label and switch back
4166 after emitting the word after it.
4168 2017-09-07 Richard Biener <rguenther@suse.de>
4170 * passes.def (pass_split_crit_edges): Remove instance before PRE.
4171 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
4172 critical edges here, after loop init.
4173 (pass_data_pre): Remove PROP_no_crit_edges flags.
4174 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
4175 for valueization of call args to avoid leaking VN_TOP.
4176 (visit_use): Assert we do not visit default defs.
4177 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
4178 Use error_mark_node to more easily detect leaking VN_TOP.
4179 All default-defs are varying, not VN_TOP. Mark them visited.
4180 (run_scc_vn): Make code match comment.
4182 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
4184 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
4185 OPTION_MASK_FLOAT128_KEYWORD.
4186 (POWERPC_MASKS): Likewise.
4187 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
4188 support for the -mfloat128-type option, and make -mfloat128
4189 default on PowerPC Linux systems. Define or undefine
4190 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
4191 Define __float128 to be __ieee128 if IEEE 128-bit support is
4192 enabled, or undefine it.
4193 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
4194 Delete defining __FLOAT128_TYPE__.
4195 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
4196 -mfloat128-type option and make -mfloat128 default on PowerPC
4198 (TARGET_FLOAT128_TYPE): Likewise.
4199 (-mfloat128-type): Likewise.
4200 * config/rs6000/rs6000.c (rs6000_option_override_internal):
4201 Delete the -mfloat128-type option and make -mfloat128 default on
4202 PowerPC Linux systems. Always use __ieee128 to be the keyword for
4203 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
4204 128-bit floating point is enabled. Change tests from using
4205 -mfloat128-type to -mfloat128.
4206 (rs6000_mangle_type): Use the correct mangling for the __float128
4207 type even if normal long double is restricted to 64-bits.
4208 (floatn_mode): Enable the _Float128 type by default on VSX Linux
4210 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
4211 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
4212 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
4214 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
4215 documentation for -mfloat128.
4217 2017-09-06 Olivier Hainque <hainque@adacore.com>
4219 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
4221 2017-09-06 Wish Wu <wishwu007@gmail.com>
4222 Jakub Jelinek <jakub@redhat.com>
4224 * asan.c (initialize_sanitizer_builtins): Add
4225 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
4226 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
4227 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
4228 BT_FN_VOID_UINT64_PTR variables.
4229 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
4230 (BT_FN_VOID_UINT16_UINT16): Likewise.
4231 (BT_FN_VOID_UINT32_UINT32): Likewise.
4232 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
4233 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
4234 (BT_FN_VOID_UINT64_PTR): Likewise.
4235 * common.opt (flag_sanitize_coverage): New variable.
4236 (fsanitize-coverage=trace-pc): Remove.
4237 (fsanitize-coverage=): Add.
4238 * flag-types.h (enum sanitize_coverage_code): New enum.
4239 * fold-const.c (fold_range_test): Disable non-short-circuit
4240 optimization if flag_sanitize_coverage.
4241 (fold_truth_andor): Likewise.
4242 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
4243 * opts.c (COVERAGE_SANITIZER_OPT): Define.
4244 (coverage_sanitizer_opts): New array.
4245 (get_closest_sanitizer_option): Add OPTS argument, handle also
4246 OPT_fsanitize_coverage_.
4247 (parse_sanitizer_options): Adjusted to also handle
4248 OPT_fsanitize_coverage_.
4249 (common_handle_option): Add OPT_fsanitize_coverage_.
4250 * sancov.c (instrument_comparison, instrument_switch): New function.
4251 (sancov_pass): Add trace-cmp support.
4252 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
4253 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
4254 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
4255 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
4256 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
4257 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
4258 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
4259 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
4260 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
4261 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
4263 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
4265 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
4266 error. Only quit immediately if parsing is complete.
4267 (BEGIN): Initialize fatal_err and parse_done.
4268 (begin fpu, end fpu): Check number of arguments.
4269 (begin arch, end arch): Likewise.
4270 (begin cpu, end cpu): Likewise.
4271 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
4272 (optalias): Likewise.
4274 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
4276 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
4277 * config/arm/arm-isa.h: Delete. Move definitions to ...
4278 * arm-cpus.in: ... here. Use new feature and fgroup values.
4279 * config/arm/arm.c (arm_option_override): Use lower case for feature
4281 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
4282 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
4283 * config/arm/parsecpu.awk (END): Add new command 'isa'.
4285 (print_isa_bits_for): New function.
4286 (gen_isa): New function.
4287 (gen_comm_data): Use print_isa_bits_for.
4288 (define feature): New keyword.
4289 (define fgroup): New keyword.
4290 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
4291 (arm-isa.h): Add rule to generate file.
4292 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
4293 case for feature bit names.
4295 2017-09-06 Richard Biener <rguenther@suse.de>
4297 * tree-ssa-pre.c (NECESSARY): Remove.
4298 (create_expression_by_pieces): Do not touch pass-local flags.
4299 (insert_into_preds_of_block): Likewise.
4300 (do_pre_regular_insertion): Likewise.
4301 (eliminate_insert): Likewise.
4302 (eliminate_dom_walker::before_dom_children): Likewise.
4303 (fini_eliminate): Do not look at inserted_exprs.
4304 (mark_operand_necessary): Remove.
4305 (remove_dead_inserted_code): Replace with simple work-list
4306 algorithm based on inserted_exprs and SSA uses.
4307 (pass_pre::execute): Re-order fini_eliminate and
4308 remove_dead_inserted_code.
4310 2017-09-06 Olivier Hainque <hainque@adacore.com>
4312 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
4313 for VxWorks 7. Adjust surrounding comments.
4315 2017-09-06 Richard Biener <rguenther@suse.de>
4317 * gimple-ssa-strength-reduction.c
4318 (find_candidates_dom_walker::before_dom_children): Also allow
4321 2017-09-06 Richard Biener <rguenther@suse.de>
4323 PR tree-optimization/82108
4324 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
4325 for gap in the non-permutation SLP case.
4327 2017-09-06 Martin Jambor <mjambor@suse.cz>
4329 PR tree-optimization/82078
4330 * tree-sra.c (sort_and_splice_var_accesses): Move call to
4331 add_access_to_work_queue...
4332 (build_accesses_from_assign): ...here.
4333 (propagate_all_subaccesses): Make sure racc is the group
4334 representative, if there is one.
4336 2017-09-06 Jakub Jelinek <jakub@redhat.com>
4339 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
4342 2017-09-06 Richard Biener <rguenther@suse.de>
4344 * gimple-ssa-strength-reduction.c
4345 (find_candidates_dom_walker::before_doom_children): Use a
4346 type and not a mode check.
4348 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
4351 * config/arm/predicates.md (arm_general_adddi_operand): Create new
4353 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
4355 2017-09-05 Jeff Law <law@redhat.com>
4357 PR tree-optimization/64910
4358 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
4359 cases where we have 3 or more operands.
4361 2017-09-05 Jakub Jelinek <jakub@redhat.com>
4364 * omp-low.c (lower_omp_for): Recompute tree invariant if
4365 gimple_omp_for_initial/final is ADDR_EXPR.
4368 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
4369 into gimple val before gimplification fo the COND_EXPR.
4371 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
4373 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
4374 REGION_COPY argument.
4375 (thread_through_all_blocks): Remove unused argument to
4376 duplicate_thread_path.
4378 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4379 Alan Hayward <alan.hayward@arm.com>
4380 David Sherwood <david.sherwood@arm.com>
4382 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
4383 Take a scalar_mode rather than a machine_mode.
4384 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
4385 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
4386 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
4387 (aarch64_gen_adjusted_ldpstp): Likewise.
4388 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
4390 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4391 Alan Hayward <alan.hayward@arm.com>
4392 David Sherwood <david.sherwood@arm.com>
4394 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
4395 Take a scalar_int_mode instead of a machine_mode.
4396 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
4397 (aarch64_output_scalar_simd_mov_immediate): Likewise.
4398 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
4399 (aarch64_simd_attr_length_rglist): Delete.
4400 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
4401 a scalar_int_mode instead of a machine_mode.
4402 (aarch64_add_offset): Likewise.
4403 (aarch64_internal_mov_immediate): Likewise
4404 (aarch64_add_constant_internal): Likewise.
4405 (aarch64_add_constant): Likewise.
4406 (aarch64_movw_imm): Likewise.
4407 (aarch64_rtx_arith_op_extract_p): Likewise.
4408 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
4409 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
4410 Remove assert that the mode isn't a vector.
4411 (aarch64_output_scalar_simd_mov_immediate): Likewise.
4412 (aarch64_expand_mov_immediate): Update calls after above changes.
4413 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
4414 (aarch64_and_bitmask_imm): Check for scalar integer modes.
4415 (aarch64_move_imm): Likewise.
4416 (aarch64_can_const_movi_rtx_p): Likewise.
4417 (aarch64_strip_extend): Likewise.
4418 (aarch64_extr_rtx_p): Likewise.
4419 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
4420 a CONST_INT when the mode parameter is VOIDmode.
4421 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
4423 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4425 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
4426 * stor-layout.c (bitwise_mode_for_mode): Likewise.
4427 (bitwise_type_for_mode): Update accordingly.
4429 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4431 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
4432 * stor-layout.c (mode_for_size_tree): Likewise.
4433 (mode_for_array): Update accordingly.
4434 (layout_decl): Likewise.
4435 (compute_record_mode): Likewise. Only set the mode once.
4437 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4439 * target.def (get_mask_mode): Change return type to opt_mode.
4441 * doc/tm.texi: Regenerate.
4442 * targhooks.h (default_get_mask_mode): Return an opt_mode.
4443 * targhooks.c (default_get_mask_mode): Likewise.
4444 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
4445 * optabs-query.c (can_vec_mask_load_store_p): Update use of
4446 targetm.get_mask_mode.
4447 * tree.c (build_truth_vector_type): Likewise.
4449 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4451 * machmode.h (mode_for_vector): Return an opt_mode.
4452 * stor-layout.c (mode_for_vector): Likewise.
4453 (mode_for_int_vector): Update accordingly.
4454 (layout_type): Likewise.
4455 * config/i386/i386.c (emit_memmov): Likewise.
4456 (ix86_expand_set_or_movmem): Likewise.
4457 (ix86_expand_vector_init): Likewise.
4458 (ix86_get_mask_mode): Likewise.
4459 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
4461 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
4462 * expmed.c (extract_bit_field_1): Likewise.
4463 * expr.c (expand_expr_real_2): Likewise.
4464 * optabs-query.c (can_vec_perm_p): Likewise.
4465 (can_vec_mask_load_store_p): Likewise.
4466 * optabs.c (expand_vec_perm): Likewise.
4467 * targhooks.c (default_get_mask_mode): Likewise.
4468 * tree-vect-stmts.c (vectorizable_store): Likewise.
4469 (vectorizable_load): Likewise.
4470 (get_vectype_for_scalar_type_and_size): Likewise.
4472 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4474 * machmode.h (mode_for_int_vector): New function.
4475 * stor-layout.c (mode_for_int_vector): Likewise.
4476 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
4477 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
4478 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
4479 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
4480 (s390_expand_vcond): Likewise.
4482 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4484 * machmode.h (opt_machine_mode): New type.
4485 (opt_mode<T>): Allow construction from anything that can be
4487 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
4488 (mode_for_size): Return an opt_machine_mode.
4489 * stor-layout.c (mode_for_size): Likewise.
4490 (mode_for_size_tree): Update call accordingly.
4491 (bitwise_mode_for_mode): Likewise.
4492 (make_fract_type): Likewise.
4493 (make_accum_type): Likewise.
4494 * caller-save.c (replace_reg_with_saved_mem): Update call
4496 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4497 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4498 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4499 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4500 * expmed.c (extract_bit_field_1): Likewise.
4501 * reload.c (get_secondary_mem): Likewise.
4502 * varasm.c (assemble_integer): Likewise.
4503 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
4506 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4508 * machmode.h (decimal_float_mode_for_size): New function.
4509 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
4510 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
4511 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
4512 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
4513 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
4514 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
4516 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4518 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
4519 (get_builtin_sync_mode): Likewise.
4520 (expand_ifn_atomic_compare_exchange): Likewise.
4521 (expand_builtin_atomic_clear): Likewise.
4522 (expand_builtin_atomic_test_and_set): Likewise.
4523 (fold_builtin_atomic_always_lock_free): Likewise.
4524 * calls.c (compute_argument_addresses): Likewise.
4525 (emit_library_call_value_1): Likewise.
4526 (store_one_arg): Likewise.
4527 * combine.c (combine_instructions): Likewise.
4528 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
4529 * config/arm/arm.c (arm_function_value): Likewise.
4530 (aapcs_allocate_return_reg): Likewise.
4531 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
4532 * config/i386/i386.c (construct_container): Likewise.
4533 (ix86_gimplify_va_arg): Likewise.
4534 (ix86_expand_sse_cmp): Likewise.
4535 (emit_memmov): Likewise.
4536 (emit_memset): Likewise.
4537 (expand_small_movmem_or_setmem): Likewise.
4538 (ix86_expand_pextr): Likewise.
4539 (ix86_expand_pinsr): Likewise.
4540 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
4541 * config/microblaze/microblaze.c (microblaze_block_move_straight):
4543 * config/mips/mips.c (mips_function_value_1) Likewise.
4544 (mips_block_move_straight): Likewise.
4545 (mips_expand_ins_as_unaligned_store): Likewise.
4546 * config/powerpcspe/powerpcspe.c
4547 (rs6000_darwin64_record_arg_advance_flush): Likewise.
4548 (rs6000_darwin64_record_arg_flush): Likewise.
4549 * config/rs6000/rs6000.c
4550 (rs6000_darwin64_record_arg_advance_flush): Likewise.
4551 (rs6000_darwin64_record_arg_flush): Likewise.
4552 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
4553 (sparc_function_value_1): Likewise.
4554 * config/spu/spu.c (adjust_operand): Likewise.
4555 (spu_emit_branch_or_set): Likewise.
4556 (arith_immediate_p): Likewise.
4557 * emit-rtl.c (gen_lowpart_common): Likewise.
4558 * expr.c (expand_expr_real_1): Likewise.
4559 * function.c (assign_parm_setup_block): Likewise.
4560 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
4561 * reload1.c (alter_reg): Likewise.
4562 * stor-layout.c (mode_for_vector): Likewise.
4563 (layout_type): Likewise.
4565 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4567 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
4568 (spu_convert_move): Likewise.
4569 * lower-subreg.c (resolve_simple_move): Likewise.
4571 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4574 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
4575 define_insn to a define_expand.
4576 (altivec_vsum2sws_direct): New define_insn.
4577 (altivec_vsumsws): Convert from a define_insn to a define_expand.
4579 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
4581 * config/arm/arm.c (arm_option_params_internal): Improve setting of
4584 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
4589 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
4590 realign stack if stack alignment needed is less than incoming
4593 2017-09-05 Marek Polacek <polacek@redhat.com>
4596 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
4599 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
4601 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
4603 2017-09-05 Richard Biener <rguenther@suse.de>
4605 PR tree-optimization/82084
4606 * fold-const.c (can_native_encode_string_p): Handle wide characters.
4608 2017-09-05 Richard Biener <rguenther@suse.de>
4610 PR tree-optimization/82102
4611 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
4613 2017-09-05 Martin Liska <mliska@suse.cz>
4615 PR tree-optimization/82032
4616 * tree-cfg.c (generate_range_test): New function.
4617 * tree-cfg.h (generate_range_test): Declared here.
4618 * tree-cfgcleanup.c (convert_single_case_switch): New function.
4619 (cleanup_control_expr_graph): Use it.
4620 * tree-switch-conversion.c (try_switch_expansion): Remove
4622 (emit_case_nodes): Use generate_range_test.
4624 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
4627 * config/i386/i386.md (*<btsc><mode>_mask): Add
4628 TARGET_USE_BT to insn constraint.
4629 (*btr<mode>_mask): Ditto.
4631 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
4633 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
4634 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
4636 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
4639 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
4640 TARGET_NEON and TARGET_IWMMXT.
4641 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
4642 TARGET_NEON and TARGET_IWMMXT.
4643 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
4645 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
4647 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
4648 (ix86_rewrite_tls_address): Ditto.
4649 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
4650 (ix86_rewrite_tls_address_1): Ditto.
4651 (ix86_rewrite_tls_address): Ditto.
4652 * config/i386/predicates.md (tls_address_pattern): New predicate.
4653 * config/i386/i386.md (TLS address splitter): New splitter.
4655 2017-09-04 Richard Biener <rguenther@suse.de>
4657 PR tree-optimization/82084
4658 * fold-const.h (can_native_encode_string_p): Declare.
4659 * fold-const.c (can_native_encode_string_p): Factor out from ...
4660 (native_encode_string): ... here.
4661 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
4662 vectorizing stores from constants we later cannot handle.
4664 2017-09-04 Marek Polacek <polacek@redhat.com>
4667 * doc/invoke.texi: Update -Wtautological-compare documentation.
4669 2017-09-04 Jeff Law <law@redhat.com>
4671 PR tree-optimization/64910
4672 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
4673 swap the first and last operand if the last is a constant.
4675 2017-09-04 Marek Polacek <polacek@redhat.com>
4678 * convert.c (do_narrow): When sanitizing signed integer overflows,
4679 bail out for signed types.
4680 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
4682 2017-09-04 Richard Biener <rguenther@suse.de>
4684 PR tree-optimization/82060
4685 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
4686 Move devirtualization after stmt folding and before EH/AB/noreturn
4687 cleanup to get the stmt refs canonicalized. Use a bool instead
4688 of gimple_modified_p since that doesn't work for NOPs. Schedule
4689 NOPs generated by folding for removal.
4691 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4692 Alan Hayward <alan.hayward@arm.com>
4693 David Sherwood <david.sherwood@arm.com>
4695 * coretypes.h (pad_direction): New enum.
4696 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
4697 (FUNCTION_ARG_PADDING): Likewise.
4698 * target.def (function_arg_padding): New hook.
4699 * targhooks.h (default_function_arg_padding): Declare.
4700 * targhooks.c (default_function_arg_padding): New function.
4701 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
4702 (TARGET_FUNCTION_ARG_PADDING): ...this.
4703 * doc/tm.texi: Regenerate.
4704 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
4705 instead of direction.
4706 (compute_argument_addresses): Likewise.
4707 (load_register_parameters): Likewise.
4708 (emit_library_call_value_1): Likewise.
4709 (store_one_arg): Use targetm.calls.function_arg_padding instead
4710 of FUNCTION_ARG_PADDING.
4711 (must_pass_in_stack_var_size_or_pad): Likewise.
4712 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
4713 (emit_group_store): Likewise.
4714 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
4715 instead of FUNCTION_ARG_PADDING.
4716 (emit_push_insn): Likewise, and propagate enum change throughout
4718 * function.h (direction): Delete.
4719 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
4721 * function.c (assign_parm_find_stack_rtl): Likewise.
4722 (assign_parm_setup_block_p): Likewise.
4723 (assign_parm_setup_block): Likewise.
4724 (gimplify_parameters): Likewise.
4725 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
4726 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
4728 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
4729 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
4730 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
4731 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
4732 (aarch64_function_arg_padding): ...this new function.
4733 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
4734 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4735 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
4736 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
4737 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
4738 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4739 (arm_pad_arg_upward): Replace with...
4740 (arm_function_arg_padding): ...this new function.
4741 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
4743 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
4744 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
4745 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4746 (ia64_hpux_function_arg_padding): Replace with...
4747 (ia64_function_arg_padding): ...this new function. Use pad_direction
4748 instead of direction. Check for TARGET_HPUX.
4749 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
4750 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4751 (iq2000_function_arg_padding): New function.
4752 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
4753 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
4754 (mips_function_arg_padding): ...this new function.
4755 (mips_pad_reg_upward): Update accordingly.
4756 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4757 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
4758 targetm.calls.function_arg_padding.
4759 (FUNCTION_ARG_PADDING): Delete.
4760 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
4761 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
4762 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
4763 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
4764 (nios2_block_reg_padding): Return pad_direction instead of direction.
4765 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
4766 instead of direction.
4767 (nios2_function_arg_padding): Likewise. Make static.
4768 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4769 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
4770 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
4771 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
4772 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4773 (pa_function_arg_padding): Make static. Return pad_direction instead
4775 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
4776 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
4777 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
4778 instead of direction. Use targetm.calls.function_arg_padding.
4779 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
4780 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
4781 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
4782 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
4783 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
4785 (function_arg_padding): Rename to...
4786 (rs6000_function_arg_padding): ...this. Make static. Return
4787 pad_direction instead of direction.
4788 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
4789 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
4790 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
4791 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
4792 instead of direction. Use targetm.calls.function_arg_padding.
4793 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
4794 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
4795 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
4796 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
4797 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4798 (function_arg_padding): Rename to...
4799 (rs6000_function_arg_padding): ...this. Make static. Return
4800 pad_direction instead of direction.
4801 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
4802 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
4803 * config/s390/s390.c (s390_function_arg_padding): New function.
4804 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4805 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
4806 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
4807 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4808 (function_arg_padding): Rename to...
4809 (sparc_function_arg_padding): ...this. Make static. Return
4810 pad_direction instead of direction.
4811 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
4812 * config/spu/spu.c (spu_function_arg_padding): New function.
4813 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4814 * system.h (FUNCTION_ARG_PADDING): Poison.
4816 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4817 Alan Hayward <alan.hayward@arm.com>
4818 David Sherwood <david.sherwood@arm.com>
4820 * target.def (modes_tieable_p): New hook.
4821 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
4822 (TARGET_MODES_TIEABLE_P): ...this.
4823 * doc/tm.texi.in: Regenerate.
4824 * hooks.h (hook_bool_mode_mode_true): Declare.
4825 * hooks.c (hook_bool_mode_mode_true): New function.
4826 * combine.c (subst): Use targetm.modes_tieable_p instead of
4828 * dse.c (find_shift_sequence): Likewise.
4829 * expmed.c (extract_low_bits): Likewise.
4830 * lower-subreg.c: Include target.h.
4831 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
4833 * rtlanal.c (rtx_cost): Likewise.
4834 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
4835 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
4836 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
4837 (TARGET_MODES_TIEABLE_P): Redefine.
4838 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
4839 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
4840 (TARGET_MODES_TIEABLE_P): Redefine.
4841 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
4842 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
4843 (arc_modes_tieable_p): New function.
4844 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
4845 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
4846 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
4847 (arm_modes_tieable_p): Make static.
4848 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
4849 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
4850 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
4851 (TARGET_MODES_TIEABLE_P): Redefine.
4852 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
4853 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
4854 (TARGET_MODES_TIEABLE_P): Redefine.
4855 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
4856 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
4857 (cr16_modes_tieable_p): New function.
4858 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
4859 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
4860 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
4861 (TRULY_NOOP_TRUNCATION): Update comment.
4862 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
4863 (TRULY_NOOP_TRUNCATION): Update comment.
4864 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
4865 (frv_modes_tieable_p): New function.
4866 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
4867 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
4868 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
4869 (TARGET_MODES_TIEABLE_P): Redefine.
4870 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
4871 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
4872 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
4873 (TARGET_MODES_TIEABLE_P): Redefine.
4874 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
4875 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
4876 (ia64_modes_tieable_p): New function.
4877 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
4878 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
4879 (iq2000_modes_tieable_p): New function.
4880 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
4881 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
4882 (lm32_modes_tieable_p): New function.
4883 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
4884 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
4885 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
4886 (TARGET_MODES_TIEABLE_P): Redefine.
4887 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
4888 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
4889 (m32r_modes_tieable_p): New function.
4890 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
4891 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
4892 (m68k_modes_tieable_p): New function.
4893 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
4894 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
4895 (mcore_modes_tieable_p): New function.
4896 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
4897 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
4899 (TARGET_MODES_TIEABLE_P): Redefine.
4900 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
4901 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
4902 * config/mips/mips.c (mips_modes_tieable_p): Make static.
4903 (TARGET_MODES_TIEABLE_P): Redefine.
4904 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
4905 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
4906 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
4907 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
4908 (mn10300_modes_tieable_p): ...this and make static.
4909 (TARGET_MODES_TIEABLE_P): Redefine.
4910 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
4911 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
4912 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
4913 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
4914 (msp430_modes_tieable_p): Make static.
4915 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
4916 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
4917 (TARGET_MODES_TIEABLE_P): Redefine.
4918 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
4919 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
4920 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
4921 (TARGET_MODES_TIEABLE_P): Redefine.
4922 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
4923 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
4924 * config/pa/pa.c (pa_modes_tieable_p): Make static.
4925 (TARGET_MODES_TIEABLE_P): Redefine.
4926 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
4927 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
4928 (pdp11_modes_tieable_p): New function.
4929 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
4930 * config/powerpcspe/powerpcspe.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/powerpcspe/powerpcspe.md: Update comment.
4934 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
4935 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
4936 (TARGET_MODES_TIEABLE_P): Redefine.
4937 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
4938 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
4939 (rl78_modes_tieable_p): New function.
4940 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
4941 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
4942 (rs6000_modes_tieable_p): New function.
4943 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
4944 * config/rs6000/rs6000.md: Update comment.
4945 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
4946 * config/rx/rx.c (rx_modes_tieable_p): New function.
4947 (TARGET_MODES_TIEABLE_P): Redefine.
4948 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
4949 * config/s390/s390.c (s390_modes_tieable_p): New function.
4950 (TARGET_MODES_TIEABLE_P): Redefine.
4951 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
4952 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
4953 (sh_modes_tieable_p): New function.
4954 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
4955 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
4956 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
4957 (sparc_modes_tieable_p): Make static.
4958 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
4959 * config/spu/spu.c (spu_modes_tieable_p): New function.
4960 (TARGET_MODES_TIEABLE_P): Redefine.
4961 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
4962 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
4963 (TARGET_MODES_TIEABLE_P): Redefine.
4964 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
4965 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
4966 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
4967 * config/v850/v850.c (v850_modes_tieable_p): New function.
4968 (TARGET_MODES_TIEABLE_P): Redefine.
4969 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
4970 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
4971 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
4972 (visium_modes_tieable_p): New function.
4973 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
4974 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
4975 (xtensa_modes_tieable_p): New function.
4976 * system.h (MODES_TIEABLE_P): Poison.
4978 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4979 Alan Hayward <alan.hayward@arm.com>
4980 David Sherwood <david.sherwood@arm.com>
4982 * target.def (hard_regno_mode_ok): New hook.
4983 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
4984 (TARGET_HARD_REGNO_MODE_OK): ...this.
4985 * doc/tm.texi.in: Regenerate.
4986 * hooks.h (hook_bool_uint_mode_true): Declare.
4987 * hooks.c (hook_bool_uint_mode_true): New function.
4988 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
4990 * genpreds.c (write_insn_preds_c): Add an include of target.h.
4991 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
4992 instead of HARD_REGNO_MODE_OK.
4993 * caller-save.c: Include target.h.
4994 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
4996 * combine.c (can_combine_p): Likewise.
4997 (combinable_i3pat): Likewise.
4998 (can_change_dest_mode): Likewise.
4999 * expr.c (init_expr_target): Likewise.
5000 (convert_move): Likewise.
5001 (convert_modes): Likewise.
5002 * ira.c (setup_prohibited_class_mode_regs): Likewise.
5003 (setup_prohibited_mode_move_regs): Likewise.
5004 * ira.h (target_ira): Likewise.
5005 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5006 * lra-constraints.c (process_alt_operands): Likewise.
5007 (split_reg): Likewise.
5008 * recog.c (peep2_find_free_register): Likewise.
5009 * ree.c (combine_reaching_defs): Likewise.
5010 * regcprop.c (maybe_mode_change): Likewise.
5011 * reginfo.c (init_reg_sets_1): Likewise.
5012 (choose_hard_reg_mode): Likewise.
5013 (simplifiable_subregs): Likewise.
5014 * regrename.c (check_new_reg_p): Likewise.
5015 * reload.c (find_valid_class): Likewise.
5016 (find_valid_class_1): Likewise.
5017 (reload_inner_reg_of_subreg): Likewise.
5018 (push_reload): Likewise.
5019 (combine_reloads): Likewise.
5020 (find_dummy_reload): Likewise.
5021 (find_reloads): Likewise.
5022 * reload1.c (find_reg): Likewise.
5023 (set_reload_reg): Likewise.
5024 (allocate_reload_reg): Likewise.
5025 (choose_reload_regs): Likewise.
5026 (reload_adjust_reg_for_temp): Likewise.
5027 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
5028 (simplify_subreg_regno): Likewise.
5029 * sel-sched.c (init_regs_for_mode): Likewise.
5030 * varasm.c (make_decl_rtl): Likewise.
5031 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
5032 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
5034 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
5035 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
5036 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5037 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
5038 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
5039 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5040 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
5041 (arc_mode_class): Delete.
5042 (HARD_REGNO_MODE_OK): Delete.
5043 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5044 (arc_hard_regno_mode_ok): Rename old array to...
5045 (arc_hard_regno_mode_ok_modes): ...this.
5046 (arc_conditional_register_usage): Update accordingly.
5047 (arc_mode_class): Make static.
5048 (arc_hard_regno_mode_ok): New function.
5049 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
5050 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
5051 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5052 (arm_hard_regno_mode_ok): Make static.
5053 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
5055 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
5056 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
5057 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
5059 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5060 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
5061 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
5062 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
5063 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
5064 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5065 * config/bfin/predicates.md (valid_reg_operand): Use
5066 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
5067 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
5068 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
5069 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5070 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
5071 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
5072 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5073 (cr16_hard_regno_mode_ok): Make static and return a bool.
5074 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
5075 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5076 (cris_hard_regno_mode_ok): New function.
5077 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
5078 (epiphany_mode_class): Delete.
5079 (HARD_REGNO_MODE_OK): Delete.
5080 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
5081 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5082 (hard_regno_mode_ok): Rename to...
5083 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
5084 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
5085 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
5087 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
5088 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
5089 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5090 (frv_hard_regno_mode_ok): Make static and return a bool.
5091 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
5093 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
5094 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
5095 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
5096 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
5098 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5099 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
5100 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
5101 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
5103 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5104 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
5105 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5106 (ia64_hard_regno_mode_ok): New function.
5107 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
5108 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5109 (iq2000_hard_regno_mode_ok): New function.
5110 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
5111 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5112 (lm32_hard_regno_mode_ok): New function.
5113 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
5114 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
5115 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
5116 instead of HARD_REGNO_MODE_OK.
5117 (m32c_hard_regno_ok): Rename to...
5118 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
5119 (m32c_cannot_change_mode_class): Update accordingly.
5120 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5121 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
5122 (m32r_mode_class): Delete.
5123 (HARD_REGNO_MODE_OK): Delete.
5124 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5125 (m32r_hard_regno_mode_ok): Rename to...
5126 (m32r_hard_regno_modes): ...this.
5127 (m32r_mode_class): Make static.
5128 (m32r_hard_regno_mode_ok): New function.
5129 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
5130 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
5131 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5132 (m68k_hard_regno_mode_ok): Make static.
5133 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
5134 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5135 (mcore_hard_regno_mode_ok): New function.
5136 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
5137 (HARD_REGNO_MODE_OK): Delete.
5138 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
5140 (microblaze_hard_regno_mode_ok_p): ...this and make static.
5141 (microblaze_hard_regno_mode_ok): New function.
5142 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5143 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
5144 (mips_hard_regno_mode_ok): Delete.
5145 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
5146 (mips_hard_regno_mode_ok_p): ...this and make static.
5147 (mips_hard_regno_mode_ok_p): Rename to...
5148 (mips_hard_regno_mode_ok_uncached): ...this.
5149 (mips_hard_regno_mode_ok): New function.
5150 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
5151 of HARD_REGNO_MODE_OK.
5152 (mips_option_override): Update after above name changes.
5153 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5154 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
5155 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
5156 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
5157 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
5158 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5159 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
5160 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
5161 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
5162 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5163 (msp430_hard_regno_mode_ok): Make static and return a bool.
5164 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
5165 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
5166 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
5168 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5169 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
5170 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
5171 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
5172 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
5173 (PA_HARD_REGNO_MODE_OK): ...this
5174 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
5175 (PA_HARD_REGNO_MODE_OK): ...this.
5176 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5177 (pa_hard_regno_mode_ok): New function.
5178 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
5179 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5180 (pdp11_hard_regno_mode_ok): New function.
5181 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
5182 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
5184 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
5186 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5187 (rs6000_hard_regno_mode_ok): Rename to...
5188 (rs6000_hard_regno_mode_ok_uncached): ...this.
5189 (rs6000_init_hard_regno_mode_ok): Update accordingly.
5190 (rs6000_hard_regno_mode_ok): New function.
5191 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
5192 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
5193 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
5194 (riscv_hard_regno_mode_ok): ...this and make static.
5195 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5196 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
5197 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
5198 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5199 (rl78_hard_regno_mode_ok): Make static and return bool.
5200 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
5201 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
5203 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
5204 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5205 (rs6000_hard_regno_mode_ok): Rename to...
5206 (rs6000_hard_regno_mode_ok_uncached): ...this.
5207 (rs6000_init_hard_regno_mode_ok): Update accordingly.
5208 (rs6000_hard_regno_mode_ok): New function.
5209 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
5210 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
5211 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5212 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
5213 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
5214 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
5215 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5216 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
5217 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
5218 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5219 (sh_hard_regno_mode_ok): Make static.
5220 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
5221 instead of HARD_REGNO_MODE_OK.
5222 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
5223 (sparc_mode_class): Delete.
5224 (HARD_REGNO_MODE_OK): Delete.
5225 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5226 (hard_regno_mode_classes): Make static.
5227 (sparc_mode_class): Likewise.
5228 (sparc_hard_regno_mode_ok): New function.
5229 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
5230 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
5231 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
5233 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5234 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
5235 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
5236 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
5237 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
5238 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5239 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
5240 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
5241 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5242 (visium_hard_regno_mode_ok): New function.
5243 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
5244 instead of HARD_REGNO_MODE_OK.
5245 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
5246 (HARD_REGNO_MODE_OK): Delete.
5247 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
5248 (xtensa_hard_regno_mode_ok_p): ...this and make static.
5249 (xtensa_option_override): Update accordingly.
5250 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5251 (xtensa_hard_regno_mode_ok): New function.
5252 * system.h (HARD_REGNO_MODE_OK): Poison.
5254 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5255 Alan Hayward <alan.hayward@arm.com>
5256 David Sherwood <david.sherwood@arm.com>
5258 * target.def (hard_regno_call_part_clobbered): New hook.
5259 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
5260 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
5261 * doc/tm.texi: Regenerate.
5262 * hooks.h (hook_bool_uint_mode_false): Declare.
5263 * hooks.c (hook_bool_uint_mode_false): New function.
5264 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5265 * cselib.c (cselib_process_insn): Use
5266 targetm.hard_regno_call_part_clobbered instead of
5267 HARD_REGNO_CALL_PART_CLOBBERED.
5268 * ira-conflicts.c (ira_build_conflicts): Likewise.
5269 * ira-costs.c (ira_tune_allocno_costs): Likewise.
5270 * lra-constraints.c (need_for_call_save_p): Likewise.
5271 * lra-lives.c: Include target.h.
5272 (check_pseudos_live_through_calls): Use
5273 targetm.hard_regno_call_part_clobbered instead of
5274 HARD_REGNO_CALL_PART_CLOBBERED.
5275 * regcprop.c: Include target.h.
5276 (copyprop_hardreg_forward_1): Use
5277 targetm.hard_regno_call_part_clobbered instead of
5278 HARD_REGNO_CALL_PART_CLOBBERED.
5279 * reginfo.c (choose_hard_reg_mode): Likewise.
5280 * regrename.c (check_new_reg_p): Likewise.
5281 * reload.c (find_equiv_reg): Likewise.
5282 * reload1.c (emit_reload_insns): Likewise.
5283 * sched-deps.c (deps_analyze_insn): Likewise.
5284 * sel-sched.c (init_regs_for_mode): Likewise.
5285 (mark_unavailable_hard_regs): Likewise.
5286 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
5287 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5288 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
5290 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5291 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5292 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
5294 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
5296 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5297 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5298 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
5300 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5301 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5302 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
5304 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5305 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
5307 * config/powerpcspe/powerpcspe.c
5308 (rs6000_hard_regno_call_part_clobbered): New function.
5309 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5310 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5311 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
5313 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5314 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5315 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
5317 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5318 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5319 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
5321 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5322 Alan Hayward <alan.hayward@arm.com>
5323 David Sherwood <david.sherwood@arm.com>
5325 * rtl.h (subreg_memory_offset): Declare.
5326 * emit-rtl.c (subreg_memory_offset): New function.
5327 * expmed.c (store_bit_field_1): Use it.
5328 * expr.c (undefined_operand_subword_p): Likewise.
5329 * simplify-rtx.c (simplify_subreg): Likewise.
5331 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
5333 PR rtl-optimization/57448
5336 * optabs.c (expand_atomic_load): Place compiler memory barriers if
5337 using atomic_load pattern.
5338 (expand_atomic_store): Likewise.
5340 2017-09-04 Jakub Jelinek <jakub@redhat.com>
5343 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
5344 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
5345 handling. Use replace_call_with_value with NULL instead of
5346 gsi_replace, unlink_stmt_vdef and release_defs.
5348 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
5351 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
5353 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5356 * rtl.h (emit_library_call_value_1): Declare.
5357 (emit_library_call): Replace declaration with a series of overloads.
5358 Remove the parameter count argument.
5359 (emit_library_call_value): Likewise.
5360 * calls.c (emit_library_call_value_1): Make global. Replace varargs
5361 with an "rtx_mode_t *".
5362 (emit_library_call_value): Delete.
5363 (emit_library_call): Likewise.
5364 * asan.c (asan_emit_stack_protection): Update calls accordingly.
5365 (asan_emit_allocas_unpoison): Likewise.
5366 * builtins.c (expand_builtin_powi): Likewise.
5367 (expand_asan_emit_allocas_unpoison): Likewise.
5368 * cfgexpand.c (expand_main_function): Likewise.
5369 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
5370 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
5371 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
5372 * config/arm/arm.c (arm_trampoline_init): Likewise.
5373 (arm_call_tls_get_addr): Likewise.
5374 (arm_expand_divmod_libfunc): Likewise.
5375 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
5376 (smulsi3_highpart): Likewise.
5377 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
5378 (c6x_expand_compare): Likewise.
5379 (c6x_expand_movmem): Likewise.
5380 * config/frv/frv.c (frv_trampoline_init): Likewise.
5381 * config/i386/i386.c (ix86_trampoline_init): Likewise.
5382 (ix86_expand_divmod_libfunc): Likewise.
5383 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
5384 (ia64_expand_compare): Likewise.
5385 (ia64_profile_hook): Likewise.
5386 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
5387 (nonlocal_goto): Likewise.
5388 (restore_stack_nonlocal): Likewise.
5389 * config/m32r/m32r.c (block_move_call): Likewise.
5390 (m32r_trampoline_init): Likewise.
5391 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
5392 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
5393 (m68k_call_m68k_read_tp): Likewise.
5394 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
5395 (microblaze_expand_divide): Likewise.
5396 * config/mips/mips.h (mips_args): Likewise.
5397 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
5398 (MIPS_ICACHE_SYNC): Likewise.
5399 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
5400 (nios2_trampoline_init): Likewise.
5401 * config/pa/pa.c (hppa_tls_call): Likewise.
5402 (pa_trampoline_init): Likewise.
5403 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
5404 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
5405 (expand_strn_compare): Likewise.
5406 (rs6000_generate_compare): Likewise.
5407 (rs6000_expand_float128_convert): Likewise.
5408 (output_profile_hook): Likewise.
5409 (rs6000_trampoline_init): Likewise.
5410 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
5411 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
5412 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
5413 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
5414 (rs6000_generate_compare): Likewise.
5415 (rs6000_expand_float128_convert): Likewise.
5416 (output_profile_hook): Likewise.
5417 (rs6000_trampoline_init): Likewise.
5418 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
5419 * config/sh/sh.c (sh_trampoline_init): Likewise.
5420 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
5421 (sparc_emit_float_lib_cmp): Likewise.
5422 (sparc32_initialize_trampoline): Likewise.
5423 (sparc64_initialize_trampoline): Likewise.
5424 (sparc_profile_hook): Likewise.
5425 * config/spu/spu.c (ea_load_store): Likewise.
5426 * config/spu/spu.md (floatunssidf2): Likewise.
5427 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
5428 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
5429 * config/visium/visium.c (expand_block_move_4): Likewise.
5430 (expand_block_move_2): Likewise.
5431 (expand_block_move_1): Likewise.
5432 (expand_block_set_4): Likewise.
5433 (expand_block_set_2): Likewise.
5434 (expand_block_set_1): Likewise.
5435 (visium_trampoline_init): Likewise.
5436 (visium_profile_hook): Likewise.
5437 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
5438 (xtensa_setup_frame_addresses): Likewise.
5439 (xtensa_trampoline_init): Likewise.
5440 * except.c (sjlj_emit_function_enter): Likewise.
5441 (sjlj_emit_function_exit): Likewise.
5442 * explow.c (allocate_dynamic_stack_space): Likewise.
5443 (probe_stack_range): Likewise.
5444 * expr.c (convert_mode_scalar): Likewise.
5445 * optabs.c (expand_binop): Likewise.
5446 (expand_twoval_binop_libfunc): Likewise.
5447 (expand_unop): Likewise.
5448 (prepare_cmp_insn): Likewise.
5449 (prepare_float_lib_cmp): Likewise.
5450 (expand_float): Likewise.
5451 (expand_fix): Likewise.
5452 (expand_fixed_convert): Likewise.
5453 (maybe_emit_sync_lock_test_and_set): Likewise.
5454 (expand_atomic_compare_and_swap): Likewise.
5455 (expand_mem_thread_fence): Likewise.
5456 (expand_atomic_fetch_op): Likewise.
5458 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
5460 * doc/generic.texi (OpenACC): Adjust URL.
5461 * doc/invoke.texi (C Dialect Options): Ditto.
5463 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
5465 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
5466 predicate for operand 1. Add (m,<S>) constraint.
5467 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
5468 Prevent memory operand 1 with register operand 2.
5470 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
5472 PR rtl-optimization/82024
5473 * combine.c (try_combine): If the combination result is a PARALLEL,
5474 and we only need to retain the SET in there that would be placed
5475 at I2, check that we can place that at I3 instead, before doing so.
5477 2017-09-01 Jakub Jelinek <jakub@redhat.com>
5480 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
5482 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
5483 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
5486 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
5487 Jeff Law <law@redhat.com>
5489 * varasm.c (bss_initializer_p): Do not put constants into .bss
5490 (categorize_decl_for_section): Handle bss_initializer_p returning
5491 false when DECL_INITIAL is NULL.
5493 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5496 * config/s390/s390.c (s390_can_inline_p): New function.
5498 2017-09-01 Jeff Law <law@redhat.com>
5500 PR tree-optimization/82052
5501 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
5502 Always initialize the returned slot after a hash table miss
5503 when INSERT is true.
5505 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
5507 * config/s390/s390.md (mem_signal_fence): Remove.
5508 * doc/md.texi (mem_signal_fence): Remove.
5509 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
5511 * target-insns.def (mem_signal_fence): Remove.
5513 2017-09-01 Jakub Jelinek <jakub@redhat.com>
5516 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
5519 * asan.c (create_odr_indicator): Strip name encoding from assembler
5520 name before appending it after __odr_asan_.
5522 2017-09-01 Martin Liska <mliska@suse.cz>
5524 PR tree-optimization/82059
5525 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
5526 frequency only when an edge is redirected.
5528 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5530 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
5531 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
5532 (arc_conditional_register_usage): Remove ARC600 lp_count
5534 (arc_file_start): Emit Tag_ARC_CPU_variation.
5535 (arc_can_use_doloop_p): New conditions to use ZOLs.
5536 (hwloop_fail): New function.
5537 (hwloop_optimize): Likewise.
5538 (hwloop_pattern_reg): Likewise.
5539 (arc_doloop_hooks): New struct, to be used with reorg_loops.
5540 (arc_reorg_loops): New function, calls reorg_loops.
5541 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
5542 (arc600_corereg_hazard): Remove ZOL checking, case handled by
5544 (arc_loop_hazard): Remove function, functionality moved into
5546 (arc_hazard): Remove arc_loop_hazard call.
5547 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
5548 into hwloop_optimize.
5549 (arc_label_align): Remove ZOL handling.
5550 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
5551 * config/arc/arc.md (doloop_begin): Remove pattern.
5552 (doloop_begin_i): Likewise.
5553 (doloop_end_i): Likewise.
5554 (doloop_fallback): Likewise.
5555 (doloop_fallback_m): Likewise.
5556 (doloop_end): Reimplement expand.
5557 (arc_lp): New pattern for LP instruction.
5558 (loop_end): New pattern.
5559 (loop_fail): Likewise.
5560 (decrement_and_branch_until_zero): Likewise.
5561 * config/arc/arc.opt (mlpc-width): New option.
5562 * doc/invoke.texi (mlpc-width): Document option.
5564 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5566 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
5567 (arc_ccfsm_advance): Fix checking for delay slots.
5568 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
5570 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5572 * config/arc/arc.md (movqi_insn): Add stores to save constant long
5574 (movhi_insn): Update store instruction constraint which are saving
5575 6-bit short immediates.
5576 (movsi_insn): Consider also short scaled load operations.
5577 (zero_extendhisi2_i): Use Usd constraint instead of T.
5578 (extendhisi2_i): Add q constraint.
5579 (arc_clzsi2): Add type and length attributes.
5580 (arc_ctzsi2): Likewise.
5581 * config/arc/constraints.md (Usc): Update constraint, the
5582 assembler can parse two relocations for a single instruction.
5584 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5586 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
5587 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
5589 2017-08-31 Olivier Hainque <hainque@adacore.com>
5591 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
5592 match as powerpc-wrs-vxworks*.
5594 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
5596 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
5597 register constraint for by-element operand.
5598 (aarch64_mls_elt_merge<mode>): Likewise.
5600 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
5602 * config/arc/arc.c (arc_can_follow_jump): Check for short
5605 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
5607 * config.gcc: Use g.opt for arc.
5608 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
5609 functionality moved to ...
5610 (legitimate_scaled_address_p): New function, ...here.
5611 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
5612 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
5613 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
5615 (arc_override_options): Handle G option.
5616 (arc_output_pic_addr_const): Correct function definition.
5617 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
5618 (arc_decl_anon_ns_mem_p): Delete.
5619 (arc_in_small_data_p): Overhaul this function to take into
5620 consideration the value given via G option.
5621 (arc_rewrite_small_data_1): Renamed and corrected old
5622 arc_rewrite_small_data function.
5623 (arc_rewrite_small_data): New function.
5624 (small_data_pattern): Don't use pic_offset_table_rtx.
5625 * config/arc/arc.h (CC1_SPEC): Recognize G option.
5626 * config/arc/simdext.md (movmisalignv2hi): Use
5627 prepare_move_operands function.
5629 (movmisalign*): Likewise.
5630 * doc/invoke.texi (ARC options): Document -G option.
5632 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
5634 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
5636 * config/arc/arc.c (arc_print_operand): Output scalled address for
5637 sdata whenever is possible.
5638 (arc_in_small_data_p): Allow sdata for 64bit datum when double
5639 load/stores are available.
5640 (compact_sda_memory_operand): Check for the alignment required by
5641 code density instructions.
5642 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
5644 * config/arc/constraints.md (Usd): Update constraint.
5645 (Us0): New constraint.
5646 (Usc): Update constraint.
5648 2017-08-31 Richard Biener <rguenther@suse.de>
5651 * dwarf2out.c (dwarf2out_early_global_decl): Process each
5654 2017-08-31 Tamar Christina <tamar.christina@arm.com>
5656 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
5657 Resize type_signature.
5659 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
5660 Alan Hayward <alan.hayward@arm.com>
5661 David Sherwood <david.sherwood@arm.com>
5663 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
5664 subregs whose inner modes can be stored in GPRs.
5665 (aarch64_classify_index): Likewise.
5667 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
5668 Alan Hayward <alan.hayward@arm.com>
5669 David Sherwood <david.sherwood@arm.com>
5671 * config/aarch64/iterators.md (V_cmp_result): Rename to...
5672 (V_INT_EQUIV): ...this.
5673 (v_cmp_result): Rename to...
5674 (v_int_equiv): ...this.
5675 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
5676 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
5677 (copysign<mode>3): Likewise.
5678 (aarch64_simd_bsl<mode>_internal): Likewise.
5679 (aarch64_simd_bsl<mode>): Likewise.
5680 (vec_cmp<mode><mode>): Likewise.
5681 (vcond<mode><mode>): Likewise.
5682 (vcond<v_cmp_mixed><mode>): Likewise.
5683 (vcondu<mode><v_cmp_mixed>): Likewise.
5684 (aarch64_cm<optab><mode>): Likewise.
5685 (aarch64_cmtst<mode>): Likewise.
5686 (aarch64_fac<optab><mode>): Likewise.
5687 (vec_perm_const<mode>): Likewise.
5688 (vcond_mask_<mode><v_cmp_result>): Rename to...
5689 (vcond_mask_<mode><v_int_equiv>): ...this.
5690 (vec_cmp<mode><v_cmp_result>): Rename to...
5691 (vec_cmp<mode><v_int_equiv>): ...this.
5693 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
5694 Alan Hayward <alan.hayward@arm.com>
5695 David Sherwood <david.sherwood@arm.com>
5697 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
5699 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
5700 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
5701 (UNSPEC_LD4_DREG): New unspecs.
5702 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
5703 (aarch64_ld2<mode>_dreg_be): Replace with...
5704 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
5706 (aarch64_ld3<mode>_dreg_le)
5707 (aarch64_ld3<mode>_dreg_be): Replace with...
5708 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
5710 (aarch64_ld4<mode>_dreg_le)
5711 (aarch64_ld4<mode>_dreg_be): Replace with...
5712 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
5715 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5717 PR tree-optimization/81987
5718 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
5719 insert an initializer in a location not dominated by the stride
5722 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
5724 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
5725 on the entire header of the finally block in the fallthru case.
5727 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
5729 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
5731 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
5733 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
5734 rs6000_emit_move_from_cr and call renamed function.
5735 (rs6000_emit_prologue): Call renamed functions.
5736 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
5737 movesi_from_cr, remove volatile CRs.
5739 2017-08-30 Jon Beniston <jon@beniston.com>
5740 Richard Biener <rguenther@suse.de>
5742 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
5743 of VECTOR_MODE_P check.
5744 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
5745 element vector types.
5747 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5749 * df.h (df_read_modify_subreg_p): Remove in favor of...
5750 * rtl.h (read_modify_subreg_p): ...this new function. Take a
5751 const_rtx instead of an rtx.
5752 * cprop.c (local_cprop_find_used_regs): Update accordingly.
5753 * df-problems.c (df_word_lr_mark_ref): Likewise.
5754 * ira-lives.c (mark_pseudo_reg_live): Likewise.
5755 (mark_pseudo_reg_dead): Likewise.
5756 (mark_ref_dead): Likewise.
5757 * reginfo.c (init_subregs_of_mode): Likewise.
5758 * sched-deps.c (sched_analyze_1): Likewise.
5759 * df-scan.c (df_def_record_1): Likewise.
5760 (df_uses_record): Likewise.
5761 (df_read_modify_subreg_p): Remove in favor of...
5762 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
5763 const_rtx instead of an rtx.
5765 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5766 Alan Hayward <alan.hayward@arm.com>
5767 David Sherwood <david.sherwood@arm.com>
5769 * rtl.h (partial_subreg_p): New function.
5770 * caller-save.c (save_call_clobbered_regs): Use it.
5771 * calls.c (expand_call): Likewise.
5772 * combine.c (combinable_i3pat): Likewise.
5773 (simplify_set): Likewise.
5774 (make_extraction): Likewise.
5775 (make_compound_operation_int): Likewise.
5776 (gen_lowpart_or_truncate): Likewise.
5777 (force_to_mode): Likewise.
5778 (make_field_assignment): Likewise.
5779 (reg_truncated_to_mode): Likewise.
5780 (record_truncated_value): Likewise.
5781 (move_deaths): Likewise.
5782 * cse.c (record_jump_cond): Likewise.
5783 (cse_insn): Likewise.
5784 * cselib.c (cselib_lookup_1): Likewise.
5785 * expmed.c (extract_bit_field_using_extv): Likewise.
5786 * function.c (assign_parm_setup_reg): Likewise.
5787 * ifcvt.c (noce_convert_multiple_sets): Likewise.
5788 * ira-build.c (create_insn_allocnos): Likewise.
5789 * lra-coalesce.c (merge_pseudos): Likewise.
5790 * lra-constraints.c (match_reload): Likewise.
5791 (simplify_operand_subreg): Likewise.
5792 (curr_insn_transform): Likewise.
5793 * lra-lives.c (process_bb_lives): Likewise.
5794 * lra.c (new_insn_reg): Likewise.
5795 (lra_substitute_pseudo): Likewise.
5796 * regcprop.c (mode_change_ok): Likewise.
5797 (maybe_mode_change): Likewise.
5798 (copyprop_hardreg_forward_1): Likewise.
5799 * reload.c (push_reload): Likewise.
5800 (find_reloads): Likewise.
5801 (find_reloads_subreg_address): Likewise.
5802 * reload1.c (alter_reg): Likewise.
5803 (eliminate_regs_1): Likewise.
5804 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5806 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
5808 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
5809 back to if statements, including unpack.
5811 2017-08-30 Martin Liska <mliska@suse.cz>
5814 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
5816 (func_checker::compare_asm_inputs_outputs): ... this function.
5817 (func_checker::compare_gimple_asm): Use the function to compare
5818 also ASM constrains.
5819 * ipa-icf-gimple.h: Rename the function.
5821 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5822 Alan Hayward <alan.hayward@arm.com>
5823 David Sherwood <david.sherwood@arm.com>
5825 * coretypes.h (complex_mode): New type.
5826 * gdbhooks.py (build_pretty_printer): Handle it.
5827 * machmode.h (complex_mode): New class.
5828 (complex_mode::includes_p): New function.
5829 (is_complex_int_mode): Likewise.
5830 (is_complex_float_mode): Likewise.
5831 * genmodes.c (get_mode_class): Handle complex mode classes.
5832 * function.c (expand_function_end): Use is_complex_int_mode.
5834 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5835 Alan Hayward <alan.hayward@arm.com>
5836 David Sherwood <david.sherwood@arm.com>
5838 * coretypes.h (scalar_mode_pod): New typedef.
5839 * gdbhooks.py (build_pretty_printer): Handle it.
5840 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
5841 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
5842 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
5843 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
5846 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5847 Alan Hayward <alan.hayward@arm.com>
5848 David Sherwood <david.sherwood@arm.com>
5850 * machmode.h (mode_for_vector): Take a scalar_mode instead
5852 * stor-layout.c (mode_for_vector): Likewise.
5853 * explow.c (promote_mode): Use as_a <scalar_mode>.
5854 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
5856 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5857 Alan Hayward <alan.hayward@arm.com>
5858 David Sherwood <david.sherwood@arm.com>
5860 * target.def (preferred_simd_mode): Take a scalar_mode
5861 instead of a machine_mode.
5862 * targhooks.h (default_preferred_simd_mode): Likewise.
5863 * targhooks.c (default_preferred_simd_mode): Likewise.
5864 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
5865 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
5866 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
5867 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
5868 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
5869 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
5870 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
5871 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
5873 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
5874 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
5875 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
5876 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
5877 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
5878 * doc/tm.texi: Regenerate.
5879 * optabs-query.c (can_vec_mask_load_store_p): Return false for
5882 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5883 Alan Hayward <alan.hayward@arm.com>
5884 David Sherwood <david.sherwood@arm.com>
5886 * target.def (scalar_mode_supported_p): Take a scalar_mode
5887 instead of a machine_mode.
5888 * targhooks.h (default_scalar_mode_supported_p): Likewise.
5889 * targhooks.c (default_scalar_mode_supported_p): Likewise.
5890 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
5891 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
5892 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
5893 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
5894 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
5895 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
5896 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
5897 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
5898 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
5899 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
5900 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
5901 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
5903 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
5904 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
5905 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
5906 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
5907 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
5909 * doc/tm.texi: Regenerate.
5911 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5912 Alan Hayward <alan.hayward@arm.com>
5913 David Sherwood <david.sherwood@arm.com>
5915 * coretypes.h (opt_scalar_mode): New typedef.
5916 * gdbhooks.py (build_pretty_printers): Handle it.
5917 * machmode.h (mode_iterator::get_2xwider): Add overload for
5919 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
5921 * expr.c (convert_mode_scalar): Likewise.
5922 * omp-low.c (omp_clause_aligned_alignment): Likewise.
5923 * optabs.c (expand_float): Likewise.
5924 (expand_fix): Likewise.
5925 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5927 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5928 Alan Hayward <alan.hayward@arm.com>
5929 David Sherwood <david.sherwood@arm.com>
5931 * optabs.c (expand_float): Explicitly check for scalars before
5932 using a branching expansion.
5933 (expand_fix): Likewise.
5935 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5936 Alan Hayward <alan.hayward@arm.com>
5937 David Sherwood <david.sherwood@arm.com>
5939 * expr.c (convert_mode): Split scalar handling out into...
5940 (convert_mode_scalar): ...this new function. Treat the modes
5943 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5944 Alan Hayward <alan.hayward@arm.com>
5945 David Sherwood <david.sherwood@arm.com>
5947 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
5949 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
5951 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5952 Alan Hayward <alan.hayward@arm.com>
5953 David Sherwood <david.sherwood@arm.com>
5955 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
5956 rather than a machine_mode.
5957 (fixed_from_string): Likewise.
5958 (fixed_convert): Likewise.
5959 (fixed_convert_from_int): Likewise.
5960 (fixed_convert_from_real): Likewise.
5961 (real_convert_from_fixed): Likewise.
5962 * fixed-value.c (fixed_from_double_int): Likewise.
5963 (fixed_from_string): Likewise.
5964 (fixed_convert): Likewise.
5965 (fixed_convert_from_int): Likewise.
5966 (fixed_convert_from_real): Likewise.
5967 (real_convert_from_fixed): Likewise.
5968 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
5970 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5971 Alan Hayward <alan.hayward@arm.com>
5972 David Sherwood <david.sherwood@arm.com>
5974 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
5975 of separate mode class checks. Do not allow vector modes here.
5976 (immed_wide_int_const): Use as_a <scalar_mode>.
5977 * explow.c (trunc_int_for_mode): Likewise.
5978 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
5979 (wi::shwi): Likewise.
5980 (wi::min_value): Likewise.
5981 (wi::max_value): Likewise.
5982 * dwarf2out.c (loc_descriptor): Likewise.
5983 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
5986 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5987 Alan Hayward <alan.hayward@arm.com>
5988 David Sherwood <david.sherwood@arm.com>
5990 * tree.h (SCALAR_TYPE_MODE): New macro.
5991 * expr.c (expand_expr_addr_expr_1): Use it.
5992 (expand_expr_real_2): Likewise.
5993 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
5994 (fold_convert_const_fixed_from_int): Likewise.
5995 (fold_convert_const_fixed_from_real): Likewise.
5996 (native_encode_fixed): Likewise
5997 (native_encode_complex): Likewise
5998 (native_encode_vector): Likewise.
5999 (native_interpret_fixed): Likewise.
6000 (native_interpret_real): Likewise.
6001 (native_interpret_complex): Likewise.
6002 (native_interpret_vector): Likewise.
6003 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
6004 (simd_clone_adjust_argument_types): Likewise.
6005 (simd_clone_init_simd_arrays): Likewise.
6006 (simd_clone_adjust): Likewise.
6007 * stor-layout.c (layout_type): Likewise.
6008 * tree.c (build_minus_one_cst): Likewise.
6009 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
6010 * tree-inline.c (estimate_move_cost): Likewise.
6011 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
6012 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
6013 (vectorizable_reduction): Likewise.
6014 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
6015 (vect_recog_mixed_size_cond_pattern): Likewise.
6016 (check_bool_pattern): Likewise.
6017 (adjust_bool_pattern): Likewise.
6018 (search_type_for_mask_1): Likewise.
6019 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
6020 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6021 (vectorizable_load): Likewise.
6022 (vectorizable_store): Likewise.
6023 * ubsan.c (ubsan_encode_value): Likewise.
6024 * varasm.c (output_constant): Likewise.
6026 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6027 Alan Hayward <alan.hayward@arm.com>
6028 David Sherwood <david.sherwood@arm.com>
6030 * coretypes.h (scalar_mode): New class.
6031 * machmode.h (scalar_mode): Likewise.
6032 (scalar_mode::includes_p): New function.
6033 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
6034 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
6035 * genmodes.c (get_mode_class): Handle remaining scalar modes.
6036 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
6037 * expmed.c (store_bit_field_1): Likewise.
6038 (extract_bit_field_1): Likewise.
6039 * expr.c (write_complex_part): Likewise.
6040 (read_complex_part): Likewise.
6041 (emit_move_complex_push): Likewise.
6042 (expand_expr_real_2): Likewise.
6043 * function.c (assign_parm_setup_reg): Likewise.
6044 (assign_parms_unsplit_complex): Likewise.
6045 * optabs.c (expand_binop): Likewise.
6046 * rtlanal.c (subreg_get_info): Likewise.
6047 * simplify-rtx.c (simplify_immed_subreg): Likewise.
6048 * varasm.c (output_constant_pool_2): Likewise.
6050 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6051 Alan Hayward <alan.hayward@arm.com>
6052 David Sherwood <david.sherwood@arm.com>
6054 * expmed.c (extract_high_half): Use scalar_int_mode and remove
6056 (expmed_mult_highpart_optab): Likewise.
6057 (expmed_mult_highpart): Likewise.
6059 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6060 Alan Hayward <alan.hayward@arm.com>
6061 David Sherwood <david.sherwood@arm.com>
6063 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
6064 instead of a machine_mode.
6065 (builtin_memset_read_str): Likewise.
6066 * builtins.c (c_readstr): Likewise.
6067 (builtin_memcpy_read_str): Likewise.
6068 (builtin_strncpy_read_str): Likewise.
6069 (builtin_memset_read_str): Likewise.
6070 (builtin_memset_gen_str): Likewise.
6071 (expand_builtin_signbit): Use scalar_int_mode for local variables.
6072 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
6073 instead of a machine_mode.
6074 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
6076 (make_extraction): Likewise.
6077 (try_widen_shift_mode): Take and return scalar_int_modes instead
6079 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
6080 a scalar_int_mode instead of a machine_mode.
6081 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
6082 (avr_addr_space_pointer_mode): Likewise.
6083 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
6084 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
6085 (msp430_unwind_word_mode): Likewise.
6086 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
6087 (spu_addr_space_pointer_mode): Likewise.
6088 (spu_addr_space_address_mode): Likewise.
6089 (spu_libgcc_cmp_return_mode): Likewise.
6090 (spu_libgcc_shift_count_mode): Likewise.
6091 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
6092 (rl78_addr_space_pointer_mode): Likewise.
6093 (fl78_unwind_word_mode): Likewise.
6094 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
6096 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
6097 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
6098 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
6099 (mips_valid_pointer_mode): Likewise.
6100 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
6101 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
6102 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
6104 (ft32_addr_space_address_mode): Likewise.
6105 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
6106 scalar_int_mode instead of a machine_mode.
6107 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
6109 (m32c_addr_space_address_mode): Likewise.
6110 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
6111 (rs6000_eh_return_filter_mode): Likewise.
6112 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
6113 (rs6000_eh_return_filter_mode): Likewise.
6114 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
6115 (s390_libgcc_shift_count_mode): Likewise.
6116 (s390_unwind_word_mode): Likewise.
6117 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
6119 * target.def (mode_rep_extended): Likewise.
6120 (valid_pointer_mode): Likewise.
6121 (addr_space.valid_pointer_mode): Likewise.
6122 (eh_return_filter_mode): Return a scalar_int_mode rather than
6124 (libgcc_cmp_return_mode): Likewise.
6125 (libgcc_shift_count_mode): Likewise.
6126 (unwind_word_mode): Likewise.
6127 (addr_space.pointer_mode): Likewise.
6128 (addr_space.address_mode): Likewise.
6129 * doc/tm.texi: Regenerate.
6130 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
6132 (do_jump): Use scalar_int_mode for local variables.
6133 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
6134 rather than a machine_mode.
6135 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
6136 (scompare_loc_descriptor_wide): Likewise.
6137 (scompare_loc_descriptor_narrow): Likewise.
6138 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
6140 * except.c (sjlj_emit_dispatch_table): Likewise.
6141 (expand_builtin_eh_copy_values): Likewise.
6142 * explow.c (convert_memory_address_addr_space_1): Likewise.
6143 Take a scalar_int_mode rather than a machine_mode.
6144 (convert_memory_address_addr_space): Take a scalar_int_mode rather
6145 than a machine_mode.
6146 (memory_address_addr_space): Use scalar_int_mode for local variables.
6147 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
6148 rather than a machine_mode.
6149 * expmed.c (mask_rtx): Likewise.
6150 (init_expmed_one_conv): Likewise.
6151 (expand_mult_highpart_adjust): Likewise.
6152 (extract_high_half): Likewise.
6153 (expmed_mult_highpart_optab): Likewise.
6154 (expmed_mult_highpart): Likewise.
6155 (expand_smod_pow2): Likewise.
6156 (expand_sdiv_pow2): Likewise.
6157 (emit_store_flag_int): Likewise.
6158 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
6160 (extract_low_bits): Likewise.
6161 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
6163 * expr.c (pieces_addr::adjust): Likewise.
6164 (can_store_by_pieces): Likewise.
6165 (store_by_pieces): Likewise.
6166 (clear_by_pieces_1): Likewise.
6167 (expand_expr_addr_expr_1): Likewise.
6168 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
6169 (expand_expr_real_1): Likewise.
6170 (try_casesi): Likewise.
6171 * final.c (shorten_branches): Likewise.
6172 * fold-const.c (fold_convert_const_int_from_fixed): Change the
6173 type of "mode" to machine_mode.
6174 * internal-fn.c (expand_arith_overflow_result_store): Take a
6175 scalar_int_mode rather than a machine_mode.
6176 (expand_mul_overflow): Use scalar_int_mode for local variables.
6177 * loop-doloop.c (doloop_modify): Likewise.
6178 (doloop_optimize): Likewise.
6179 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
6180 than a machine_mode.
6181 (expand_doubleword_shift_condmove): Likewise.
6182 (expand_doubleword_shift): Likewise.
6183 (expand_doubleword_clz): Likewise.
6184 (expand_doubleword_popcount): Likewise.
6185 (expand_doubleword_parity): Likewise.
6186 (expand_absneg_bit): Use scalar_int_mode for local variables.
6187 (prepare_float_lib_cmp): Likewise.
6188 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
6189 rather than a machine_mode.
6190 (convert_memory_address_addr_space): Likewise.
6191 (get_mode_bounds): Likewise.
6192 (get_address_mode): Return a scalar_int_mode rather than a
6194 * rtlanal.c (get_address_mode): Likewise.
6195 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
6196 than a machine_mode.
6197 * targhooks.c (default_mode_rep_extended): Likewise.
6198 (default_valid_pointer_mode): Likewise.
6199 (default_addr_space_valid_pointer_mode): Likewise.
6200 (default_eh_return_filter_mode): Return a scalar_int_mode rather
6201 than a machine_mode.
6202 (default_libgcc_cmp_return_mode): Likewise.
6203 (default_libgcc_shift_count_mode): Likewise.
6204 (default_unwind_word_mode): Likewise.
6205 (default_addr_space_pointer_mode): Likewise.
6206 (default_addr_space_address_mode): Likewise.
6207 * targhooks.h (default_eh_return_filter_mode): Likewise.
6208 (default_libgcc_cmp_return_mode): Likewise.
6209 (default_libgcc_shift_count_mode): Likewise.
6210 (default_unwind_word_mode): Likewise.
6211 (default_addr_space_pointer_mode): Likewise.
6212 (default_addr_space_address_mode): Likewise.
6213 (default_mode_rep_extended): Take a scalar_int_mode rather than
6215 (default_valid_pointer_mode): Likewise.
6216 (default_addr_space_valid_pointer_mode): Likewise.
6217 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
6219 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
6220 rather than a machine_mode.
6221 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
6222 for local variables.
6223 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6224 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
6225 than a machine_mode.
6227 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6228 Alan Hayward <alan.hayward@arm.com>
6229 David Sherwood <david.sherwood@arm.com>
6231 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
6232 the mode argument to scalar_int_mode.
6233 (do_jump_by_parts_zero_rtx): Likewise.
6234 (do_jump_by_parts_equality_rtx): Likewise.
6235 (do_jump_by_parts_greater): Take a mode argument.
6236 (do_jump_by_parts_equality): Likewise.
6237 (do_jump_1): Update calls accordingly.
6239 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6240 Alan Hayward <alan.hayward@arm.com>
6241 David Sherwood <david.sherwood@arm.com>
6243 * is-a.h (safe_dyn_cast): New function.
6244 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
6245 (jump_table_for_label): Likewise.
6246 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
6247 instead of an rtx_insn *.
6248 (shorten_branches): Use dyn_cast instead of LABEL_P and
6249 JUMP_TABLE_DATA_P. Use jump_table_for_label and
6250 rtx_jump_table_data::get_data_mode.
6251 (final_scan_insn): Likewise.
6253 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6254 Alan Hayward <alan.hayward@arm.com>
6255 David Sherwood <david.sherwood@arm.com>
6257 * combine.c (try_combine): Use is_a <scalar_int_mode> when
6258 trying to combine a full-register integer set with a subreg
6261 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6262 Alan Hayward <alan.hayward@arm.com>
6263 David Sherwood <david.sherwood@arm.com>
6265 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
6266 that is always either address_mode or pointer_mode.
6268 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6269 Alan Hayward <alan.hayward@arm.com>
6270 David Sherwood <david.sherwood@arm.com>
6272 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
6273 when the two are known to be equal.
6275 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6276 Alan Hayward <alan.hayward@arm.com>
6277 David Sherwood <david.sherwood@arm.com>
6279 * simplify-rtx.c (simplify_const_unary_operation): Use
6280 is_a <scalar_int_mode> instead of checking for a nonzero
6281 precision. Forcibly convert op_mode to a scalar_int_mode
6282 in that case. More clearly differentiate the operand and
6283 result modes and use the former when deciding what the value
6284 of a count-bits operation should be. Use is_int_mode instead
6285 of checking for a MODE_INT. Remove redundant check for whether
6286 this mode has a zero precision.
6288 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6289 Alan Hayward <alan.hayward@arm.com>
6290 David Sherwood <david.sherwood@arm.com>
6292 * optabs.c (widen_leading): Change the type of the mode argument
6293 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
6294 (widen_bswap): Likewise.
6295 (expand_parity): Likewise.
6296 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
6297 (expand_ffs): Likewise.
6298 (epand_unop): Check for scalar integer modes before calling the
6301 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6302 Alan Hayward <alan.hayward@arm.com>
6303 David Sherwood <david.sherwood@arm.com>
6305 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
6307 (expand_expr_real_1): Update call accordingly.
6309 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6310 Alan Hayward <alan.hayward@arm.com>
6311 David Sherwood <david.sherwood@arm.com>
6313 * expmed.c (store_bit_field_using_insv): Add op0_mode and
6314 value_mode arguments. Use scalar_int_mode internally.
6315 (store_bit_field_1): Rename the new integer mode from imode
6316 to op0_mode and use it instead of GET_MODE (op0). Update calls
6317 to store_split_bit_field, store_bit_field_using_insv and
6318 store_fixed_bit_field.
6319 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
6320 Use scalar_int_mode internally. Use a bit count rather than a mode
6321 when calculating the largest bit size for get_best_mode.
6322 Update calls to store_split_bit_field and store_fixed_bit_field_1.
6323 (store_fixed_bit_field_1): Add mode and value_mode arguments.
6324 Remove assertion that OP0 has a scalar integer mode.
6325 (store_split_bit_field): Add op0_mode and value_mode arguments.
6326 Update calls to extract_fixed_bit_field.
6327 (extract_bit_field_using_extv): Add an op0_mode argument.
6328 Use scalar_int_mode internally.
6329 (extract_bit_field_1): Rename the new integer mode from imode to
6330 op0_mode and use it instead of GET_MODE (op0). Update calls to
6331 extract_split_bit_field, extract_bit_field_using_extv and
6332 extract_fixed_bit_field.
6333 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
6334 to extract_split_bit_field and extract_fixed_bit_field_1.
6335 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
6336 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
6338 (extract_split_bit_field): Add an op0_mode argument. Update call
6339 to extract_fixed_bit_field.
6341 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6342 Alan Hayward <alan.hayward@arm.com>
6343 David Sherwood <david.sherwood@arm.com>
6345 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
6346 * explow.c (hard_function_value): Likewise.
6347 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
6348 convert_to_mode call outside the loop.
6349 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
6350 for the mode iterator. Require the mode specified by max_pieces
6352 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
6354 (copy_blkmode_to_reg): Likewise.
6355 (set_storage_via_setmem): Likewise.
6356 * optabs.c (prepare_cmp_insn): Likewise.
6357 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6358 * stor-layout.c (finish_bitfield_representative): Likewise.
6360 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6361 Alan Hayward <alan.hayward@arm.com>
6362 David Sherwood <david.sherwood@arm.com>
6364 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
6365 * expr.c (convert_move): Use them.
6366 (convert_modes): Likewise.
6367 (store_expr_with_bounds): Likewise.
6369 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6370 Alan Hayward <alan.hayward@arm.com>
6371 David Sherwood <david.sherwood@arm.com>
6373 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
6374 parameter for the mode of "x". Remove the "known_x", "known_mode"
6375 and "known_ret" arguments. Change the type of the mode argument
6377 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
6378 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
6379 (reg_num_sign_bit_copies_for_combine): Likewise.
6380 * rtlanal.c (nonzero_bits1): Likewise.
6381 (num_sign_bit_copies1): Likewise.
6382 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
6383 (reg_num_sign_bit_copies_general): Likewise.
6384 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
6385 (reg_nonzero_bits_general): Likewise.
6387 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6388 Alan Hayward <alan.hayward@arm.com>
6389 David Sherwood <david.sherwood@arm.com>
6391 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
6392 than in subroutines. Return 1 for non-integer modes.
6393 (cached_num_sign_bit_copies): Change the type of the mode parameter
6395 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
6396 classes. Handle CONST_INT_P first and then check whether X also
6397 has a scalar integer mode. Check the same thing for inner registers
6398 of a SUBREG and for values that are being extended or truncated.
6400 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6401 Alan Hayward <alan.hayward@arm.com>
6402 David Sherwood <david.sherwood@arm.com>
6404 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
6405 in subroutines. Return the mode mask for non-integer modes.
6406 (cached_nonzero_bits): Change the type of the mode parameter
6408 (nonzero_bits1): Likewise. Remove early exit for other mode
6409 classes. Handle CONST_INT_P first and then check whether X
6410 also has a scalar integer mode.
6412 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6413 Alan Hayward <alan.hayward@arm.com>
6414 David Sherwood <david.sherwood@arm.com>
6416 * expr.c (widest_int_mode_for_size): Make the comment match the code.
6417 Return a scalar_int_mode and assert that the size is greater than
6419 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
6420 (op_by_pieces_d::op_by_pieces_d): Likewise.
6421 (op_by_pieces_d::run): Likewise.
6422 (can_store_by_pieces): Likewise.
6424 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6425 Alan Hayward <alan.hayward@arm.com>
6426 David Sherwood <david.sherwood@arm.com>
6428 * combine.c (extract_left_shift): Add a mode argument and update
6430 (make_compound_operation_int): Change the type of the mode parameter
6431 to scalar_int_mode and update the call to extract_left_shift.
6433 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6434 Alan Hayward <alan.hayward@arm.com>
6435 David Sherwood <david.sherwood@arm.com>
6437 * combine.c (simplify_and_const_int): Change the type of the mode
6438 parameter to scalar_int_mode.
6439 (simplify_and_const_int_1): Likewise. Update recursive call.
6441 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6442 Alan Hayward <alan.hayward@arm.com>
6443 David Sherwood <david.sherwood@arm.com>
6445 * combine.c (simplify_compare_const): Check that the mode is a
6446 scalar_int_mode (rather than VOIDmode) before testing its
6448 (simplify_comparison): Move COMPARISON_P handling out of the
6449 loop and restrict the latter part of the loop to scalar_int_modes.
6450 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
6451 and when considering SUBREG_REGs. Use is_int_mode instead of
6452 checking GET_MODE_CLASS against MODE_INT.
6454 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6455 Alan Hayward <alan.hayward@arm.com>
6456 David Sherwood <david.sherwood@arm.com>
6458 * combine.c (try_widen_shift_mode): Move check for equal modes to...
6459 (simplify_shift_const_1): ...here. Use scalar_int_mode for
6460 shift_unit_mode and for modes involved in scalar shifts.
6462 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6463 Alan Hayward <alan.hayward@arm.com>
6464 David Sherwood <david.sherwood@arm.com>
6466 * combine.c (force_int_to_mode): New function, split out from...
6467 (force_to_mode): ...here. Keep xmode up-to-date and use it
6468 instead of GET_MODE (x).
6470 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6471 Alan Hayward <alan.hayward@arm.com>
6472 David Sherwood <david.sherwood@arm.com>
6474 * optabs-query.h (extraction_insn::struct_mode): Change type to
6475 opt_scalar_int_mode and update comment.
6476 (extraction_insn::field_mode): Change type to scalar_int_mode.
6477 (extraction_insn::pos_mode): Likewise.
6478 * combine.c (make_extraction): Update accordingly.
6479 * optabs-query.c (get_traditional_extraction_insn): Likewise.
6480 (get_optab_extraction_insn): Likewise.
6481 * recog.c (simplify_while_replacing): Likewise.
6482 * expmed.c (narrow_bit_field_mem): Change the type of the mode
6483 parameter to opt_scalar_int_mode.
6485 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6486 Alan Hayward <alan.hayward@arm.com>
6487 David Sherwood <david.sherwood@arm.com>
6489 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
6490 to a scalar_int_mode instead of a machine_mode.
6491 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
6492 (get_best_mode): Return a boolean and use a pointer argument to store
6493 the selected mode. Replace the limit mode parameter with a bit limit.
6494 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
6495 for the values returned by bit_field_mode_iterator::next_mode.
6496 (store_bit_field): Update call to get_best_mode.
6497 (store_fixed_bit_field): Likewise.
6498 (extract_fixed_bit_field): Likewise.
6499 * expr.c (optimize_bitfield_assignment_op): Likewise.
6500 * fold-const.c (optimize_bit_field_compare): Likewise.
6501 (fold_truth_andor_1): Likewise.
6502 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
6503 Update for new type of m_mode.
6504 (get_best_mode): As above.
6506 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6507 Alan Hayward <alan.hayward@arm.com>
6508 David Sherwood <david.sherwood@arm.com>
6510 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
6511 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
6512 (store_bit_field): Check is_a <scalar_int_mode> before calling
6513 strict_volatile_bitfield_p.
6514 (extract_bit_field): Likewise.
6516 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6517 Alan Hayward <alan.hayward@arm.com>
6518 David Sherwood <david.sherwood@arm.com>
6520 * target.def (cstore_mode): Return a scalar_int_mode.
6521 * doc/tm.texi: Regenerate.
6522 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
6523 * targhooks.h (default_cstore_mode): Likewise.
6524 * targhooks.c (default_cstore_mode): Likewise, using a forced
6526 * expmed.c (emit_cstore): Expect the target of the cstore to be
6529 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6530 Alan Hayward <alan.hayward@arm.com>
6531 David Sherwood <david.sherwood@arm.com>
6533 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
6535 (niter_desc): Likewise mode.
6536 (iv_analyze): Add a mode parameter.
6538 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
6539 and change its type to scalar_int_mode.
6540 * loop-iv.c: Update commentary at head of file.
6541 (iv_constant): Pass the mode paraeter before the rtx it describes
6542 and change its type to scalar_int_mode. Remove VOIDmode handling.
6543 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
6544 (iv_extend): Likewise.
6545 (shorten_into_mode): Likewise.
6546 (iv_add): Use scalar_int_mode.
6547 (iv_mult): Likewise.
6548 (iv_shift): Likewise.
6549 (canonicalize_iv_subregs): Likewise.
6550 (get_biv_step_1): Pass the outer_mode parameter before the rtx
6551 it describes and change its mode to scalar_int_mode. Also change
6552 the type of the returned inner_mode to scalar_int_mode.
6553 (get_biv_step): Likewise, turning outer_mode from a pointer
6554 into a direct parameter. Update call to get_biv_step_1.
6555 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
6556 iv_constant and get_biv_step.
6557 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
6558 and change its type to scalar_int_mode. Don't initialise iv->mode
6559 to VOIDmode and remove later checks for its still being VOIDmode.
6560 Update calls to iv_analyze_op and iv_analyze_expr. Check
6561 is_a <scalar_int_mode> when changing the mode under consideration.
6562 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
6563 Update call to iv_analyze_expr.
6564 (iv_analyze_op): Add a mode parameter. Reject subregs whose
6565 inner register is not also a scalar_int_mode. Update call to
6567 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
6568 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
6569 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
6570 separate mode class checks. Update calls to iv_analyze. Remove
6571 fix-up of VOIDmodes after iv_analyze_biv.
6572 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
6573 don't have a scalar_int_mode. Update call to biv_p.
6575 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6576 Alan Hayward <alan.hayward@arm.com>
6577 David Sherwood <david.sherwood@arm.com>
6579 * cfgexpand.c (convert_debug_memory_address): Use
6580 as_a <scalar_int_mode>.
6581 * combine.c (expand_compound_operation): Likewise.
6582 (make_extraction): Likewise.
6583 (change_zero_ext): Likewise.
6584 (simplify_comparison): Likewise.
6585 * cse.c (cse_insn): Likewise.
6586 * dwarf2out.c (minmax_loc_descriptor): Likewise.
6587 (mem_loc_descriptor): Likewise.
6588 (loc_descriptor): Likewise.
6589 * expmed.c (init_expmed_one_mode): Likewise.
6590 (synth_mult): Likewise.
6591 (emit_store_flag_1): Likewise.
6592 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
6593 of a comparison with size.
6594 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
6595 (reduce_to_bit_field_precision): Likewise.
6596 * function.c (expand_function_end): Likewise.
6597 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
6598 * loop-doloop.c (doloop_modify): Likewise.
6599 * optabs.c (expand_binop): Likewise.
6600 (expand_unop): Likewise.
6601 (expand_copysign_absneg): Likewise.
6602 (prepare_cmp_insn): Likewise.
6603 (maybe_legitimize_operand): Likewise.
6604 * recog.c (const_scalar_int_operand): Likewise.
6605 * rtlanal.c (get_address_mode): Likewise.
6606 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6607 (simplify_cond_clz_ctz): Likewise.
6608 * tree-nested.c (get_nl_goto_field): Likewise.
6609 * tree.c (build_vector_type_for_mode): Likewise.
6610 * var-tracking.c (use_narrower_mode): Likewise.
6612 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6613 Alan Hayward <alan.hayward@arm.com>
6614 David Sherwood <david.sherwood@arm.com>
6616 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
6617 * builtins.c (expand_builtin_signbit): Use it.
6618 * cfgexpand.c (expand_debug_expr): Likewise.
6619 * dojump.c (do_jump): Likewise.
6620 (do_compare_and_jump): Likewise.
6621 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
6622 * expmed.c (make_tree): Likewise.
6623 * expr.c (expand_expr_real_2): Likewise.
6624 (expand_expr_real_1): Likewise.
6625 (try_casesi): Likewise.
6626 * fold-const-call.c (fold_const_call_ss): Likewise.
6627 * fold-const.c (unextend): Likewise.
6628 (extract_muldiv_1): Likewise.
6629 (fold_single_bit_test): Likewise.
6630 (native_encode_int): Likewise.
6631 (native_encode_string): Likewise.
6632 (native_interpret_int): Likewise.
6633 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
6634 * internal-fn.c (expand_addsub_overflow): Likewise.
6635 (expand_neg_overflow): Likewise.
6636 (expand_mul_overflow): Likewise.
6637 (expand_arith_overflow): Likewise.
6638 * match.pd: Likewise.
6639 * stor-layout.c (layout_type): Likewise.
6640 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
6641 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
6642 * tree-ssanames.c (get_range_info): Likewise.
6643 * tree-switch-conversion.c (array_value_type) Likewise.
6644 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
6645 (vect_recog_divmod_pattern): Likewise.
6646 (vect_recog_mixed_size_cond_pattern): Likewise.
6647 * tree-vrp.c (extract_range_basic): Likewise.
6648 (simplify_float_conversion_using_ranges): Likewise.
6649 * tree.c (int_fits_type_p): Likewise.
6650 * ubsan.c (instrument_bool_enum_load): Likewise.
6651 * varasm.c (mergeable_string_section): Likewise.
6652 (narrowing_initializer_constant_valid_p): Likewise.
6653 (output_constant): Likewise.
6655 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6656 Alan Hayward <alan.hayward@arm.com>
6657 David Sherwood <david.sherwood@arm.com>
6659 * machmode.h (NARROWEST_INT_MODE): New macro.
6660 * expr.c (alignment_for_piecewise_move): Use it instead of
6661 GET_CLASS_NARROWEST_MODE (MODE_INT).
6662 (push_block): Likewise.
6663 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
6665 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
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 * postreload.c (move2add_valid_value_p): Change the type of the
6672 mode parameter to scalar_int_mode.
6673 (move2add_use_add2_insn): Add a mode parameter and use it instead
6675 (move2add_use_add3_insn): Likewise.
6676 (reload_cse_move2add): Update accordingly.
6678 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6679 Alan Hayward <alan.hayward@arm.com>
6680 David Sherwood <david.sherwood@arm.com>
6682 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
6684 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
6685 * optabs.c (expand_unop): Likewise.
6687 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6688 Alan Hayward <alan.hayward@arm.com>
6689 David Sherwood <david.sherwood@arm.com>
6691 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
6692 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
6693 (popcount_loc_descriptor): Likewise.
6694 (bswap_loc_descriptor): Likewise.
6695 (rotate_loc_descriptor): Likewise.
6696 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
6697 calling the functions above.
6699 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6700 Alan Hayward <alan.hayward@arm.com>
6701 David Sherwood <david.sherwood@arm.com>
6703 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
6705 (try_combine): Likewise.
6706 (simplify_if_then_else): Likewise.
6707 * cse.c (cse_insn): Likewise.
6708 * dwarf2out.c (mem_loc_descriptor): Likewise.
6709 * emit-rtl.c (gen_lowpart_common): Likewise.
6710 * simplify-rtx.c (simplify_truncation): Likewise.
6711 (simplify_binary_operation_1): Likewise.
6712 (simplify_const_relational_operation): Likewise.
6713 (simplify_ternary_operation): Likewise.
6714 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
6716 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6717 Alan Hayward <alan.hayward@arm.com>
6718 David Sherwood <david.sherwood@arm.com>
6720 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
6721 * reload.c (push_reload): Likewise.
6722 (find_reloads): Likewise.
6724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6725 Alan Hayward <alan.hayward@arm.com>
6726 David Sherwood <david.sherwood@arm.com>
6728 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
6729 (make_compound_operation_int): Likewise.
6730 (change_zero_ext): Likewise.
6731 * expr.c (convert_move): Likewise.
6732 (convert_modes): Likewise.
6733 * fwprop.c (forward_propagate_subreg): Likewise.
6734 * loop-iv.c (get_biv_step_1): Likewise.
6735 * optabs.c (widen_operand): Likewise.
6736 * postreload.c (move2add_valid_value_p): Likewise.
6737 * recog.c (simplify_while_replacing): Likewise.
6738 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6739 (simplify_binary_operation_1): Likewise. Remove redundant
6740 mode equality check.
6742 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6743 Alan Hayward <alan.hayward@arm.com>
6744 David Sherwood <david.sherwood@arm.com>
6746 * combine.c (combine_simplify_rtx): Add checks for
6747 is_a <scalar_int_mode>.
6748 (simplify_if_then_else): Likewise.
6749 (make_field_assignment): Likewise.
6750 (simplify_comparison): Likewise.
6751 * ifcvt.c (noce_try_bitop): Likewise.
6752 * loop-invariant.c (canonicalize_address_mult): Likewise.
6753 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6755 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6756 Alan Hayward <alan.hayward@arm.com>
6757 David Sherwood <david.sherwood@arm.com>
6759 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
6760 is_a <scalar_int_mode> instead of != BLKmode.
6762 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6763 Alan Hayward <alan.hayward@arm.com>
6764 David Sherwood <david.sherwood@arm.com>
6766 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
6767 instead of != VOIDmode.
6768 * combine.c (if_then_else_cond): Likewise.
6769 (change_zero_ext): Likewise.
6770 * dwarf2out.c (mem_loc_descriptor): Likewise.
6771 (loc_descriptor): Likewise.
6772 * rtlanal.c (canonicalize_condition): Likewise.
6773 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
6775 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6776 Alan Hayward <alan.hayward@arm.com>
6777 David Sherwood <david.sherwood@arm.com>
6779 * simplify-rtx.c (simplify_binary_operation_1): Use
6780 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
6782 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6783 Alan Hayward <alan.hayward@arm.com>
6784 David Sherwood <david.sherwood@arm.com>
6786 * wide-int.h (int_traits<unsigned char>) New class.
6787 (int_traits<unsigned short>) Likewise.
6788 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
6789 Use GET_MODE_UNIT_PRECISION and remove redundant test for
6791 * combine.c (set_nonzero_bits_and_sign_copies): Use
6792 is_a <scalar_int_mode>.
6793 (find_split_point): Likewise.
6794 (combine_simplify_rtx): Likewise.
6795 (simplify_logical): Likewise.
6796 (expand_compound_operation): Likewise.
6797 (expand_field_assignment): Likewise.
6798 (make_compound_operation): Likewise.
6799 (extended_count): Likewise.
6800 (change_zero_ext): Likewise.
6801 (simplify_comparison): Likewise.
6802 * dwarf2out.c (scompare_loc_descriptor): Likewise.
6803 (ucompare_loc_descriptor): Likewise.
6804 (minmax_loc_descriptor): Likewise.
6805 (mem_loc_descriptor): Likewise.
6806 (loc_descriptor): Likewise.
6807 * expmed.c (init_expmed_one_mode): Likewise.
6808 * lra-constraints.c (lra_constraint_offset): Likewise.
6809 * optabs.c (prepare_libcall_arg): Likewise.
6810 * postreload.c (move2add_note_store): Likewise.
6811 * reload.c (operands_match_p): Likewise.
6812 * rtl.h (load_extend_op): Likewise.
6813 * rtlhooks.c (gen_lowpart_general): Likewise.
6814 * simplify-rtx.c (simplify_truncation): Likewise.
6815 (simplify_unary_operation_1): Likewise.
6816 (simplify_binary_operation_1): Likewise.
6817 (simplify_const_binary_operation): Likewise.
6818 (simplify_const_relational_operation): Likewise.
6819 (simplify_subreg): Likewise.
6820 * stor-layout.c (bitwise_mode_for_mode): Likewise.
6821 * var-tracking.c (adjust_mems): Likewise.
6822 (prepare_call_arguments): Likewise.
6824 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6825 Alan Hayward <alan.hayward@arm.com>
6826 David Sherwood <david.sherwood@arm.com>
6828 * machmode.h (is_int_mode): New fuction.
6829 * combine.c (find_split_point): Use it.
6830 (combine_simplify_rtx): Likewise.
6831 (simplify_if_then_else): Likewise.
6832 (simplify_set): Likewise.
6833 (simplify_shift_const_1): Likewise.
6834 (simplify_comparison): Likewise.
6835 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
6836 * cse.c (notreg_cost): Likewise.
6837 (cse_insn): Likewise.
6838 * cselib.c (cselib_lookup_1): Likewise.
6839 * dojump.c (do_jump_1): Likewise.
6840 (do_compare_rtx_and_jump): Likewise.
6841 * dse.c (get_call_args): Likewise.
6842 * dwarf2out.c (rtl_for_decl_init): Likewise.
6843 (native_encode_initializer): Likewise.
6844 * expmed.c (emit_store_flag_1): Likewise.
6845 (emit_store_flag): Likewise.
6846 * expr.c (convert_modes): Likewise.
6847 (store_field): Likewise.
6848 (expand_expr_real_1): Likewise.
6849 * fold-const.c (fold_read_from_constant_string): Likewise.
6850 * gimple-ssa-sprintf.c (get_format_string): Likewise.
6851 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
6852 * optabs.c (expand_binop): Likewise.
6853 (expand_unop): Likewise.
6854 (expand_abs_nojump): Likewise.
6855 (expand_one_cmpl_abs_nojump): Likewise.
6856 * simplify-rtx.c (mode_signbit_p): Likewise.
6857 (val_signbit_p): Likewise.
6858 (val_signbit_known_set_p): Likewise.
6859 (val_signbit_known_clear_p): Likewise.
6860 (simplify_relational_operation_1): Likewise.
6861 * tree.c (vector_type_mode): Likewise.
6863 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6864 Alan Hayward <alan.hayward@arm.com>
6865 David Sherwood <david.sherwood@arm.com>
6867 * machmode.h (smallest_mode_for_size): Fix formatting.
6868 (smallest_int_mode_for_size): New function.
6869 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
6870 instead of smallest_mode_for_size.
6871 * combine.c (make_extraction): Likewise.
6872 * config/arc/arc.c (arc_expand_movmem): Likewise.
6873 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
6874 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
6875 * config/s390/s390.c (s390_expand_insv): Likewise.
6876 * config/sparc/sparc.c (assign_int_registers): Likewise.
6877 * config/spu/spu.c (spu_function_value): Likewise.
6878 (spu_function_arg): Likewise.
6879 * coverage.c (get_gcov_type): Likewise.
6880 (get_gcov_unsigned_t): Likewise.
6881 * dse.c (find_shift_sequence): Likewise.
6882 * expmed.c (store_bit_field_1): Likewise.
6883 * expr.c (convert_move): Likewise.
6884 (store_field): Likewise.
6885 * internal-fn.c (expand_arith_overflow): Likewise.
6886 * optabs-query.c (get_best_extraction_insn): Likewise.
6887 * optabs.c (expand_twoval_binop_libfunc): Likewise.
6888 * stor-layout.c (layout_type): Likewise.
6889 (initialize_sizetypes): Likewise.
6890 * targhooks.c (default_get_mask_mode): Likewise.
6891 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
6893 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6894 Alan Hayward <alan.hayward@arm.com>
6895 David Sherwood <david.sherwood@arm.com>
6897 * machmode.h (opt_mode::else_blk): New function.
6898 (int_mode_for_mode): Declare.
6899 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
6900 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
6902 * cfgexpand.c (expand_debug_expr): Likewise.
6903 * combine.c (gen_lowpart_or_truncate): Likewise.
6904 (gen_lowpart_for_combine): Likewise.
6905 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
6906 * config/avr/avr.c (avr_to_int_mode): Likewise.
6907 (avr_out_plus_1): Likewise.
6908 (avr_out_plus): Likewise.
6909 (avr_out_round): Likewise.
6910 * config/i386/i386.c (ix86_split_to_parts): Likewise.
6911 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
6912 (s390_expand_vcond): Likewise.
6913 * config/spu/spu.c (spu_split_immediate): Likewise.
6914 (spu_expand_mov): Likewise.
6915 * dse.c (get_stored_val): Likewise.
6916 * expmed.c (store_bit_field_1): Likewise.
6917 (convert_extracted_bit_field): Use int_mode_for_mode instead of
6919 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
6920 (extract_low_bits): Likewise.
6921 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
6922 handling rather than repeating the check.
6923 (emit_group_store): Likewise.
6924 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
6925 * optabs.c (expand_absneg_bit): Likewise.
6926 (expand_copysign_absneg): Likewise.
6927 (expand_copysign_bit): Likewise.
6928 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
6929 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
6930 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
6931 * var-tracking.c (prepare_call_arguments): Likewise.
6932 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
6933 int_mode_for_mode instead of mode_for_size.
6934 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
6936 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6937 Alan Hayward <alan.hayward@arm.com>
6938 David Sherwood <david.sherwood@arm.com>
6940 * machmode.h (int_mode_for_size): New function.
6941 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
6942 instead of mode_for_size.
6943 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
6945 * combine.c (expand_field_assignment): Use int_mode_for_size
6946 instead of mode_for_size.
6947 (make_extraction): Likewise.
6948 (simplify_shift_const_1): Likewise.
6949 (simplify_comparison): Likewise.
6950 * dojump.c (do_jump): Likewise.
6951 * dwarf2out.c (mem_loc_descriptor): Likewise.
6952 * emit-rtl.c (init_derived_machine_modes): Likewise.
6953 * expmed.c (flip_storage_order): Likewise.
6954 (convert_extracted_bit_field): Likewise.
6955 * expr.c (copy_blkmode_from_reg): Likewise.
6956 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
6957 * internal-fn.c (expand_mul_overflow): Likewise.
6958 * lower-subreg.c (simple_move): Likewise.
6959 * optabs-libfuncs.c (init_optabs): Likewise.
6960 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6961 * tree.c (vector_type_mode): Likewise.
6962 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
6963 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
6964 * tree-vect-generic.c (expand_vector_parallel): Likewise.
6965 * tree-vect-stmts.c (vectorizable_load): Likewise.
6966 (vectorizable_store): Likewise.
6968 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6969 Alan Hayward <alan.hayward@arm.com>
6970 David Sherwood <david.sherwood@arm.com>
6972 * coretypes.h (pod_mode): New type.
6973 (scalar_int_mode_pod): New typedef.
6974 * machmode.h (pod_mode): New class.
6975 (int_n_data_t::m): Change type to scalar_int_mode_pod.
6976 * genmodes.c (emit_mode_int_n): Update accordingly.
6977 * lower-subreg.h (target_lower_subreg): Change type to
6978 scalar_int_mode_pod.
6979 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
6980 scalar_int_mode_pod.
6982 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6983 Alan Hayward <alan.hayward@arm.com>
6984 David Sherwood <david.sherwood@arm.com>
6986 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
6987 machine_mode to scalar_int_mode.
6988 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
6989 (rs6000_option_override_internal): Remove cast to int.
6990 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
6991 machine_mode to scalar_int_mode.
6992 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
6993 (rs6000_option_override_internal): Remove cast to int.
6994 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
6995 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
6997 * config/s390/s390.c (s390_expand_builtin): Likewise.
6998 * coretypes.h (scalar_int_mode): New type.
6999 (opt_scalar_int_mode): New typedef.
7000 * machmode.h (scalar_int_mode): New class.
7001 (scalar_int_mode::includes_p): New function.
7002 (byte_mode): Change type to scalar_int_mode.
7003 (word_mode): Likewise.
7004 (ptr_mode): Likewise.
7005 * emit-rtl.c (byte_mode): Likewise.
7006 (word_mode): Likewise.
7007 (ptr_mode): Likewise.
7008 (init_derived_machine_modes): Update accordingly.
7009 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
7010 and MODE_PARTIAL_INT.
7011 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
7012 opt_scalar_int_mode.
7014 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7015 Alan Hayward <alan.hayward@arm.com>
7016 David Sherwood <david.sherwood@arm.com>
7018 * target.def (libgcc_floating_mode_supported_p): Take a
7020 * doc/tm.texi: Regenerate.
7021 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
7023 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
7024 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
7027 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7028 Alan Hayward <alan.hayward@arm.com>
7029 David Sherwood <david.sherwood@arm.com>
7031 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
7032 * doc/tm.texi: Regenerate.
7033 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
7034 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
7035 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
7036 * targhooks.h (default_floatn_mode): Likewise.
7037 * targhooks.c (default_floatn_mode): Likewise.
7038 * tree.c (build_common_tree_nodes): Update accordingly.
7040 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7041 Alan Hayward <alan.hayward@arm.com>
7042 David Sherwood <david.sherwood@arm.com>
7044 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
7045 (mode_iterator::iterate_p): Likewise.
7046 (mode_iterator::get_wider): Likewise.
7047 * expr.c (init_expr_target): Use 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 * coretypes.h (opt_scalar_float_mode): New typedef.
7054 * machmode.h (float_mode_for_size): New function.
7055 * emit-rtl.c (double_mode): Delete.
7056 (init_emit_once): Use float_mode_for_size.
7057 * stor-layout.c (layout_type): Likewise.
7058 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
7060 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7061 Alan Hayward <alan.hayward@arm.com>
7062 David Sherwood <david.sherwood@arm.com>
7064 * output.h (assemble_real): Take a scalar_float_mode.
7065 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
7066 * config/arm/arm.md (consttable_4): Likewise.
7067 (consttable_8): Likewise.
7068 (consttable_16): Likewise.
7069 * config/mips/mips.md (consttable_float): Likewise.
7070 * config/s390/s390.c (s390_output_pool_entry): Likewise.
7071 * varasm.c (assemble_real): Take a scalar_float_mode.
7072 (output_constant_pool_2): Update accordingly.
7073 (output_constant): Likewise.
7075 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7076 Alan Hayward <alan.hayward@arm.com>
7077 David Sherwood <david.sherwood@arm.com>
7079 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
7080 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
7081 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
7082 (native_encode_real): Likewise.
7083 (native_interpret_real): Likewise.
7084 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
7085 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7087 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7088 Alan Hayward <alan.hayward@arm.com>
7089 David Sherwood <david.sherwood@arm.com>
7091 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
7092 <scalar_float_mode>. Simplify.
7093 (gen_extend_conv_libfunc): Likewise.
7095 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7096 Alan Hayward <alan.hayward@arm.com>
7097 David Sherwood <david.sherwood@arm.com>
7099 * coretypes.h (scalar_float_mode): New type.
7100 * machmode.h (mode_traits::from_int): Use machine_mode if
7101 USE_ENUM_MODES is defined.
7102 (is_a): New function.
7104 (dyn_cast): Likewise.
7105 (scalar_float_mode): New class.
7106 (scalar_float_mode::includes_p): New function.
7107 (is_float_mode): Likewise.
7108 * gdbhooks.py (MachineModePrinter): New class.
7109 (build_pretty_printer): Use it for scalar_float_mode.
7110 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
7111 (format_helper::format_helper): Turn into a template.
7112 * genmodes.c (get_mode_class): New function.
7113 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
7114 or machine_mode if none.
7115 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
7116 as_a <scalar_float_mode>.
7117 * dwarf2out.c (mem_loc_descriptor): Likewise.
7118 (insert_float): Likewise.
7119 (add_const_value_attribute): Likewise.
7120 * simplify-rtx.c (simplify_immed_subreg): Likewise.
7121 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
7122 (expand_unop): Update accordingly.
7123 (expand_abs_nojump): Likewise.
7124 (expand_copysign_absneg): Take a scalar_float_mode.
7125 (expand_copysign_bit): Likewise.
7126 (expand_copysign): Update accordingly.
7128 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7129 Alan Hayward <alan.hayward@arm.com>
7130 David Sherwood <david.sherwood@arm.com>
7132 * coretypes.h (opt_mode): New class.
7133 * machmode.h (opt_mode): Likewise.
7134 (opt_mode::else_void): New function.
7135 (opt_mode::require): Likewise.
7136 (opt_mode::exists): Likewise.
7137 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
7138 (GET_MODE_2XWIDER_MODE): Likewise.
7139 (mode_iterator::get_wider): Update accordingly.
7140 (mode_iterator::get_2xwider): Likewise.
7141 (mode_iterator::get_known_wider): Likewise, turning into a template.
7142 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
7143 forcing a wider mode to exist.
7144 * config/cr16/cr16.h (LONG_REG_P): Likewise.
7145 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
7146 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
7147 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
7148 * lower-subreg.c (init_lower_subreg): Likewise.
7149 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
7150 on the final iteration.
7151 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
7152 a wider mode exists before asking for a move pattern.
7153 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
7154 forcing a wider mode to exist.
7155 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
7156 returning false if no such mode exists.
7157 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
7158 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
7159 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
7160 Avoid checking for a MODE_INT if we already know the mode is not a
7162 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
7163 forcing a wider mode to exist.
7164 (expmed_mult_highpart_optab): Likewise.
7165 (expmed_mult_highpart): Likewise.
7166 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
7168 * lto-streamer-in.c (lto_input_mode_table): Likewise.
7169 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
7170 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
7171 * internal-fn.c (expand_mul_overflow): Update use of
7172 GET_MODE_2XWIDER_MODE.
7173 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7174 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
7175 GET_MODE_WIDER_MODE.
7176 (convert_plusminus_to_widen): Likewise.
7177 * tree-switch-conversion.c (array_value_type): Likewise.
7178 * var-tracking.c (emit_note_insn_var_location): Likewise.
7179 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7180 Return false inside rather than outside the loop if no wider mode
7182 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
7183 and GET_MODE_2XWIDER_MODE
7184 (can_compare_p): Use else_void.
7185 * gdbhooks.py (OptMachineModePrinter): New class.
7186 (build_pretty_printer): Use it for opt_mode.
7188 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7189 Alan Hayward <alan.hayward@arm.com>
7190 David Sherwood <david.sherwood@arm.com>
7192 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
7193 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
7195 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7196 Alan Hayward <alan.hayward@arm.com>
7197 David Sherwood <david.sherwood@arm.com>
7199 * machmode.h (mode_traits): New structure.
7200 (get_narrowest_mode): New function.
7201 (mode_iterator::start): Likewise.
7202 (mode_iterator::iterate_p): Likewise.
7203 (mode_iterator::get_wider): Likewise.
7204 (mode_iterator::get_known_wider): Likewise.
7205 (mode_iterator::get_2xwider): Likewise.
7206 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
7207 (FOR_EACH_MODE): Likewise.
7208 (FOR_EACH_MODE_FROM): Likewise.
7209 (FOR_EACH_MODE_UNTIL): Likewise.
7210 (FOR_EACH_WIDER_MODE): Likewise.
7211 (FOR_EACH_2XWIDER_MODE): Likewise.
7212 * builtins.c (expand_builtin_strlen): Use new mode iterators.
7213 * combine.c (simplify_comparison): Likewise
7214 * config/i386/i386.c (type_natural_mode): Likewise.
7215 * cse.c (cse_insn): Likewise.
7216 * dse.c (find_shift_sequence): Likewise.
7217 * emit-rtl.c (init_derived_machine_modes): Likewise.
7218 (init_emit_once): Likewise.
7219 * explow.c (hard_function_value): Likewise.
7220 * expmed.c (extract_fixed_bit_field_1): Likewise.
7221 (extract_bit_field_1): Likewise.
7222 (expand_divmod): Likewise.
7223 (emit_store_flag_1): Likewise.
7224 * expr.c (init_expr_target): Likewise.
7225 (convert_move): Likewise.
7226 (alignment_for_piecewise_move): Likewise.
7227 (widest_int_mode_for_size): Likewise.
7228 (emit_block_move_via_movmem): Likewise.
7229 (copy_blkmode_to_reg): Likewise.
7230 (set_storage_via_setmem): Likewise.
7231 (compress_float_constant): Likewise.
7232 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7233 * optabs-query.c (get_best_extraction_insn): Likewise.
7234 * optabs.c (expand_binop): Likewise.
7235 (expand_twoval_unop): Likewise.
7236 (expand_twoval_binop): Likewise.
7237 (widen_leading): Likewise.
7238 (widen_bswap): Likewise.
7239 (expand_parity): Likewise.
7240 (expand_unop): Likewise.
7241 (prepare_cmp_insn): Likewise.
7242 (prepare_float_lib_cmp): Likewise.
7243 (expand_float): Likewise.
7244 (expand_fix): Likewise.
7245 (expand_sfix_optab): Likewise.
7246 * postreload.c (move2add_use_add2_insn): Likewise.
7247 * reg-stack.c (reg_to_stack): Likewise.
7248 * reginfo.c (choose_hard_reg_mode): Likewise.
7249 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
7250 * stor-layout.c (mode_for_size): Likewise.
7251 (smallest_mode_for_size): Likewise.
7252 (mode_for_vector): Likewise.
7253 (finish_bitfield_representative): Likewise.
7254 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
7255 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
7256 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7257 * var-tracking.c (prepare_call_arguments): Likewise.
7259 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7260 Alan Hayward <alan.hayward@arm.com>
7261 David Sherwood <david.sherwood@arm.com>
7263 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
7264 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
7265 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
7266 * machmode.h (mode_size): Move earlier in file.
7267 (mode_precision): Likewise.
7268 (mode_inner): Likewise.
7269 (mode_nunits): Likewise.
7270 (mode_unit_size): Likewise.
7271 (unit_unit_precision): Likewise.
7272 (mode_wider): Likewise.
7273 (mode_2xwider): Likewise.
7274 (machine_mode): New class.
7275 (mode_to_bytes): New function.
7276 (mode_to_bits): Likewise.
7277 (mode_to_precision): Likewise.
7278 (mode_to_inner): Likewise.
7279 (mode_to_unit_size): Likewise.
7280 (mode_to_unit_precision): Likewise.
7281 (mode_to_nunits): Likewise.
7282 (GET_MODE_SIZE): Use mode_to_bytes.
7283 (GET_MODE_BITSIZE): Use mode_to_bits.
7284 (GET_MODE_PRECISION): Use mode_to_precision.
7285 (GET_MODE_INNER): Use mode_to_inner.
7286 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
7287 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
7288 (GET_MODE_NUNITS): Use mode_to_nunits.
7289 * system.h (ALWAYS_INLINE): New macro.
7290 * config/powerpcspe/powerpcspe-c.c
7291 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
7292 int for arg1_mode and arg2_mode.
7294 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7295 Alan Hayward <alan.hayward@arm.com>
7296 David Sherwood <david.sherwood@arm.com>
7298 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
7299 Prefix mode names with E_ in case statements.
7300 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7301 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
7302 (aarch64_split_simd_move): Likewise.
7303 (aarch64_gen_storewb_pair): Likewise.
7304 (aarch64_gen_loadwb_pair): Likewise.
7305 (aarch64_gen_store_pair): Likewise.
7306 (aarch64_gen_load_pair): Likewise.
7307 (aarch64_get_condition_code_1): Likewise.
7308 (aarch64_constant_pool_reload_icode): Likewise.
7309 (get_rsqrte_type): Likewise.
7310 (get_rsqrts_type): Likewise.
7311 (get_recpe_type): Likewise.
7312 (get_recps_type): Likewise.
7313 (aarch64_gimplify_va_arg_expr): Likewise.
7314 (aarch64_simd_container_mode): Likewise.
7315 (aarch64_emit_load_exclusive): Likewise.
7316 (aarch64_emit_store_exclusive): Likewise.
7317 (aarch64_expand_compare_and_swap): Likewise.
7318 (aarch64_gen_atomic_cas): Likewise.
7319 (aarch64_emit_bic): Likewise.
7320 (aarch64_emit_atomic_swap): Likewise.
7321 (aarch64_emit_atomic_load_op): Likewise.
7322 (aarch64_evpc_trn): Likewise.
7323 (aarch64_evpc_uzp): Likewise.
7324 (aarch64_evpc_zip): Likewise.
7325 (aarch64_evpc_ext): Likewise.
7326 (aarch64_evpc_rev): Likewise.
7327 (aarch64_evpc_dup): Likewise.
7328 (aarch64_gen_ccmp_first): Likewise.
7329 (aarch64_gen_ccmp_next): Likewise.
7330 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
7331 (alpha_emit_xfloating_libcall): Likewise.
7332 (emit_insxl): Likewise.
7333 (alpha_arg_type): Likewise.
7334 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
7335 (arc_preferred_simd_mode): Likewise.
7336 (arc_secondary_reload): Likewise.
7337 (get_arc_condition_code): Likewise.
7338 (arc_print_operand): Likewise.
7339 (arc_legitimate_constant_p): Likewise.
7340 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7341 * config/arc/arc.md (casesi_load): Likewise.
7342 (casesi_compact_jump): Likewise.
7343 * config/arc/predicates.md (proper_comparison_operator): Likewise.
7344 (cc_use_register): Likewise.
7345 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7346 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
7347 (arm_init_iwmmxt_builtins): Likewise.
7348 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
7349 (neon_expand_vector_init): Likewise.
7350 (arm_attr_length_move_neon): Likewise.
7351 (maybe_get_arm_condition_code): Likewise.
7352 (arm_emit_vector_const): Likewise.
7353 (arm_preferred_simd_mode): Likewise.
7354 (arm_output_iwmmxt_tinsr): Likewise.
7355 (thumb1_output_casesi): Likewise.
7356 (thumb2_output_casesi): Likewise.
7357 (arm_emit_load_exclusive): Likewise.
7358 (arm_emit_store_exclusive): Likewise.
7359 (arm_expand_compare_and_swap): Likewise.
7360 (arm_evpc_neon_vuzp): Likewise.
7361 (arm_evpc_neon_vzip): Likewise.
7362 (arm_evpc_neon_vrev): Likewise.
7363 (arm_evpc_neon_vtrn): Likewise.
7364 (arm_evpc_neon_vext): Likewise.
7365 (arm_validize_comparison): Likewise.
7366 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
7367 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
7368 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
7369 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
7370 (c6x_preferred_simd_mode): Likewise.
7371 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
7372 (epiphany_rtx_costs): Likewise.
7373 * config/epiphany/predicates.md (proper_comparison_operator):
7375 * config/frv/frv.c (condexec_memory_operand): Likewise.
7376 (frv_emit_move): Likewise.
7377 (output_move_single): Likewise.
7378 (output_condmove_single): Likewise.
7379 (frv_hard_regno_mode_ok): Likewise.
7380 (frv_matching_accg_mode): Likewise.
7381 * config/h8300/h8300.c (split_adds_subs): Likewise.
7382 (h8300_rtx_costs): Likewise.
7383 (h8300_print_operand): Likewise.
7384 (compute_mov_length): Likewise.
7385 (output_logical_op): Likewise.
7386 (compute_logical_op_length): Likewise.
7387 (compute_logical_op_cc): Likewise.
7388 (h8300_shift_needs_scratch_p): Likewise.
7389 (output_a_shift): Likewise.
7390 (compute_a_shift_length): Likewise.
7391 (compute_a_shift_cc): Likewise.
7392 (expand_a_rotate): Likewise.
7393 (output_a_rotate): Likewise.
7394 * config/i386/i386.c (classify_argument): Likewise.
7395 (function_arg_advance_32): Likewise.
7396 (function_arg_32): Likewise.
7397 (function_arg_64): Likewise.
7398 (function_value_64): Likewise.
7399 (ix86_gimplify_va_arg): Likewise.
7400 (ix86_legitimate_constant_p): Likewise.
7401 (put_condition_code): Likewise.
7402 (split_double_mode): Likewise.
7403 (ix86_avx256_split_vector_move_misalign): Likewise.
7404 (ix86_expand_vector_logical_operator): Likewise.
7405 (ix86_split_idivmod): Likewise.
7406 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
7407 (ix86_build_const_vector): Likewise.
7408 (ix86_build_signbit_mask): Likewise.
7409 (ix86_match_ccmode): Likewise.
7410 (ix86_cc_modes_compatible): Likewise.
7411 (ix86_expand_branch): Likewise.
7412 (ix86_expand_sse_cmp): Likewise.
7413 (ix86_expand_sse_movcc): Likewise.
7414 (ix86_expand_int_sse_cmp): Likewise.
7415 (ix86_expand_vec_perm_vpermi2): Likewise.
7416 (ix86_expand_vec_perm): Likewise.
7417 (ix86_expand_sse_unpack): Likewise.
7418 (ix86_expand_int_addcc): Likewise.
7419 (ix86_split_to_parts): Likewise.
7420 (ix86_vectorize_builtin_gather): Likewise.
7421 (ix86_vectorize_builtin_scatter): Likewise.
7422 (avx_vpermilp_parallel): Likewise.
7423 (inline_memory_move_cost): Likewise.
7424 (ix86_tieable_integer_mode_p): Likewise.
7425 (x86_maybe_negate_const_int): Likewise.
7426 (ix86_expand_vector_init_duplicate): Likewise.
7427 (ix86_expand_vector_init_one_nonzero): Likewise.
7428 (ix86_expand_vector_init_one_var): Likewise.
7429 (ix86_expand_vector_init_concat): Likewise.
7430 (ix86_expand_vector_init_interleave): Likewise.
7431 (ix86_expand_vector_init_general): Likewise.
7432 (ix86_expand_vector_set): Likewise.
7433 (ix86_expand_vector_extract): Likewise.
7434 (emit_reduc_half): Likewise.
7435 (ix86_emit_i387_round): Likewise.
7436 (ix86_mangle_type): Likewise.
7437 (ix86_expand_round_sse4): Likewise.
7438 (expand_vec_perm_blend): Likewise.
7439 (canonicalize_vector_int_perm): Likewise.
7440 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
7441 (expand_vec_perm_1): Likewise.
7442 (expand_vec_perm_interleave3): Likewise.
7443 (expand_vec_perm_even_odd_pack): Likewise.
7444 (expand_vec_perm_even_odd_1): Likewise.
7445 (expand_vec_perm_broadcast_1): Likewise.
7446 (ix86_vectorize_vec_perm_const_ok): Likewise.
7447 (ix86_expand_vecop_qihi): Likewise.
7448 (ix86_expand_mul_widen_hilo): Likewise.
7449 (ix86_expand_sse2_abs): Likewise.
7450 (ix86_expand_pextr): Likewise.
7451 (ix86_expand_pinsr): Likewise.
7452 (ix86_preferred_simd_mode): Likewise.
7453 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
7454 * config/i386/sse.md (*andnot<mode>3): Likewise.
7455 (<mask_codefor><code><mode>3<mask_name>): Likewise.
7456 (*<code><mode>3): Likewise.
7457 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
7458 (ia64_expand_atomic_op): Likewise.
7459 (ia64_arg_type): Likewise.
7460 (ia64_mode_to_int): Likewise.
7461 (ia64_scalar_mode_supported_p): Likewise.
7462 (ia64_vector_mode_supported_p): Likewise.
7463 (expand_vec_perm_broadcast): Likewise.
7464 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
7465 (iq2000_function_arg_advance): Likewise.
7466 (iq2000_function_arg): Likewise.
7467 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
7468 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
7469 (m68k_libcall_value): Likewise.
7470 (m68k_function_value): Likewise.
7471 (sched_attr_op_type): Likewise.
7472 * config/mcore/mcore.c (mcore_output_move): Likewise.
7473 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
7475 (microblaze_function_arg): Likewise.
7476 * config/mips/mips.c (mips16_build_call_stub): Likewise.
7477 (mips_print_operand): Likewise.
7478 (mips_mode_ok_for_mov_fmt_p): Likewise.
7479 (mips_vector_mode_supported_p): Likewise.
7480 (mips_preferred_simd_mode): Likewise.
7481 (mips_expand_vpc_loongson_even_odd): Likewise.
7482 (mips_expand_vec_unpack): Likewise.
7483 (mips_expand_vi_broadcast): Likewise.
7484 (mips_expand_vector_init): Likewise.
7485 (mips_expand_vec_reduc): Likewise.
7486 (mips_expand_msa_cmp): Likewise.
7487 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
7488 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
7489 (cc_flags_for_mode): Likewise.
7490 * config/msp430/msp430.c (msp430_print_operand): Likewise.
7491 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
7492 (nds32_output_casesi_pc_relative): Likewise.
7493 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7494 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
7495 (nvptx_gen_unpack): Likewise.
7496 (nvptx_gen_pack): Likewise.
7497 (nvptx_gen_shuffle): Likewise.
7498 (nvptx_gen_wcast): Likewise.
7499 (nvptx_preferred_simd_mode): Likewise.
7500 * config/pa/pa.c (pa_secondary_reload): Likewise.
7501 * config/pa/predicates.md (base14_operand): Likewise.
7502 * config/powerpcspe/powerpcspe-c.c
7503 (altivec_resolve_overloaded_builtin): Likewise.
7504 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
7506 (rs6000_preferred_simd_mode): Likewise.
7507 (output_vec_const_move): Likewise.
7508 (rs6000_expand_vector_extract): Likewise.
7509 (rs6000_split_vec_extract_var): Likewise.
7510 (reg_offset_addressing_ok_p): Likewise.
7511 (rs6000_legitimate_offset_address_p): Likewise.
7512 (rs6000_legitimize_address): Likewise.
7513 (rs6000_emit_set_const): Likewise.
7514 (rs6000_const_vec): Likewise.
7515 (rs6000_emit_move): Likewise.
7516 (spe_build_register_parallel): Likewise.
7517 (rs6000_darwin64_record_arg_recurse): Likewise.
7518 (swap_selector_for_mode): Likewise.
7519 (spe_init_builtins): Likewise.
7520 (paired_init_builtins): Likewise.
7521 (altivec_init_builtins): Likewise.
7522 (do_load_for_compare): Likewise.
7523 (rs6000_generate_compare): Likewise.
7524 (rs6000_expand_float128_convert): Likewise.
7525 (emit_load_locked): Likewise.
7526 (emit_store_conditional): Likewise.
7527 (rs6000_output_function_epilogue): Likewise.
7528 (rs6000_handle_altivec_attribute): Likewise.
7529 (rs6000_function_value): Likewise.
7530 (emit_fusion_gpr_load): Likewise.
7531 (emit_fusion_p9_load): Likewise.
7532 (emit_fusion_p9_store): Likewise.
7533 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
7534 (fusion_gpr_mem_load): Likewise.
7535 (fusion_addis_mem_combo_load): Likewise.
7536 (fusion_addis_mem_combo_store): Likewise.
7537 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
7538 (fusion_gpr_mem_load): Likewise.
7539 (fusion_addis_mem_combo_load): Likewise.
7540 (fusion_addis_mem_combo_store): Likewise.
7541 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7543 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
7544 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
7545 (rs6000_preferred_simd_mode): Likewise.
7546 (output_vec_const_move): Likewise.
7547 (rs6000_expand_vector_extract): Likewise.
7548 (rs6000_split_vec_extract_var): Likewise.
7549 (reg_offset_addressing_ok_p): Likewise.
7550 (rs6000_legitimate_offset_address_p): Likewise.
7551 (rs6000_legitimize_address): Likewise.
7552 (rs6000_emit_set_const): Likewise.
7553 (rs6000_const_vec): Likewise.
7554 (rs6000_emit_move): Likewise.
7555 (rs6000_darwin64_record_arg_recurse): Likewise.
7556 (swap_selector_for_mode): Likewise.
7557 (paired_init_builtins): Likewise.
7558 (altivec_init_builtins): Likewise.
7559 (rs6000_expand_float128_convert): Likewise.
7560 (emit_load_locked): Likewise.
7561 (emit_store_conditional): Likewise.
7562 (rs6000_output_function_epilogue): Likewise.
7563 (rs6000_handle_altivec_attribute): Likewise.
7564 (rs6000_function_value): Likewise.
7565 (emit_fusion_gpr_load): Likewise.
7566 (emit_fusion_p9_load): Likewise.
7567 (emit_fusion_p9_store): Likewise.
7568 * config/rx/rx.c (rx_gen_move_template): Likewise.
7569 (flags_from_mode): Likewise.
7570 * config/s390/predicates.md (s390_alc_comparison): Likewise.
7571 (s390_slb_comparison): Likewise.
7572 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
7573 (s390_vector_mode_supported_p): Likewise.
7574 (s390_cc_modes_compatible): Likewise.
7575 (s390_match_ccmode_set): Likewise.
7576 (s390_canonicalize_comparison): Likewise.
7577 (s390_emit_compare_and_swap): Likewise.
7578 (s390_branch_condition_mask): Likewise.
7579 (s390_rtx_costs): Likewise.
7580 (s390_secondary_reload): Likewise.
7581 (__SECONDARY_RELOAD_CASE): Likewise.
7582 (s390_expand_cs): Likewise.
7583 (s390_preferred_simd_mode): Likewise.
7584 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
7585 * config/sh/sh.c (sh_print_operand): Likewise.
7586 (dump_table): Likewise.
7587 (sh_secondary_reload): Likewise.
7588 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7589 * config/sh/sh.md (casesi_worker_1): Likewise.
7590 (casesi_worker_2): Likewise.
7591 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
7592 (fcc_comparison_operator): Likewise.
7593 * config/sparc/sparc.c (sparc_expand_move): Likewise.
7594 (emit_soft_tfmode_cvt): Likewise.
7595 (sparc_preferred_simd_mode): Likewise.
7596 (output_cbranch): Likewise.
7597 (sparc_print_operand): Likewise.
7598 (sparc_expand_vec_perm_bmask): Likewise.
7599 (vector_init_bshuffle): Likewise.
7600 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
7601 (spu_vector_mode_supported_p): Likewise.
7602 (spu_expand_insv): Likewise.
7603 (spu_emit_branch_or_set): Likewise.
7604 (spu_handle_vector_attribute): Likewise.
7605 (spu_builtin_splats): Likewise.
7606 (spu_builtin_extract): Likewise.
7607 (spu_builtin_promote): Likewise.
7608 (spu_expand_sign_extend): Likewise.
7609 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
7610 (tilegx_simd_int): Likewise.
7611 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
7612 (tilepro_simd_int): Likewise.
7613 * config/v850/v850.c (const_double_split): Likewise.
7614 (v850_print_operand): Likewise.
7615 (ep_memory_offset): Likewise.
7616 * config/vax/vax.c (vax_rtx_costs): Likewise.
7617 (vax_output_int_move): Likewise.
7618 (vax_output_int_add): Likewise.
7619 (vax_output_int_subtract): Likewise.
7620 * config/visium/predicates.md (visium_branch_operator): Likewise.
7621 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
7622 (visium_print_operand_address): Likewise.
7623 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7624 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
7625 (xtensa_expand_conditional_branch): Likewise.
7626 (xtensa_copy_incoming_a7): Likewise.
7627 (xtensa_output_literal): Likewise.
7628 * dfp.c (decimal_real_maxval): Likewise.
7629 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
7631 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7632 Alan Hayward <alan.hayward@arm.com>
7633 David Sherwood <david.sherwood@arm.com>
7635 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
7636 (mode_nunits_inline): Likewise.
7637 (mode_inner_inline): Likewise.
7638 (mode_unit_size_inline): Likewise.
7639 (mode_unit_precision_inline): Likewise.
7640 (emit_insn_modes_h): Likewise. Also emit a #define of the
7642 (emit_mode_wider): Add an E_ prefix to mode names.
7643 (emit_mode_complex): Likewise.
7644 (emit_mode_inner): Likewise.
7645 (emit_mode_adjustments): Likewise.
7646 (emit_mode_int_n): Likewise.
7647 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
7648 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
7649 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
7650 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
7651 (CRC32_BUILTIN, ENTRY): Likewise.
7652 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
7653 (aarch64_pop_regs): Likewise.
7654 (aarch64_process_components): Likewise.
7655 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
7656 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
7657 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
7658 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
7659 * config/arm/arm.c (arm_init_libfuncs): Likewise.
7660 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
7662 * config/i386/i386-builtin.def (pcmpestr): Likewise.
7663 (pcmpistr): Likewise.
7664 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
7665 * config/mmix/mmix.c (mmix_output_condition): Likewise.
7666 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
7668 * config/rl78/rl78.c (mduc_regs): Likewise.
7669 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
7670 (htm_expand_builtin): Likewise.
7671 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
7672 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
7673 * config/xtensa/xtensa.c (print_operand): Likewise.
7674 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
7675 (NUM_MODE_VECTOR_INT): Likewise.
7676 * genoutput.c (null_operand): Likewise.
7677 (output_operand_data): Likewise.
7678 * genrecog.c (print_parameter_value): Likewise.
7679 * lra.c (debug_operand_data): Likewise.
7681 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7683 * dwarf2out.c (scompare_loc_descriptor_wide)
7684 (scompare_loc_descriptor_narrow): New functions, split out from...
7685 (scompare_loc_descriptor): ...here.
7686 * expmed.c (emit_store_flag_int): New function, split out from...
7687 (emit_store_flag): ...here.
7689 2017-08-30 Richard Biener <rguenther@suse.de>
7691 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
7692 (dwarf2out_early_finish): Move setting of AT_pubnames from
7693 early debug output to early finish.
7695 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
7697 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
7698 and -mdata-region to the assembler.
7700 2017-08-30 Richard Biener <rguenther@suse.de>
7702 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
7704 (gen_subprogram_die): Add DW_AT_object_pointer only early.
7705 (dwarf2out_early_global_decl): Only generate a DIE for the
7706 abstract origin if it doesn't already exist or is a declaration DIE.
7707 (resolve_addr): Do not add the linkage name twice when
7708 generating a stub DIE for the DW_TAG_GNU_call_site target.
7710 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7712 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7713 Use machine_mode rather than int for arg1_mode.
7715 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
7718 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
7719 that the second argument of the built-in functions to unpack
7720 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
7721 switch statement instead a lot of if statements.
7722 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
7723 Allow 64-bit values to be in Altivec registers as well as
7724 traditional floating point registers.
7725 (pack<mode>, FMOVE128_VSX iterator): Likewise.
7727 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
7729 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
7730 MAX_REGS_PER_ADDRESS == 1.
7732 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
7734 * config/i386/i386.opt (flag_fentry): Do not init to -1.
7735 * config/i386/i386.c (ix86_option_override_internal): Simplify
7736 setting of opts->x_flag_entry.
7738 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7739 Jakub Jelinek <jakub@redhat.com>
7740 Richard Biener <rguenther@suse.de>
7742 PR tree-optimization/81503
7743 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
7744 folded constant fits in the target type; reorder tests for clarity.
7746 2017-08-29 Martin Liska <mliska@suse.cz>
7748 * passes.def: Include pass_lower_switch.
7749 * stmt.c (dump_case_nodes): Remove and move to
7750 tree-switch-conversion.
7751 (case_values_threshold): Likewise.
7752 (expand_switch_as_decision_tree_p): Likewise.
7753 (emit_case_decision_tree): Likewise.
7754 (expand_case): Likewise.
7755 (balance_case_nodes): Likewise.
7756 (node_has_low_bound): Likewise.
7757 (node_has_high_bound): Likewise.
7758 (node_is_bounded): Likewise.
7759 (emit_case_nodes): Likewise.
7760 (struct simple_case_node): New struct.
7761 (add_case_node): Remove.
7762 (emit_case_dispatch_table): Use vector instead of case_list.
7763 (reset_out_edges_aux): Remove.
7764 (compute_cases_per_edge): Likewise.
7765 (expand_case): Build list of simple_case_node.
7766 (expand_sjlj_dispatch_table): Use it.
7767 * tree-switch-conversion.c (struct case_node): Moved from
7768 stmt.c and adjusted.
7769 (emit_case_nodes): Likewise.
7770 (node_has_low_bound): Likewise.
7771 (node_has_high_bound): Likewise.
7772 (node_is_bounded): Likewise.
7773 (case_values_threshold): Likewise.
7774 (reset_out_edges_aux): Likewise.
7775 (compute_cases_per_edge): Likewise.
7776 (add_case_node): Likewise.
7777 (dump_case_nodes): Likewise.
7778 (balance_case_nodes): Likewise.
7779 (expand_switch_as_decision_tree_p): Likewise.
7780 (emit_jump): Likewise.
7781 (emit_case_decision_tree): Likewise.
7782 (try_switch_expansion): Likewise.
7783 (do_jump_if_equal): Likewise.
7784 (emit_cmp_and_jump_insns): Likewise.
7785 (fix_phi_operands_for_edge): New function.
7786 (record_phi_operand_mapping): Likewise.
7787 (class pass_lower_switch): New pass.
7788 (pass_lower_switch::execute): New function.
7789 (make_pass_lower_switch): Likewise.
7790 (conditional_probability):
7791 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
7792 * tree-pass.h: Add make_pass_lower_switch.
7794 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
7797 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
7800 2017-08-29 Richard Biener <rguenther@suse.de>
7802 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
7803 we do not add a DW_AT_inline attribute twice.
7804 (gen_subprogram_die): Remove code setting DW_AT_inline on
7805 DECL_ABSTRACT_P nodes.
7807 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
7809 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
7810 calls to internal functions.
7811 (gimplify_modify_expr): Likewise.
7812 * tree-call-cdce.c (use_internal_fn): Likewise.
7813 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
7814 (convert_to_divmod): Set the nothrow flag.
7815 * tree-if-conv.c (predicate_mem_writes): Likewise.
7816 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7817 (vectorizable_call): Likewise.
7818 (vectorizable_store): Likewise.
7819 (vectorizable_load): Likewise.
7820 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
7821 (vect_recog_mask_conversion_pattern): Likewise.
7823 2017-08-29 Martin Liska <mliska@suse.cz>
7826 * gcc.c (driver_handle_option): Add new argument.
7827 * opts-common.c (handle_option): Pass
7828 target_option_override_hook.
7829 * opts-global.c (lang_handle_option): Add new option.
7830 (set_default_handlers): Add new argument.
7831 (decode_options): Likewise.
7832 * opts.c (target_handle_option): Likewise.
7833 (common_handle_option): Call target_option_override_hook.
7834 * opts.h (struct cl_option_handler_func): Add hook for
7835 target option override.
7836 (struct cl_option_handlers): Likewise.
7837 (set_default_handlers): Add new argument.
7838 (decode_options): Likewise.
7839 (common_handle_option): Likewise.
7840 (target_handle_option): Likewise.
7841 * toplev.c (toplev::main): Pass targetm.target_option.override
7844 2017-08-29 Richard Biener <rguenther@suse.de>
7845 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7847 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
7848 life to the active subtree.
7850 2017-08-28 Jeff Law <law@redhat.com>
7852 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
7853 derive_equivalences.
7854 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
7856 (edge_info::derive_equivalences): New private member function
7858 * tree-ssa-dom.c (class edge_info): Changed from a struct
7859 to a class. Add ctor/dtor, methods and data members.
7860 (edge_info::edge_info): Renamed from allocate_edge_info.
7861 Initialize additional members.
7862 (edge_info::~edge_info): New.
7863 (free_dom_edge_info): Delete the edge info.
7864 (record_edge_info): Use new class & associated member functions.
7865 Tighten forms for testing for edge equivalences.
7866 (record_temporary_equivalences): Iterate over the simple
7867 equivalences rather than assuming there's only one per edge.
7868 (cprop_into_successor_phis): Iterate over the simple
7869 equivalences rather than assuming there's only one per edge.
7870 (optimize_stmt): Use operand_equal_p rather than pointer
7871 equality for mini-DSE code.
7873 2017-08-28 Nathan Sidwell <nathan@acm.org>
7875 * gcc.c (execute): Fold SIGPIPE handling into switch
7876 statement. Adjust internal error message.
7878 2017-08-28 Richard Biener <rguenther@suse.de>
7881 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
7882 Do nothing for removed DIEs.
7884 2017-08-28 Richard Biener <rguenther@suse.de>
7886 PR tree-optimization/81977
7887 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
7890 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
7893 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
7894 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
7895 using targetm.gen_mem_thread_fence.
7897 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
7900 * config/i386/i386.md (*<btsc><mode>): Change operand 2
7901 predicate to register_operand. Reorder operands.
7902 (*btr<mode>): Ditto.
7903 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
7904 (*btr<mode>_mask): Ditto.
7906 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
7908 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
7909 * config/rs6000/xmmintrin.h: New file.
7910 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
7912 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7915 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
7916 parameter and_insn and return it.
7917 (recombine_lvx_pattern): Insert a copy to ensure availability of
7918 the base register of the copied masking operation at the point of
7919 the instruction replacement.
7920 (recombine_stvx_pattern): Likewise.
7922 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
7924 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
7925 undocumented switches.
7926 (-mpower9-dform-vector): Likewise.
7927 (-mpower9-dform): Likewise.
7928 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
7929 comments to delete references to -mpower9-dform* switches.
7930 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
7931 Delete reference to -mpower9-dform* switches, test for
7932 -mpower9-vector instead.
7933 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
7934 (OTHER_P9_VECTOR_MASKS): Likewise.
7935 (POWERPC_MASKS): Likewise.
7936 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
7937 tests against -mpower9-dform* to -mpower9-vector. Delete code
7938 that checked for -mpower9-dform* consistancy with other options.
7939 Add test for -mpower9-misc to enable other power9 switches.
7940 (rs6000_init_hard_regno_mode_ok): Likewise.
7941 (rs6000_option_override_internal): Likewise.
7942 (rs6000_emit_prologue): Likewise.
7943 (rs6000_emit_epilogue): Likewise.
7944 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
7945 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
7946 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
7948 (emit_fusion_p9_store): Likewise.
7949 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
7950 resetting these macros if the assembler does not support ISA 3.0
7952 (TARGET_P9_DFORM_VECTOR): Likewise.
7953 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
7954 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
7956 2017-08-25 Alan Modra <amodra@gmail.com>
7958 PR rtl-optimization/81747
7959 * cse.c (cse_extended_basic_block): Don't attempt to record
7960 equivalences for degenerate conditional branches that branch
7961 to their fall-through.
7963 2017-08-24 Martin Sebor <msebor@redhat.com>
7966 * gimple-fold.c (size_must_be_zero_p): New function.
7967 (gimple_fold_builtin_memory_op): Call it.
7969 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
7971 * config/rs6000/mm_malloc.h: New file.
7973 2017-08-24 Bin Cheng <bin.cheng@arm.com>
7975 PR tree-optimization/81913
7976 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
7977 analysis when either IVs in condition can wrap.
7979 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
7981 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
7982 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
7984 2017-08-24 Richard Biener <rguenther@suse.de>
7987 * targhooks.c (default_target_can_inline_p): Properly
7988 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
7989 is present and always compare.
7990 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
7991 infer -mfpmath=sse from TARGET_SSE_P.
7992 (ix86_can_inline_p): Properly use target_option_default_node when
7993 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
7995 2017-08-24 Richard Biener <rguenther@suse.de>
7998 * dwarf2out.c (output_die): Handle flag_generate_offload like
8000 (output_comp_unit): Likewise.
8001 (gen_array_type_die): Likewise.
8002 (dwarf2out_early_finish): Likewise.
8003 (note_variable_value_in_expr): Likewise.
8004 (dwarf2out_finish): Likewise. Adjust assert.
8005 * cgraphunit.c (symbol_table::compile): Move setting of
8006 flag_generate_offload earlier ...
8007 (symbol_table::finalize_compilation_unit): ... here, before
8008 early debug finalization.
8010 2017-08-24 Richard Biener <rguenther@suse.de>
8012 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
8013 and ipa-fnsummary.h.
8014 (ix86_can_inline_p): When ix86_fpmath flags do not match
8015 check whether the callee uses FP math at all.
8017 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
8020 * tree-ssanames.c (get_nonzero_bits): Use element_precision
8021 instead of TYPE_PRECISION.
8023 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
8024 Alan Hayward <alan.hayward@arm.com>
8025 David Sherwood <david.sherwood@arm.com>
8027 * combine.c (make_extraction): Use subreg_offset_from_lsb.
8029 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
8031 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
8033 (ix86_frame::stack_realign_allocate): New field.
8034 (struct machine_frame_state): Modify comments.
8035 (machine_frame_state::sp_realigned_fp_end): New field.
8036 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
8038 (sp_valid_at): Add assertion to assure no attempt to access invalid
8039 offset of a realigned stack.
8040 (fp_valid_at): Likewise.
8041 (choose_baseaddr): Modify comments.
8042 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
8043 ix86_expand_prologue.
8044 (ix86_expand_prologue): Modify stack realignment and allocation.
8045 (ix86_expand_epilogue): Modify comments.
8046 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
8047 avx2_runtime, avx512f, and avx512f_runtime.
8049 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
8051 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
8052 (mstackrealign): Do not init to -1.
8053 * config/i386/i386.c (ix86_option_override_internal):
8054 Check opts_set, not opts when setting default value of
8055 opts->x_ix86_force_align_arg_pointer.
8057 2017-08-23 Richard Biener <rguenther@suse.de>
8059 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
8060 lang_hooks.decl_printable_name.
8061 * print-rtl-function.c (print_rtx_function): Likewise.
8062 * tree-pretty-print.c (dump_function_header): Likewise.
8064 2017-08-23 Richard Biener <rguenther@suse.de>
8067 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
8068 -g0 at compile-time.
8070 2017-08-23 Tamar Christina <tamar.christina@arm.com>
8073 * doc/sourcebuild.texi (Other hardware attributes):
8076 2017-08-23 Tamar Christina <tamar.christina@arm.com>
8079 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
8080 Fix single-use check.
8082 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
8084 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
8086 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
8088 * doc/install.texi: Modify to add more details on running selected
8091 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
8093 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
8094 is combined with -mabi=ms.
8095 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
8098 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8100 PR tree-optimization/81488
8101 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
8102 and cached_basis fields.
8103 (MAX_SPREAD): New constant.
8104 (alloc_cand_and_find_basis): Initialize new fields.
8105 (clear_visited): New function.
8106 (create_phi_basis_1): Rename from create_phi_basis, set visited
8107 and cached_basis fields.
8108 (create_phi_basis): New wrapper function.
8109 (phi_add_costs_1): Rename from phi_add_costs, add spread
8110 parameter, set visited field, short-circuit when limits reached.
8111 (phi_add_costs): New wrapper function.
8112 (record_phi_increments_1): Rename from record_phi_increments, set
8114 (record_phi_increments): New wrapper function.
8115 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
8116 (phi_incr_cost): New wrapper function.
8117 (all_phi_incrs_profitable_1): Rename from
8118 all_phi_incrs_profitable, set visited field.
8119 (all_phi_incrs_profitable): New wrapper function.
8121 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
8122 Alan Hayward <alan.hayward@arm.com>
8123 David Sherwood <david.sherwood@arm.com>
8125 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
8126 that takes the outer and inner modes.
8127 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
8128 comparison as the canonical test for a paradoxical subreg.
8129 * combine.c (simplify_set): Use paradoxical_subreg_p.
8130 (make_extraction): Likewise.
8131 (force_to_mode): Likewise.
8132 (rtx_equal_for_field_assignment_p): Likewise.
8133 (gen_lowpart_for_combine): Likewise.
8134 (simplify_comparison): Likewise.
8135 * cse.c (equiv_constant): Likewise.
8136 * expmed.c (store_bit_field_1): Likewise.
8137 * final.c (alter_subreg): Likewise.
8138 * fwprop.c (propagate_rtx): Likewise.
8139 (forward_propagate_subreg): Likewise.
8140 * ira-conflicts.c (ira_build_conflicts): Likewise.
8141 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
8142 * lra-constraints.c (curr_insn_transform): Likewise.
8143 (split_reg): Likewise.
8144 * lra-eliminations.c (move_plus_up): Likewise.
8145 (lra_eliminate_regs_1): Likewise.
8146 * recog.c (general_operand): Likewise.
8147 * ree.c (combine_reaching_defs): Likewise.
8148 * reload.c (push_reload): Likewise.
8149 (find_reloads): Likewise.
8150 * reload1.c (elimination_effects): Likewise.
8151 (compute_reload_subreg_offset): Likewise.
8152 (choose_reload_regs): Likewise.
8153 * rtlanal.c (subreg_lsb_1): Likewise.
8154 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8155 (simplify_subreg): Likewise.
8156 * var-tracking.c (track_loc_p): Likewise.
8157 * emit-rtl.c (byte_lowpart_offset): Likewise.
8158 (paradoxical_subreg_p): Delete out-of-line definition.
8160 2017-08-22 Jeff Law <law@redhat.com>
8162 PR tree-optimization/81741
8163 PR tree-optimization/71947
8164 * tree-ssa-dom.c: Include tree-inline.h.
8165 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
8166 equivalences if one is more expensive to compute than the other.
8167 * tree-ssa-scopedtables.h (class const_or_copies): Make
8168 record_const_or_copy_raw method private.
8169 (class avail_exprs_stack): New method simplify_binary_operation.
8170 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
8171 avail_exprs_stack::simplify_binary_operation as needed.
8172 (avail_exprs_stack::simplify_binary_operation): New function.
8174 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
8176 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
8177 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
8178 (DOT_SYMBOLS): Likewise.
8179 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
8180 (RELOCATABLE_NEEDS_FIXUP): Likewise.
8181 (RS6000_ABI_NAME): Likewise.
8182 (TARGET_CMODEL): Likewise.
8183 (TOC_SECTION_ASM_OP): Likewise.
8184 (SET_CMODEL): New macro.
8185 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
8187 2017-08-22 Richard Biener <rguenther@suse.de>
8189 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
8190 to allow for free-lang-data replacements similar to verify_type_variant.
8192 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
8193 Alan Hayward <alan.hayward@arm.com>
8194 David Sherwood <david.sherwood@arm.com>
8196 * config/aarch64/aarch64.md (casesi): Use DImode rather than
8197 VOIDmode for the LABEL_REF.
8199 2017-08-22 Richard Biener <rguenther@suse.de>
8201 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
8203 2017-08-22 Richard Biener <rguenther@suse.de>
8205 * common.opt (feliminate-dwarf2-dups): Ignore.
8206 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
8207 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
8208 same_die_p_wrap, compute_section_prefix,
8209 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
8210 (comdat_symbol_id, comdat_symbol_number): Likewise.
8211 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
8213 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
8214 (output_die): Mark unreachable path unreachable.
8215 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
8216 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
8217 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
8218 (dwarf2out_early_finish): Likewise.
8220 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
8222 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
8224 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
8227 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
8228 not VAR_P. Filter attribute warnings with OPT_Wattributes.
8229 (avr_attribute_table) <io, io_low, address>: Initialize
8230 .decl_required with true.
8232 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
8234 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
8235 undocumented debugging options.
8236 (-mvsx-scalar-double): Likewise.
8237 (-mallow-df-permute): Likewise.
8238 (-mvectorize-builtins): Likewise.
8239 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8240 (rs6000_builtin_vectorized_function): Likewise.
8241 (rs6000_builtin_md_vectorized_function): Likewise.
8242 (rs6000_opt_vars): Likewise.
8244 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
8247 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
8248 (*btrq_imm): Rename from *btrq.
8249 (*btcq_imm): Rename from *btcq.
8250 (btsc): New code attribute.
8251 (*<btsc><mode>): New insn pattern.
8252 (*btr<mode>): Ditto.
8253 (*<btsc><mode>_mask): New insn_and_split pattern.
8254 (*btr<mode>_mask): Ditto.
8256 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8257 Alan Hayward <alan.hayward@arm.com>
8258 David Sherwood <david.sherwood@arm.com>
8260 * function.c (pad_below): Simplify padding calculation.
8262 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8263 Alan Hayward <alan.hayward@arm.com>
8264 David Sherwood <david.sherwood@arm.com>
8266 * target.def (function_prologue): Remove frame size argument.
8267 (function_epilogue): Likewise.
8268 * doc/tm.texi: Regenerate.
8269 * final.c (final_start_function): Update call to function_prologue.
8270 (final_end_function): Update call to function_epilogue.
8271 (default_function_pro_epilogue): Remove frame size argument.
8272 * output.h (default_function_pro_epilogue): Likewise.
8273 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
8274 (arm_output_function_prologue): Likewise.
8275 * config/frv/frv.c (frv_function_prologue): Likewise.
8276 (frv_function_epilogue): Likewise.
8277 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
8278 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
8279 (ia64_output_function_epilogue): Likewise.
8280 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
8281 (m32r_output_function_epilogue): Likewise.
8282 * config/microblaze/microblaze.c (microblaze_function_prologue)
8283 (microblaze_function_epilogue): Likewise.
8284 * config/mips/mips.c (mips_output_function_prologue): Likewise.
8285 (mips_output_function_epilogue): Likewise.
8286 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
8287 (mmix_target_asm_function_epilogue): Likewise.
8288 * config/msp430/msp430.c (msp430_start_function): Likewise.
8289 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
8290 (nds32_asm_function_epilogue): Likewise.
8291 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
8292 * config/pa/pa.c (pa_output_function_prologue): Likewise.
8293 (pa_output_function_epilogue): Likewise.
8294 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
8295 (rs6000_output_function_epilogue): Likewise.
8296 * config/rl78/rl78.c (rl78_start_function): Likewise.
8297 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
8298 (rs6000_output_function_epilogue): Likewise.
8299 * config/rx/rx.c (rx_output_function_prologue): Likewise.
8300 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
8301 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
8302 (sparc_asm_function_epilogue): Likewise.
8304 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8306 * tree.h (type_has_mode_precision_p): New function.
8307 * convert.c (convert_to_integer_1): Use it.
8308 * expr.c (expand_expr_real_2): Likewise.
8309 (expand_expr_real_1): Likewise.
8310 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
8311 * match.pd: Likewise.
8312 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
8313 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
8314 * tree-tailcall.c (process_assignment): Likewise.
8315 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8316 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
8317 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
8318 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8319 (vectorizable_assignment): Likewise.
8320 (vectorizable_shift): Likewise.
8321 (vectorizable_operation): Likewise.
8322 * tree-vrp.c (register_edge_assert_for_2): Likewise.
8324 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
8326 * match.pd: Add pow (C, x) simplification.
8328 2017-08-21 Richard Biener <rguenther@suse.de>
8330 PR tree-optimization/81900
8331 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
8332 for blocks with abnormal predecessors.
8333 (compute_antic): Do not set visited flag prematurely.
8335 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
8338 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
8340 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8342 * stor-layout.h (vector_type_mode): Move to...
8343 * tree.h (vector_type_mode): ...here.
8344 * stor-layout.c (vector_type_mode): Move to...
8345 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
8347 2017-08-21 Richard Biener <rguenther@suse.de>
8349 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
8350 register_external_die hooks.
8351 (debug_false_tree_charstarstar_uhwistar): Declare.
8352 (debug_nothing_tree_charstar_uhwi): Likewise.
8353 * debug.c (do_nothing_debug_hooks): Adjust.
8354 (debug_false_tree_charstarstar_uhwistar): New do nothing.
8355 (debug_nothing_tree_charstar_uhwi): Likewise.
8356 * dbxout.c (dbx_debug_hooks): Adjust.
8357 (xcoff_debug_hooks): Likewise.
8358 * sdbout.c (sdb_debug_hooks): Likewise.
8359 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
8360 * dwarf2out.c (macinfo_label_base): New global.
8361 (dwarf2out_register_external_die): New function for the
8362 register_external_die hook.
8363 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
8364 (dwarf2_debug_hooks): Use them.
8365 (dwarf2_lineno_debug_hooks): Adjust.
8366 (struct die_struct): Add with_offset flag.
8367 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
8368 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
8369 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
8370 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
8371 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
8372 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
8373 defining section names for the early LTO debug variants.
8374 (reset_indirect_string): New helper.
8375 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
8376 (print_dw_val): Add support for offsetted symbol references.
8377 (get_ultimate_context): Split out from is_cxx.
8378 (is_cxx): Use get_ultimate_context.
8379 (is_fortran): Add decl overload.
8380 (compute_comp_unit_symbol): Split out worker from
8381 compute_section_prefix.
8382 (compute_section_prefix): Call compute_comp_unit_symbol and
8383 set comdat_type_p here.
8384 (output_die): Skip DIE symbol output for the LTO added one.
8385 Handle DIE symbol references with offset.
8386 (output_comp_unit): Guard section name mangling properly.
8387 For LTO debug sections emit a symbol at the section beginning
8388 which we use to refer to its DIEs.
8389 (add_abstract_origin_attribute): For DIEs registered via
8390 dwarf2out_register_external_die directly refer to the early
8391 DIE rather than indirectly through the shadow one we created.
8392 Remove obsolete call to dwarf2out_abstract_function for
8393 non-function/block origins.
8394 (gen_array_type_die): When generating early LTO debug do
8395 not emit DW_AT_string_length.
8396 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
8397 late when in LTO. As suggested place a gcc_unreachable for
8398 the DECL_ABSTRACT_P case.
8399 (gen_subprogram_die): Avoid another specification DIE
8400 for early built declarations/definitions for the late LTO case.
8401 (gen_variable_die): Add type references for late duplicated VLA dies
8403 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
8404 we have the abstract instance already.
8405 (process_scope_var): Adjust decl DIE contexts in LTO which
8406 first puts them in limbo.
8407 (gen_decl_die): Do not generate type DIEs late apart from
8408 types for VLAs or for decls we do not yet have a DIE. Do not
8409 call dwarf2out_abstract_function late.
8410 (dwarf2out_early_global_decl): Make sure to create DIEs
8411 for abstract instances of a decl first.
8412 (dwarf2out_late_global_decl): Adjust comment.
8413 (output_macinfo_op): With multiple macro sections use
8414 macinfo_label_base to distinguish labels.
8415 (output_macinfo): Likewise. Update macinfo_label_base.
8416 Pass in the line info label.
8417 (note_variable_value_in_expr): When generating LTO resolve
8418 all variable values here by generating DIEs as needed.
8419 (init_sections_and_labels): Add early LTO debug flag parameter
8420 and generate different sections and names if set. Add generation
8421 counter for the labels so we can have multiple of them.
8422 (reset_dies): Helper to allow DIEs to be output multiple times.
8423 (dwarf2out_finish): When outputting DIEs to the fat part of an
8424 LTO object first reset DIEs.
8425 (dwarf2out_early_finish): Output early DIEs when generating LTO.
8426 (modified_type_die): Check for decl_ultimate_origin being self
8428 (gen_type_die_with_usage): Likewise.
8429 (gen_typedef_die): Allow decl_ultimate_origin being self.
8430 (set_decl_abstract_flags): Remove.
8431 (set_block_abstract_flags): Likewise.
8432 (dwarf2out_abstract_function): Treat the early generated DIEs
8433 as the abstract copy and only add DW_AT_inline and
8434 DW_AT_artificial here and call set_decl_origin_self.
8435 If the DIE has an abstract origin don't do anything.
8436 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
8437 if we have none yet (Go fails to build one, PR78628).
8438 (variably_modified_type_p): Prevent endless recursion for Ada
8439 cyclic pointer types.
8440 * lto-streamer-in.c: Include debug.h.
8441 (dref_queue): New global.
8442 (lto_read_tree_1): Stream in DIE references.
8443 (lto_input_tree): Register DIE references.
8444 (input_function): Stream DECL_DEBUG_ARGS.
8445 * lto-streamer-out.c: Include debug.h.
8446 (lto_write_tree_1): Output DIE references.
8447 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
8448 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
8449 (output_function): Stream DECL_DEBUG_ARGS.
8450 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
8451 Stream DECL_ABSTRACT_ORIGIN.
8452 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
8453 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
8454 DECL_CONTEXT for file-scope decls.
8455 * lto-streamer.h (struct dref_entry): Declare.
8456 (dref_queue): Likewise.
8457 * cfgexpand.c (pass_expand::execute): Do not call the
8458 outlining_inline_function hook here.
8459 * lto-wrapper.c (debug_obj): New global.
8460 (tool_cleanup): Unlink it if required.
8461 (debug_objcopy): New function.
8462 (run_gcc): Handle early debug sections in the IL files by
8463 extracting them to separate files, partially linkin them and
8464 feeding the result back as result to the linker.
8465 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
8466 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
8467 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
8468 sections into a separate segment.
8469 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
8471 (darwin_asm_dwarf_section): Likewise.
8472 (darwin_asm_output_dwarf_offset): Likewise.
8473 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
8475 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8476 Alan Hayward <alan.hayward@arm.com>
8477 David Sherwood <david.sherwood@arm.com>
8479 * read-md.h (md_reader::record_potential_iterator_use): Replace
8480 pointer argument with an rtx and an index.
8481 * read-rtl.c (iterator_group::apply_iterator): Likewise.
8482 (apply_mode_iterator): Likewise.
8483 (apply_code_iterator): Likewise.
8484 (apply_int_iterator): Likewise.
8485 (apply_subst_iterator): Likewise.
8486 (record_iterator_use): Likewise.
8487 (record_attribute_use): Likewise.
8488 (md_reader::record_potential_iterator_use): Likewise. Update calls
8489 to record_iterator_use and apply_iterator.
8490 (iterator_use): Replace ptr with x and index.
8491 (attribute_use): Likewise.
8492 (apply_attribute_uses): Update calls to apply_iterator.
8493 (apply_iterators): Likewise. Update initialization of iterator_use.
8494 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
8495 and record_potential_iterator_use.
8496 (rtx_reader::read_rtx_operand): Likewise.
8498 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8499 Alan Hayward <alan.hayward@arm.com>
8500 David Sherwood <david.sherwood@arm.com>
8502 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
8505 2017-08-21 Richard Biener <rguenther@suse.de>
8508 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
8509 at struct end conservatively when comparing common bases.
8511 2017-08-21 Richard Biener <rguenther@suse.de>
8513 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
8514 (mem_ref_in_stmt): Remove.
8515 (determine_max_movement): Use ref index to get at the reference.
8516 (invariantness_dom_walker::before_dom_children): Deal with
8517 lim data already initialized.
8518 (gather_mem_refs_stmt): Initialize lim data and record ref index.
8520 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
8522 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
8523 (TARGET_ISA_ROUND): Ditto.
8524 (TARGET_ROUND): Ditto.
8525 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
8526 * config/i386/i386.md: Ditto.
8527 * config/i386/sse.md: Ditto.
8528 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
8529 with OPTION_MASK_ISA_SSE4_1.
8531 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
8534 * doc/extend.texi (x86 Built-in Functions): Correct the name of
8535 __builtin_ia32_lzcnt_u16.
8537 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
8540 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
8541 (rs6000_set_current_function): Rewrite function to use it.
8543 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
8546 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
8547 and TYPE_WARN_IF_NOT_ALIGN.
8548 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
8549 (handle_warn_if_not_align): New.
8550 (place_union_field): Call handle_warn_if_not_align.
8551 (place_field): Call handle_warn_if_not_align.
8552 Copy TYPE_WARN_IF_NOT_ALIGN.
8553 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
8554 (layout_type): Likewise.
8555 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
8557 (tree_decl_common): Add warn_if_not_align.
8558 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
8559 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
8560 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
8561 (DECL_WARN_IF_NOT_ALIGN): Likewise.
8562 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
8563 * doc/extend.texi: Document warn_if_not_aligned attribute.
8564 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
8566 2017-08-17 Martin Liska <mliska@suse.cz>
8569 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
8570 (get_data_dependence): Use it as pointer type.
8571 (distribute_loop): Likewise.
8573 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8575 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
8576 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
8577 (p8_vmrgow_<mode>_direct): New define_insn.
8578 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
8579 handle endianness for vmrgew and vmrgow permute patterns.
8581 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
8583 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
8584 * config/rs6000/rs6000-cpus.def: Remove comment.
8585 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
8586 (POWERPC_MASKS): Likewise.
8587 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
8588 use of TARGET_VSX_TIMODE.
8589 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
8590 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
8591 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
8592 (rs6000_option_override_internal): Remove dead code.
8593 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
8594 (rs6000_legitimize_reload_address): Likewise.
8595 (rs6000_legitimate_address_p): Likewise.
8596 (rs6000_opt_masks): Delete "vsx-timode".
8597 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
8598 from function comment.
8599 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
8600 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
8601 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
8603 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
8604 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
8605 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
8608 (define_peephole2): Likewise.
8610 2017-08-17 Martin Sebor <msebor@redhat.com>
8613 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
8614 past the end of an array.
8615 (test_pp_format): Add test cases.
8617 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
8619 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
8620 missing ECF_NOTHROW flags.
8622 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
8625 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
8626 operands residing in integer registers.
8627 (*vsx_le_perm_load_<mode>): Likewise.
8628 (*vsx_le_perm_store_<mode>): Likewise.
8629 (define_peephole2): Add peepholes to optimize the above.
8631 2017-08-17 Marek Polacek <polacek@redhat.com>
8634 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
8635 to mimic what shorten_compare did. Change the return type to bool.
8636 (fold_cond_expr_with_comparison): Update call to
8637 operand_equal_for_comparison_p.
8638 (fold_ternary_loc): Likewise.
8640 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
8642 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
8644 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
8646 2017-08-17 Richard Biener <rguenther@suse.de>
8648 * tree-ssa-structalias.c (solve_graph): When propagating
8649 to successors update the graphs succ edges and avoid duplicate work.
8651 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
8654 * config/i386/i386.c (ix86_option_override_internal): Save target
8655 specific options after ix86_stack_protector_guard_reg was changed.
8657 2017-08-17 Richard Biener <rguenther@suse.de>
8659 PR tree-optimization/81827
8660 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
8661 (new_var_info): Initialize it conservatively.
8662 (get_call_vi): Mark register vars.
8663 (new_scalar_tmp_constraint_exp): Likewise.
8664 (handle_rhs_call): Likewise.
8665 (handle_const_call): Likewise.
8666 (create_function_info_for): Likewise.
8667 (solve_constraints): Sort varinfos to separate register from
8668 non-register vars to pack points-to solution bitmaps during
8671 2017-08-17 Marek Polacek <polacek@redhat.com>
8673 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
8675 2017-08-17 Richard Biener <rguenther@suse.de>
8677 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
8678 to true when overflow is undefined and we saturated the result.
8680 2017-08-17 Alan Modra <amodra@gmail.com>
8683 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
8684 Don't use store multiple if only one reg needs saving.
8685 (interesting_frame_related_regno): New function.
8686 (rs6000_frame_related): Don't emit frame info for regs that
8688 (rs6000_emit_epilogue): Likewise.
8690 2017-08-16 Nathan Sidwell <nathan@acm.org>
8692 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
8693 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
8694 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
8695 * tree.c (free_lang_data_in_type): Use else-if chain. Always
8696 clear TYPE_LANG_1. Remove obsolete member-function stripping.
8697 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
8698 (verify_type): Adjust for TYPE_BINFO move.
8699 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
8700 process TYPE_BINFO directly.
8701 (hash_tree): Likewise.
8702 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
8704 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
8707 2017-08-16 David Malcolm <dmalcolm@redhat.com>
8709 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
8711 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
8714 * config/i386/i386.md (*anddi_1_btr): Change predicates of
8715 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
8716 Add ix86_binary_operator_ok to insn constraint.
8717 (*iordi_1_bts): Ditto.
8718 (*xordi_1_btc): Ditto.
8719 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
8720 Update corresponding peephole2 pattern.
8724 2017-08-16 Bin Cheng <bin.cheng@arm.com>
8726 PR tree-optimization/81832
8727 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
8728 copy loop header which has IFN_LOOP_DIST_ALIAS call.
8730 2017-08-16 Marek Polacek <polacek@redhat.com>
8733 * fold-const.c (fold_indirect_ref_1): Restore original behavior
8734 regarding size_zero_node.
8736 2017-08-16 Martin Liska <mliska@suse.cz>
8739 * config.gcc: Respect previously set extra_objs in case
8742 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
8744 PR tree-optimization/81835
8745 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
8746 the phi SSA_NAME. Check that the condition in a COND_EXPR does
8747 not depend on the phi.
8749 2017-08-16 Alan Modra <amodra@gmail.com>
8751 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
8754 2017-08-16 Alan Modra <amodra@gmail.com>
8756 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
8757 (save_reg_p): ..into this. Update all callers.
8758 (first_reg_to_save): Simplify.
8760 2017-08-16 Alan Modra <amodra@gmail.com>
8762 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
8765 2017-08-15 Joseph Myers <joseph@codesourcery.com>
8769 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
8770 constant count if that count is less than 32.
8772 2017-08-15 Nathan Sidwell <nathan@acm.org>
8774 * gcc.c (execute): Emit friendlier message if inferior is killed
8775 by an external cause.
8777 2017-08-15 Richard Biener <rguenther@suse.de>
8779 PR tree-optimization/81790
8780 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
8781 CONSTRUCTORs from simplifying and VN.
8783 2017-08-14 Martin Sebor <msebor@redhat.com>
8785 * builtin-attrs.def: Add comments.
8787 2017-08-14 Martin Sebor <msebor@redhat.com>
8790 * doc/extend.texi (attribute nonstring): Document new attribute.
8792 2017-08-14 Martin Sebor <msebor@redhat.com>
8795 * tree-diagnostic.c (default_tree_printer): Handle %G.
8796 * gimple-pretty-print.h (percent_G_format): Declare new function.
8797 * gimple-pretty-print.c (percent_G_format): Define.
8798 * tree-pretty-print.c (percent_K_format): Add argument.
8800 2017-08-14 Martin Sebor <msebor@redhat.com>
8802 PR translation/79998
8803 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
8804 Remove a stray space.
8806 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
8809 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
8810 (*iordi_1_bts): Ditto.
8811 (*xordi_1_btc): Ditto.
8813 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8816 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
8817 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8819 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
8820 quoted strings, and make more translator-friendly.
8821 (darwin_rs6000_override_options): Likewise.
8822 (rs6000_option_override_internal): Likewise.
8823 (rs6000_return_in_memory): Fix overlong line.
8824 (init_cmulative_args): Use quoted strings, and make more
8825 translator-friendly.
8826 (rs6000_pass_by_reference): Fix overlong line.
8827 (def_builtin): Use quoted strings.
8828 (altivec_expand_predicate_builtin): Use quoted strings, and make
8829 more translator-friendly.
8830 (htm_expand_builtin): Use quoted strings.
8831 (cpu_expand_builtin): Use quoted strings, and make more
8832 translator-friendly.
8833 (altivec_expand_builtin): Likewise.
8834 (paired_expand_predicate_builtin): Likewise.
8835 (rs6000_invalid_builtin): Likewise.
8836 (builtin_function_type): Use quoted strings.
8837 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
8838 more translator-friendly.
8839 (rs6000_trampoline_init): Likewise.
8840 (rs6000_handle_altivec_attribute): Likewise.
8841 (rs6000_inner_target_options): Use quoted strings.
8842 (rs6000_disable_incompatible_switches): Likewise.
8843 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
8844 strings, and make more translator-friendly.
8845 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
8847 2017-08-14 Bin Cheng <bin.cheng@arm.com>
8849 PR tree-optimization/81799
8850 * tree-loop-distribution.c (version_loop_by_alias_check): Force
8851 cond_expr to simple gimple operand.
8853 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
8856 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
8858 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
8862 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
8863 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
8864 TARGET_GASISR_PROLOGUES.
8865 * config/avr/avr.c (avr_option_override): Same.
8866 (avr_pass_pre_proep::execute): Same.
8868 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
8871 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
8872 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
8874 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
8876 * config/i386/i386.md (*load_tp_<mode>): Redefine as
8877 define_insn_and_split. Split to a memory load from 0 in
8878 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
8879 using PTR mode iterator.
8880 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
8881 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
8882 (*add_tp_<mode>): Redefine as define_insn_and_split.
8883 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
8884 address space. Merge with *add_tp_x32 using PTR mode iterator.
8885 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
8886 Split to an add with a memory load from 0 in
8887 DEFAULT_TLS_SEG_REG address space.
8889 2017-08-12 Andrew Pinski <apinski@cavium.com>
8891 * config/aarch64/aarch64-option-extensions.def (rdma):
8892 Fix feature string to what Linux prints out in /proc/cpuinfo.
8894 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
8897 * dwarf2out.c (modified_type_die): For C typedef types that have
8898 an ultimate origin, process the ultimate origin instead of the
8900 (gen_typedef_die): Assert that input DECLs have no ultimate
8902 (gen_type_die_with_usage): For typedef variants that have an
8903 ultimate origin, just call gen_decl_die on the original DECL.
8904 (process_scope_var): Avoid creating DIEs for local typedefs and
8905 concrete static variables.
8907 2017-08-12 Alan Modra <amodra@gmail.com>
8911 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
8912 match gnu-user.h startfile.
8913 (ENDFILE_LINUX_SPEC): Similarly.
8915 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
8918 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
8920 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
8922 2017-08-11 Tamar Christina <tamar.christina@arm.com>
8923 * config/aarch64/aarch64.md (mov<mode>): Change.
8924 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
8925 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
8926 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
8928 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
8930 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
8931 for storage order barriers.
8933 2017-08-11 Martin Liska <mliska@suse.cz>
8936 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
8937 variables of void type.
8939 2017-08-11 Martin Liska <mliska@suse.cz>
8941 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
8942 TARGET_SUPPORTS_ALIASES.
8943 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
8944 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
8945 (optimize_weakref): Likewise.
8946 * symtab.c (symtab_node::noninterposable_alias): Likewise.
8947 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
8948 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
8950 2017-08-11 Martin Liska <mliska@suse.cz>
8953 * config/i386/i386.c (make_resolver_func): Do complete
8954 refactoring of the function.
8956 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
8959 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
8960 * config/i386/i386.c (ix86_stack_protect_guard): Use
8961 ix86_stack_protect_guard_symbol_str to generate varible declaration.
8962 * doc/invoke.texi (x86 Options): Document
8963 -mstack-protector-guard-symbol= option.
8965 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
8967 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
8968 * config/i386/i386.c (ix86_split_stack_guard): New function.
8969 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
8970 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
8971 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
8972 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
8973 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
8974 (split_stack_space_check): Call ix86_split_stack_guard.
8976 2017-08-10 Martin Sebor <msebor@redhat.com>
8978 * print-tree.c (print_node): Print location using the established
8980 Replace spaces with colons.
8981 (debug_raw, debug): Ditto.
8983 2017-08-10 Martin Sebor <msebor@redhat.com>
8986 * pretty-print.c (pp_format): Correct the handling of %s precision.
8988 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
8991 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
8993 (ix86_finalize_stack_frame_flags): This. Also clear
8994 frame_pointer_needed if -fno-omit-frame-pointer is used without
8996 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
8997 with ix86_finalize_stack_frame_flags.
8998 (ix86_expand_epilogue): Likewise.
8999 (ix86_expand_split_stack_prologue): Likewise.
9000 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
9002 2017-08-10 Martin Liska <mliska@suse.cz>
9005 * c-attribs.c (handle_target_attribute):
9006 Report warning for an empty string argument of target attribute.
9008 2017-08-09 Jakub Jelinek <jakub@redhat.com>
9011 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
9013 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
9014 or DECL_NONLOCAL labels.
9015 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
9016 or DECL_NONLOCAL labels here.
9018 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
9020 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
9021 to indicate when early gimple folding has been disabled.
9022 (rs6000_gimple_fold_builtin): Add debug content.
9023 (rs6000_invalid_builtin): Fix whitespace.
9024 (rs6000_expand_builtin): Fix whitespace.
9025 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
9027 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
9030 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
9031 SAVE_MULTIPLE if not all the registers that saves, should be saved.
9033 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
9035 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
9036 (qdf24xx): Likewise.
9037 * config/aarch64/aarch64.md: Include falkor.md.
9038 * config/aarch64/falkor.md: New.
9040 2017-08-09 Marek Polacek <polacek@redhat.com>
9043 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
9044 * diagnostic.c (emit_diagnostic): Add a comment.
9045 (emit_diagnostic_valist): New function.
9047 2017-08-09 Marek Polacek <polacek@redhat.com>
9050 * input.c (make_location): New overload.
9051 * input.h (make_location): Declare.
9053 2017-08-08 Alan Modra <amodra@gmail.com>
9054 H.J. Lu <hongjiu.lu@intel.com>
9057 * gcc.c (NO_PIE_SPEC): Delete.
9058 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
9060 (LINK_PIE_SPEC): ..to here.
9061 (LINK_COMMAND_SPEC): Support -no-pie.
9062 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
9063 chain of crtbegin*.o selection, update for PIE_SPEC changes and
9065 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
9066 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
9067 (ENDFILE_CRTEND_SPEC): Similarly.
9069 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
9072 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
9073 (mstack-protector-guard-offset=): Ditto.
9074 * config/i386/i386.c (ix86_option_override): Handle
9075 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
9077 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
9078 ix86_stack_protect_guard_offset variables.
9079 (TARGET_STACK_PROTECT_GUARD): Always define.
9080 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
9081 and -mstack-protector-guard-offset= options.
9083 2017-08-08 Bin Cheng <bin.cheng@arm.com>
9085 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
9086 boundary case for the last candidate.
9088 2017-08-08 Bin Cheng <bin.cheng@arm.com>
9090 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
9091 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
9093 2017-08-08 Tamar Christina <tamar.christina@arm.com>
9096 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
9097 * config/aarch64/aarch64-builtins.c
9098 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
9099 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
9100 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
9102 2017-08-08 Tamar Christina <tamar.christina@arm.com>
9103 Andrew Pinski <pinskia@gmail.com>
9106 * internal-fn.def (XORSIGN): New.
9107 * optabs.def (xorsign_optab): New.
9108 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
9109 (convert_expand_mult_copysign): New.
9110 (pass_optimize_widening_mul::execute): Call
9111 convert_expand_mult_copysign.
9113 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9115 PR tree-optimization/81354
9116 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
9117 Insert on edges rather than explicitly creating landing pads.
9118 (analyze_candidates_and_replace): Commit edge inserts.
9120 2017-08-08 Richard Biener <rguenther@suse.de>
9123 * tree-ssa-loop-niter.c: Include tree-dfa.h.
9124 (expand_simple_operations): Also look through ADDR_EXPRs with
9125 MEM_REF bases treating them as POINTER_PLUS_EXPR.
9127 2017-08-08 Richard Biener <rguenther@suse.de>
9129 PR tree-optimization/81723
9130 * tree-vect-slp.c (struct bst_traits): New hash traits.
9131 (bst_fail): New global.
9132 (vect_build_slp_tree_2): New worker, split out from ...
9133 (vect_build_slp_tree): ... this now wrapping it with using
9134 bst_fail set to cache SLP tree build fails. Properly handle
9136 (vect_analyze_slp_instance): Allocate and free bst_fail.
9138 2017-08-08 Martin Liska <mliska@suse.cz>
9141 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
9142 LABEL_DECLs that can be from a different function.
9144 2017-08-08 Bin Cheng <bin.cheng@arm.com>
9146 PR tree-optimization/81744
9147 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
9148 loop's number of iterations.
9150 2017-08-08 Martin Liska <mliska@suse.cz>
9152 * asan.c: Include header files.
9153 * attribs.c (build_decl_attribute_variant): New function moved
9155 (build_type_attribute_qual_variant): Likewise.
9156 (cmp_attrib_identifiers): Likewise.
9157 (simple_cst_list_equal): Likewise.
9158 (omp_declare_simd_clauses_equal): Likewise.
9159 (attribute_value_equal): Likewise.
9160 (comp_type_attributes): Likewise.
9161 (build_type_attribute_variant): Likewise.
9162 (lookup_ident_attribute): Likewise.
9163 (remove_attribute): Likewise.
9164 (merge_attributes): Likewise.
9165 (merge_type_attributes): Likewise.
9166 (merge_decl_attributes): Likewise.
9167 (merge_dllimport_decl_attributes): Likewise.
9168 (handle_dll_attribute): Likewise.
9169 (attribute_list_equal): Likewise.
9170 (attribute_list_contained): Likewise.
9171 * attribs.h (lookup_attribute): New function moved from tree.[ch].
9172 (lookup_attribute_by_prefix): Likewise.
9173 * bb-reorder.c: Include header files.
9174 * builtins.c: Likewise.
9175 * calls.c: Likewise.
9176 * cfgexpand.c: Likewise.
9177 * cgraph.c: Likewise.
9178 * cgraphunit.c: Likewise.
9179 * convert.c: Likewise.
9180 * dwarf2out.c: Likewise.
9181 * final.c: Likewise.
9182 * fold-const.c: Likewise.
9183 * function.c: Likewise.
9184 * gimple-expr.c: Likewise.
9185 * gimple-fold.c: Likewise.
9186 * gimple-pretty-print.c: Likewise.
9187 * gimple.c: Likewise.
9188 * gimplify.c: Likewise.
9189 * hsa-common.c: Likewise.
9190 * hsa-gen.c: Likewise.
9191 * internal-fn.c: Likewise.
9192 * ipa-chkp.c: Likewise.
9193 * ipa-cp.c: Likewise.
9194 * ipa-devirt.c: Likewise.
9195 * ipa-fnsummary.c: Likewise.
9196 * ipa-inline.c: Likewise.
9197 * ipa-visibility.c: Likewise.
9199 * lto-cgraph.c: Likewise.
9200 * omp-expand.c: Likewise.
9201 * omp-general.c: Likewise.
9202 * omp-low.c: Likewise.
9203 * omp-offload.c: Likewise.
9204 * omp-simd-clone.c: Likewise.
9205 * opts-global.c: Likewise.
9206 * passes.c: Likewise.
9207 * predict.c: Likewise.
9208 * sancov.c: Likewise.
9209 * sanopt.c: Likewise.
9210 * symtab.c: Likewise.
9211 * toplev.c: Likewise.
9212 * trans-mem.c: Likewise.
9213 * tree-chkp.c: Likewise.
9214 * tree-eh.c: Likewise.
9215 * tree-into-ssa.c: Likewise.
9216 * tree-object-size.c: Likewise.
9217 * tree-parloops.c: Likewise.
9218 * tree-profile.c: Likewise.
9219 * tree-ssa-ccp.c: Likewise.
9220 * tree-ssa-live.c: Likewise.
9221 * tree-ssa-loop.c: Likewise.
9222 * tree-ssa-sccvn.c: Likewise.
9223 * tree-ssa-structalias.c: Likewise.
9224 * tree-ssa.c: Likewise.
9225 * tree-streamer-in.c: Likewise.
9226 * tree-vectorizer.c: Likewise.
9227 * tree-vrp.c: Likewise.
9229 * ubsan.c: Likewise.
9230 * varasm.c: Likewise.
9231 * varpool.c: Likewise.
9232 * tree.c: Remove functions moved to attribs.[ch].
9234 * config/aarch64/aarch64.c: Add attrs.h header file.
9235 * config/alpha/alpha.c: Likewise.
9236 * config/arc/arc.c: Likewise.
9237 * config/arm/arm.c: Likewise.
9238 * config/avr/avr.c: Likewise.
9239 * config/bfin/bfin.c: Likewise.
9240 * config/c6x/c6x.c: Likewise.
9241 * config/cr16/cr16.c: Likewise.
9242 * config/cris/cris.c: Likewise.
9243 * config/darwin.c: Likewise.
9244 * config/epiphany/epiphany.c: Likewise.
9245 * config/fr30/fr30.c: Likewise.
9246 * config/frv/frv.c: Likewise.
9247 * config/ft32/ft32.c: Likewise.
9248 * config/h8300/h8300.c: Likewise.
9249 * config/i386/winnt.c: Likewise.
9250 * config/ia64/ia64.c: Likewise.
9251 * config/iq2000/iq2000.c: Likewise.
9252 * config/lm32/lm32.c: Likewise.
9253 * config/m32c/m32c.c: Likewise.
9254 * config/m32r/m32r.c: Likewise.
9255 * config/m68k/m68k.c: Likewise.
9256 * config/mcore/mcore.c: Likewise.
9257 * config/microblaze/microblaze.c: Likewise.
9258 * config/mips/mips.c: Likewise.
9259 * config/mmix/mmix.c: Likewise.
9260 * config/mn10300/mn10300.c: Likewise.
9261 * config/moxie/moxie.c: Likewise.
9262 * config/msp430/msp430.c: Likewise.
9263 * config/nds32/nds32-isr.c: Likewise.
9264 * config/nds32/nds32.c: Likewise.
9265 * config/nios2/nios2.c: Likewise.
9266 * config/nvptx/nvptx.c: Likewise.
9267 * config/pa/pa.c: Likewise.
9268 * config/pdp11/pdp11.c: Likewise.
9269 * config/powerpcspe/powerpcspe.c: Likewise.
9270 * config/riscv/riscv.c: Likewise.
9271 * config/rl78/rl78.c: Likewise.
9272 * config/rx/rx.c: Likewise.
9273 * config/s390/s390.c: Likewise.
9274 * config/sh/sh.c: Likewise.
9275 * config/sol2.c: Likewise.
9276 * config/sparc/sparc.c: Likewise.
9277 * config/spu/spu.c: Likewise.
9278 * config/stormy16/stormy16.c: Likewise.
9279 * config/tilegx/tilegx.c: Likewise.
9280 * config/tilepro/tilepro.c: Likewise.
9281 * config/v850/v850.c: Likewise.
9282 * config/vax/vax.c: Likewise.
9283 * config/visium/visium.c: Likewise.
9284 * config/xtensa/xtensa.c: Likewise.
9286 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9289 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
9290 constraints since the -mupper-regs-* switches have been
9292 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
9293 into a vector from a double word element that was extracted from
9294 another vector, and eliminate extra XXPERMDI instructions.
9295 (vsx_concat_<mode>_2): Likewise.
9296 (vsx_concat_<mode>_3): Likewise.
9297 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
9298 concat to allow optimizing inserts from previous extracts.
9300 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
9302 * config/i386/i386.c (ix86_stack_protect_guard): Generate
9303 memory reference to a SSP offset in TLS address space.
9304 (ix86_print_operand) <case '@'>: Remove.
9305 (ix86_print_operand_punct_valid_p): Remove '@' code.
9306 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
9308 (stack_tls_protect_set_<mode>): Remove.
9309 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
9310 (stack_tls_protect_test_<mode>): Remove.
9311 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
9313 2017-08-07 Olivier Hainque <hainque@adacore.com>
9316 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
9318 2017-08-07 Douglas Rupp <rupp@adacore.com>
9320 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
9321 variable was referenced as multidir in command.
9323 2017-08-07 Jakub Jelinek <jakub@redhat.com>
9326 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
9329 2017-08-07 Martin Liska <mliska@suse.cz>
9331 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
9332 * config/rl78/rl78.c: Add include of attribs.h.
9333 * config/sh/sh.c: Likewise.
9334 * config/v850/v850.c: Likewise.
9336 2017-08-07 Tom de Vries <tom@codesourcery.com>
9339 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
9341 2017-08-07 Martin Liska <mliska@suse.cz>
9343 * config/mips/mips.c: Include attribs.h.
9345 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
9348 * doc/invoke.texi: Document change in behvaior for -Ofast for
9351 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
9353 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
9355 (aarch64_pop_regs): Likewise.
9356 (aarch64_gen_load_pair): Likewise.
9357 (aarch64_save_callee_saves): Likewise.
9358 (aarch64_restore_callee_saves): Likewise.
9360 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
9362 * config/i386/i386.c: Revert the last change.
9364 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
9367 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
9369 (ix86_finalize_stack_frame_flags): This. Also clear
9370 frame_pointer_needed if -fno-omit-frame-pointer is used without
9372 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
9373 with ix86_finalize_stack_frame_flags.
9374 (ix86_expand_epilogue): Likewise.
9375 (ix86_expand_split_stack_prologue): Likewise.
9377 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
9380 * config/i386/i386.c (get_builtin_code_for_version): Set priority
9381 to P_AES for Westmere.
9383 2017-08-07 Jonathan Yong <10walls@gmail.com>
9385 * config/i386/mingw.opt (fset-stack-executable): Removed.
9386 * config/i386/cygming.opt (fset-stack-executable): Moved
9388 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
9390 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
9392 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
9394 2017-08-07 Marek Polacek <polacek@redhat.com>
9397 * fold-const.c (fold_indirect_ref_1): Check type_domain.
9399 2017-08-07 Martin Liska <mliska@suse.cz>
9401 * attribs.h (canonicalize_attr_name): New function.
9402 (cmp_attribs): Move from c-format.c and adjusted.
9403 (is_attribute_p): Moved from tree.h.
9404 * tree-inline.c: Add new includes.
9405 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
9406 (private_is_attribute_p): Remove.
9407 (private_lookup_attribute): Likewise.
9408 (private_lookup_attribute_by_prefix): Simplify.
9409 (remove_attribute): Use is_attribute_p.
9410 * tree.h: Remove removed declarations.
9412 2017-08-07 Jakub Jelinek <jakub@redhat.com>
9415 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
9416 instead of computing it in the function. Formatting fix.
9417 (expand_case): Don't rely on default_edge being the first edge,
9418 clear it if removing it, pass default_edge to
9419 emit_case_dispatch_table.
9420 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
9423 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
9425 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
9426 insn in the function, emit NOP after the insn.
9428 2017-08-06 Tom de Vries <tom@codesourcery.com>
9430 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
9433 2017-08-06 Tom de Vries <tom@codesourcery.com>
9435 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
9438 2017-08-04 Yury Gribov <tetra2005@gmail.com>
9440 PR tree-optimization/57371
9441 * match.pd: New pattern.
9443 2017-08-04 Marek Polacek <polacek@redhat.com>
9446 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
9447 perform the computation in offset_int.
9449 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
9451 PR tree-optimization/81136
9452 * tree-vectorizer.h: Include tree-hash-traits.h.
9453 (vec_base_alignments): New typedef.
9454 (vec_info): Add a base_alignments field.
9455 (vect_record_base_alignments): Declare.
9456 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
9458 (DR_IS_CONDITIONAL_IN_STMT): New macro.
9459 (create_data_ref): Add an is_conditional_in_stmt argument.
9460 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
9461 the is_conditional_in_stmt field.
9462 (data_ref_loc): Add an is_conditional_in_stmt field.
9463 (get_references_in_stmt): Set the is_conditional_in_stmt field.
9464 (find_data_references_in_stmt): Update call to create_data_ref.
9465 (graphite_find_data_references_in_stmt): Likewise.
9466 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
9467 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
9468 (vect_record_base_alignment): New function.
9469 (vect_record_base_alignments): Likewise.
9470 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
9471 for nested statements even if we fail to compute a misalignment.
9472 Use pooled base alignments for unconditional references.
9473 (vect_find_same_alignment_drs): Compare base addresses instead
9475 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
9476 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
9478 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
9480 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
9481 Add an explicit name for the enum. Use auto_vec for slp_instances
9483 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
9485 (_bb_vec_info): Add a constructor and destructor.
9486 (vinfo_for_stmt): Return NULL for uids of -1 as well.
9487 (destroy_loop_vec_info): Delete.
9488 (vect_destroy_datarefs): Likewise.
9489 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
9490 (vec_info::vec_info): New function.
9491 (vec_info::~vec_info): Likewise.
9492 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
9493 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
9494 destroy_loop_vec_info.
9495 * tree-vect-loop.c (new_loop_vec_info): Replace with...
9496 (_loop_vec_info::_loop_vec_info): ...this.
9497 (destroy_loop_vec_info): Replace with...
9498 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
9499 the stmt_vec_infos. Leave handling of vec_info information to its
9500 destructor. Remove explicit vector releases.
9501 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
9502 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
9503 * tree-vect-slp.c (new_bb_vec_info): Replace with...
9504 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
9505 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
9506 (destroy_bb_vec_info): Replace with...
9507 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
9508 information to its destructor.
9509 (vect_slp_analyze_bb_1): Use new and delete instead of
9510 new_bb_vec_info and destroy_bb_vec_info.
9511 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
9514 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
9516 * tree-data-ref.h (subscript): Add access_fn field.
9517 (data_dependence_relation): Add could_be_independent_p.
9518 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
9519 (same_access_functions): Move to tree-data-ref.c.
9520 * tree-data-ref.c (ref_contains_union_access_p): New function.
9521 (access_fn_component_p): Likewise.
9522 (access_fn_components_comparable_p): Likewise.
9523 (dr_analyze_indices): Add a reference to access_fn_component_p.
9524 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
9526 (constant_access_functions): Likewise.
9527 (add_other_self_distances): Likewise.
9528 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
9529 (initialize_data_dependence_relation): Use XCNEW and remove
9530 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
9531 of access functions that have the same type. Allow the
9532 subsequence to end with different bases in some circumstances.
9533 Record the chosen access functions in SUB_ACCESS_FN.
9534 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
9535 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
9536 (subscript_dependence_tester_1): Likewise dra and drb.
9537 (build_classic_dist_vector): Update calls accordingly.
9538 (subscript_dependence_tester): Likewise.
9539 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
9540 DDR_COULD_BE_INDEPENDENT_P.
9541 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
9542 comp_alias_ddrs instead of may_alias_ddrs.
9543 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
9545 (vect_analyze_data_ref_dependence): Use it if
9546 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
9547 distance vectors if that fails.
9548 (dependence_distance_ge_vf): New function.
9549 (vect_prune_runtime_alias_test_list): Use it. Don't clear
9550 LOOP_VINFO_MAY_ALIAS_DDRS.
9552 2017-08-04 Richard Biener <rguenther@suse.de>
9555 * fold-const.c (fold_binary_loc): Properly restrict
9556 minus_var0 && minus_var1 case when associating undefined overflow
9559 2017-08-04 Tom de Vries <tom@codesourcery.com>
9561 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
9563 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9565 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9566 Don't start diagnostic messages with a capital letter.
9567 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9569 (rs6000_invalid_builtin): Likewise.
9570 (rs6000_trampoline_init): Likewise.
9572 2017-08-03 Jakub Jelinek <jakub@redhat.com>
9575 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
9576 after setting changeable df flags.
9578 2017-08-03 Richard Biener <rguenther@suse.de>
9580 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
9581 up if the use is in USE - X.
9583 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
9585 * toplev.c (dumpfile.h): New include.
9586 (internal_error_reentered): New static function. Use it...
9587 (internal_error_function): ...here to handle reentered internal_error.
9589 2017-08-03 Richard Biener <rguenther@suse.de>
9592 * fold-const.c (split_tree): Add minus_var and minus_con
9593 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
9594 here but always use minus_*.
9595 (associate_trees): Assert we never associate with MINUS_EXPR
9596 and NULL first operand. Do not recurse for PLUS_EXPR operands
9597 when associating as MINUS_EXPR either.
9598 (fold_binary_loc): Track minus_var and minus_con.
9600 2017-08-03 Tom de Vries <tom@codesourcery.com>
9603 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
9604 ACCEL_COMPILER, apply finish_options on
9605 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
9607 2017-08-03 Tom de Vries <tom@codesourcery.com>
9610 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
9611 function_entry_patch_area_size > 0.
9613 2017-08-03 Jakub Jelinek <jakub@redhat.com>
9616 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
9617 instead of 10??LU, perform unit multiplication in wide_int,
9618 don't change alloc_object_size_limit if the limit is larger
9621 PR tree-optimization/81655
9622 PR tree-optimization/81588
9623 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
9624 the case when ranges[i].low and high are 1 for unsigned type with
9628 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
9629 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
9631 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9633 * tree-vrp.h: Add include guard.
9635 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
9638 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
9639 (ud2): New insn pattern.
9640 * config/i386/i386.c (ix86_expand_epilogue):
9641 For naked functions, generate ud2 instead of trap insn.
9643 2017-08-02 Marek Polacek <polacek@redhat.com>
9646 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
9648 2017-08-02 Tom de Vries <tom@codesourcery.com>
9649 Cesar Philippidis <cesar@codesourcery.com>
9651 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
9652 Add missing edge probabilities.
9654 2017-08-02 Tamar Christina <tamar.christina@arm.com>
9656 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
9659 2017-08-02 Jakub Jelinek <jakub@redhat.com>
9662 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
9663 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
9664 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
9665 prologue_seq sequences - if any.
9667 2017-08-02 Richard Biener <rguenther@suse.de>
9669 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
9670 via vectors if supported, integer extracts via punning if supported
9671 or otherwise vector extracts.
9673 2017-08-02 Richard Biener <rguenther@suse.de>
9675 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
9677 (bitmap_insert_into_set): ... this.
9679 2017-08-02 Richard Biener <rguenther@suse.de>
9681 PR tree-optimization/81633
9683 2015-08-17 Alan Hayward <alan.hayward@arm.com>
9685 PR tree-optimization/71752
9686 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
9688 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
9690 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
9691 (machine_function::call_ms2sysv_pad_out): Remove field.
9692 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
9693 (ix86_compute_frame_layout): Likewise.
9695 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
9698 * config/i386/i386.c (ix86_set_func_type): Disallow naked
9699 attribute with interrupt attribute.
9701 2017-08-01 Andrew Pinski <apinski@cavium.com>
9703 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
9704 BIT_INSERT_EXPR's operand 1
9705 to see if the types precision matches.
9707 2017-08-01 Martin Liska <mliska@suse.cz>
9710 * builtins.c (expand_builtin_memcpy_args): Remove.
9711 (expand_builtin_memcpy): Call newly added function
9712 expand_builtin_memory_copy_args.
9713 (expand_builtin_memcpy_with_bounds): Likewise.
9714 (expand_builtin_mempcpy): Remove last argument.
9715 (expand_builtin_mempcpy_with_bounds): Likewise.
9716 (expand_builtin_memory_copy_args): New function created from
9717 expand_builtin_mempcpy_args with small modifications.
9718 (expand_builtin_mempcpy_args): Remove.
9719 (expand_builtin_stpcpy): Remove unused argument.
9720 (expand_builtin): Likewise.
9721 (expand_builtin_with_bounds): Likewise.
9723 2017-08-01 Martin Liska <mliska@suse.cz>
9726 Make mempcpy more optimal (PR middle-end/70140).
9728 2017-08-01 Jakub Jelinek <jakub@redhat.com>
9731 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9732 __builtin_vec_cmpne verify both arguments are compatible vectors
9733 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
9734 verify arg1_type is a pointer or array type. For __builtin_vec_st,
9735 move computation of aligned to after checking the argument types.
9739 * config/rs6000/vsx.md (vextract_fp_from_shorth,
9740 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
9743 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
9744 Doug Rupp <rupp@adacore.com>
9745 Olivier Hainque <hainque@adacore.com>
9747 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
9748 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
9750 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
9751 for TARGET_OS_CPP_BUILTIN.
9752 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
9753 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
9755 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
9756 passing required abi options to the assembler for EABI configurations.
9757 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
9758 of .text.hot and .text.unlikely sections for kernel modules when
9759 using ARM style exceptions.
9760 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
9761 options. Add EXTRA_CC1_SPEC.
9762 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
9764 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
9766 (ARM_TARGET2_DWARF_FORMAT): Define.
9767 * config/arm/t-vxworks: Adjust multilib control to removal of the
9768 Diab command line options.
9770 2017-08-01 Martin Liska <mliska@suse.cz>
9772 PR gcov-profile/81561
9773 * gcov.c (unblock): Make unblocking safe as we need to preserve
9774 index correspondence of blocks and block_lists.
9776 2017-08-01 Richard Biener <rguenther@suse.de>
9778 PR tree-optimization/81181
9779 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9780 (compute_antic): ... end of iteration here.
9782 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
9784 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
9785 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
9786 (ftree-slp-vectorize): Likewise.
9787 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
9788 can no longer be set independent of flag_tree_loop_vectorize.
9789 * omp-general.c (emp_max_vf): Likewise.
9790 * opts.c (enable_fdo_optimizations): Remove references to
9791 flag_tree_vectorize, these are now implicit.
9792 (common_handle_option): Remove handling for OPT_ftree_vectorize,
9793 and leave it for the options machinery.
9795 2017-08-01 Martin Liska <mliska@suse.cz>
9798 * builtins.c (expand_builtin_memcpy_args): Remove.
9799 (expand_builtin_memcpy): Call newly added function
9800 expand_builtin_memory_copy_args.
9801 (expand_builtin_memcpy_with_bounds): Likewise.
9802 (expand_builtin_mempcpy): Remove last argument.
9803 (expand_builtin_mempcpy_with_bounds): Likewise.
9804 (expand_builtin_memory_copy_args): New function created from
9805 expand_builtin_mempcpy_args with small modifications.
9806 (expand_builtin_mempcpy_args): Remove.
9807 (expand_builtin_stpcpy): Remove unused argument.
9808 (expand_builtin): Likewise.
9809 (expand_builtin_with_bounds): Likewise.
9811 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
9814 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
9815 print "ds:" only for immediates in generic address space.
9817 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
9820 * config/i386/i386.c (ix86_funciton_naked): New prototype.
9821 (ix86_function_ok_for_sibcall): Return false for naked functions.
9823 2017-08-01 Richard Biener <rguenther@suse.de>
9825 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
9826 (compute_antic): Seed worklist with exit block predecessors.
9827 * cfganal.c (dfs_find_deadend): For a cycle return the source
9828 of the edge closing it.
9830 2017-08-01 Tamar Christina <tamar.christina@arm.com>
9832 * config/aarch64/aarch64.c
9833 (aarch64_can_const_movi_rtx_p): Move 0 check.
9835 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9837 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
9838 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
9840 * match.pd: Ditto in address comparison pattern.
9842 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9844 PR tree-optimization/81627
9845 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
9846 closed ssa form for store-store chain.
9848 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9850 PR tree-optimization/81620
9851 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
9852 for store-store chain.
9854 2017-08-01 Jakub Jelinek <jakub@redhat.com>
9856 PR tree-optimization/81588
9857 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
9858 ranges[i].in_p, invert comparison code ccode. For >/>=,
9859 swap rhs1 and rhs2 and comparison code unconditionally,
9860 for </<= don't do that. Don't swap rhs1/rhs2 again if
9861 ranges[i].in_p, instead invert comparison code ccode if
9862 opcode or oe->rank is BIT_IOR_EXPR.
9865 * optabs.def (vec_extract_optab, vec_init_optab): Change from
9866 a direct optab to conversion optab.
9867 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
9868 with GET_MODE_INNER as last argument instead of optab_handler.
9869 * expmed.c (extract_bit_field_1): Likewise. Use vector from
9870 vector extraction if possible and optab is available.
9871 * expr.c (store_constructor): Use convert_optab_handler instead
9872 of optab_handler. Use vector initialization from smaller
9873 vectors if possible and optab is available.
9874 * tree-vect-stmts.c (vectorizable_load): Likewise.
9875 * doc/md.texi (vec_extract, vec_init): Document that the optabs
9877 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
9878 of vec_init from half-sized vectors with the same element mode.
9879 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
9880 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
9881 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
9882 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
9883 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
9884 after mode in gen_vec_extract* calls.
9885 (vec_extract<mode>): Renamed to ...
9886 (vec_extract<mode><ssescalarmodelower>): ... this.
9887 (vec_extract<mode><ssehalfvecmodelower>): New expander.
9888 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
9889 element mode after mode in gen_vec_init* calls.
9890 (VEC_INIT_HALF_MODE): New mode iterator.
9891 (vec_init<mode>): Renamed to ...
9892 (vec_init<mode><ssescalarmodelower>): ... this.
9893 (vec_init<mode><ssehalfvecmodelower>): New expander.
9894 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
9895 (vec_extractv2sfsf): ... this.
9896 (vec_initv2sf): Renamed to ...
9897 (vec_initv2sfsf): ... this.
9898 (vec_extractv2si): Renamed to ...
9899 (vec_extractv2sisi): ... this.
9900 (vec_initv2si): Renamed to ...
9901 (vec_initv2sisi): ... this.
9902 (vec_extractv4hi): Renamed to ...
9903 (vec_extractv4hihi): ... this.
9904 (vec_initv4hi): Renamed to ...
9905 (vec_initv4hihi): ... this.
9906 (vec_extractv8qi): Renamed to ...
9907 (vec_extractv8qiqi): ... this.
9908 (vec_initv8qi): Renamed to ...
9909 (vec_initv8qiqi): ... this.
9910 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
9911 (vec_init<mode>): Renamed to ...
9912 (vec_init<mode><VEC_base_l>): ... this.
9913 (vec_extract<mode>): Renamed to ...
9914 (vec_extract<mode><VEC_base_l>): ... this.
9915 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
9916 (vec_initv2sfsf): ... this.
9917 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
9918 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
9919 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
9920 element mode after mode in gen_vec_init* calls.
9921 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
9922 (vec_init<mode><Vel>): ... this.
9923 (vec_extract<mode>): Renamed to ...
9924 (vec_extract<mode><Vel>): ... this.
9925 * config/aarch64/iterators.md (Vel): New mode attribute.
9926 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
9927 Add element mode after mode in gen_vec_extract* calls.
9928 * config/s390/vector.md (non_vec_l): New mode attribute.
9929 (vec_extract<mode>): Renamed to ...
9930 (vec_extract<mode><non_vec_l>): ... this.
9931 (vec_init<mode>): Renamed to ...
9932 (vec_init<mode><non_vec_l>): ... this.
9933 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
9934 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
9936 * config/arm/iterators.md (V_elem_l): New mode attribute.
9937 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
9938 (vec_extract<mode><V_elem_l>): ... this.
9939 (vec_extractv2di): Renamed to ...
9940 (vec_extractv2didi): ... this.
9941 (vec_init<mode>): Renamed to ...
9942 (vec_init<mode><V_elem_l>): ... this.
9943 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
9944 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
9945 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
9946 Add element mode after gen_vec_extract* calls.
9947 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
9948 (vec_init<mode><unitmode>): ... this.
9949 (vec_extract<mode>): Renamed to ...
9950 (vec_extract<mode><unitmode>): ... this.
9951 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
9952 (vec_init<mode><unitmode>): ... this.
9953 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
9954 (vec_initv2sfsf): ... this.
9955 (vec_extractv2sf): Renamed to ...
9956 (vec_extractv2sfsf): ... this.
9957 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
9958 Add element mode after gen_vec_extract* calls.
9959 * config/mips/mips.md (unitmode): New mode iterator.
9960 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
9961 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
9962 * config/spu/spu.md (inner_l): New mode attribute.
9963 (vec_init<mode>): Renamed to ...
9964 (vec_init<mode><inner_l>): ... this.
9965 (vec_extract<mode>): Renamed to ...
9966 (vec_extract<mode><inner_l>): ... this.
9967 * config/sparc/sparc.md (veltmode): New mode iterator.
9968 (vec_init<VMALL:mode>): Renamed to ...
9969 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
9970 * config/ia64/vect.md (vec_initv2si): Renamed to ...
9971 (vec_initv2sisi): ... this.
9972 (vec_initv2sf): Renamed to ...
9973 (vec_initv2sfsf): ... this.
9974 (vec_extractv2sf): Renamed to ...
9975 (vec_extractv2sfsf): ... this.
9976 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
9977 (vec_init<mode>): Renamed to ...
9978 (vec_init<mode><VEC_base_l>): ... this.
9979 (vec_extract<mode>): Renamed to ...
9980 (vec_extract<mode><VEC_base_l>): ... this.
9981 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
9982 (vec_initv2sfsf): ... this.
9983 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
9984 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
9985 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
9986 gen_vec_init* calls.
9988 2017-08-01 Richard Biener <rguenther@suse.de>
9990 PR tree-optimization/81297
9991 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
9992 TREE_OVERFLOW from INTEGER_CSTs.
9994 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
9996 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
9998 2017-07-31 Carl Love <cel@us.ibm.com>
10000 * config/rs6000/rs6000-c: Add support for built-in functions
10001 vector signed char vec_xl_be (signed long long, signed char *);
10002 vector unsigned char vec_xl_be (signed long long, unsigned char *);
10003 vector signed int vec_xl_be (signed long long, signed int *);
10004 vector unsigned int vec_xl_be (signed long long, unsigned int *);
10005 vector signed long long vec_xl_be (signed long long, signed long long *);
10006 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
10007 vector signed short vec_xl_be (signed long long, signed short *);
10008 vector unsigned short vec_xl_be (signed long long, unsigned short *);
10009 vector double vec_xl_be (signed long long, double *);
10010 vector float vec_xl_be (signed long long, float *);
10011 * config/rs6000/altivec.h (vec_xl_be): Add #define.
10012 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
10013 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
10015 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
10016 (altivec_expand_builtin): Add switch statement to call
10017 altivec_expand_xl_be for each builtin.
10018 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
10019 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
10020 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
10021 __builtin_vsx_le_be_v16qi.
10022 * doc/extend.texi: Update the built-in documentation file for the
10023 new built-in functions.
10025 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
10028 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
10030 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
10032 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10034 * config.gcc: Add z14.
10035 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
10036 CPU model numbers for z13s and z14.
10037 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
10039 * config/s390/s390-opts.h (enum processor_type): Rename
10040 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
10041 * config/s390/s390.c (processor_table): Add field for CPU name to
10042 be passed to Binutils.
10043 (s390_asm_output_machine_for_arch): Use the new field in
10044 processor_table for Binutils.
10045 (s390_expand_builtin): Replace arch12 with z14.
10046 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
10047 (s390_get_sched_attrmask): Likewise.
10048 (s390_get_unit_mask): Likewise.
10049 * config/s390/s390.opt: Add z14 to processor_type enum.
10051 2017-07-31 Martin Jambor <mjambor@suse.cz>
10054 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
10055 regardless of optimization level.
10057 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
10058 Martin Liska <mliska@suse.cz>
10060 * predict.def: Remove old comment and adjust probability.
10061 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
10062 PREDICT statements.
10064 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
10067 * config/i386/i386.c (ix86_function_naked): New function.
10068 (ix86_can_use_return_insn_p): Return false for naked functions.
10069 (ix86_expand_prologue): Skip prologue for naked functions.
10070 (ix86_expand_epilogue): Skip epilogue for naked functions
10071 and emit trap instruction.
10072 (ix86_warn_func_return): New function.
10073 (ix86_attribute_table): Add "naked" attribute specification.
10074 (TARGET_WARN_FUNC_RETURN): Define.
10075 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
10077 2017-07-31 Martin Liska <mliska@suse.cz>
10079 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
10080 (dump_gimple_bb_header): Always dump BB info.
10081 (pp_cfg_jump): Do not append info about BB when dumping a jump.
10083 2017-07-31 Martin Liska <mliska@suse.cz>
10086 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
10087 also with current_function_decl non-null equality.
10089 2017-07-31 Jakub Jelinek <jakub@redhat.com>
10092 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
10093 change type to the element type, instead add eltype variable and
10094 use it where we are interested in the element type.
10096 PR tree-optimization/81603
10097 * ipa-polymorphic-call.c
10098 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
10099 offset arithmetic in offset_int, bail out if the resulting bit offset
10100 doesn't fit into shwi.
10102 2017-07-31 Martin Liska <mliska@suse.cz>
10104 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
10105 (gimplify_save_expr): Fix comment.
10107 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
10110 * config/i386/i386.c (ix86_function_arg): Update arguments for
10112 (ix86_compute_frame_layout): Set the initial stack offset to
10113 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
10114 INCOMING_FRAME_SP_OFFSET.
10115 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
10116 stack before exception handler returns.
10117 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
10118 the 'ERROR_CODE' for exception handler.
10120 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
10122 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
10123 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
10124 (ASM_OUTPUT_REG_POP): Ditto.
10125 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
10126 instead of asm_fprintf to output pure string.
10128 2017-07-29 Jakub Jelinek <jakub@redhat.com>
10130 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
10131 to imported_module_or_decl hook.
10132 (debug_nothing_tree_tree_tree_bool): Remove.
10133 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
10134 * debug.c (do_nothing_debug_hooks): Use
10135 debug_nothing_tree_tree_tree_bool_bool instead of
10136 debug_nothing_tree_tree_tree_bool.
10137 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
10138 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
10139 * sdbout.c (sdb_debug_hooks): Likewise.
10140 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
10141 (gen_namespace_die): Add DW_AT_export_symbols attribute if
10143 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
10144 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
10145 attribute, don't add anything.
10147 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10149 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
10150 (fold_build2_stat_loc): Likewise.
10151 (fold_build3_stat_loc): Likewise.
10152 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
10153 (fold_build1_loc): Remove macro.
10154 (fold_build2_loc): Likewise.
10155 (fold_build3_loc): Likewise.
10157 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10159 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
10160 (gimple_build_debug_bind_source_stat): Likewise.
10161 * gimple.h (gimple_build_debug_bind): Remove macro.
10162 (gimple_build_debug_bind_source): Likewise.
10164 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10166 * bitmap.c (bitmap_alloc): Adjust.
10167 (bitmap_gc_alloc): Likewise.
10168 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
10170 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10172 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
10173 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
10174 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
10175 (bitmap_gc_alloc_stat): Likewise.
10176 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
10178 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10180 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
10181 * rtl.h (shallow_copy_rtx): Remove macro.
10183 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10185 * emit-rtl.c (gen_raw_REG): Adjust.
10186 * gengenrtl.c (gendef): Likewise.
10187 * rtl.c (rtx_alloc_stat): Remove _stat from name.
10188 * rtl.h (rtx_alloc): Remove macro.
10190 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10192 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
10193 (build_tree_list_stat): Likewise.
10194 * tree.h (build_tree_list): Remove macro.
10195 (build_tree_list_vec): Likewise.
10197 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10199 * tree.c (make_vector_stat): Remove _stat from name.
10200 (build_vector_stat): Likewise.
10201 * tree.h (make_vector_stat): Remove macro.
10202 (build_vector_stat): Likewise.
10204 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10206 * tree.h (build_var_debug_value): Remove prototype.
10208 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10210 * tree.c (tree_cons_stat): Remove _stat from name.
10211 * tree.h (tree_cons): Remove macro.
10213 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10215 * tree.c (build_vl_exp_stat): Remove _stat from name.
10216 * tree.h (build_vl_exp): Remove macro.
10218 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10220 * tree.c (build_decl_stat): Remove _stat from name.
10221 * tree.h (build_decl): Remove macro.
10223 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10225 * gimple.c (gimple_build_with_ops_stat): Adjust.
10226 (gimple_alloc_stat): Remove _stat from name.
10227 * gimple.h (gimple_alloc): Remove macro.
10229 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10231 * tree.c (make_tree_vec_stat): Remove _stat from name.
10232 (grow_tree_vec_stat): Likewise.
10233 * tree.h (make_tree_vec_stat): Adjust prototype.
10234 (grow_tree_vec_stat): Likewise.
10235 (make_tree_vec): Remove macro.
10236 (grow_tree_vec): Likewise.
10238 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10240 * fold-const.c (fold_build1_stat_loc): Adjust.
10241 (fold_build2_stat_loc): Likewise.
10242 (fold_build3_stat_loc): Likewise.
10243 * tree.c (build0_stat): Remove _stat from name.
10244 (build1_stat): Likewise.
10245 (build2_stat): Likewise.
10246 (build3_stat): Likewise.
10247 (build4_stat): Likewise.
10248 (build5_stat): Likewise.
10249 * tree.h (build1_loc): Remove macro, and rename _stat function
10251 (build2_loc): Likewise.
10252 (build3_loc): Likewise.
10253 (build4_loc): Likewise.
10254 (build5_loc): Likewise.
10256 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10258 * tree.c (make_int_cst_stat): Remove _stat from name.
10259 * tree.h (make_int_cst_stat): Adjust prototype.
10260 (make_int_cst): Remove macro.
10262 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10264 * tree.c (make_tre_binfo_stat): Remove _stat from name.
10265 * tree.h (make_tree_binfo_stat): Adjust prototype.
10266 (make_tree_binfo): Remove.
10268 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10270 * tree.c (copy_node_stat): Rename to copy_node.
10271 (build_distinct_type_copy): Adjust.
10272 * tree.h (copy_node_stat): Adjust prototype.
10273 (copy_node): Remove macro.
10275 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10277 * tree.c (make_node_stat): rename to make_node.
10278 (build_tree_list_stat): Adjust.
10279 (build0_stat): Likewise.
10280 (build2_stat): Likewise.
10281 (build3_stat): Likewise.
10282 (build4_stat): Likewise.
10283 (build5_stat): Likewise.
10284 (build_decl_stat): Likewise.
10285 * tree.h (make_node_stat): Adjust prototype.
10286 (make_node): remove macro.
10288 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
10290 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
10291 (PPC_FEATURE2_SCV): Likewise.
10292 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
10294 2017-07-28 Tamar Christina <tamar.christina@arm.com>
10296 * config/aarch64/aarch64.c
10297 (aarch64_internal_mov_immediate): Add new special pattern.
10298 * config/aarch64/aarch64.md (*movdi_aarch64):
10299 Add reg/32bit const mov case.
10301 2017-07-28 Tamar Christina <tamar.christina@arm.com>
10302 Richard Sandiford <richard.sandiford@linaro.org>
10304 * config/aarch64/aarch64.md (mov<mode>): Generalize.
10305 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
10306 Add integer and movi cases.
10307 (movi-split-hf-df-sf split, fp16): New.
10308 (enabled): Added TARGET_FP_F16INST.
10309 * config/aarch64/iterators.md (GPF_HF): New.
10310 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
10312 2017-07-28 Tamar Christina <tamar.christina@arm.com>
10314 * config/aarch64/aarch64.c
10315 (aarch64_simd_container_mode): Add prototype.
10316 (aarch64_expand_mov_immediate): Add HI support.
10317 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
10318 (aarch64_can_const_movi_rtx_p): New.
10319 (aarch64_preferred_reload_class):
10320 Remove restrictions of using FP registers for certain SIMD operations.
10321 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
10322 (aarch64_valid_floating_const): Add integer move validation.
10323 (aarch64_simd_imm_scalar_p): Remove.
10324 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
10325 (aarch64_legitimate_constant_p): Expand list of supported cases.
10326 * config/aarch64/aarch64-protos.h
10327 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
10328 (aarch64_reinterpret_float_as_int): New.
10329 (aarch64_simd_imm_scalar_p): Remove.
10330 * config/aarch64/constraints.md (Uvi): New.
10331 (Dd): Split into Ds and new Dd.
10332 * config/aarch64/aarch64.md (*movsi_aarch64):
10334 (*movdi_aarch64): Add SIMD mov case.
10336 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10338 * tree-predcom.c: (struct chain): Handle store-store chain in which
10339 stores for elimination only store loop invariant values.
10340 (execute_pred_commoning_chain): Ditto.
10341 (prepare_initializers_chain_store_elim): Ditto.
10342 (prepare_finalizers): Ditto.
10343 (is_inv_store_elimination_chain): New function.
10344 (initialize_root_vars_store_elim_1): New function.
10346 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10348 * tree-predcom.c: Revise general description of the pass.
10349 (enum chain_type): New enum type for store elimination.
10350 (struct chain): New field supporting store elimination.
10351 (struct component): Ditto.
10352 (dump_chain): Dump store-stores chain.
10353 (release_chain): Release resources.
10354 (split_data_refs_to_components): Compute and create component
10355 contains only stores for elimination.
10356 (get_chain_last_ref_at): New function.
10357 (make_invariant_chain): Initialization.
10358 (make_rooted_chain): Specify chain type in parameter and record it.
10359 (add_looparound_copies): Skip for store-stores chain.
10360 (determine_roots_comp): Compute type of chain and pass it to
10362 (initialize_root_vars_store_elim_2): New function.
10363 (finalize_eliminated_stores): New function.
10364 (remove_stmt): Handle store for elimination.
10365 (execute_pred_commoning_chain): Execute predictive commoning on
10366 store-store chains.
10367 (determine_unroll_factor): Skip unroll for store-stores chain.
10368 (prepare_initializers_chain_store_elim): New function.
10369 (prepare_initializers_chain): Hanlde store-store chain.
10370 (prepare_finalizers_chain, prepare_finalizers): New function.
10371 (tree_predictive_commoning_loop): Return integer value indicating
10372 if loop is unrolled or lcssa form is corrupted.
10373 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
10375 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10377 * tree-predcom.c (initialize_root): Delete.
10378 (execute_pred_commoning_chain): Initialize root vars and replace
10379 reference of non-combined chain directly, rather than call above
10382 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10384 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
10385 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
10387 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10389 * tree-predcom.c (struct chain): New field init_seq.
10390 (release_chain): Release init_seq.
10391 (prepare_initializers_chain): Record intialization stmts in above
10393 (insert_init_seqs): New function.
10394 (tree_predictive_commoning_loop): Call insert_init_seqs.
10396 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10398 * tree-predcom.c (determine_roots_comp): Skip trivial components.
10400 2017-07-28 Richard Biener <rguenther@suse.de>
10402 * match.pd: Remove superfluous :c.
10403 * genmatch.c (simplify::id): Add member.
10404 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
10406 (current_id): New global.
10407 (dt_node::parent): Move from ...
10408 (dt_operand::parent): ... here. Add for_id member.
10409 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
10410 (decision_tree::find_node): Relax order requirement when
10411 merging DT_TRUE nodes to ones inbetween the current simplify
10412 and the one we try to merge with. Add diagnostic whenever
10413 we need to enforce pattern order by not merging.
10414 (decision_tree::insert): Set current_id.
10415 (decision_tree::print_node): Dump parent node and for_id.
10416 (parser::last_id): Add member.
10417 (parser::push_simplify): Assign unique id.
10418 (parser::parser): Initialize last_id.
10420 2017-07-28 Martin Liska <mliska@suse.cz>
10423 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
10424 gimple_build_debug_bind.
10426 2017-07-28 Richard Biener <rguenther@suse.de>
10428 PR tree-optimization/81502
10429 * match.pd: Add pattern combining BIT_INSERT_EXPR with
10431 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
10433 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
10434 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
10435 for BIT_FIELD_REF args.
10436 * fold-const.c (make_bit_field_ref): Likewise.
10437 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
10439 2017-07-28 Jakub Jelinek <jakub@redhat.com>
10442 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
10443 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
10444 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
10445 Or it into SANITIZER_UNDEFINED.
10446 * ubsan.c: Include gimple-fold.h and varasm.h.
10447 (ubsan_expand_ptr_ifn): New function.
10448 (instrument_pointer_overflow): New function.
10449 (maybe_instrument_pointer_overflow): New function.
10450 (instrument_object_size): Formatting fix.
10451 (pass_ubsan::execute): Call instrument_pointer_overflow
10452 and maybe_instrument_pointer_overflow.
10453 * internal-fn.c (expand_UBSAN_PTR): New function.
10454 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
10455 * sanitizer.def (__ubsan_handle_pointer_overflow,
10456 __ubsan_handle_pointer_overflow_abort): New builtins.
10457 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
10458 * internal-fn.def (UBSAN_PTR): New internal function.
10459 * opts.c (sanitizer_opts): Add pointer-overflow.
10460 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
10461 * fold-const.c (build_range_check): Compute pointer range check in
10462 integral type if pointer arithmetics would be needed. Formatting
10465 2017-07-28 Martin Liska <mliska@suse.cz>
10468 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
10469 parameters that are of a variable-length.
10471 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
10473 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
10475 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
10476 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
10477 (CRT_CALL_STATIC_FUNCTION): Likewise.
10478 (ASM_DEFAULT_SPEC): New define.
10479 (ASM_SPEC32): Likewise.
10480 (ASM_SPEC64): Likewise.
10481 (ASM_SPEC_COMMON): Likewise.
10482 (ASM_SPEC): Likewise.
10483 (INVALID_64BIT): Likewise.
10484 (LINK_OS_DEFAULT_SPEC): Likewise.
10485 (LINK_OS_SPEC32): Likewise.
10486 (LINK_OS_SPEC64): Likewise.
10487 (POWERPC_LINUX): Likewise.
10488 (PTRDIFF_TYPE): Likewise.
10489 (RESTORE_FP_PREFIX): Likewise.
10490 (RESTORE_FP_SUFFIX): Likewise.
10491 (SAVE_FP_PREFIX): Likewise.
10492 (SAVE_FP_SUFFIX): Likewise.
10493 (SIZE_TYPE): Likewise.
10494 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
10495 (TARGET_64BIT): Likewise.
10496 (TARGET_64BIT): Likewise.
10497 (TARGET_AIX): Likewise.
10498 (WCHAR_TYPE_SIZE): Likewise.
10499 (WCHAR_TYPE): Undefine.
10500 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
10501 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
10502 (CPP_OS_RTEMS_SPEC): Delete.
10503 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
10504 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
10506 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
10508 2017-07-28 Jakub Jelinek <jakub@redhat.com>
10510 PR tree-optimization/81578
10511 * tree-parloops.c (build_new_reduction): Bail out if
10512 reduction_code isn't one of the standard OpenMP reductions.
10513 Move the details printing after that decision.
10515 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
10517 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
10518 related to reload_in_progress.
10519 (splat_input_operand): Likewise.
10520 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
10522 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
10524 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
10525 (TARGET_INSTANTIATE_DECLS): Likewise.
10526 (legitimate_indexed_address_p): Delete reload_in_progress code.
10527 (rs6000_debug_legitimate_address_p): Likewise.
10528 (rs6000_eliminate_indexed_memrefs): Likewise.
10529 (rs6000_emit_le_vsx_store): Likewise.
10530 (rs6000_emit_move_si_sf_subreg): Likewise.
10531 (rs6000_emit_move): Likewise.
10532 (register_to_reg_type): Likewise.
10533 (rs6000_pre_atomic_barrier): Likewise.
10534 (rs6000_machopic_legitimize_pic_address): Likewise.
10535 (rs6000_allocate_stack_temp): Likewise.
10536 (rs6000_address_for_fpconvert): Likewise.
10537 (rs6000_address_for_altivec): Likewise.
10538 (rs6000_secondary_memory_needed_rtx): Delete function.
10539 (rs6000_check_sdmode): Likewise.
10540 (rs6000_alloc_sdmode_stack_slot): Likewise.
10541 (rs6000_instantiate_decls): Likewise.
10542 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
10543 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
10544 Delete reload_in_progress.
10545 (*vec_reload_and_plus_<mptrsize>): Likewise.
10546 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
10547 (vsx_div_v2di): Likewise.
10548 (vsx_udiv_v2di): Likewise.
10550 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
10552 * config/rs6000/rs6000.opt (mlra): Replace with stub.
10553 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
10554 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
10555 (rs6000_debug_reg_global): Delete print of LRA status.
10556 (rs6000_option_override_internal): Delete dead LRA related code.
10557 (rs6000_lra_p): Delete function.
10558 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
10560 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
10562 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
10563 * config/riscv/rtems.h: New file.
10565 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10566 Sudakshina Das <sudi.das@arm.com>
10568 * config/aarch64/aarch64.md
10569 (define_split for and<mode>3nr_compare): Move
10570 non aarch64_logical_operand to a register.
10571 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
10572 register immediate operand to a register.
10573 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
10575 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
10577 PR middle-end/81564
10578 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
10580 2017-07-27 Richard Biener <rguenther@suse.de>
10582 PR tree-optimization/81573
10583 PR tree-optimization/81494
10584 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
10585 multi defuse cycle case.
10587 2017-07-27 Richard Biener <rguenther@suse.de>
10589 PR tree-optimization/81571
10590 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
10593 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
10595 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
10596 earlier and only if MASK_FPU is set. Adjust formatting.
10598 2017-07-27 Martin Liska <mliska@suse.cz>
10600 * opt-functions.awk: Add validation of value of Init.
10601 * optc-gen.awk: Pass new argument.
10603 2017-07-27 Martin Liska <mliska@suse.cz>
10605 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
10606 Fix wrong condition.
10608 2017-07-27 Martin Liska <mliska@suse.cz>
10610 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
10611 BBs and edges seen by autoFDO.
10613 2017-07-27 Richard Biener <rguenther@suse.de>
10615 PR tree-optimization/81502
10616 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
10617 with incompatible but same sized type.
10618 (execute_update_addresses_taken): Likewise.
10620 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
10622 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
10623 flag_tree_loop_vectorize rather than flag_tree_vectorize.
10625 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10628 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
10629 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
10630 Change s_operand to memory_operand.
10632 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
10634 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
10635 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
10636 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
10637 Emit instructions rather than returning an expression. Handle TFmode
10638 and KFmode by casting to TImode.
10639 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
10640 (rs6000_emit_le_vsx_store): Likewise.
10641 * config/rs6000/vsx.md (VSX_TI): New iterator.
10642 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
10643 (*vsx_le_undo_permute_<mode>): Likewise.
10644 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
10645 emit the split sequence.
10646 (*vsx_le_perm_store_<mode>): Likewise.
10648 2017-07-27 Jakub Jelinek <jakub@redhat.com>
10650 PR tree-optimization/81555
10651 PR tree-optimization/81556
10652 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
10653 if true, force CHANGED for the recursive invocation.
10654 (reassociate_bb): Remember original length of ops array, pass
10655 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
10657 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
10658 attributes for noipa attribute. For naked attribute use
10659 lookup_attribute first before lookup_attribute_spec.
10660 * final.c (rest_of_handle_final): Disable IPA RA for functions with
10662 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
10663 for functions with noipa attribute.
10664 (cgraph_externally_visible_p): Return true for functions with noipa
10666 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
10667 for functions with noipa attribute.
10668 * doc/extend.texi: Document noipa function attribute.
10669 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
10670 also for functions with noipa attribute.
10671 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
10673 2017-07-26 Andrew Pinski <apinski@cavium.com>
10675 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
10676 vec_unalign_load_cost and vec_unalign_store_cost.
10678 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10680 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
10681 -mvsx-small-integer option.
10682 (ISA_3_0_MASKS_IEEE): Likewise.
10683 (OTHER_VSX_VECTOR_MASKS): Likewise.
10684 (POWERPC_MASKS): Likewise.
10685 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
10686 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
10687 code, only testing for DImode being allowed in non-VSX floating
10689 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
10690 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
10692 (rs6000_option_override_internal): Delete -mvsx-small-integer.
10693 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
10694 TARGET_P8_VECTOR test.
10695 (rs6000_secondary_reload_simple_move): Likewise.
10696 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
10697 since TARGET_P9_VECTOR was already tested.
10698 (rs6000_opt_masks): Remove -mvsx-small-integer.
10699 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
10700 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
10702 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
10703 test for TARGET_VEXTRACTUB was used, and that uses
10705 (p9 extract splitter): Likewise.
10706 (vsx_extract_<mode>_di_p9): Likewise.
10707 (vsx_extract_<mode>_store_p9): Likewise.
10708 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
10709 for TARGET_P9_VECTOR was used. Delete code that is now dead with
10710 the elimination of TARGET_VSX_SMALL_INTEGER.
10711 (vsx_extract_<mode>_p8): Likewise.
10712 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
10713 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
10714 (vsx_set_<mode>_p9): Likewise.
10715 (vsx_set_v4sf_p9): Likewise.
10716 (vsx_set_v4sf_p9_zero): Likewise.
10717 (vsx_insert_extract_v4sf_p9): Likewise.
10718 (vsx_insert_extract_v4sf_p9_2): Likewise.
10719 * config/rs6000/rs6000.md (sign extend splitter): Change
10720 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
10721 (floatsi<mode>2_lfiwax_mem): Likewise.
10722 (floatunssi<mode>2_lfiwzx_mem): Likewise.
10723 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
10724 since a test for TARGET_P9_VECTOR was used.
10725 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10726 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
10727 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10728 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
10729 TARGET_P8_VECTOR test.
10730 (fix_trunc<mode>si2_stfiwx): Likewise.
10731 (fix_trunc<mode>si2_internal): Likewise.
10732 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
10733 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
10735 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
10736 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
10737 TARGET_P8_VECTOR test.
10738 (fixuns_trunc<mode>si2_stfiwx): Likewise.
10739 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
10740 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
10742 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
10743 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
10744 since a test for TARGET_P9_VECTOR was used.
10745 (splitter for loading small constants): Likewise.
10747 2017-07-26 Andrew Pinski <apinski@cavium.com>
10749 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
10752 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
10755 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
10756 (fp_valid_at): Likewise.
10758 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
10760 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
10761 (qdf24xx_addrcost_table): Likewise.
10762 (cortexa57_tunings): Update to use generic_branch_cost.
10763 (cortexa72_tunings): Likewise.
10764 (cortexa73_tunings): Likewise.
10765 (qdf24xx_tunings): Likewise.
10767 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
10769 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
10770 (thunderx2t99_branch_cost): Likewise.
10771 (cortexa35_tunings): Update to use generic_branch_cost.
10772 (cortexa53_tunings): Likewise.
10773 (cortexa57_tunings): Likewise.
10774 (cortexa72_tunings): Likewise.
10775 (cortexa73_tunings): Likewise.
10776 (thunderx2t99_tunings): Likewise.
10778 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
10780 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
10781 (sparc_option_override): Honour MASK_FSMULD.
10782 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
10783 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
10784 * config/sparc/sparc.opt (mfsmuld): New option.
10785 * doc/invoke.texi (mfsmuld): Document option.
10787 2017-07-26 Marek Polacek <polacek@redhat.com>
10789 PR middle-end/70992
10790 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
10792 2017-07-26 Richard Biener <rguenther@suse.de>
10794 * gimple-match-head.c (do_valueize): Return OP if valueize
10796 (get_def): New helper to get at the def stmt of a SSA name
10797 if valueize allows.
10798 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
10799 do_valueize to get at the def stmt.
10800 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
10802 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
10804 PR middle-end/46932
10805 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
10807 2017-07-26 Martin Liska <mliska@suse.cz>
10810 * function.c (expand_function_start): Make expansion of
10811 nonlocal_goto_save_area after parm_birth_insn.
10813 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
10815 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
10816 from all CPU target flags enable members.
10818 2017-07-26 Richard Biener <rguenther@suse.de>
10820 * genmatch.c (dt_simplify::gen): Make iterator vars const.
10821 (decision_tree::gen): Make 'type' const.
10822 (write_predicate): Likewise.
10824 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
10826 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
10827 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
10828 (rs6000_option_override_internal): Likewise.
10829 (rs6000_expand_vector_set): Likewise.
10830 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
10831 (TARGET_UPPER_REGS_SF): Likewise.
10832 (TARGET_UPPER_REGS_DI): Likewise.
10833 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
10834 (TARGET_DIRECT_MOVE_64BIT): Likewise.
10835 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
10836 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10837 (Splitters for DI constants in Altivec registers): Likewise.
10838 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
10839 (vsx_set_v4sf_p9): Likewise.
10840 (vsx_set_v4sf_p9_zero): Likewise.
10841 (vsx_insert_extract_v4sf_p9): Likewise.
10842 (vsx_insert_extract_v4sf_p9_2): Likewise.
10844 2017-07-25 Carl Love <cel@us.ibm.com>
10846 * doc/extend.texi: Update the built-in documentation file for the
10847 existing built-in functions
10848 vector signed char vec_cnttz (vector signed char);
10849 vector unsigned char vec_cnttz (vector unsigned char);
10850 vector signed short vec_cnttz (vector signed short);
10851 vector unsigned short vec_cnttz (vector unsigned short);
10852 vector signed int vec_cnttz (vector signed int);
10853 vector unsigned int vec_cnttz (vector unsigned int);
10854 vector signed long long vec_cnttz (vector signed long long);
10855 vector unsigned long long vec_cnttz (vector unsigned long long);
10857 2017-07-25 Andrew Pinski <apinski@cavium.com>
10859 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
10860 accesses where the use is for the first operand of a BIT_INSERT.
10862 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
10865 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
10866 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
10868 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
10870 * config/i386/gstabs.h: Delete.
10871 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
10873 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
10875 * config/i386/i386.c (ix86_decompose_address): Do not check for
10876 register RTX when looking at index_reg or base_reg.
10877 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
10879 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
10881 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
10882 to update EH info here.
10884 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10886 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
10888 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10890 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
10892 2017-07-25 Torsten Duwe <duwe@suse.de>
10894 * common.opt: Introduce -fpatchable-function-entry
10895 command line option, and its variables function_entry_patch_area_size
10896 and function_entry_patch_area_start.
10897 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
10898 including a two-value parser.
10899 * target.def (print_patchable_function_entry): New target hook.
10900 * targhooks.h (default_print_patchable_function_entry): New function.
10901 * targhooks.c (default_print_patchable_function_entry): Likewise.
10902 * toplev.c (process_options): Switch off IPA-RA if
10903 patchable function entries are being generated.
10904 * varasm.c (assemble_start_function): Look at the
10905 patchable-function-entry command line switch and current
10906 function attributes and maybe generate NOP instructions by
10907 calling the print_patchable_function_entry hook.
10908 * doc/extend.texi: Document patchable_function_entry attribute.
10909 * doc/invoke.texi: Document -fpatchable_function_entry
10910 command line option.
10911 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
10913 * doc/tm.texi: Re-generate.
10915 2017-07-25 Jakub Jelinek <jakub@redhat.com>
10918 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
10919 TARGET_AVX512DQ rather than TARGET_AVX512BW.
10921 2017-07-25 Tamar Christina <tamar.christina@arm.com>
10923 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
10925 2017-07-25 Richard Biener <rguenther@suse.de>
10927 PR tree-optimization/81455
10928 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
10929 not walk in cycles when looking for guards.
10931 2017-07-25 Richard Biener <rguenther@suse.de>
10933 PR tree-optimization/81529
10934 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
10935 when optimizing backedge uses.
10937 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
10939 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
10941 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
10942 to dl_section_ref. On AIX, append an expression to subtract
10943 the size of the section length to dl_section_ref.
10945 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
10947 * configure.ac: If any of the config.* scripts fail, exit 1.
10948 * configure: Regenerate.
10950 2017-07-25 Richard Biener <rguenther@suse.de>
10952 PR middle-end/81546
10953 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
10954 of immediate uses, be more verbose on errors.
10956 2017-07-25 Richard Biener <rguenther@suse.de>
10958 PR tree-optimization/81510
10959 * tree-vect-loop.c (vect_is_simple_reduction): When the
10960 reduction stmt is not inside the loop bail out.
10962 2017-07-25 Richard Biener <rguenther@suse.de>
10964 PR tree-optimization/81303
10965 * tree-vect-loop-manip.c (vect_loop_versioning): Build
10966 profitability check against LOOP_VINFO_NITERSM1.
10968 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10970 * domwalk.c (cmp_bb_postorder): Simplify.
10971 (sort_bbs_postorder): New function. Use it...
10972 (dom_walker::walk): ...here to optimize common cases.
10974 2017-07-25 Martin Liska <mliska@suse.cz>
10977 * ipa-visibility.c (function_and_variable_visibility): Make the
10978 redirection just on target that supports aliasing.
10979 Fix GNU coding style.
10981 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
10984 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
10986 * config/arm/rtems.h: Likewise.
10987 * config/bfin/rtems.h: Likewise.
10988 * config/i386/rtemself.h: Likewise.
10989 * config/lm32/rtems.h: Likewise.
10990 * config/m32c/rtems.h: Likewise.
10991 * config/m68k/rtemself.h: Likewise.
10992 * config/microblaze/rtems.h: Likewise.
10993 * config/mips/rtems.h: Likewise.
10994 * config/moxie/rtems.h: Likewise.
10995 * config/nios2/rtems.h: Likewise.
10996 * config/powerpcspe/rtems.h: Likewise.
10997 * config/rs6000/rtems.h: Likewise.
10998 * config/rtems.h: Likewise.
10999 * config/sh/rtems.h: Likewise.
11000 * config/sh/rtemself.h: Likewise.
11001 * config/sparc/rtemself.h: Likewise.
11003 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
11006 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
11007 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
11008 * tree-ssa-structalias.c (alias_get_name): Same.
11010 2017-07-25 Bin Cheng <bin.cheng@arm.com>
11013 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
11014 instructions if no du chain is found.
11016 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
11018 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
11020 2017-07-25 Richard Biener <rguenther@suse.de>
11022 PR middle-end/81505
11023 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
11026 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
11028 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
11029 upper-regs options.
11030 (ISA_2_7_MASKS_SERVER): Likewise.
11031 (ISA_3_0_MASKS_IEEE): Likewise.
11032 (OTHER_P8_VECTOR_MASKS): Likewise.
11033 (OTHER_VSX_VECTOR_MASKS): Likewise.
11034 (POWERPC_MASKS): Likewise.
11035 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
11036 duplicate list of options.
11037 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
11038 explicit -mupper-regs options.
11039 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
11040 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
11041 alias for -mupper-regs-df.
11042 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
11043 (rs6000_init_hard_regno_mode_ok): Likewise.
11044 (rs6000_option_override_internal): Likewise.
11045 (rs6000_opt_masks): Likewise.
11046 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
11047 options in terms of whether -mvsx or -mpower8-vector was used.
11048 (TARGET_UPPER_REGS_DI): Likewise.
11049 (TARGET_UPPER_REGS_SF): Likewise.
11050 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
11051 -mupper-regs-* options.
11053 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
11055 * passes.c (emergency_dump_function): Print some empty lines and a
11056 header before the RTL dump.
11058 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
11060 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
11062 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
11065 * config/aarch64/aarch64.c (aarch64_classify_symbol):
11066 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
11068 2017-07-24 Carl Love <cel@us.ibm.com>
11070 * config/rs6000/rs6000-c.c: Add support for built-in functions
11071 vector float vec_extract_fp32_from_shorth (vector unsigned short);
11072 vector float vec_extract_fp32_from_shortl (vector unsigned short);
11073 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
11074 vec_extract_fp_from_shortl): Add defines for the two builtins.
11075 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
11076 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
11078 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
11079 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
11080 * doc/extend.texi: Update the built-in documentation file for the
11081 new built-in function.
11083 2017-07-24 Jakub Jelinek <jakub@redhat.com>
11086 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
11088 * doc/generic.texi: Likewise.
11089 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
11090 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
11092 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
11094 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
11095 (aarch64_mls_elt_merge<mode>): Likewise.
11097 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
11099 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
11100 having __cxa_atexit.
11102 2017-07-23 Michael Collison <michael.collison@arm.com>
11104 * config/arm/arm.c (arm_option_override): Deprecate
11105 use of -mstructure-size-boundary.
11106 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
11107 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
11109 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11112 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
11113 Reduce cost estimate for direct moves.
11115 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
11118 * config/i386/i386.c (ix86_option_override_internal): Disable
11119 BMI, BMI2 and TBM instructions for -m16.
11121 2017-07-21 Carl Love <cel@us.ibm.com>
11123 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11124 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11125 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11126 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11127 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11128 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11129 VMULOSW): New enum "unspec" values.
11130 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11131 altivec_vmulosw): New patterns.
11132 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11133 VMULOSW): Add definitions.
11135 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
11137 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
11138 (qdf24xx): Likewise.
11139 * config/aarch64/aarch64-options-extensions.def (rdma); New.
11140 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
11141 (AARCH64_FL_V8_1): Renumber.
11142 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
11143 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
11144 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
11145 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
11146 rdma to feature modifiers list.
11148 2017-07-21 Yury Gribov <tetra2005@gmail.com>
11150 PR middle-end/56727
11151 * ipa-visibility (function_and_variable_visibility): Convert
11152 recursive PLT call to direct call if appropriate.
11154 2017-07-21 Andrew Pinski <apinski@cavium.com>
11156 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
11157 operand 1 to see if the types precision matches.
11158 * fold-const.c (operand_equal_p): Likewise.
11160 2017-07-21 Richard Biener <rguenther@suse.de>
11162 PR tree-optimization/81303
11163 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
11164 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
11165 (vect_peeling_hash_get_lowest_cost): Adjust.
11166 (vect_enhance_data_refs_alignment): Likewise. Use
11167 vect_get_peeling_costs_all_drs to compute the penalty for no
11168 peeling to match up costs.
11170 2017-07-21 Richard Biener <rguenther@suse.de>
11172 PR tree-optimization/81500
11173 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
11174 we didn't identify a reduction path.
11176 2017-07-21 Tom de Vries <tom@codesourcery.com>
11177 Cesar Philippidis <cesar@codesourcery.com>
11179 PR gcov-profile/81442
11180 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
11183 2017-07-21 Tom de Vries <tom@codesourcery.com>
11186 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
11188 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
11189 nvptx_override_options_after_change.
11191 2017-07-21 Ulrich Drepper <drepper@redhat.com>
11193 * dwarf2out.c (output_file_names): Avoid double testing for
11194 dwarf_version >= 5.
11196 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
11198 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
11200 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
11202 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
11204 (try_crossjump_to_edge): Do not punt on partitioned functions.
11206 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
11208 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
11209 Put all BBs reachable only via paths crossing cold region to cold
11211 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
11213 2016-07-21 Richard Biener <rguenther@suse.de>
11215 PR tree-optimization/81303
11216 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
11217 into account prologue and epilogue iterations when raising
11218 min_profitable_iters to sth at least covering one vector iteration.
11220 2017-07-21 Tamar Christina <tamar.christina@arm.com>
11222 * config/arm/arm.c (arm_test_cpu_arch_dat):
11225 2017-07-20 Nathan Sidwell <nathan@acm.org>
11227 Remove TYPE_METHODS.
11228 * tree.h (TYPE_METHODS): Delete.
11229 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
11230 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
11231 (dbxout_type_methods): Scan TYPE_FIELDS.
11232 (dbxout_type): Don't check TYPE_METHODS here.
11233 * function.c (use_register_for_decl): Always ignore register for
11234 class types when not optimizing.
11235 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
11236 * tree.c (free_lang_data_in_type): Stitch out member functions and
11237 templates from TYPE_FIELDS.
11238 (build_distinct_type_copy, verify_type_variant,
11239 verify_type): Member fns are on TYPE_FIELDS.
11240 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
11241 * tree-pretty-print.c (dump_generic_node): Likewise.
11243 2017-07-20 Jakub Jelinek <jakub@redhat.com>
11246 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
11247 V2TImode and V4TImode.
11248 (ix86_expand_vector_extract): Likewise.
11249 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
11250 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
11251 (ssescalarmode): Handle V4TImode and V2TImode.
11252 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
11253 (*vec_extractv2ti, *vec_extractv4ti): New insns.
11254 (VEXTRACTI128_MODE): New mode iterator.
11255 (splitter for *vec_extractv?ti first element): New.
11256 (VEC_INIT_MODE): New mode iterator.
11257 (vec_init<mode>): Consolidate 3 expanders into one using
11258 VEC_INIT_MODE mode iterator.
11260 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
11262 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
11263 non_spilled_static_chain_regno_p.
11265 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
11267 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
11269 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
11271 * bb-reorder.c (connect_traces): Allow copying of blocks within
11274 2017-07-20 Richard Biener <rguenther@suse.de>
11276 * gimple.h (gimple_phi_result): Add gphi * overload.
11277 (gimple_phi_result_ptr): Likewise.
11278 (gimple_phi_arg): Likewise. Adjust index assert to only
11279 allow actual argument accesses rather than all slots available
11281 (gimple_phi_arg_def): Add gphi * overload.
11282 * tree-phinodes.c (make_phi_node): Initialize only actual
11284 (resize_phi_node): Clear memory not covered by old node,
11285 do not initialize excess argument slots.
11286 (reserve_phi_args_for_new_edge): Initialize new argument slot
11289 2017-07-20 Bin Cheng <bin.cheng@arm.com>
11291 PR tree-optimization/81388
11293 2016-07-21 Bin Cheng <bin.cheng@arm.com>
11295 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
11296 by removing computation of may_be_zero.
11298 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
11299 Tom de Vries <tom@codesourcery.com>
11301 PR middle-end/81030
11302 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
11303 when gimple level profile disagrees with what RTL expander did.
11305 2017-07-20 Richard Biener <rguenther@suse.de>
11307 PR tree-optimization/61171
11308 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
11309 (vect_analyze_stmt): Add slp instance parameter.
11310 (vectorizable_reduction): Likewise.
11311 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
11312 (vect_is_simple_reduction): Deal with chains not detected
11313 as SLP reduction chain, specifically not properly associated
11314 chains containing a mix of plus/minus.
11315 (get_reduction_op): Remove.
11316 (get_initial_defs_for_reduction): Simplify, pass in whether
11317 this is a reduction chain, pass in the SLP node for the PHIs.
11318 (vect_create_epilog_for_reduction): Get the SLP instance as
11320 (vectorizable_reduction): Get the SLP instance as arg.
11321 During analysis remember the SLP node with the PHIs in the
11322 instance. Simplify getting at the vectorized reduction PHIs.
11323 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
11324 through SLP instance.
11325 (vect_slp_analyze_operations): Likewise.
11326 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
11327 (vect_transform_stmt): Likewise.
11329 2017-07-20 Tom de Vries <tom@codesourcery.com>
11331 PR tree-optimization/81489
11332 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
11333 read of phi arg location to before loop that modifies phi.
11335 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11337 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
11340 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
11342 PR middle-end/81331
11343 * except.c (execute): Fix ordering issue.
11345 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
11347 PR rtl-optimization/81423
11348 * combine.c (make_compound_operation_int): Don't try to optimize
11349 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
11351 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
11353 PR rtl-optimization/81423
11354 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
11355 with a constant that is -1 in the truncated to mode.
11357 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
11359 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
11360 (determine_unlikely_bbs): ... here.
11361 * predict.h (propagate_unlikely_bbs_forward): Declare.
11362 * cfgexpand.c (pass_expand::execute): Use it.
11363 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
11365 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
11366 propagate_unlikely_bbs_forward.
11368 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
11370 PR middle-end/81331
11371 * except.c (maybe_add_nop_after_section_switch): New function.
11374 2017-07-19 Tom de Vries <tom@codesourcery.com>
11376 * gimple.h (gimple_phi_set_arg): Make assert more strict.
11378 2017-07-19 Tom de Vries <tom@codesourcery.com>
11380 * gimple.h (gimple_phi_arg): Make assert more strict.
11382 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
11384 * config.gcc (powerpc*-*-*): Add mmintrin.h.
11385 * config/rs6000/mmintrin.h: New file.
11386 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
11388 2017-07-19 Jakub Jelinek <jakub@redhat.com>
11390 PR tree-optimization/81346
11391 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
11393 2017-07-19 Tom de Vries <tom@codesourcery.com>
11395 * config/nvptx/nvptx.md (VECIM): Add V2DI.
11397 2017-07-19 Tom de Vries <tom@codesourcery.com>
11399 * config/nvptx/nvptx-modes.def: Add V2DImode.
11400 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
11401 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
11402 (nvptx_output_mov_insn): Handle lack of mov.b128.
11403 (nvptx_print_operand): Handle 'H' and 'L' codes.
11404 (nvptx_vector_mode_supported): Allow V2DImode.
11405 (nvptx_preferred_simd_mode): New function.
11406 (nvptx_data_alignment): New function.
11407 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
11408 nvptx_preferred_simd_mode.
11409 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
11411 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
11413 2017-07-19 Tom de Vries <tom@codesourcery.com>
11415 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
11416 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
11417 (nvptx_vector_mode_supported): New function. Allow V2SImode.
11418 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
11419 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
11420 (mov<VECIM>_insn): New define_insn.
11421 (define_expand "mov<VECIM>): New define_expand.
11423 2017-07-19 Tom de Vries <tom@codesourcery.com>
11425 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
11427 2017-07-19 Jakub Jelinek <jakub@redhat.com>
11429 PR tree-optimization/81346
11430 * fold-const.h (fold_div_compare, range_check_type): Declare.
11431 * fold-const.c (range_check_type): New function.
11432 (build_range_check): Use range_check_type.
11433 (fold_div_compare): No longer static, rewritten into
11434 a match.pd helper function.
11435 (fold_comparison): Don't call fold_div_compare here.
11436 * match.pd (X / C1 op C2): New optimization using fold_div_compare
11437 as helper function.
11439 2017-07-19 Nathan Sidwell <nathan@acm.org>
11441 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
11442 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
11443 * tree.c (find_decls_types_r, verify_type): Use
11444 TYPE_{MIN,MAX}_VALUE_RAW.
11445 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
11446 (hash_tree): Likewise.
11447 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
11449 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
11452 2017-07-18 Tom de Vries <tom@codesourcery.com>
11454 PR middle-end/81464
11455 * omp-expand.c (expand_omp_for_static_chunk): Handle
11456 equal-argument loop exit phi.
11458 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
11461 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
11462 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
11463 operand 2 predicate.
11464 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
11465 operand 2 predicate.
11466 (ror,rol -> rorx splitters): Use const_int_operand as
11467 operand 2 predicate.
11469 2017-06-18 Richard Biener <rguenther@suse.de>
11471 PR tree-optimization/81410
11472 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
11473 the gap in the ! slp_perm SLP case after each group.
11475 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
11477 PR middle-end/81463
11478 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
11481 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
11483 PR middle-end/81462
11484 * predict.c (set_even_probabilities): Cleanup; do not affect
11485 probabilities that are already known.
11486 (combine_predictions_for_bb): Call even when count is set.
11488 2017-07-18 Nathan Sidwell <nathan@acm.org>
11490 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
11493 2017-07-18 Bin Cheng <bin.cheng@arm.com>
11496 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
11497 optimization for loop niter analysis.
11499 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
11502 * config/avr/avr.c (avr_optimize_casesi): Don't use
11503 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
11505 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11507 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
11508 body_cost_vec from _vect_peel_extended_info.
11509 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
11510 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
11513 2017-07-18 Bin Cheng <bin.cheng@arm.com>
11515 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
11517 2017-07-18 Richard Biener <rguenther@suse.de>
11519 PR tree-optimization/80620
11520 PR tree-optimization/81403
11521 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
11522 info when re-using a VN table entry.
11524 2017-07-18 Richard Biener <rguenther@suse.de>
11526 PR tree-optimization/81418
11527 * tree-vect-loop.c (vectorizable_reduction): Properly compute
11528 vectype_in. Verify that with lane-reducing reduction operations
11529 we have a single def-use cycle.
11531 2017-07-17 Carl Love <cel@us.ibm.com>
11533 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
11535 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11536 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11537 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11538 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11539 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11540 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11541 VMULOSW): New enum "unspec" values.
11542 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
11543 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
11544 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11545 altivec_vmulosw): New patterns.
11546 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11547 VMULOSW): Add definitions.
11548 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
11550 * config/alpha/alpha.c: Include predict.h.
11552 2017-07-17 Yury Gribov <tetra2005@gmail.com>
11554 * tree-vrp.c (compare_assert_loc): Fix comparison function
11555 to return predictable results.
11557 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11559 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
11561 (subdi3): Likewise.
11562 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
11563 * doc/invoke.texi (mexpand-adddi): Update text.
11565 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11567 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
11568 that also clobbers the CC register. The old expand code is moved
11570 (*arc_clzsi2): ... here.
11571 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
11572 the CC register. The old expand code is moved to ...
11573 (arc_ctzsi2): ... here.
11575 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11577 * config/arc/arc.opt (mindexed-loads): Use initial value
11578 TARGET_INDEXED_LOADS_DEFAULT.
11579 (mauto-modify-reg): Use initial value
11580 TARGET_AUTO_MODIFY_REG_DEFAULT.
11581 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
11582 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
11583 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
11584 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
11586 2017-07-17 Martin Liska <mliska@suse.cz>
11589 * opts.c (finish_options): Do not allow -fgnu-tm
11590 w/ -fsanitize={kernel-,}address. Say sorry.
11592 2017-07-17 Bin Cheng <bin.cheng@arm.com>
11595 * tree-loop-distribution.c (classify_partition): Only assert on
11596 numer of iterations.
11597 (merge_dep_scc_partitions): Delete prameter. Update function call.
11598 (distribute_loop): Remove code handling loop with unknown niters.
11599 (pass_loop_distribution::execute): Skip loop with unknown niters.
11601 2017-07-17 Bin Cheng <bin.cheng@arm.com>
11604 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
11605 function sort_partitions_by_post_order.
11607 2017-07-17 Bin Cheng <bin.cheng@arm.com>
11609 PR tree-optimization/81374
11610 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
11611 the max index of basic blocks, rather than number of basic blocks.
11613 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11615 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
11617 (arc_legitimate_pic_operand_p): Likewise.
11618 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
11620 (arc_needs_pcl_p): Likewise.
11621 (arc_legitimate_pc_offset_p): Likewise.
11622 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
11623 function is also used in constrains.md.
11624 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
11625 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
11626 PLUS. Only return true/false in known cases, otherwise assert.
11627 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
11628 is already called in arc_legitimate_constant_p.
11629 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
11631 (LEGITIMATE_PIC_OPERAND_P): Use
11632 arc_raw_symbolic_reference_mentioned_p function.
11633 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
11638 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11639 Andrew Burgess <andrew.burgess@embecosm.com>
11641 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
11642 (arc_return_address_register): New function.
11643 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
11644 (arc_handle_fndecl_attribute): Add naked attribute.
11645 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
11646 (TARGET_WARN_FUNC_RETURN): Likewise.
11647 (arc_allocate_stack_slots_for_args): New function.
11648 (arc_warn_func_return): Likewise.
11649 (machine_function): Change type fn_type.
11650 (arc_compute_function_type): Consider new naked function type,
11651 change function return type.
11652 (arc_must_save_register): Adapt to handle new
11653 arc_compute_function_type's return type.
11654 (arc_expand_prologue): Likewise.
11655 (arc_expand_epilogue): Likewise.
11656 (arc_return_address_regs): Delete.
11657 (arc_return_address_register): New function.
11658 (arc_epilogue_uses): Use above function.
11659 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
11660 (arc_function_type): Change encoding, add naked type.
11661 (ARC_INTERRUPT_P): Change to handle the new encoding.
11662 (ARC_FAST_INTERRUPT_P): Likewise.
11663 (ARC_NORMAL_P): Define.
11664 (ARC_NAKED_P): Likewise.
11665 (arc_compute_function_type): Delete prototype.
11666 * config/arc/arc.md (in_ret_delay_slot): Use
11667 arc_return_address_register function.
11668 (simple_return): Likewise.
11669 (p_return_i): Likewise.
11671 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11673 PR tree-optimization/81428
11674 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
11675 can't be built for those types.
11677 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
11679 Remove stuff dead since r239246.
11681 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
11682 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
11683 (avr_inform_devices): Remove dead stuff.
11685 2017-07-17 Tamar Christina <tamar.christina@arm.com>
11687 * config/arm/arm_neon.h: Fix softp typo.
11689 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11691 PR tree-optimization/81365
11692 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
11693 aggregate moves onto bb predecessor edges, make sure there are no
11694 loads that could alias the lhs in between the start of bb and the
11697 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
11700 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
11701 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
11702 [LSHIFTRT, outer_code = TRUNCATE]: Same.
11704 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11706 PR tree-optimization/81396
11707 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
11708 (init_symbolic_number): Initialize it to 1.
11709 (perform_symbolic_merge): Add n_ops from both operands into the new
11711 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
11712 without base_addr as useless if they need more than one operation.
11713 (bswap_replace): Handle !bswap case for NULL base_addr.
11715 2017-07-17 Tom de Vries <tom@codesourcery.com>
11718 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
11721 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
11723 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
11724 conditional builtin define __FIX_LEON3FT_B2BST.
11726 2017-07-17 Daniel Cederman <cederman@gaisler.com>
11728 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
11729 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
11732 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
11734 PR rtl-optimization/81424
11735 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
11736 to remove potential trapping from operands if -fnon-call-exceptions.
11738 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11740 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
11741 profile_proability for scalling.
11742 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
11744 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11746 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
11748 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11750 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
11751 fixpoint arithmetics.
11753 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11755 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
11756 fixpoint arithmetics.
11758 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11760 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
11761 fixpoint arithmetics.
11763 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11765 * profile-count.h (profile_probability::from_reg_br_prob_note,
11766 profile_probability::to_reg_br_prob_note): New functions.
11767 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
11768 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
11769 * predict.c (probability_reliable_p): Update.
11770 (edge_probability_reliable_p): Update.
11771 (br_prob_note_reliable_p): Update.
11772 (invert_br_probabilities): Update.
11773 (add_reg_br_prob_note): New function.
11774 (combine_predictions_for_insn): Update.
11775 * asan.c (asan_clear_shadow): Update.
11776 * cfgbuild.c (compute_outgoing_frequencies): Update.
11777 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
11778 (update_br_prob_note): Update.
11779 (rtl_verify_edges): Update.
11780 (purge_dead_edges): Update.
11781 (fixup_reorder_chain): Update.
11782 * emit-rtl.c (try_split): Update.
11783 * ifcvt.c (cond_exec_process_insns): Update.
11784 (cond_exec_process_if_block): Update.
11785 (dead_or_predicable): Update.
11786 * internal-fn.c (expand_addsub_overflow): Update.
11787 (expand_neg_overflow): Update.
11788 (expand_mul_overflow): Update.
11789 * loop-doloop.c (doloop_modify): Update.
11790 * loop-unroll.c (compare_and_jump_seq): Update.
11791 * optabs.c (emit_cmp_and_jump_insn_1): Update.
11792 * predict.h: Update.
11793 * reorg.c (mostly_true_jump): Update.
11795 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
11796 * config/alpha/alpha.c (emit_unlikely_jump): Update.
11797 * config/arc/arc.c: (emit_unlikely_jump): Update.
11798 * config/arm/arm.c: (emit_unlikely_jump): Update.
11799 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
11800 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
11801 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
11802 (ix86_print_operand): Update.
11803 (ix86_split_fp_branch): Update.
11804 (predict_jump): Update.
11805 * config/ia64/ia64.c (ia64_print_operand): Update.
11806 * config/mmix/mmix.c (mmix_print_operand): Update.
11807 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
11808 (rs6000_expand_split_stack_prologue): Update.
11809 * config/rs6000/rs6000.c: Update.
11810 * config/s390/s390.c (s390_expand_vec_strlen): Update.
11811 (s390_expand_vec_movstr): Update.
11812 (s390_expand_cs_tdsi): Update.
11813 (s390_expand_split_stack_prologue): Update.
11814 * config/sh/sh.c (sh_print_operand): Update.
11815 (expand_cbranchsi4): Update.
11816 (expand_cbranchdi4): Update.
11817 * config/sparc/sparc.c (output_v9branch): Update.
11818 * config/spu/spu.c (get_branch_target): Update.
11819 (ea_load_store_inline): Update.
11820 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
11821 * config/tilepro/tilepro.c: Update.
11823 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
11825 * gimplify.c (mostly_copy_tree_r): Revert latest change.
11826 (gimplify_save_expr): Likewise.
11828 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11830 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
11832 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11834 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
11836 * timevar.def (TV_IPA_FNSUMMARY): Define.
11838 2017-07-16 Daniel Cederman <cederman@gaisler.com>
11840 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
11841 to back store errata sensitive sequence from being generated.
11842 (sqrtdf2_fix): Likewise.
11844 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11846 * tree-ssa-threadupdate.c (compute_path_counts,
11847 update_joiner_offpath_counts): Use profile_probability.
11849 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
11852 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11854 * config/arm/arm-c.c (arm_cpu_builtins): Define
11855 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
11857 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
11859 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11860 array entries to represent __ieee128 versions of the
11861 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
11862 scalar_extract_sig, and scalar_insert_exp built-in functions.
11863 (altivec_resolve_overloaded_builtin): Add special case handling
11864 for the __builtin_scalar_insert_exp function, as represented by
11865 the P9V_BUILTIN_VEC_VSIEDP constant.
11866 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
11867 exponent support for __ieee128 argument.
11868 (VSESQP): Add scalar extract signature support for __ieee128
11870 (VSTDCNQP): Add scalar test negative support for __ieee128
11872 (VSIEQP): Add scalar insert exponent support for __int128 argument
11873 with __ieee128 result.
11874 (VSIEQPF): Add scalar insert exponent support for __ieee128
11875 argument with __ieee128 result.
11876 (VSTDCQP): Add scalar test data class support for __ieee128
11878 (VSTDCNQP): Add overload support for scalar test negative with
11879 __ieee128 argument.
11880 (VSTDCQP): Add overload support for scalar test data class
11881 __ieee128 argument.
11882 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
11883 UNSPEC_VSX_SXSIGDP.
11884 (UNSPEC_VSX_SIEXPQP): New constant.
11885 (xsxexpqp): New insn for VSX scalar extract exponent quad
11887 (xsxsigqp): New insn for VSX scalar extract significand quad
11889 (xsiexpqpf): New insn for VSX scalar insert exponent quad
11890 precision with floating point argument.
11891 (xststdcqp): New expand for VSX scalar test data class quad
11893 (xststdcnegqp): New expand for VSX scalar test negative quad
11895 (xststdcqp): New insn to match expansions for VSX scalar test data
11896 class quad precision and VSX scalar test negative quad precision.
11897 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
11898 special case operand checking to enforce that second operand of
11899 VSX scalar test data class with quad precision argument is a 7-bit
11901 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
11902 prototypes and descriptions of __ieee128 versions of
11903 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
11904 scalar_test_data_class, and scalar_test_neg built-in functions.
11906 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11908 PR tree-optimization/81162
11909 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
11910 replace a negate with an add.
11912 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
11914 * doc/invoke.texi (arm/-mcpu): Document +crypto.
11916 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11918 * config/arm/arm-c.c (arm_cpu_builtins): Define
11919 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
11921 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11923 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
11924 (armv8-r): Set ARM Cortex-R52 as default CPU.
11925 * config/arm/arm-tables.opt: Regenerate.
11926 * config/arm/arm-tune.md: Regenerate.
11927 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
11929 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
11930 extension for -mcpu=cortex-r52.
11932 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11934 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
11935 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
11936 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
11937 (fp-armv8): Define it as FP_ARMv8 only.
11938 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
11939 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
11941 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
11942 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
11943 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
11944 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
11945 than TARGET_FPU_ARMV8.
11946 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
11947 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
11948 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
11950 * config/arm/neon.md (neon_vrint): Likewise.
11951 (neon_vcvt): Likewise.
11952 (neon_<fmaxmin_op><mode>): Likewise.
11953 (<fmaxmin><mode>3): Likewise.
11954 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
11955 * config/arm/predicates.md (arm_cond_move_operator): Check against
11956 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
11958 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
11960 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
11961 to top of function.
11963 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11965 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
11966 loop in comment with memset.
11968 2017-07-14 Martin Liska <mliska@suse.cz>
11970 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
11971 * dwarf2out.c (is_java): Remove the function.
11972 (output_pubname): Remove usage of the function.
11973 (lower_bound_default): Remove usage of DW_LANG_Java.
11974 (gen_compile_unit_die): Likewise.
11975 * gcc.c: Remove compiler defaults for .java and .zip files.
11976 * gimple-expr.c (remove_suffix): Change as there's no longer
11977 extension than 4-letter one.
11978 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
11979 (gimplify_save_expr): Likewise.
11980 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
11981 as it's possible even for other languages than Java.
11982 * langhooks.h (struct lang_hooks): Remove Java from a comment.
11983 * lto-opts.c (lto_write_options): Remove reference to Java.
11984 * opts.c (strip_off_ending): Update file extension handling.
11985 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
11986 * tree-eh.c (lower_resx): Likewise.
11987 * tree.c (free_lang_data_in_type): Remove dead code.
11988 (find_decls_types_r): Likewise.
11989 (build_common_builtin_nodes): Remove Java from a comment.
11990 (verify_type): Remove dead code.
11991 * varasm.c (assemble_external): Remove Java from a comment.
11993 2017-07-14 Martin Liska <mliska@suse.cz>
11995 * opts.c (finish_options): Add quotes.
11996 (common_handle_option): Likewise.
11998 2017-07-14 Martin Liska <mliska@suse.cz>
12000 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
12001 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
12002 Remove N_SO_PASCAL.
12003 * dwarf2out.c (lower_bound_default): Do not handle
12005 (gen_compile_unit_die): Likewise.
12006 * gcc.c: Remove default extension binding for GNU Pascal.
12007 * stmt.c: Remove Pascal language from a comment.
12008 * xcoffout.c: Likewise.
12010 2017-07-13 David Malcolm <dmalcolm@redhat.com>
12013 * diagnostic-show-locus.c (fixit_cmp): New function.
12014 (layout::layout): Sort m_fixit_hints.
12015 (column_range::column_range): Assert that the values are valid.
12016 (struct char_span): New struct.
12017 (correction::overwrite): New method.
12018 (struct source_line): New struct.
12019 (line_corrections::add_hint): Add assertions. Reimplement memcpy
12020 calls in terms of classes source_line and char_span, and
12021 correction::overwrite.
12022 (selftest::test_overlapped_fixit_printing_2): New function.
12023 (selftest::diagnostic_show_locus_c_tests): Call it.
12025 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
12027 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
12028 early if there is no lhs.
12030 2017-07-13 Martin Liska <mliska@suse.cz>
12032 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
12033 (gen_reference_type_die): Likewise.
12034 * stor-layout.c: Remove Pascal-related comment.
12036 2017-07-13 Martin Liska <mliska@suse.cz>
12038 * opts.c (finish_options): Add quotes to error messages.
12039 (parse_sanitizer_options): Likewise.
12041 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
12043 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
12045 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
12047 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
12049 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
12051 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
12053 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
12055 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
12058 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
12059 provides the hardware capability bits, define the macro
12060 __BUILTIN_CPU_SUPPORTS__.
12061 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
12062 if GLIBC does not provide the hardware capability bits. Add a
12063 gcc_unreachable call if the built-in cpu function is neither
12064 __builtin_cpu_is nor __builtin_cpu_supports.
12065 (rs6000_get_function_versions_dispatcher): Change the warning
12066 that an old GLIBC is used which does not export the capability
12067 bits to be an error.
12068 * doc/extend.texi (target_clones attribute): Document the
12069 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
12070 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
12071 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
12072 the macros defined by GCC if the newer GLIBC is available.
12074 2017-07-12 Jeff Law <law@redhat.com>
12076 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
12077 remaining includes slightly.
12078 * config/riscv/riscv-builtins.c: Include profile-count.h.
12080 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
12083 * config/avr/avr.c (avr_set_current_function): In diagnostic
12084 messages: Quote keywords and (parts of) identifiers.
12085 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
12088 2017-07-12 Carl Love <cel@us.ibm.com>
12090 * config/rs6000/rs6000-c.c: Add support for built-in functions
12091 vector bool char vec_revb (vector bool char);
12092 vector bool short vec_revb (vector short char);
12093 vector bool int vec_revb (vector bool int);
12094 vector bool long long vec_revb (vector bool long long);
12095 * doc/extend.texi: Update the built-in documentation file for the
12096 new built-in functions.
12098 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12100 * config/s390/s390.md: Remove movcc splitter.
12102 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12104 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
12105 load/store on condition.
12107 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
12110 * config/avr/avr.c (avr_encode_section_info)
12111 [progmem && !TREE_READONLY]: Error if progmem object needs
12114 2017-07-11 Michael Collison <michael.collison@arm.com>
12116 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
12119 2017-07-11 Carl Love <cel@us.ibm.com>
12121 * config/rs6000/rs6000-c.c: Add support for builtins
12122 vector unsigned int vec_parity_lsbb (vector signed int);
12123 vector unsigned int vec_parity_lsbb (vector unsigned int);
12124 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
12125 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
12126 vector unsigned long long vec_parity_lsbb (vector signed long long);
12127 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
12128 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
12129 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
12130 * doc/extend.texi: Update the built-in documentation file for the
12131 new built-in functions.
12133 2017-07-11 David Malcolm <dmalcolm@redhat.com>
12135 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
12136 (layout::m_primary_loc): New field.
12137 (layout::layout): Initialize new field. Move location filtering
12138 logic from here to...
12139 (layout::maybe_add_location_range): ...this new method. Add
12140 support for filtering to just the lines already specified by other
12142 (layout::will_show_line_p): New method.
12143 (gcc_rich_location::add_location_if_nearby): New method.
12144 (selftest::test_add_location_if_nearby): New test function.
12145 (selftest::diagnostic_show_locus_c_tests): Call it.
12146 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
12149 2017-07-11 Tom de Vries <tom@codesourcery.com>
12151 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
12152 (bb_first_real_insn): New function.
12153 (nvptx_single): Add extra initialization of broadcasted condition
12156 2017-07-11 Nathan Sidwell <nathan@acm.org>
12158 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
12160 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
12162 * doc/extend.texi (AVR Function Attributes): Remove weblink to
12163 Binutils doc as TEXI will mess them up.
12164 * doc/invoke.texi (AVR Options): Same here.
12166 2017-07-11 Daniel Cederman <cederman@gaisler.com>
12168 * config/sparc/sparc.opt (mfix-ut700): New option.
12169 (mfix-gr712rc): Likewise.
12170 (sparc_fix_b2bst): New variable.
12171 * doc/invoke.texi (SPARC options): Document them.
12172 (ARM options): Fix warnings.
12173 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
12174 instructions to prevent sequences that can trigger the store-store
12175 errata for certain LEON3FT processors.
12176 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
12177 (sparc_option_override): Set sparc_fix_b2bst appropriately.
12178 * config/sparc/sparc.md (fix_b2bst): New attribute.
12179 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
12181 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
12184 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
12186 (*rsqrtsf2_sse): Ditto.
12188 (div<mode>3): Macroize insn from divdf3 and divsf3
12189 using MODEF mode iterator.
12191 2017-07-10 Martin Sebor <msebor@redhat.com>
12193 PR tree-optimization/80397
12194 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
12195 instead of testing for equality to INTEGER_TYPE.
12197 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
12199 * config.gcc: Remove uclibc from arc target spec.
12201 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
12203 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
12205 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12208 * lto-wrapper.c (remove_option): New function.
12209 (merge_and_complain): Merge PIC/PIE options more realistically.
12211 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
12213 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
12217 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
12218 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
12219 * config.in: Regenerate.
12220 * configure: Regenerate.
12221 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
12222 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
12223 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
12224 (TARGET_GASISR_PROLOGUES): ...target mask.
12225 * common/config/avr/avr-common.c
12226 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
12227 Set -mgas-isr-prologues.
12228 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
12229 INSERT_PASS_BEFORE for it.
12230 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
12231 * config/avr/avr.c (avr_option_override)
12232 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
12233 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
12234 (avr_attribute_table) <no_gccisr>: Add new function attribute.
12235 (avr_set_current_function) <is_no_gccisr>: Init machine field.
12236 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
12238 (make_avr_pass_pre_proep): New function.
12239 (emit_push_sfr) <treg>: Add argument to function and use it
12240 instead of TMP_REG.
12241 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
12242 and set machine->gasisr.yes.
12243 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
12244 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
12245 __gcc_isr.n_pushed to .L__stack_usage.
12246 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
12247 (avr_asm_final_postscan_insn): ...this new static function.
12248 * config/avr/avr.h (machine_function)
12249 <is_no_gccisr, use_L__stack_usage>: New fields.
12250 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
12251 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
12252 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
12253 (gasisr, *gasisr): New expander and insn.
12254 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
12255 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
12256 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
12258 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
12260 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
12263 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
12265 Move jump-tables out of .text again.
12268 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
12269 (ASM_OUTPUT_ADDR_VEC): New function.
12270 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
12271 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
12272 INSN_ADDRESSes as asm comment.
12273 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
12274 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
12275 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
12276 * config/avr/avr.md (*tablejump): Adjust comment.
12277 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
12278 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
12280 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
12281 (avr_output_addr_vec): New proto.
12282 (avr_log_t) <insn_addresses>: New field.
12284 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
12287 * config/i386/i386.c (ix86_function_arg_advance): Set
12288 outgoing_args_on_stack to true if there are outgoing arguments
12290 (ix86_function_arg): Likewise.
12291 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
12292 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
12293 * config/i386/i386.h (machine_function): Add
12294 outgoing_args_on_stack.
12296 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
12298 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
12299 supporting pthreds.
12300 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
12302 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
12304 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
12305 (REAL_H): Remove $(MACHMODE_H).
12306 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
12308 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
12309 $(MACHMODE_H) and double-int.h.
12310 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
12312 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
12315 2017-07-07 Andrew Pinski <apinski@cavium.com>
12317 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
12318 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
12320 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
12322 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
12323 Add warning if GCC was not configured to link against a GLIBC that
12324 exports the hardware capability bits.
12325 (make_resolver_func): Make resolver function private and not a
12326 COMDAT function. Create the name with clone_function_name instead
12327 of make_unique_name.
12330 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
12331 correct operand in doing the split.
12333 2017-07-07 Carl Love <cel@us.ibm.com>
12335 * config/rs6000/rs6000-c: Add support for built-in function
12336 vector unsigned short vec_pack_to_short_fp32 (vector float,
12338 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
12339 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
12340 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
12341 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
12342 (convert_4f32_8i16): Add define_expand.
12343 * doc/extend.texi: Update the built-in documentation file for the
12344 new built-in function.
12346 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12348 * config/sparc/m8.md: New file.
12349 * config/sparc/sparc.md: Include m8.md.
12351 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12353 * config/sparc/sparc.opt: New option -mvis4b.
12354 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
12355 (sparc_option_override): Handle VIS4B.
12356 (enum sparc_builtins): Define
12357 SPARC_BUILTIN_DICTUNPACK{8,16,32},
12358 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
12359 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
12360 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
12361 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
12362 (check_constant_argument): New function.
12363 (sparc_vis_init_builtins): Define builtins
12364 __builtin_vis_dictunpack{8,16,32},
12365 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
12366 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
12367 __builtin_vis_fpcmpde{8,16,32}shl and
12368 __builtin_vis_fpcmpur{8,16,32}shl.
12369 (sparc_expand_builtin): Check that the constant operands to
12370 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
12371 constant and in range.
12372 * config/sparc/sparc-c.c (sparc_target_macros): Handle
12374 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
12375 (SPARC_IMM5_P): Likewise.
12376 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
12377 (enabled): Handle vis4b.
12378 (UNSPEC_DICTUNPACK): New unspec.
12379 (UNSPEC_FPCMPSHL): Likewise.
12380 (UNSPEC_FPUCMPSHL): Likewise.
12381 (UNSPEC_FPCMPDESHL): Likewise.
12382 (UNSPEC_FPCMPURSHL): Likewise.
12383 (cpu_feature): New CPU feature `vis4b'.
12384 (dictunpack{8,16,32}): New insns.
12385 (FPCSMODE): New mode iterator.
12386 (fpcscond): New code iterator.
12387 (fpcsucond): Likewise.
12388 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
12389 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
12390 (fpcmpde{8,16,32}{si,di}shl): Likewise.
12391 (fpcmpur{8,16,32}{si,di}shl): Likewise.
12392 * config/sparc/constraints.md: Define constraints `q' for unsigned
12393 2-bit integer constants and `t' for unsigned 5-bit integer
12395 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
12397 (imm5_operand_dictunpack16): Likewise.
12398 (imm5_operand_dictunpack32): Likewise.
12399 (imm2_operand): Likewise.
12400 * doc/invoke.texi (SPARC Options): Document -mvis4b.
12401 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
12402 ditunpack* and fpcmp*shl builtins.
12404 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12406 * config.gcc: Handle m8 in --with-{cpu,tune} options.
12407 * config.in: Add HAVE_AS_SPARC6 define.
12408 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
12410 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
12412 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
12413 (CPP_CPU_SPEC): Handle m8.
12414 (ASM_CPU_SPEC): Likewise.
12415 * config/sparc/sparc-opts.h (enum processor_type): Add
12417 * config/sparc/sparc.c (m8_costs): New struct.
12418 (sparc_option_override): Handle TARGET_CPU_m8.
12419 (sparc32_initialize_trampoline): Likewise.
12420 (sparc64_initialize_trampoline): Likewise.
12421 (sparc_issue_rate): Likewise.
12422 (sparc_register_move_cost): Likewise.
12423 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
12424 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
12425 (ASM_CPU64_DEFAULT_SPEC): Likewise.
12426 (CPP_CPU_SPEC): Handle M8.
12427 (ASM_CPU_SPEC): Likewise.
12428 (AS_M8_FLAG): Define.
12429 * config/sparc/sparc.md: Add m8 to the cpu attribute.
12430 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
12431 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
12433 * configure: Regenerate.
12434 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
12437 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12439 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
12441 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
12442 ("*movdi_insn_sp32"): Do not set v3pipe.
12443 ("*movsi_insn"): Likewise.
12444 ("*movdi_insn_sp64"): Likewise.
12445 ("*movsf_insn"): Likewise.
12446 ("*movdf_insn_sp32"): Likewise.
12447 ("*movdf_insn_sp64"): Likewise.
12448 ("*zero_extendsidi2_insn_sp64"): Likewise.
12449 ("*sign_extendsidi2_insn"): Likewise.
12450 ("*mov<VM32:mode>_insn"): Likewise.
12451 ("*mov<VM64:mode>_insn_sp64"): Likewise.
12452 ("*mov<VM64:mode>_insn_sp32"): Likewise.
12453 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
12454 ("<vlop:code><VL:mode>3"): Likewise.
12455 ("*not_<vlop:code><VL:mode>3"): Likewise.
12456 ("*nand<VL:mode>_vis"): Likewise.
12457 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
12458 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
12459 ("one_cmpl<VL:mode>2"): Likewise.
12460 ("faligndata<VM64:mode>_vis"): Likewise.
12461 ("alignaddrsi_vis"): Likewise.
12462 ("alignaddrdi_vis"): Likweise.
12463 ("alignaddrlsi_vis"): Likewise.
12464 ("alignaddrldi_vis"): Likewise.
12465 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
12466 ("bmaskdi_vis"): Likewise.
12467 ("bmasksi_vis"): Likewise.
12468 ("bshuffle<VM64:mode>_vis"): Likewise.
12469 ("cmask8<P:mode>_vis"): Likewise.
12470 ("cmask16<P:mode>_vis"): Likewise.
12471 ("cmask32<P:mode>_vis"): Likewise.
12472 ("pdistn<P:mode>_vis"): Likewise.
12473 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
12475 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12477 * config/sparc/sparc.md ("subtype"): New insn attribute.
12478 ("*wrgsr_sp64"): Set insn subtype.
12479 ("*rdgsr_sp64"): Likewise.
12480 ("alignaddrsi_vis"): Likewise.
12481 ("alignaddrdi_vis"): Likewise.
12482 ("alignaddrlsi_vis"): Likewise.
12483 ("alignaddrldi_vis"): Likewise.
12484 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
12485 ("fexpand_vis"): Likewise.
12486 ("fpmerge_vis"): Likewise.
12487 ("faligndata<VM64:mode>_vis"): Likewise.
12488 ("bshuffle<VM64:mode>_vis"): Likewise.
12489 ("cmask8<P:mode>_vis"): Likewise.
12490 ("cmask16<P:mode>_vis"): Likewise.
12491 ("cmask32<P:mode>_vis"): Likewise.
12492 ("fchksm16_vis"): Likewise.
12493 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
12494 ("fmean16_vis"): Likewise.
12495 ("fp<plusminus_insn>64_vis"): Likewise.
12496 ("<plusminus_insn>v8qi3"): Likewise.
12497 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
12498 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
12499 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
12500 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
12501 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
12502 ("*movqi_insn"): Likewise.
12503 ("*movhi_insn"): Likewise.
12504 ("*movsi_insn"): Likewise.
12505 ("movsi_pic_gotdata_op"): Likewise.
12506 ("*movdi_insn_sp32"): Likewise.
12507 ("*movdi_insn_sp64"): Likewise.
12508 ("movdi_pic_gotdata_op"): Likewise.
12509 ("*movsf_insn"): Likewise.
12510 ("*movdf_insn_sp32"): Likewise.
12511 ("*movdf_insn_sp64"): Likewise.
12512 ("*zero_extendhisi2_insn"): Likewise.
12513 ("*zero_extendqihi2_insn"): Likewise.
12514 ("*zero_extendqisi2_insn"): Likewise.
12515 ("*zero_extendqidi2_insn"): Likewise.
12516 ("*zero_extendhidi2_insn"): Likewise.
12517 ("*zero_extendsidi2_insn_sp64"): Likewise.
12518 ("ldfsr"): Likewise.
12519 ("prefetch_64"): Likewise.
12520 ("prefetch_32"): Likewise.
12521 ("tie_ld32"): Likewise.
12522 ("tie_ld64"): Likewise.
12523 ("*tldo_ldub_sp32"): Likewise.
12524 ("*tldo_ldub1_sp32"): Likewise.
12525 ("*tldo_ldub2_sp32"): Likewise.
12526 ("*tldo_ldub_sp64"): Likewise.
12527 ("*tldo_ldub1_sp64"): Likewise.
12528 ("*tldo_ldub2_sp64"): Likewise.
12529 ("*tldo_ldub3_sp64"): Likewise.
12530 ("*tldo_lduh_sp32"): Likewise.
12531 ("*tldo_lduh1_sp32"): Likewise.
12532 ("*tldo_lduh_sp64"): Likewise.
12533 ("*tldo_lduh1_sp64"): Likewise.
12534 ("*tldo_lduh2_sp64"): Likewise.
12535 ("*tldo_lduw_sp32"): Likewise.
12536 ("*tldo_lduw_sp64"): Likewise.
12537 ("*tldo_lduw1_sp64"): Likewise.
12538 ("*tldo_ldx_sp64"): Likewise.
12539 ("*mov<VM32:mode>_insn"): Likewise.
12540 ("*mov<VM64:mode>_insn_sp64"): Likewise.
12541 ("*mov<VM64:mode>_insn_sp32"): Likewise.
12543 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12545 * config/sparc/sparc.md ("type"): New insn type viscmp.
12546 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
12548 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
12549 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
12550 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
12551 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
12553 ("n7_vis_logical_11cycle"): Likewise.
12554 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
12555 * config/sparc/niagara2.md ("niag3_vis": Likewise.
12556 * config/sparc/niagara.md ("niag_vis"): Likewise.
12557 * config/sparc/ultra3.md ("us3_fga"): Likewise.
12558 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
12560 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12562 * config/sparc/sparc.md: New instruction type `bmask'.
12563 (bmaskdi_vis): Use the `bmask' type.
12564 (bmasksi_vis): Likewise.
12565 * config/sparc/ultra3.md (us3_array): Likewise.
12566 * config/sparc/niagara7.md (n7_array): Likewise.
12567 * config/sparc/niagara4.md (n4_array): Likewise.
12568 * config/sparc/niagara2.md (niag2_vis): Likewise.
12569 (niag3_vis): Likewise.
12570 * config/sparc/niagara.md (niag_vis): Likewise.
12572 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12574 * ipa-comdats.c: Remove optimize check from gate.
12575 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
12576 for functions not optimized.
12577 (ipa_fn_summary_read): Skip optimize check.
12578 (ipa_fn_summary_write): Likewise.
12579 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
12581 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
12583 (can_inline_edge_p): Check flag_pcc_struct_return for match.
12584 (check_callers): Give up on caller which is not optimized.
12585 (inline_small_functions): Likewise.
12586 (ipa_inline): Do not give up when not optimizing.
12587 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
12588 away unoptimizes cdtors.
12589 (whole_program_function_and_variable_visibility): Do
12590 ipa_discover_readonly_nonaddressable_vars in LTO mode.
12591 * ipa.c (process_references): Do not check optimize.
12592 (symbol_table::remove_unreachable_nodes): Update optimize check.
12593 (set_writeonly_bit): Update optimize check.
12594 (pass_ipa_cdtor_merge::gate): Do not check optimize.
12595 (pass_ipa_single_use::gate): Remove.
12597 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12599 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
12600 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
12601 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
12602 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
12603 permute_load, permute_store, adjust_extract, adjust_splat,
12604 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
12605 replace_swap_with_copy, dump_swap_insn_table,
12606 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
12607 recombine_lvx_pattern, recombine_stvx_pattern,
12608 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
12609 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
12610 to file rs6000-p8swap.c.
12611 * config/rs6000/rs6000-p8swap.c: New file.
12612 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
12613 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
12614 and rs6000*-*-* targets.
12616 2017-07-06 David Malcolm <dmalcolm@redhat.com>
12618 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
12620 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12622 * lto-wrapper.c (merge_and_complain): Do not merge
12623 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
12624 fsigned_zeros, ftrapping_math, fwrapv.
12625 (append_compiler_options): Do not track these options.
12626 (append_linker_options): Likewie
12628 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12630 * cgraphunit.c (cgraph_node::finalize_function): When
12631 !flag_toplevel_reorde set no_reorder flag.
12632 (varpool_node::finalize_decl): Likewise.
12633 (symbol_table::compile): Drop no toplevel reorder path.
12635 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12637 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
12638 edges; zero probability is not better than uninitialized.
12640 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
12642 * asan.h (asan_sanitize_allocas_p): Declare.
12643 * asan.c (asan_sanitize_allocas_p): New function.
12644 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
12645 (handle_builtin_alloca): Likewise.
12646 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
12647 if !asan_sanitize_allocas_p.
12648 * params.def (asan-instrument-allocas): Add new option.
12649 * params.h (ASAN_PROTECT_ALLOCAS): Define.
12650 * opts.c (common_handle_option): Disable allocas sanitization for
12653 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
12655 * asan.c: Include gimple-fold.h.
12656 (get_last_alloca_addr): New function.
12657 (handle_builtin_stackrestore): Likewise.
12658 (handle_builtin_alloca): Likewise.
12659 (asan_emit_allocas_unpoison): Likewise.
12660 (get_mem_refs_of_builtin_call): Add new parameter, remove const
12661 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
12662 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
12663 (instrument_builtin_call): Pass gimple iterator to
12664 get_mem_refs_of_builtin_call.
12665 (last_alloca_addr): New global.
12666 * asan.h (asan_emit_allocas_unpoison): Declare.
12667 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
12668 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
12669 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
12670 if function calls alloca.
12671 * gimple-fold.c (replace_call_with_value): Remove static keyword.
12672 * gimple-fold.h (replace_call_with_value): Declare.
12673 * internal-fn.c: Include asan.h.
12674 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
12675 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
12677 2017-07-06 David Malcolm <dmalcolm@redhat.com>
12679 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
12680 (C_SELFTEST_FLAGS): New.
12681 (CPP_SELFTEST_FLAGS): New.
12682 (SELFTEST_DEPS): New, from deps of s-selftest.
12683 (C_SELFTEST_DEPS): New, from deps of s-selftest.
12684 (CPP_SELFTEST_DEPS): New.
12685 (selftest): Add dependency on s-selftest-c++.
12686 (s-selftest): Rename to...
12687 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
12688 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
12689 than SELFTEST_FLAGS.
12690 (selftest-gdb): Rename to...
12691 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
12693 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
12694 (selftest-valgrind): Rename to...
12695 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
12697 (selftest-valgrind): Reintroduce as an alias for
12698 selftest-c-valgrind.
12699 (s-selftest-c++): New.
12700 (selftest-c++-gdb): New.
12701 (selftest-c++-valgrind): New.
12703 2017-07-06 Olivier Hainque <hainque@adacore.com>
12705 * gcc.c (process_command): When deciding if undefined variables
12706 should be ignored when processing specs, accept "gcc -v" as well.
12708 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12710 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
12711 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
12713 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
12715 * config/arm/arm-cpus.in (armv8-r): Add new entry.
12716 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
12717 * config/arm/arm-tables.opt: Regenerate.
12718 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
12720 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
12722 2017-07-06 Carl Love <cel@us.ibm.com>
12724 * ChangeLog: Clean up from mid air collision
12726 2017-07-06 Carl Love <cel@us.ibm.com>
12728 * config/rs6000/rs6000-c.c: Add support for built-in functions
12729 vector signed int vec_subc (vector signed int, vector signed int);
12730 vector signed __int128 vec_subc (vector signed __int128,
12731 vector signed __int128);
12732 vector unsigned __int128 vec_subc (vector unsigned __int128,
12733 vector unsigned __int128);
12734 vector signed int vec_sube (vector signed int, vector signed int,
12735 vector signed int);
12736 vector unsigned int vec_sube (vector unsigned int,
12737 vector unsigned int,
12738 vector unsigned int);
12739 vector signed __int128 vec_sube (vector signed __int128,
12740 vector signed __int128,
12741 vector signed__int128);
12742 vector unsigned __int128 vec_sube (vector unsigned __int128,
12743 vector unsigned __int128,
12744 vector unsigned __int128);
12745 vector signed int vec_subec (vector signed int, vector signed int,
12746 vector signed int);
12747 vector unsigned int vec_subec (vector unsigned int,
12748 vector unsigned int,
12749 vector unsigned int);
12750 vector signed __int128 vec_subec (vector signed __int128,
12751 vector signed __int128,
12752 vector signed__int128);
12753 vector unsigned __int128 vec_subec (vector unsigned __int128,
12754 vector unsigned __int128,
12755 vector unsigned __int128);
12756 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
12757 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
12758 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
12759 BU_ALTIVEC_OVERLOAD_X definitions.
12760 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
12761 * doc/extend.texi: Update the built-in documentation file for the new
12762 built-in functions.
12764 2017-07-06 David Malcolm <dmalcolm@redhat.com>
12767 * diagnostic-show-locus.c (layout::layout): Use start and finish
12768 spelling location for the start and finish of each range.
12769 * genmatch.c (linemap_client_expand_location_to_spelling_point):
12770 Add unused aspect param.
12771 * input.c (expand_location_1): Add "aspect" param, and use it
12772 to access the correct part of the location.
12773 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
12775 (expand_location_to_spelling_point): Likewise.
12776 (linemap_client_expand_location_to_spelling_point): Add "aspect"
12777 param, and pass it to expand_location_1.
12779 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
12781 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
12782 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
12783 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
12784 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
12785 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
12786 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
12787 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
12788 _mm_maskz_getmant_ss): New intrinsics.
12789 (__builtin_ia32_getexpss128_mask): Changed to ...
12790 __builtin_ia32_getexpss128_round ... this.
12791 (__builtin_ia32_getexpsd128_mask): Changed to ...
12792 __builtin_ia32_getexpsd128_round ... this.
12793 * config/i386/i386-builtin-types.def
12794 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
12795 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
12796 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
12797 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
12798 __builtin_ia32_getmantss_mask_round): New builtins.
12799 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
12800 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
12801 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
12802 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
12803 * config/i386/sse.md
12804 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
12805 avx512f_sgetexp<mode><mask_scalar_name>
12806 <round_saeonly_scalar_name> ... this.
12807 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
12808 %0, %1, %2<round_saeonly_op3>}): Changed to ...
12809 vgetexp<ssescalarmodesuffix>
12810 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12811 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
12812 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
12813 avx512f_vgetmant<mode><mask_scalar_name>
12814 <round_saeonly_scalar_name> ... this.
12815 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
12816 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
12817 vgetmant<ssescalarmodesuffix>
12818 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
12819 %0<mask_scalar_operand4>, %1, %2
12820 <round_saeonly_scalar_mask_op4>, %3} ... this.
12821 * config/i386/subst.md (mask_scalar_operand4,
12822 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
12823 round_saeonly_scalar_nimm_predicate): New subst attributes.
12825 2017-07-06 Julia Koval <julia.koval@intel.com>
12827 * config/i386/i386.c (ix86_erase_embedded_rounding):
12828 Remove code for old rounding pattern.
12830 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
12832 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
12834 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
12836 * doc/sourcebuild.texi (Test Directives, Variants of
12837 dg-require-support): Add documentation for dg-require-stack-check.
12839 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
12841 * config/i386/subst.md (mask_scalar, round_scalar,
12842 round_saeonly_scalar): New meta-templates.
12843 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
12844 round_scalar_mask_operand3, round_scalar_mask_op3,
12845 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
12846 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
12847 round_saeonly_scalar_constraint,
12848 round_saeonly_scalar_prefix): New subst attribute.
12849 * config/i386/sse.md
12850 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
12851 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
12852 <round_scalar_name> ... this.
12853 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
12854 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
12855 <round_scalar_name> ... this.
12856 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
12857 <sse>_vm<code><mode>3<mask_scalar_name>
12858 <round_saeonly_scalar_name> ... this.
12859 (v<plusminus_mnemonic><ssescalarmodesuffix>
12860 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
12861 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
12862 v<plusminus_mnemonic><ssescalarmodesuffix>
12863 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12864 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
12865 (v<multdiv_mnemonic><ssescalarmodesuffix>
12866 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
12867 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
12868 v<multdiv_mnemonic><ssescalarmodesuffix>
12869 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12870 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
12871 (v<maxmin_float><ssescalarmodesuffix>
12872 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
12873 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
12874 v<maxmin_float><ssescalarmodesuffix>
12875 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12876 %0<mask_scalar_operand3>, %1, %<iptr>2
12877 <round_saeonly_scalar_mask_op3>} ... this.
12879 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
12881 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
12882 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
12884 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
12885 Alan Hayward <alan.hayward@arm.com>
12886 David Sherwood <david.sherwood@arm.com>
12888 * combine.c (simplify_if_then_else): Remove "enum" before
12890 * compare-elim.c (can_eliminate_compare): Likewise.
12891 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
12893 (aarch64_lookup_simd_builtin_type): Likewise.
12894 (aarch64_simd_builtin_type): Likewise.
12895 (aarch64_init_simd_builtin_types): Likewise.
12896 (aarch64_simd_expand_args): Likewise.
12897 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
12899 (aarch64_reverse_mask): Likewise.
12900 (aarch64_simd_emit_reg_reg_move): Likewise.
12901 (aarch64_gen_adjusted_ldpstp): Likewise.
12902 (aarch64_ccmp_mode_to_code): Likewise.
12903 (aarch64_operands_ok_for_ldpstp): Likewise.
12904 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12905 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
12907 (aarch64_min_divisions_for_recip_mul): Likewise.
12908 (aarch64_reassociation_width): Likewise.
12909 (aarch64_get_condition_code_1): Likewise.
12910 (aarch64_simd_emit_reg_reg_move): Likewise.
12911 (aarch64_simd_attr_length_rglist): Likewise.
12912 (aarch64_reverse_mask): Likewise.
12913 (aarch64_operands_ok_for_ldpstp): Likewise.
12914 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12915 (aarch64_gen_adjusted_ldpstp): Likewise.
12916 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
12918 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
12919 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
12920 (arm_lookup_simd_builtin_type): Likewise.
12921 (arm_simd_builtin_type): Likewise.
12922 (arm_init_simd_builtin_types): Likewise.
12923 (arm_expand_builtin_args): Likewise.
12924 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
12925 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
12926 (ft32_setup_incoming_varargs): Likewise.
12927 (ft32_function_arg): Likewise.
12928 (ft32_function_arg_advance): Likewise.
12929 (ft32_pass_by_reference): Likewise.
12930 (ft32_arg_partial_bytes): Likewise.
12931 (ft32_valid_pointer_mode): Likewise.
12932 (ft32_addr_space_pointer_mode): Likewise.
12933 (ft32_addr_space_legitimate_address_p): Likewise.
12934 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
12936 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
12937 (ix86_emit_outlined_ms2sysv_restore): Likewise.
12938 (iamcu_alignment): Likewise.
12939 (canonicalize_vector_int_perm): Likewise.
12940 (ix86_noce_conversion_profitable_p): Likewise.
12941 (ix86_mpx_bound_mode): Likewise.
12942 (ix86_operands_ok_for_move_multiple): Likewise.
12943 * config/microblaze/microblaze-protos.h
12944 (microblaze_expand_conditional_branch_reg): Likewise.
12945 * config/microblaze/microblaze.c
12946 (microblaze_expand_conditional_branch_reg): Likewise.
12947 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12949 (rs6000_reassociation_width): Likewise.
12950 (rs6000_invalid_binary_op): Likewise.
12951 (fusion_p9_p): Likewise.
12952 (emit_fusion_p9_load): Likewise.
12953 (emit_fusion_p9_store): Likewise.
12954 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
12956 (riscv_hard_regno_mode_ok_p): Likewise.
12957 (riscv_address_insns): Likewise.
12958 (riscv_split_symbol): Likewise.
12959 (riscv_legitimize_move): Likewise.
12960 (riscv_function_value): Likewise.
12961 (riscv_hard_regno_nregs): Likewise.
12962 (riscv_expand_builtin): Likewise.
12963 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
12964 (riscv_build_integer): Likewise.
12965 (riscv_split_integer): Likewise.
12966 (riscv_legitimate_constant_p): Likewise.
12967 (riscv_cannot_force_const_mem): Likewise.
12968 (riscv_regno_mode_ok_for_base_p): Likewise.
12969 (riscv_valid_base_register_p): Likewise.
12970 (riscv_valid_offset_p): Likewise.
12971 (riscv_valid_lo_sum_p): Likewise.
12972 (riscv_classify_address): Likewise.
12973 (riscv_legitimate_address_p): Likewise.
12974 (riscv_address_insns): Likewise.
12975 (riscv_load_store_insns): Likewise.
12976 (riscv_force_binary): Likewise.
12977 (riscv_split_symbol): Likewise.
12978 (riscv_force_address): Likewise.
12979 (riscv_legitimize_address): Likewise.
12980 (riscv_move_integer): Likewise.
12981 (riscv_legitimize_const_move): Likewise.
12982 (riscv_legitimize_move): Likewise.
12983 (riscv_address_cost): Likewise.
12984 (riscv_subword): Likewise.
12985 (riscv_output_move): Likewise.
12986 (riscv_canonicalize_int_order_test): Likewise.
12987 (riscv_emit_int_order_test): Likewise.
12988 (riscv_function_arg_boundary): Likewise.
12989 (riscv_pass_mode_in_fpr_p): Likewise.
12990 (riscv_pass_fpr_single): Likewise.
12991 (riscv_pass_fpr_pair): Likewise.
12992 (riscv_get_arg_info): Likewise.
12993 (riscv_function_arg): Likewise.
12994 (riscv_function_arg_advance): Likewise.
12995 (riscv_arg_partial_bytes): Likewise.
12996 (riscv_function_value): Likewise.
12997 (riscv_pass_by_reference): Likewise.
12998 (riscv_setup_incoming_varargs): Likewise.
12999 (riscv_print_operand): Likewise.
13000 (riscv_elf_select_rtx_section): Likewise.
13001 (riscv_save_restore_reg): Likewise.
13002 (riscv_for_each_saved_reg): Likewise.
13003 (riscv_register_move_cost): Likewise.
13004 (riscv_hard_regno_mode_ok_p): Likewise.
13005 (riscv_hard_regno_nregs): Likewise.
13006 (riscv_class_max_nregs): Likewise.
13007 (riscv_memory_move_cost): Likewise.
13008 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
13009 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
13010 (rl78_addr_space_address_mode): Likewise.
13011 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13013 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13014 (rs6000_reassociation_width): Likewise.
13015 (rs6000_invalid_binary_op): Likewise.
13016 (fusion_p9_p): Likewise.
13017 (emit_fusion_p9_load): Likewise.
13018 (emit_fusion_p9_store): Likewise.
13019 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
13020 (ok_for_simple_move_operands): Likewise.
13021 (ok_for_simple_move_strict_operands): Likewise.
13022 (ok_for_simple_arith_logic_operands): Likewise.
13023 (visium_legitimize_reload_address): Likewise.
13024 (visium_select_cc_mode): Likewise.
13025 (output_cbranch): Likewise.
13026 (visium_split_double_move): Likewise.
13027 (visium_expand_copysign): Likewise.
13028 (visium_expand_int_cstore): Likewise.
13029 (visium_expand_fp_cstore): Likewise.
13030 * config/visium/visium.c (visium_pass_by_reference): Likewise.
13031 (visium_function_arg): Likewise.
13032 (visium_function_arg_advance): Likewise.
13033 (visium_libcall_value): Likewise.
13034 (visium_setup_incoming_varargs): Likewise.
13035 (visium_legitimate_constant_p): Likewise.
13036 (visium_legitimate_address_p): Likewise.
13037 (visium_legitimize_address): Likewise.
13038 (visium_secondary_reload): Likewise.
13039 (visium_register_move_cost): Likewise.
13040 (visium_memory_move_cost): Likewise.
13041 (prepare_move_operands): Likewise.
13042 (ok_for_simple_move_operands): Likewise.
13043 (ok_for_simple_move_strict_operands): Likewise.
13044 (ok_for_simple_arith_logic_operands): Likewise.
13045 (visium_function_value_1): Likewise.
13046 (rtx_ok_for_offset_p): Likewise.
13047 (visium_legitimize_reload_address): Likewise.
13048 (visium_split_double_move): Likewise.
13049 (visium_expand_copysign): Likewise.
13050 (visium_expand_int_cstore): Likewise.
13051 (visium_expand_fp_cstore): Likewise.
13052 (visium_split_cstore): Likewise.
13053 (visium_select_cc_mode): Likewise.
13054 (visium_split_cbranch): Likewise.
13055 (output_cbranch): Likewise.
13056 (visium_print_operand_address): Likewise.
13057 * expmed.c (flip_storage_order): Likewise.
13058 * expmed.h (emit_cstore): Likewise.
13059 (flip_storage_order): Likewise.
13060 * genrecog.c (validate_pattern): Likewise.
13061 * hsa-gen.c (gen_hsa_addr): Likewise.
13062 * internal-fn.c (expand_arith_overflow): Likewise.
13063 * ira-color.c (allocno_copy_cost_saving): Likewise.
13064 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13065 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
13066 (process_invariant_for_inheritance): Likewise.
13067 * lra-eliminations.c (move_plus_up): Likewise.
13068 * omp-low.c (lower_oacc_reductions): Likewise.
13069 * simplify-rtx.c (simplify_subreg): Likewise.
13070 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
13071 (TARGET_CHKP_BOUND_MODE): Likewise..
13072 * targhooks.c (default_chkp_bound_mode): Likewise.
13073 (default_setup_incoming_vararg_bounds): Likewise.
13074 * targhooks.h (default_chkp_bound_mode): Likewise.
13075 (default_setup_incoming_vararg_bounds): Likewise.
13076 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
13077 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
13078 (have_whole_vector_shift): Likewise.
13079 * tree-vect-stmts.c (vectorizable_load): Likewise.
13080 * doc/tm.texi: Regenerate.
13082 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
13084 Graceful degrade if Binutils PR21472 is not available.
13087 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
13088 .rodata in flash test fails.
13089 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
13090 * confgure: Regenerate.
13091 * config.in: Regenerate.
13092 * config/avr/avr.c (avr_asm_named_section)
13093 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
13094 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
13095 (avr_asm_init_sections): Same.
13097 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13099 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
13100 (fma<VH:mode>4_intrinsic): Likewise.
13101 (*fmsub<VCVTF:mode>4): Likewise.
13102 (*fmsub<VH:mode>4_intrinsic): Likewise.
13104 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
13107 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
13108 Don't depend on "optimize > 0".
13109 (out_movhi_r_mr, out_movqi_mr_r): Same.
13110 (out_movhi_mr_r, out_movqi_r_mr): Same.
13111 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
13112 io_address_operand on "optimize > 0".
13114 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13116 * tree-loop-distribution.c: Add general explanantion on the pass.
13117 (generate_loops_for_partition): Mark distributed loop.
13118 (pg_add_dependence_edges): New parameter. Handle alias data
13119 dependence specially and record it in the parameter if asked.
13120 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
13121 (init_partition_graph_vertices, add_partition_graph_edge): New.
13122 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
13123 (free_partition_graph_vdata, build_partition_graph): New.
13124 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
13125 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
13126 (data_ref_segment_size, latch_dominated_by_data_ref): New.
13127 (compute_alias_check_pairs, version_loop_by_alias_check): New.
13128 (version_for_distribution_p, finalize_partitions): New.
13129 (distribute_loop): Handle alias data dependence specially. Factor
13130 out loop fusion code as functions and call these functions.
13132 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13134 * tree-loop-distribution.c (classify_partition): New parameter and
13135 better handle reduction statement.
13136 (rdg_build_partitions): Revise comment.
13137 (distribute_loop): Compute statements in all partitions and pass it
13138 to classify_partition.
13140 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13142 * tree-loop-distribution.c (enum partition_type): New.
13143 (struct partition): New field type.
13144 (partition_merge_into): Add parameter. Update partition type.
13145 (data_dep_in_cycle_p, update_type_for_merge): New functions.
13146 (build_rdg_partition_for_vertex): Compute partition type.
13147 (rdg_build_partitions): Dump partition type.
13148 (distribute_loop): Update calls to partition_merge_into.
13150 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13152 * tree-loop-distribution.c (struct ddr_hasher): New.
13153 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
13155 (classify_partition): Call get_data_dependence.
13156 (pg_add_dependence_edges): Ditto.
13157 (distribute_loop): Release data dependence hash table.
13159 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13161 * tree-loop-distribution.c (ref_base_address): Delete.
13162 (similar_memory_accesses): Rename ...
13163 (share_memory_accesses): ... to this. Check if partitions access
13164 the same memory reference.
13165 (distribute_loop): Call share_memory_accesses.
13167 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13169 * tree-loop-distribution.c (struct partition): New field recording
13170 its data reference.
13171 (partition_alloc, partition_free): Init and release data refs.
13172 (partition_merge_into): Merge data refs.
13173 (build_rdg_partition_for_vertex): Collect data refs for partition.
13174 (pg_add_dependence_edges): Change parameters from vector to bitmap.
13176 (distribute_loop): Remve data refs from vertice data of partition
13179 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13181 * tree-loop-distribution.c (params.h): Include header file.
13182 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
13183 (datarefs_vec): New global var.
13184 (create_rdg_vertices): Use datarefs_vec directly.
13185 (free_rdg): Don't free data references.
13186 (build_rdg): Update use. Don't free data references.
13187 (distribute_loop): Compute global variable for data references.
13188 Bail out if there are too many data references.
13190 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13192 * tree-loop-distribution.c (loop_nest): New global var.
13193 (build_rdg): Use loop directly, rather than loop nest.
13194 (pg_add_dependence_edges): Remove loop nest parameter. Use global
13196 (distribute_loop): Compute global variable loop nest. Update use.
13198 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13200 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
13201 (partition_merge_into): New parameter. Dump reason for fusion.
13202 (distribute_loop): Update use of partition_merge_into.
13204 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13206 * tree-loop-distribution.c (bb_top_order_index): New.
13207 (bb_top_order_index_size, bb_top_order_cmp): New.
13208 (stmts_from_loop): Use topological order.
13209 (pass_loop_distribution::execute): Compute and release topological
13210 order for basic blocks.
13212 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13214 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
13217 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13219 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
13220 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
13221 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
13222 * internal-fn.def (LOOP_DIST_ALIAS): New.
13223 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
13224 (fold_loop_internal_call): ... this.
13225 (vect_loop_dist_alias_call): New function.
13226 (set_uid_loop_bbs): Call fold_loop_internal_call.
13227 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
13230 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
13233 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
13234 Require dead FLAGS_REG at the beginning of a peephole.
13236 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
13239 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
13240 arguments in the call to __builtin_ia32_sbb_u32.
13241 (_subborrow_u64): Swap _X and _Y arguments in the call to
13242 __builtin_ia32_sbb_u64.
13244 2017-07-04 Jakub Jelinek <jakub@redhat.com>
13247 * tree-vrp.c (compare_assert_loc): Turn into a function template
13248 with stable template parameter. Only test if a->e is NULL,
13249 !a->e == !b->e has been verified already. Use e == NULL or
13250 e != NULL instead of e or ! e tests. If stable is true, don't use
13251 iterative_hash_expr, on the other side allow a or b or both NULL
13252 and sort the NULLs last.
13253 (process_assert_insertions): Sort using compare_assert_loc<false>
13254 instead of compare_assert_loc, later sort using
13255 compare_assert_loc<true> before calling process_assert_insertions_for
13256 in a loop. Use break instead of continue once seen NULL pointer.
13258 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
13260 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
13261 Cortex-R7 and Cortex-R8 processors.
13263 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
13265 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
13266 uninitialized while src is not.
13268 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
13270 * common/config/arm/arm-common.c: Adjust include path for
13272 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
13273 (arm-cpu.h): Create in build directory. Adjust dependency rules.
13274 (arm-cpu-data.h): Likewise.
13275 (arm-cpu-cdata.h): Likewise.
13276 * config/arm/arm-cpu.h: Delete.
13277 * config/arm/arm-cpu-cdata.h: Delete.
13278 * config/arm/arm-cpu-data.h: Delete.
13280 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
13282 * config/arm/arm-cpus.in (cortex-a55): New.
13283 (cortex-a75): Likewise.
13284 (cortex-a75.cortex-a55): Likewise.
13285 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
13287 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
13288 * config/arm/arm-cpu-cdata.h: Regenerate.
13289 * config/arm/arm-cpu-data.h: Regenerate.
13290 * config/arm/arm-cpu.h: Regenerate.
13291 * config/arm/arm-tables.opt: Regenerate.
13292 * config/arm/arm-tune.md: Regenerate.
13294 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
13296 * haifa-sched.c (sched_create_recovery_edges): Update profile.
13298 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
13300 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
13303 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
13305 PR tree-optimization/81292
13306 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
13307 full_string_p, also call adjust_related_strinfos if the adjustment
13308 is simple, otherwise invalidate related strinfos.
13310 2017-07-04 Martin Liska <mliska@suse.cz>
13313 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
13314 newly created variable as DECL_IGNORED_P.
13316 2017-07-04 Martin Liska <mliska@suse.cz>
13319 * ipa-inline.c (inline_small_functions):
13320 Use xstrdup_for_dump.
13322 2017-07-04 Tom de Vries <tom@codesourcery.com>
13324 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
13326 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
13329 * config/darwin.c (darwin_function_switched_text_sections):
13332 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
13334 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
13336 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
13338 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
13340 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13342 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
13343 min_profitable_iters, and th as inclusive lower bounds.
13344 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
13345 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
13346 for min_profitable_iters and min_profitable_estimate.
13347 (vect_transform_loop): Treat th as an inclusive lower bound.
13348 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
13350 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
13353 * config/darwin.c (darwin_function_switched_text_sections):
13354 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
13355 in two pieces, and suppress the use of buf.
13357 2017-07-03 Nathan Sidwell <nathan@acm.org>
13359 * hash-table.h (hash_table_mod1): Fix indentation.
13361 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13363 PR middle-end/81290
13364 * predict.c (force_edge_cold): Be more careful about propagation
13366 * profile-count.h (profile_probability::guessed,
13367 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
13369 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
13371 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
13373 * doc/invoke.texi (rcpc architecture extension): Document it.
13375 2017-07-03 Richard Biener <rguenther@suse.de>
13377 PR tree-optimization/60510
13378 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
13379 the scalar reduction PHI and use it.
13380 (vectorizable_reduction): Properly guard the single_defuse_cycle
13381 path for non-SLP reduction chains where we cannot use it.
13382 Rework reduc_def/index and vector type deduction. Rework
13383 vector operand gathering during reduction op code-gen.
13384 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
13385 chains dissolve the chain and leave it to non-SLP reduction
13388 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13390 * tree-data-ref.h (dr_alignment): Declare.
13391 * tree-data-ref.c (dr_alignment): New function.
13392 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
13393 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
13395 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
13397 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13399 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
13400 and base_misalignment fields.
13401 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
13402 * tree-data-ref.c: Include builtins.h.
13403 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
13404 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
13405 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
13406 * tree-vect-data-refs.c: Include tree-cfg.h.
13407 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
13408 fields instead of calculating an alignment here.
13409 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
13410 innermost_loop_behavior fields.
13412 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13414 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
13416 (DR_STEP_ALIGNMENT): New macro.
13417 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
13418 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
13419 (create_data_ref): Print it.
13420 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
13421 to tell whether the step preserves vector (mis)alignment.
13422 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
13423 Move the check for an integer step and generalise to all INTEGER_CST.
13424 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
13425 Print the outer step alignment.
13427 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13429 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
13430 with offset_alignment.
13431 (DR_ALIGNED_TO): Delete.
13432 (DR_OFFSET_ALIGNMENT): New macro.
13433 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
13434 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
13435 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
13436 (create_data_ref): Likewise.
13437 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
13438 (vect_analyze_data_refs): Likewise.
13439 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
13440 creating dummy innermost behavior.
13442 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13444 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
13445 with a "innermost_loop_behavior *" and refeence tree.
13446 * tree-data-ref.c (dr_analyze_innermost): Likewise.
13447 (create_data_ref): Update call accordingly.
13448 * tree-predcom.c (find_looparound_phi): Likewise.
13450 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13452 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
13453 fields with dr_wrt_vec_loop.
13454 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
13455 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
13456 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
13457 (vect_dr_behavior): New function.
13458 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
13459 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
13460 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
13461 track whether the step preserves the misalignment.
13462 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
13463 Use vect_dr_behavior.
13464 (vect_setup_realignment): Update call accordingly.
13465 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
13466 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
13467 call to vect_create_addr_base_for_vector_ref.
13468 (vect_create_cond_for_align_checks): Likewise.
13469 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
13470 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
13471 (vect_recog_mask_conversion_pattern): Likewise.
13472 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
13473 (new_stmt_vec_info): Remove redundant zeroing.
13475 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
13477 * common/config/arm/arm-common.c (arm_be8_option): New function.
13478 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
13479 (ISA_ARMv6): Add isa_bit_be8.
13480 * config/arm/arm.h (arm_be8_option): Add prototype.
13481 (BE8_SPEC_FUNCTION): New define.
13482 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
13483 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
13484 (mlittle-endian): Similarly.
13485 (mbe8, mbe32): New options.
13486 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
13487 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
13489 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13491 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
13493 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13495 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
13496 (cleanup_tree_cfg_bb): Use it.
13497 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
13499 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
13501 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13504 * loop-doloop.c (add_test): Update profile.
13506 2017-07-03 Martin Liska <mliska@suse.cz>
13509 * sanopt.c (rewrite_usage_of_param): New function.
13510 (sanitize_rewrite_addressable_params): Likewise.
13511 (pass_sanopt::execute): Call rewrite_usage_of_param.
13513 2017-07-03 Richard Biener <rguenther@suse.de>
13515 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
13516 back to using VIEW_CONVERT_EXPR.
13518 2017-07-03 Martin Liska <mliska@suse.cz>
13521 * doc/extend.texi: Document when a resolver function is
13522 generated for target_clones.
13524 2017-07-03 Martin Liska <mliska@suse.cz>
13526 * asan.c (asan_emit_stack_protection): Unpoison just red zones
13527 and shadow memory of auto variables which are subject of
13528 use-after-scope sanitization.
13529 (asan_expand_mark_ifn): Add do set only when is_poison.
13531 2016-07-03 Richard Biener <rguenther@suse.de>
13533 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
13535 (vect_force_simple_reduction): Record reduction def -> phi mapping.
13536 (vectorizable_reduction): Perform reduction PHI creation when
13537 visiting a reduction PHI and adjust and simplify code generation
13538 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
13539 (vect_transform_loop): Visit reduction PHIs.
13540 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
13541 defs into the SLP tree.
13542 (vect_build_slp_tree): Reduction defs terminate the recursion.
13543 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
13545 (vect_get_vec_defs_for_stmt_copy): Export.
13546 (vect_get_vec_defs): Likewise.
13547 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
13549 (vect_get_vec_defs_for_stmt_copy): Declare.
13550 (vect_get_vec_defs): Likewise.
13552 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13554 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
13555 parameter with a "loop" parameter and use it instead of the
13556 loop containing DR_STMT. Don't check simple_iv when doing
13557 BB analysis. Describe the two analysis modes in the comment.
13559 2017-07-03 Tom de Vries <tom@codesourcery.com>
13561 PR tree-optimization/69468
13562 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
13563 (find_same_succ_bb): Handle ignore_edge_flags.
13565 2017-07-03 Tom de Vries <tom@codesourcery.com>
13567 PR tree-optimization/81192
13568 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
13570 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
13572 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
13574 2017-07-03 Tom de Vries <tom@codesourcery.com>
13576 PR tree-optimization/81192
13577 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
13578 BB_SAME_SUCC (bb) == NULL.
13580 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13582 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
13585 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13587 * dumpfile.c: Include profile-count.h
13588 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
13590 (insert_cond_bb): Update profile.
13591 * tree-cfg.h (insert_cond_bb): Update prototype.
13592 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
13593 * tree-dump.c: Do not include tree-cfg.
13595 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13597 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
13599 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13601 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
13604 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13606 * tree-complex.c (expand_complex_div_wide): update profile.
13608 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13609 Alan Hayward <alan.hayward@arm.com>
13610 David Sherwood <david.sherwood@arm.com>
13612 * Makefile.in (MACHMODE_H): Remove insn-modes.h
13613 (CORETYPES_H): New define.
13614 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
13615 (insn-modes-inline.h, s-modes-inline-h): New rules.
13616 (generated_files): Add insn-modes-inline.h.
13617 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
13618 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
13619 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
13620 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
13621 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
13622 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
13623 (build/gencodes.o, build/genconditions.o): Likewise.
13624 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
13625 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
13626 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
13627 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
13628 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
13629 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
13630 * coretypes.h: Include everything up to real.h for generators.
13631 Include insn-modes.h first. Include wide-int-print.h after
13632 wide-int.h. Include insn-modes-inline.h and then machmode.h.
13633 * machmode.h: Don't include insn-modes.h here.
13634 * function-tests.c: Remove includes of signop.h, machmode.h,
13635 double-int.h and wide-int.h.
13637 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
13639 * optc-save-gen.awk: Likewise.
13640 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
13641 * godump.c: Remove include of wide-int-print.h.
13642 * pretty-print.h: Likewise.
13643 * wide-int-print.cc: Likewise.
13644 * wide-int.cc: Likewise.
13645 * hash-map-tests.c: Remove include of signop.h.
13646 * hash-set-tests.c: Likewise.
13647 * rtl-tests.c: Likewise.
13648 * mkconfig.sh: Remove include of machmode.h.
13649 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
13651 (emit_insn_modes_inline_h): ...this new function. Emit the code
13652 into an insn-modes-inline.h header file, adding appropriate
13653 include guards and end comments.
13654 (emit_insn_modes_c_header): Remove include of machmode.h.
13655 (emit_min_insn_modes_c_header): Include coretypes.h rather than
13657 (main): Handle -i flag and call emit_insn_modes_inline_h when
13660 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13662 * tree-ssa-strlen.c (strinfo): Rename the length field to
13663 nonzero_chars. Add a full_string_p field.
13664 (compare_nonzero_chars, zero_length_string_p): New functions.
13665 (get_addr_stridx): Add an offset_out parameter.
13666 Use compare_nonzero_chars.
13667 (get_stridx): Update accordingly. Use compare_nonzero_chars.
13668 (new_strinfo): Update after above changes to strinfo.
13669 (set_endptr_and_length): Set full_string_p.
13670 (get_string_length): Update after above changes to strinfo.
13671 (unshare_strinfo): Update call to new_strinfo.
13672 (maybe_invalidate): Likewise.
13673 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
13674 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
13675 as a uhwi instead of an shwi. Update after above changes to
13676 strinfo and new_strinfo.
13677 (zero_length_string): Assert that chainsi contains full strings.
13678 Use zero_length_string_p. Update call to new_strinfo.
13679 (adjust_related_strinfos): Update after above changes to strinfo.
13680 Copy full_string_p from origsi.
13681 (adjust_last_stmt): Use zero_length_string_p.
13682 (handle_builtin_strlen): Update after above changes to strinfo and
13683 new_strinfo. Install the lhs as the string length if the previous
13684 entry didn't describe a full string.
13685 (handle_builtin_strchr): Update after above changes to strinfo
13687 (handle_builtin_strcpy): Likewise.
13688 (handle_builtin_strcat): Likewise.
13689 (handle_builtin_malloc): Likewise.
13690 (handle_pointer_plus): Likewise.
13691 (handle_builtin_memcpy): Likewise. Track nonzero characters
13692 that aren't necessarily followed by a nul terminator.
13693 (handle_char_store): Likewise.
13695 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13697 PR tree-optimization/80769
13698 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
13699 for malloc and calloc. Document the new invariant that all related
13700 strinfos have delayed lengths or none do.
13701 (verify_related_strinfos): Move earlier in file.
13702 (set_endptr_and_length): New function, split out from...
13703 (get_string_length): ...here. Also set the lengths of related
13705 (zero_length_string): Assert that chainsi has known (rather than
13707 (adjust_related_strinfos): Likewise.
13709 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13711 PR tree-optimization/81136
13712 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
13713 assert that two references with the same misalignment have the same
13714 compile-time misalignment if those compile-time misalignments
13717 2017-07-01 Andi Kleen <ak@linux.intel.com>
13719 * print-tree.c (print_node): Print all attributes.
13721 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13723 * cfg.c (scale_bbs_frequencies): New function.
13724 * cfg.h (scale_bbs_frequencies): Declare it.
13725 * cfgloopanal.c (single_likely_exit): Cleanup.
13726 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
13728 (scale_loop_profile): Likewise.
13729 (loop_version): Likewise.
13730 (create_empty_loop_on_edge): Update.
13731 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
13732 scale_loop_frequencies, scale_loop_profile, loopify,
13733 loop_version): Update prototypes.
13734 * modulo-sched.c (sms_schedule): Update.
13735 * predict.c (unlikely_executed_edge_p): Also check probability.
13736 (probably_never_executed_edge_p): Fix typo.
13737 * tree-if-conv.c (version_loop_for_if_conversion): Update.
13738 * tree-parloops.c (gen_parallel_loop): Update.
13739 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
13740 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13741 * tree-ssa-loop-split.c (split_loop): Update.
13742 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
13743 * tree-vect-loop-manip.c (vect_do_peeling): Update.
13744 (vect_loop_versioning): Update.
13745 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13747 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13749 * trans-mem.c (split_bb_make_tm_edge): Update profile.
13751 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13753 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
13754 to keep profile consistent.
13756 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13758 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
13759 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
13760 * profile-count.h (max_safe_multiplier): Make unsigned.
13761 (profile_count::guessed_zero): New.
13763 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13765 * bb-reorder.c (fix_up_crossing_landing_pad,
13766 fix_crossing_conditional_branches): Use make_single_succ_edge
13767 to keep profile consistent.
13769 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13771 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
13774 2017-07-01 Jakub Jelinek <jakub@redhat.com>
13777 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
13778 the right scopes, make sure cond_jump isn't preserved between multiple
13779 iterations. Search for fallthru edge whenever there are 3+ edges and
13780 use find_fallthru_edge for it.
13782 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13784 Patch by Alexander Monakov <amonakov@ispras.ru>
13785 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
13786 probabilities consistently.
13788 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13790 * pa.c (pa_expand_compare_and_swap_loop): Update call of
13791 emit_cmp_and_jump_insns.
13793 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13796 * tree-inline.c (expand_call_inline): Combine profile statuses.
13798 2017-06-30 Andrew Pinski <apinski@cavium.com>
13800 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
13801 fold_stmt returned true.
13803 2017-06-30 Nathan Sidwell <nathan@acm.org>
13805 * ggc.h (empty_string): Delete.
13806 * cfgexpand.c (expand_asm_stmt): Use plain "".
13807 * optabs.c (expand_asm_memory_barrier): Likewise.
13808 * stringpool.c (empty_string): Delete.
13809 (digit_vector, digit_string): Delete.
13810 (ggc_alloc_string): Use plain "", don't optimize single digit
13811 strings. Use ggc_alloc_atomic.
13813 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
13815 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
13816 comparison set and one other set, use the cost of the non-comparison
13819 2017-06-30 Nathan Sidwell <nathan@acm.org>
13821 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
13824 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
13826 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
13827 loops. Remove now unneeded calls to gimple_switch_set_label() that
13828 just set removed labels to NULL_TREE.
13830 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
13832 * tree-ssanames.c (set_range_info_raw): Abstract from ...
13833 (set_range_info): ...here. Only call set_range_info_raw if domain
13835 (set_nonzero_bits): Call set_range_info_raw.
13836 * tree-ssanames.h (set_range_info_raw): New.
13838 2017-06-30 Jakub Jelinek <jakub@redhat.com>
13841 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
13842 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
13843 of nonimmediate_operand and <store_mask_constraint> instead of m
13844 for the input operand. For V8FI iterator, always split if input
13845 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
13846 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
13847 <store_mask_predicate> instead of register_operand and
13848 <store_mask_constraint> instead of v for the input operand. Make
13849 sure both operands aren't MEMs for if not <mask_applied>.
13851 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
13853 * lto-wrapper.c (copy_file) Close both file descriptors before
13856 2017-06-30 Martin Liska <mliska@suse.cz>
13859 * multiple_target.c (create_dispatcher_calls): Make ifunc
13860 also for function that don't have calls or are not referenced.
13862 2017-06-30 Richard Biener <rguenther@suse.de>
13864 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
13865 analyze the first scalar stmt. Move vector type computation
13866 for the BB case here from ...
13867 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
13868 live operation processing in the SLP case properly.
13870 2017-06-30 Richard Biener <rguenther@suse.de>
13872 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
13874 2017-06-30 Martin Liska <mliska@suse.cz>
13877 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
13878 before BUILT_IN_UNWIND_RESUME when ASAN is used.
13880 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
13882 * doc/invoke.texi (AArch64): Add missing options and remove redundant
13885 2017-06-30 Richard Biener <rguenther@suse.de>
13887 PR tree-optimization/81249
13888 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
13889 condition reduction result to original scalar type.
13891 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13893 * profile-count.h (enum profile_quality): Fix typos and whitespace
13896 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13898 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
13899 type for branch probabilities.
13901 2017-06-29 Julian Brown <julian@codesourcery.com>
13902 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13904 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
13905 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
13906 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
13907 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
13909 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13911 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
13912 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
13913 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
13914 CC usage from generic code to here.
13915 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
13916 CC usage into the target macros.
13918 2017-06-29 Maya Rashish <coypu@sdf.org>
13920 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
13923 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13925 * arm/arm-builtins.c: Include profile-count.h
13926 * except.c (sjlj_emit_function_enter): Use
13927 profile_probability::unlikely.
13929 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13931 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
13932 and tocrel_offset be pointer args rather than implicitly using
13934 (legitimate_constant_pool_address_p, rs6000_emit_move,
13935 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
13936 tocrel_offset and use in toc_relative_expr_p call.
13937 (print_operand, print_operand_address): Use static tocrel_base_oac
13938 and tocrel_offset_oac.
13939 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
13942 2017-06-29 Maya Rashish <coypu@sdf.org>
13944 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
13946 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
13948 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
13949 objects, take into account only the alignment of 'op0' and 'mode1' if
13952 2017-06-29 Steve Ellcey <sellcey@cavium.com>
13954 * ccmp.c (ccmp_tree_comparison_p): New function.
13955 (ccmp_candidate_p): Update to use above function.
13956 (get_compare_parts): New function.
13957 (expand_ccmp_next): Update to use new functions.
13958 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
13960 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
13961 take mode as argument.
13962 * ccmp.h (expand_ccmp_expr): Add mode as argument.
13963 * expr.c (expand_expr_real_1): Pass mode as argument.
13965 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
13967 * combine.c (combine_instructions): Print insns to dump_file, together
13970 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13972 * asan.c (asan_emit_stack_protection): Update.
13973 (create_cond_insert_point): Update.
13974 * auto-profile.c (afdo_propagate_circuit): Update.
13975 * basic-block.h (struct edge_def): Turn probability to
13976 profile_probability.
13977 (EDGE_FREQUENCY): Update.
13978 * bb-reorder.c (find_traces_1_round): Update.
13979 (better_edge_p): Update.
13980 (sanitize_hot_paths): Update.
13981 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
13982 (make_single_succ_edge): Update.
13983 (check_bb_profile): Update.
13984 (dump_edge_info): Update.
13985 (update_bb_profile_for_threading): Update.
13986 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
13987 probabilitycount to 0.
13988 * cfgbuild.c (compute_outgoing_frequencies): Update.
13989 * cfgcleanup.c (try_forward_edges): Update.
13990 (outgoing_edges_match): Update.
13991 (try_crossjump_to_edge): Update.
13992 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
13993 (expand_gimple_tailcall): Update.
13994 (construct_init_block): Use make_single_succ_edge.
13995 (construct_exit_block): Use make_single_succ_edge.
13996 * cfghooks.c (verify_flow_info): Update.
13997 (redirect_edge_succ_nodup): Update.
13998 (split_edge): Update.
13999 (account_profile_record): Update.
14000 * cfgloopanal.c (single_likely_exit): Update.
14001 * cfgloopmanip.c (scale_loop_profile): Update.
14002 (set_zero_probability): Remove.
14003 (duplicate_loop_to_header_edge): Update.
14004 * cfgloopmanip.h (loop_version): Update prototype.
14005 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
14006 (force_nonfallthru_and_redirect): Update.
14007 (update_br_prob_note): Update.
14008 (rtl_verify_edges): Update.
14009 (purge_dead_edges): Update.
14010 (rtl_lv_add_condition_to_bb): Update.
14011 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
14012 * cgraphunit.c (init_lowered_empty_function): Update.
14013 (cgraph_node::expand_thunk): Update.
14014 * cilk-common.c: Include profile-count.h
14015 * dojump.c (inv): Remove.
14016 (jumpifnot): Update.
14017 (jumpifnot_1): Update.
14018 (do_jump_1): Update.
14020 (do_jump_by_parts_greater_rtx): Update.
14021 (do_compare_rtx_and_jump): Update.
14022 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
14023 do_jump_1. do_compare_rtx_and_jump): Update prototype.
14024 * dwarf2cfi.c: Include profile-count.h
14025 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
14026 (sjlj_emit_dispatch_table): Likewise.
14027 * explow.c: Include profile-count.h
14028 * expmed.c (emit_store_flag_force): Update.
14029 (do_cmp_and_jump): Update.
14030 * expr.c (compare_by_pieces_d::generate): Update.
14031 (compare_by_pieces_d::finish_mode): Update.
14032 (emit_block_move_via_loop): Update.
14033 (store_expr_with_bounds): Update.
14034 (store_constructor): Update.
14035 (expand_expr_real_2): Update.
14036 (expand_expr_real_1): Update.
14037 * expr.h (try_casesi, try_tablejump): Update prototypes.
14038 * gimple-pretty-print.c (dump_probability): Update.
14039 (dump_profile): New.
14040 (dump_gimple_label): Update.
14041 (dump_gimple_bb_header): Update.
14042 * graph.c (draw_cfg_node_succ_edges): Update.
14043 * hsa-gen.c (convert_switch_statements): Update.
14044 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
14045 (find_if_case_1): Update.
14046 (find_if_case_2): Update.
14047 * internal-fn.c (expand_arith_overflow_result_store): Update.
14048 (expand_addsub_overflow): Update.
14049 (expand_neg_overflow): Update.
14050 (expand_mul_overflow): Update.
14051 (expand_vector_ubsan_overflow): Update.
14052 * ipa-cp.c (good_cloning_opportunity_p): Update.
14053 * ipa-split.c (split_function): Use make_single_succ_edge.
14054 * ipa-utils.c (ipa_merge_profiles): Update.
14055 * loop-doloop.c (add_test): Update.
14056 (doloop_modify): Update.
14057 * loop-unroll.c (compare_and_jump_seq): Update.
14058 (unroll_loop_runtime_iterations): Update.
14059 * lra-constraints.c (lra_inheritance): Update.
14060 * lto-streamer-in.c (input_cfg): Update.
14061 * lto-streamer-out.c (output_cfg): Update.
14062 * mcf.c (adjust_cfg_counts): Update.
14063 * modulo-sched.c (sms_schedule): Update.
14064 * omp-expand.c (expand_omp_for_init_counts): Update.
14065 (extract_omp_for_update_vars): Update.
14066 (expand_omp_ordered_sink): Update.
14067 (expand_omp_for_ordered_loops): Update.
14068 (expand_omp_for_generic): Update.
14069 (expand_omp_for_static_nochunk): Update.
14070 (expand_omp_for_static_chunk): Update.
14071 (expand_cilk_for): Update.
14072 (expand_omp_simd): Update.
14073 (expand_omp_taskloop_for_outer): Update.
14074 (expand_omp_taskloop_for_inner): Update.
14075 * omp-simd-clone.c (simd_clone_adjust): Update.
14076 * optabs.c (expand_doubleword_shift): Update.
14077 (expand_abs): Update.
14078 (emit_cmp_and_jump_insn_1): Update.
14079 (expand_compare_and_swap_loop): Update.
14080 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
14081 * predict.c (predictable_edge_p): Update.
14082 (edge_probability_reliable_p): Update.
14083 (set_even_probabilities): Update.
14084 (combine_predictions_for_insn): Update.
14085 (combine_predictions_for_bb): Update.
14086 (propagate_freq): Update.
14087 (estimate_bb_frequencies): Update.
14088 (force_edge_cold): Update.
14089 * profile-count.c (profile_count::dump): Add missing space into dump.
14090 (profile_count::debug): Add newline.
14091 (profile_count::differs_from_p): Explicitly convert to unsigned.
14092 (profile_count::stream_in): Update.
14093 (profile_probability::dump): New member function.
14094 (profile_probability::debug): New member function.
14095 (profile_probability::differs_from_p): New member function.
14096 (profile_probability::differs_lot_from_p): New member function.
14097 (profile_probability::stream_in): New member function.
14098 (profile_probability::stream_out): New member function.
14099 * profile-count.h (profile_count_quality): Rename to ...
14100 (profile_quality): ... this one.
14101 (profile_probability): New.
14102 (profile_count): Update.
14103 * profile.c (compute_branch_probabilities): Update.
14104 * recog.c (peep2_attempt): Update.
14105 * sched-ebb.c (schedule_ebbs): Update.
14106 * sched-rgn.c (find_single_block_region): Update.
14107 (compute_dom_prob_ps): Update.
14108 (schedule_region): Update.
14109 * sel-sched-ir.c (compute_succs_info): Update.
14110 * stmt.c (struct case_node): Update.
14111 (do_jump_if_equal): Update.
14112 (get_outgoing_edge_probs): Update.
14113 (conditional_probability): Update.
14114 (emit_case_dispatch_table): Update.
14115 (expand_case): Update.
14116 (expand_sjlj_dispatch_table): Update.
14117 (emit_case_nodes): Update.
14118 * targhooks.c: Update.
14119 * tracer.c (better_p): Update.
14120 (find_best_successor): Update.
14121 * trans-mem.c (expand_transaction): Update.
14122 * tree-call-cdce.c: Update.
14123 * tree-cfg.c (gimple_split_edge): Upate.
14124 (move_sese_region_to_fn): Upate.
14125 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
14126 * tree-eh.c (lower_resx): Upate.
14127 (cleanup_empty_eh_move_lp): Upate.
14128 * tree-if-conv.c (version_loop_for_if_conversion): Update.
14129 * tree-inline.c (copy_edges_for_bb): Update.
14130 (copy_cfg_body): Update.
14131 * tree-parloops.c (gen_parallel_loop): Update.
14132 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
14133 (gimple_gen_time_profiler): Update.
14134 * tree-ssa-dce.c (remove_dead_stmt): Update.
14135 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
14136 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
14137 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
14138 (unloop_loops): Update.
14139 (try_peel_loop): Update.
14140 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14141 * tree-ssa-loop-split.c (connect_loops): Update.
14142 (split_loop): Update.
14143 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
14144 (hoist_guard): Update.
14145 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
14146 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
14147 (value_replacement): Update.
14148 * tree-ssa-reassoc.c (branch_fixup): Update.
14149 * tree-ssa-tail-merge.c (replace_block_by): Update.
14150 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
14151 (create_edge_and_update_destination_phis): Update.
14152 (compute_path_counts): Update.
14153 (recompute_probabilities): Update.
14154 (update_joiner_offpath_counts): Update.
14155 (freqs_to_counts_path): Update.
14156 (duplicate_thread_path): Update.
14157 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
14158 (struct switch_conv_info): Update.
14159 (gen_inbound_check): Update.
14160 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
14161 (vect_do_peeling): Update.
14162 (vect_loop_versioning): Update.
14163 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14164 (optimize_mask_stores): Update.
14165 * ubsan.c (ubsan_expand_null_ifn): Update.
14166 * value-prof.c (gimple_divmod_fixed_value): Update.
14167 (gimple_divmod_fixed_value_transform): Update.
14168 (gimple_mod_pow2): Update.
14169 (gimple_mod_pow2_value_transform): Update.
14170 (gimple_mod_subtract): Update.
14171 (gimple_mod_subtract_transform): Update.
14172 (gimple_ic): Update.
14173 (gimple_stringop_fixed_value): Update.
14174 (gimple_stringops_transform): Update.
14175 * value-prof.h: Update.
14177 2017-06-29 Carl Love <cel@us.ibm.com>
14179 * config/rs6000/rs6000-c.c: Add support for built-in functions
14180 vector signed int vec_signed (vector float);
14181 vector signed long long vec_signed (vector double);
14182 vector signed int vec_signed2 (vector double, vector double);
14183 vector signed int vec_signede (vector double);
14184 vector signed int vec_signedo (vector double);
14185 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
14186 instruction generator.
14187 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
14188 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
14189 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
14191 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
14192 vunsignede_v2df): Add define_expands.
14193 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
14194 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
14195 VEC_UNSIGNEDO): Add definitions.
14196 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
14197 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
14198 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
14199 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
14200 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
14201 * config/rs6000/altivec.h (vec_signed, vec_signed2,
14202 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
14203 vec_unsignede, vec_unsignedo): Add builtin defines.
14204 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
14206 * doc/extend.texi: Update the built-in documentation file for the
14207 new built-in functions.
14209 2017-06-29 Richard Biener <rguenther@suse.de>
14211 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
14212 reduction chains to LOOP_VINFO_REDUCTIONS.
14213 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
14214 SLP reductions after processing reduction chains.
14216 2017-06-29 Nathan Sidwell <nathan@acm.org>
14218 * builtins.c (fold_builtin_FUNCTION): Use
14219 lang_hooks.decl_printable_name.
14221 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
14223 PR middle-end/81194
14224 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
14225 with only one label.
14226 * stmt.c (expand_case): Assert NCASES is greater than one.
14228 2017-06-29 Richard Biener <rguenther@suse.de>
14230 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
14232 (group_case_labels): Likewise.
14233 (find_taken_edge): Push sanity checking on val to workers...
14234 (find_taken_edge_cond_expr): ... here
14235 (find_taken_edge_switch_expr): ... and here, handle cases
14236 with just a default label.
14237 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
14238 (group_case_labels): Likewise.
14239 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
14240 group_case_labels does anything cleanup the CFG again.
14242 2017-06-29 Bin Cheng <bin.cheng@arm.com>
14244 PR tree-optimization/81196
14245 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
14246 exit condition comparing two IVs.
14248 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
14250 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
14251 profile to the dummy entry at the end of the list of architectures.
14252 * config/arm/arm-cpu-cdata.h: Regenerated.
14254 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14255 Michael Collison <michael.collison@arm.com>
14258 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
14260 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
14261 (*aarch64_reg_<mode>3_minus_mask): New pattern.
14262 (*aarch64_<optab>_reg_di3_mask2): New pattern.
14263 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
14264 of shift when the shift amount is masked with constant equal to
14265 the size of the mode.
14266 * config/aarch64/predicates.md (subreg_lowpart_operator): New
14269 2017-06-29 Martin Liska <mliska@suse.cz>
14271 * config/i386/i386.opt: Change range from [1,5] to [0,5].
14273 2017-06-29 Yury Gribov <tetra2005@gmail.com>
14276 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
14279 (edge_growth_cache_entry::edge_growth_cache_entry): New
14281 (reset_edge_growth_cache): Update to use constructor.
14283 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14285 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
14286 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
14287 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
14289 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
14291 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
14292 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
14294 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
14296 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
14297 (*-linux-uclibc*): Add t-uclibc tmake_file.
14298 * config/t-musl: New.
14299 * config/t-uclibc: New.
14301 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
14303 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
14305 (gen_comm_data): Emit architectural setting of arch_prof.
14306 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
14308 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
14309 (armv8-m.base, armv8-m.main): Likewise.
14310 * arm-protos.h (arm_build_target): Add profile field.
14311 (arch_option): Likewise.
14312 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
14314 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
14315 arm_active_target.profile.
14317 2017-06-28 Richard Biener <rguenther@suse.de>
14319 PR middle-end/81227
14320 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
14321 TYPE_OVERFLOW_WRAPS.
14322 * match.pd (negate_expr_p): Likewise.
14323 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
14324 fold_build2, not fold_binary.
14326 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
14328 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
14329 Convert memory address to Pmode.
14330 (aarch64_print_operand): Assert MEM operands are always Pmode.
14332 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
14335 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
14336 Remove redundant if.
14337 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
14338 * config/arm/aarch-common-protos.h
14339 (aarch_forward_to_shift_is_not_shifted_re): Remove.
14340 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
14342 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
14345 * multiple_target.c (create_dispatcher_calls): Set the default
14346 clone to be static, not public.
14348 2017-06-28 Richard Biener <rguenther@suse.de>
14350 * tree-vect-loop.c (vectorizable_reduction): Move special
14351 cond reduction IV var creation ...
14352 (vect_create_epilog_for_reduction): ... here. Remove induction_index
14353 parameter. Use STMT_VINFO_VECTYPE.
14354 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
14357 2017-06-28 Martin Liska <mliska@suse.cz>
14360 * ipa-visibility.c (non_local_p): Handle visibility.
14362 2017-06-28 Martin Liska <mliska@suse.cz>
14365 * common.opt: Add IntegerRange to various options.
14366 * opt-functions.awk (integer_range_info): New function.
14367 * optc-gen.awk: Add integer_range_info to cl_options struct.
14368 * opts-common.c (decode_cmdline_option): Handle
14369 CL_ERR_INT_RANGE_ARG.
14370 (cmdline_handle_error): Likewise.
14371 * opts.c (print_filtered_help): Show valid interval in
14372 when --help is provided.
14373 * opts.h (struct cl_option): Add range_min and range_max fields.
14374 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
14376 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
14378 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
14379 (x * C EQ/NE y * C): New transformation.
14381 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
14383 * genmultilib (combination_space): Accept '+' in option names.
14385 2017-06-28 Martin Liska <mliska@suse.cz>
14388 * asan.c (instrument_derefs): Bail out inner references
14389 that are hard register variables.
14391 2017-06-28 Jakub Jelinek <jakub@redhat.com>
14394 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
14395 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
14397 2017-06-28 Richard Biener <rguenther@suse.de>
14399 * tree-vectorizer.h (vect_get_vec_defs): Remove.
14400 (vect_get_slp_defs): Adjust.
14401 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
14403 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
14405 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
14406 get_initial_defs_for_reduction instead of vect_get_vec_defs.
14407 (vectorizable_reduction): Adjust.
14408 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
14410 (vect_get_slp_defs): Likewise.
14411 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
14412 (vectorizable_bswap): Adjust.
14413 (vectorizable_call): Likewise.
14414 (vectorizable_conversion): Likewise.
14415 (vectorizable_assignment): Likewise.
14416 (vectorizable_shift): Likewise.
14417 (vectorizable_operation): Likewise.
14418 (vectorizable_store): Likewise.
14419 (vectorizable_condition): Likewise.
14420 (vectorizable_comparison): Likewise.
14422 2017-06-28 Michael Collison <michael.collison@arm.com>
14425 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
14427 (arm_gen_movmemqi): Removed unused variable 'i'.
14428 Convert 'for' loop into 'while' loop.
14429 (arm_expand_prologue): Remove last unnecessary set of insn.
14430 (thumb_pop): Remove unused variable 'pushed_words'.
14431 (thumb_exit): Remove last unnecessary set of regs_to_pop.
14433 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14435 * config/s390/predicates.md: Use s390_rel_address_ok_p.
14436 * config/s390/s390-protos.h: Add prototype of
14437 s390_rel_address_ok_p.
14438 * config/s390/s390.c (s390_got_symbol): New function.
14439 (s390_rel_address_ok_p): New function.
14440 (legitimize_pic_address): Use s390_rel_address_ok_p.
14441 (s390_load_got): Use s390_got_symbol.
14442 (s390_option_override): Issue error if
14443 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
14444 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
14446 * config/s390/s390.opt: New option mpic-data-is-text-relative.
14448 2017-06-27 Andrew Pinski <apinski@cavium.com>
14450 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
14451 (X * copysign (1.0, X)): New pattern.
14452 (X * copysign (1.0, -X)): New pattern.
14453 (copysign (-1.0, CST)): New pattern.
14455 2017-06-27 Joseph Myers <joseph@codesourcery.com>
14457 * genmultilib (combination_space): Remove variable.
14458 Validate reuse rules against regular expression for any sequence
14459 of multilib options in any order.
14461 2017-06-27 Michael Collison <michael.collison@arm.com>
14463 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
14464 call aarch64_split_simd_combine.
14465 * (aarch64_combine_internal<mode>): Delete pattern.
14466 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
14467 Allow register and subreg operands.
14469 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14471 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
14472 specific need, just fallback on defaults.
14473 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
14475 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14476 Olivier Hainque <hainque@adacore.com>
14478 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
14480 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
14481 targets. Pick a default if no particular attempt applied.
14482 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
14485 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14487 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
14488 (x86_64-wrs-vxworks7): Likewise.
14490 2017-06-27 Marek Polacek <polacek@redhat.com>
14493 * ubsan.c (instrument_null): Check get_base_address's result for null.
14495 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
14497 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
14499 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
14501 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
14502 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
14503 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
14504 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
14505 New function types.
14506 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
14507 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
14508 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
14509 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
14510 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
14511 BUILT_IN_FEUPDATEENV): New builtins.
14512 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
14513 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
14514 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
14515 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
14517 (builtin_structptr_types): Adjust size.
14518 * tree.c (builtin_structptr_types): Add four entries.
14520 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14521 Olivier Hainque <hainque@adacore.com>
14523 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
14524 (TLS_SYM): New local macro, forcing reference to __tls__ on
14525 link command lines for VxWorks 7 RTPs, triggering initialization
14527 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
14528 OS features TLS support, true for RTPs on VxWorks 7.
14529 * config/vxworks.c (vxworks_override_options): Setup emutls
14532 2017-06-27 Jakub Jelinek <jakub@redhat.com>
14534 * predict.c (test_prediction_value_range): Use -1U instead of -1
14535 to avoid narrowing conversion warning.
14536 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
14537 to avoid narrowing conversion warning.
14538 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
14540 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
14542 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14544 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
14545 64bit configurations.
14546 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
14547 (SIZE_TYPE): Likewise.
14548 * config/vxworks.c (vxworks_emutls_var_fields): Use
14549 long_unsigned_type_node instead of unsigned_type_node as the offset
14550 field type, which is "pointer" mode in emutls.c.
14552 2017-06-27 Jakub Jelinek <jakub@redhat.com>
14555 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
14557 PR middle-end/81207
14558 * gimple-fold.c (replace_call_with_call_and_fold): Handle
14559 gimple_vuse copying separately from gimple_vdef copying.
14561 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14563 * value-prof.c (free_hist): Remove call to memset and the enclosing if
14566 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
14567 Olivier Hainque <hainque@adacore.com>
14569 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
14570 for all vxworks7 targets.
14571 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
14572 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
14573 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
14574 variations for VX6/VX7 and 32/64bits later on in ...
14575 (VXWORKS_LIB_SPEC): Leverage new macros.
14576 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
14577 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
14579 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
14581 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
14582 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
14584 2017-06-26 Carl Love <cel@us.ibm.com>
14586 * config/rs6000/rs6000-c.c: Add support for built-in functions
14587 vector bool char vec_reve (vector bool char);
14588 vector signed char vec_reve (vector signed char);
14589 vector unsigned char vec_reve (vector unsigned char);
14590 vector bool int vec_reve (vector bool int);
14591 vector signed int vec_reve (vector signed int);
14592 vector unsigned int vec_reve (vector unsigned int);
14593 vector bool long long vec_reve (vector bool long long);
14594 vector signed long long vec_reve (vector signed long long);
14595 vector unsigned long long vec_reve (vector unsigned long long);
14596 vector bool short vec_reve (vector bool short);
14597 vector signed short vec_reve (vector signed short);
14598 vector double vec_reve (vector double);
14599 vector float vec_reve (vector float);
14600 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
14601 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
14602 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
14603 (altivec_vreve): New pattern.
14604 * config/rs6000/altivec.h (vec_reve): New define.
14605 * doc/extend.texi (vec_rev): Update the built-in documentation file
14606 for the new built-in functions.
14608 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14610 PR tree-optimization/71815
14611 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
14613 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
14615 (slsr_process_phi): Likewise.
14616 (replace_uncond_cands_and_profitable_phis): Don't replace a
14617 multiply candidate with a stride of 1 (copy or cast).
14618 (phi_incr_cost): Call uses_consumed_by_stmt rather than
14620 (lowest_cost_path): Likewise.
14621 (total_savings): Likewise.
14623 2017-06-26 Richard Biener <rguenther@suse.de>
14626 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
14627 Use def_builtin_pure for all gather builtins.
14629 2017-06-26 Richard Biener <rguenther@suse.de>
14631 PR tree-optimization/81203
14632 * tree-tailcall.c (find_tail_calls): Do not move stmts into
14633 non-dominating BBs.
14635 2017-06-26 Marek Polacek <polacek@redhat.com>
14638 * doc/invoke.texi: Document -Wmultistatement-macros.
14640 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
14642 * doc/sourcebuild.texi (ARM-specific attributes): Document new
14643 arm_neon_ok_no_float_abi effective target.
14645 2017-06-26 Richard Biener <rguenther@suse.de>
14647 PR tree-optimization/80928
14648 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
14649 (copy_bbs): Set BB_DUPLICATED flag early.
14650 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
14652 (execute_on_shrinking_pred): Likewise.
14653 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
14654 BB_DUPLICATED blocks.
14655 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
14656 iterate over all PHIs considering removal of *gsi.
14658 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
14660 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
14663 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14665 * config/rs6000/rs6000-string.c: (expand_block_clear,
14666 do_load_for_compare, select_block_compare_mode,
14667 compute_current_alignment, expand_block_compare,
14668 expand_strncmp_align_check, expand_strn_compare,
14669 expand_block_move, rs6000_output_load_multiple)
14670 Move functions related to string/block move/compare
14671 to a separate file.
14672 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
14673 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
14674 for this function which is now used in two files.
14675 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
14676 * config.gcc: Add rs6000-string.o to extra_objs for
14677 targets powerpc*-*-* and rs6000*-*-*.
14679 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
14682 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
14683 32-bit, since indexed is not valid for DImode.
14684 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
14685 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
14686 (define_peephole2 for Altivec d-form load): Add 32-bit support.
14687 (define_peephole2 for Altivec d-form store): Likewise.
14690 * multiple_target.c (create_dispatcher_calls): Only create the
14691 dispatcher call if the function is the default clone of a
14692 versioned function.
14694 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
14696 PR middle-end/80902
14697 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
14698 a call, force the call to not be a tail call.
14700 2017-06-23 Jeff Law <law@redhat.com>
14702 * doc/contrib.texi: Add entry for Steven Pemberton's work on
14705 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
14707 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
14708 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
14709 handling for early expansion of vector shifts (sl,sr,sra,rl).
14710 (builtin_function_type): Add vector shift right instructions
14711 to the unsigned argument list.
14713 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
14715 rtl-optimizatoin/79286
14716 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
14717 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
14718 trap. PIC register plus a const unspec without offset can never trap.
14720 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
14722 * tree.h (builtin_structptr_type): New type.
14723 (builtin_structptr_types): Declare new array.
14724 * tree.c (builtin_structptr_types): New array.
14725 (free_lang_data, build_common_tree_nodes): Use it.
14727 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
14730 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
14733 2017-06-22 Matt Turner <mattst88@gmail.com>
14735 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
14736 Lake models to skylake case. Assume skylake for unknown
14737 models with clflushopt.
14739 2017-06-22 Jeff Law <law@redhat.com>
14741 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
14742 frame sizes that do not satisfy aarch64_uimm12_shift.
14744 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
14746 * profile-count.h (apply_probability,
14747 apply_scale, probability_in): Fix checks for zero.
14749 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14751 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
14752 * doc/cppdiropts.texi (-I @var{dir}): Document it.
14754 2016-06-22 Richard Biener <rguenther@suse.de>
14756 * tree-vect-loop.c (vect_model_reduction_cost): Handle
14757 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
14758 REDUC_MAX_EXPR support.
14759 (vectorizable_reduction): Likewise.
14760 (vect_create_epilog_for_reduction): Likewise.
14762 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
14764 * match.pd (A / (1 << B) -> A >> B): New.
14765 * generic-match-head.c: Include optabs-tree.h.
14766 * gimple-match-head.c: Likewise.
14767 * optabs-tree.h (target_supports_op_p): New.
14768 * optabs-tree.c (target_supports_op_p): New.
14770 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14772 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
14773 $gcc_cv_ld --help output.
14774 (gcc_cv_ld_demangle): Likewise.
14775 (gcc_cv_ld_eh_frame_hdr): Likewise.
14776 (gcc_cv_ld_pie): Likewise.
14777 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
14778 (gcc_cv_ld_buildid): Likewise.
14779 (gcc_cv_ld_sysroot): Likewise.
14780 (ld_bndplt_support): Likewise.
14781 (ld_pushpopstate_support): Likewise.
14782 * configure: Regenerate.
14783 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
14785 2017-06-21 Jakub Jelinek <jakub@redhat.com>
14788 * config/i386/sse.md (round<mode>2): Renumber match_dup and
14789 operands indexes to avoid gap between operands and match_dups.
14791 2017-06-21 Andrew Pinski <apinski@cavium.com>
14793 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
14794 Increment Arith_shift and Arith_shift_reg by 1.
14795 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
14797 * config/aarch64/aarch64.c (thunderx_tunings): Enable
14798 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
14799 (aarch64_strip_extend): Add new argument and test for it.
14800 (aarch64_cheap_mult_shift_p): New function.
14801 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
14802 add a cost if it is true.
14803 Update calls to aarch64_strip_extend.
14804 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
14806 2017-06-21 Andrew Pinski <apinski@cavium.com>
14808 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
14810 (thunderxt88): Likewise.
14811 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
14812 (thunderx_prefetch_tune): New variable.
14813 (thunderx2t99_prefetch_tune): Update for the correct values.
14814 (thunderxt88_tunings): New variable.
14815 (thunderx_tunings): Use thunderx_prefetch_tune instead of
14816 generic_prefetch_tune.
14817 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
14819 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14821 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
14822 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
14823 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
14824 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
14825 (aarch64_atomic_cas<mode>, GPI): Likewise.
14827 2017-06-21 Martin Liska <mliska@suse.cz>
14829 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
14830 statements on cold and hot labels.
14831 * predict.c (tree_estimate_probability_bb): Remove the
14832 prediction from this place.
14834 2017-06-21 Martin Liska <mliska@suse.cz>
14836 PR tree-optimization/79489
14837 * gimplify.c (maybe_add_early_return_predict_stmt): New
14839 (gimplify_return_expr): Call the function.
14840 * predict.c (tree_estimate_probability_bb): Remove handling
14842 * predict.def: Update comment about early return predictor.
14843 * gimple-predict.h (is_gimple_predict): New function.
14844 * predict.def: Change default value of early return to 66.
14845 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
14847 * passes.def: Put pass_strip_predict_hints to the beginning of
14850 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
14852 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
14853 FUNCTION_DECL declarations.
14854 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
14856 (dwaf2out_decl): Likewise.
14857 * godump.c (go_early_global_decl): Skip call to the real debug hook
14858 for FUNCTION_DECL declarations.
14859 * passes.c (rest_of_decl_compilation): Skip call to the
14860 early_global_decl debug hook for FUNCTION_DECL declarations, unless
14861 -fdump-go-spec is passed.
14863 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
14865 * config/i386/i386.c (struct builtin_isa): New field pure_p.
14866 Reorder for compactness.
14867 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
14868 (def_builtin_pure, def_builtin_pure2): New functions.
14869 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
14871 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
14873 * match.pd (nop_convert): New predicate.
14874 ((A +- CST1) +- CST2): Allow some NOP conversions.
14876 2017-06-21 Jakub Jelinek <jakub@redhat.com>
14879 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
14880 with ctors/dtors if GOVD_SHARED is set.
14882 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14884 * config/aarch64/aarch64.md (movti_aarch64):
14885 Emit mov rather than orr.
14886 (movtf_aarch64): Likewise.
14887 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
14888 Emit mov rather than orr.
14890 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14892 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
14893 Swap alternatives, make integer dup more expensive.
14895 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14897 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
14898 Return true for non-tls symbols.
14900 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
14902 * config/aarch64/aarch64-cores.def (cortex-a55): New.
14903 (cortex-a75): Likewise.
14904 (cortex-a75.cortex-a55): Likewise.
14905 * config/aarch64/aarch64-tune.md: Regenerate.
14906 * doc/invoke.texi (-mtune): Document new values for -mtune.
14908 2017-06-21 Tom de Vries <tom@codesourcery.com>
14910 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
14911 stack_size feature.
14912 (Effective-Target Keywords, Other attributes): Suggest using
14913 dg-add-options stack_size feature to get stack limit in stack_size
14914 effective target documentation.
14916 2017-06-21 Julian Brown <julian@codesourcery.com>
14917 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14919 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
14920 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
14921 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
14923 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
14924 attribute type list for neon_multiply.
14925 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
14926 attribute type list for neon_multiply.
14927 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
14928 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
14929 attribute type list for neon_multiply.
14930 * config/arm/types.md (crypto_pmull): Add.
14931 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
14932 attribute type list.
14934 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
14936 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
14939 2017-06-20 Jakub Jelinek <jakub@redhat.com>
14941 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
14942 to make sure not to dereference a NULL cost_classes_ptr pointer.
14944 2017-06-20 Carl Love <cel@us.ibm.com>
14946 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14947 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14948 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14949 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14950 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14951 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
14952 VMULOSW): New enum "unspec" values.
14953 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
14954 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
14955 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
14956 altivec_vmulosw): New patterns.
14957 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14958 VMULOSW): Add definitions.
14960 2017-06-20 Julia Koval <julia.koval@intel.com>
14962 * config/i386/i386.c: Fix rounding expand for new pattern.
14963 * config/i386/subst.md: Fix pattern (parallel -> unspec).
14965 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14967 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
14968 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
14970 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14972 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
14975 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14977 * config/aarch64/aarch64-cores.def: Rearrange to sort by
14978 architecture, then by implementer ID.
14979 * config/aarch64/aarch64-tune.md: Regenerate.
14981 2017-06-20 Richard Biener <rguenther@suse.de>
14983 PR middle-end/81097
14984 * fold-const.c (split_tree): Fold to type before negating.
14986 2017-06-20 David Malcolm <dmalcolm@redhat.com>
14988 * diagnostic-show-locus.c
14989 (selftest::test_fixit_deletion_affecting_newline): New function.
14990 (selftest::diagnostic_show_locus_c_tests): Call it.
14992 2017-06-20 Andreas Schwab <schwab@suse.de>
14995 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
14998 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15000 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
15001 __ARM_FEATURE_COPROC according to support.
15003 2017-06-20 Jakub Jelinek <jakub@redhat.com>
15005 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
15006 Rewritten to avoid overflow for > 32-bit pointers.
15009 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
15010 by removing enum keyword.
15011 (ubsan_type_descriptor): Likewise. Formatting fix.
15014 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
15015 splitter): Require TARGET_SSE2 in the condition.
15017 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
15020 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
15021 for doing vector set of SFmode on ISA 3.0.
15022 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
15023 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
15025 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
15026 SFmode value into a V4SF variable that was extracted from another
15027 V4SF variable without converting the element to double precision
15028 and back to single precision vector format.
15029 (vsx_insert_extract_v4sf_p9_2): Likewise.
15031 2017-06-19 Jakub Jelinek <jakub@redhat.com>
15033 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
15034 in UWHI to avoid undefined overflow.
15037 * ubsan.h (enum ubsan_encode_value_phase): New.
15038 (ubsan_encode_value): Change second argument to
15039 enum ubsan_encode_value_phase with default value of
15040 UBSAN_ENCODE_VALUE_GENERIC.
15041 * ubsan.c (ubsan_encode_value): Change second argument to
15042 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
15043 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
15044 create_tmp_var_raw instead of create_tmp_var and use a
15046 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
15047 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
15048 ubsan_encode_value callers.
15051 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
15052 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
15053 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
15055 2017-06-19 Richard Biener <rguenther@suse.de>
15057 PR middle-end/81118
15058 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
15059 estimates if we changed anything.
15061 2017-06-19 Richard Biener <rguenther@suse.de>
15063 PR tree-optimization/80887
15064 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
15065 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
15066 simplified lookups, then reset mprts_hook.
15067 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
15069 (try_to_simplify): Likewise.
15071 2017-06-19 Martin Liska <mliska@suse.cz>
15074 * gimplify.c (gimplify_switch_expr):
15075 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
15077 2017-06-19 Martin Liska <mliska@suse.cz>
15079 * doc/install.texi: Document that PGO runs in 4 stages.
15081 2017-06-19 Martin Liska <mliska@suse.cz>
15084 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
15085 to dispatcher function name.
15086 * multiple_target.c (replace_function_decl): New function.
15087 (create_dispatcher_calls): Redirect both edges and references.
15089 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
15091 * profile-count.c (profile_count::dump): Dump quality.
15092 (profile_count::differs_from_p): Update for unsigned val.
15093 * profile-count.h (profile_count_quality): New enum.
15094 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
15096 2017-06-19 Richard Biener <rguenther@suse.de>
15098 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
15099 struct function as arg.
15100 (estimate_numbers_of_iterations): Export overload with loop arg.
15101 (free_numbers_of_iterations_estimates_loop): Use an overload of
15102 free_numbers_of_iterations_estimates instead.
15103 * tree-cfg.c (remove_bb): Adjust.
15104 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
15105 * tree-parloops.c (gen_parallel_loop): Likewise.
15106 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
15108 (tree_unroll_loops_completely): Likewise.
15109 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
15110 Use an overload instead and export.
15111 (estimated_loop_iterations): Adjust.
15112 (max_loop_iterations): Likewise.
15113 (likely_max_loop_iterations): Likewise.
15114 (estimate_numbers_of_iterations): Take struct function as arg
15116 (loop_exits_before_overflow): Adjust.
15117 (free_numbers_of_iterations_estimates_loop): Use an overload.
15118 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
15119 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
15121 2017-06-19 Richard Biener <rguenther@suse.de>
15124 * ipa-prop.c (find_constructor_constant_at_offset): Handle
15125 RANGE_EXPR conservatively.
15127 2017-06-16 Carl Love <cel@us.ibm.com>
15129 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15130 definitions for vec_float, vec_float2, vec_floato,
15131 vec_floate built-ins.
15132 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
15133 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
15135 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
15136 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
15137 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
15138 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
15139 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
15140 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
15141 vec_floato): Add builtin defines.
15142 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
15143 Update the built-in documentation file for the new built-in
15146 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15148 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
15149 (mthumb): Mark as the negative of -marm.
15151 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15153 * doc/invoke.texi (ARM Options, -mcpu): Document supported
15155 (ARM Options, -mtune): Document that this accepts the same
15156 extension options as -mcpu.
15157 (ARM Options, -mfpu): Document addition of -mfpu=auto.
15159 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15161 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
15162 permitted extensions.
15164 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15166 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
15167 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
15168 (armv8-m.main): Add option +nodsp.
15169 * config/arm/arm-cpu-cdata.h: Regenerated.
15171 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15173 * config/arm/t-fuchsia: New file.
15174 * config.gcc (arm*-*-fuchsia*): Use it.
15176 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15178 * config/arm/t-symbian: Rewrite for new option infrastructure.
15180 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15182 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
15183 (MULTILIB_REQUIRED): Likewise.
15185 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15187 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
15188 (MULTILIB_RESUE): Likewise.
15189 (MULTILIB_MATCHES): Likewise.
15190 (MULTLIB_REQUIRED): Likewise.
15192 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15194 * config/arm/t-rtems: Rewrite for new option framework.
15196 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15198 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
15199 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
15200 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
15201 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
15202 * config/arm/t-multilib: ... here.
15203 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
15204 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
15205 armv7-a and armv8*-a when A-profile libraries have not been built.
15206 * config/arm/t-rmprofile: Rewrite.
15208 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15210 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
15211 with a backslash. Remove the backslash after substituting unescaped
15213 * doc/fragments.texi (MULTILIB_REUSE): Document it.
15215 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15217 * config.gcc: (arm*-*-*): When building a-profile libraries, force
15218 the driver to pass through the default setting of -mfloat-abi.
15219 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
15221 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
15222 (all_feat_combs): New rule.
15223 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
15225 * config/arm/t-aprofile: Rewrite.
15227 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15229 * config/arm/arm.h (FPUTYPE_AUTO): Define.
15230 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
15231 fpu is not specified by the user/command-line.
15232 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
15233 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
15234 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
15235 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
15236 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
15237 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
15239 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15241 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
15242 * config/arm/t-arm-elf: Rewritten.
15244 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15246 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
15247 have some floating-point instructions.
15248 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
15249 (TARGET_MAYBE_HARD_FLOAT): New macro.
15250 * config/arm/arm-builtins.c (arm_init_builtins): Use
15251 TARGET_MAYBE_HARD_FLOAT.
15252 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
15254 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15256 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
15257 (configargs.h): Include it.
15258 (arm_print_hint_for_fpu_option): New function.
15259 (arm_parse_fpu_option): New function.
15260 (candidate_extension): New class.
15261 (arm_canon_for_multilib): New function.
15262 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
15263 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
15264 (ARCH_CANONICAL_SPECS): New macro.
15265 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
15267 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15269 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
15270 are set after handling multilib fragments. Set target_cpu_default2
15273 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15275 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
15277 (arm*-*-*): Set target_cpu_default2 to a quoted string.
15278 * config/arm/parsecpu.awk (check_cpu): Validate any extension
15280 (check_arch): Likewise.
15281 * config/arm/arm.c (arm_configure_build_target): Handle
15282 TARGET_CPU_DEFAULT being a string constant. Scan any feature
15283 options in the default.
15285 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15287 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
15288 when an option is an alias of another.
15289 * config/arm/parsecpu.awk (optalias): New parser token.
15290 (gen_comm_data): Mark non-alias options as such. Emit entries
15291 for extension aliases.
15292 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
15293 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
15294 (armv6kz, armv6zk, armv6t2): Likewise.
15295 (armv7): Make vfpv3-d16 an alias.
15296 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
15298 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
15299 Sort in canonical order.
15300 (armv8-a): Sort in canonical order.
15301 (armv8.1-a, armv8.2-a): Likewise.
15302 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
15304 (cortex-a9): Sort in canonical order.
15305 * config/arm/arm.c (selftests.h): Include it.
15306 (arm_test_cpu_arch_data): New function.
15307 (arm_run_self_tests): New function.
15308 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
15309 (targetm): Move declaration to the end of the file.
15310 * arm-cpu-cdata.h: Regenerated.
15312 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15314 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
15315 call to target_mode_check describing the type of option passed.
15316 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
15317 (arm_target_thumb_only): Use arm_parse_arch_option_name or
15318 arm_parse_cpu_option_name to match parameters against list of
15320 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
15321 arm_arch_core_flags data structure.
15322 * config/arm/arm-cpu_cdata.h: Regenerated.
15324 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15326 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
15328 (arm_print_hint_for_cpu_option): Likewise.
15329 (arm_print_hint_for_arch_option): Likewise.
15330 (arm_parse_cpu_option_name): Likewise.
15331 (arm_parse_arch_option_name): Likewise.
15332 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
15333 of entries in the all_fpus list.
15334 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
15335 (arm_parse_cpu_option_name): Declare.
15336 (arm_parse_arch_option_name): Declare.
15337 (arm_parse_option_features): Declare.
15338 (arm_intialize_isa): Declare.
15339 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
15341 (gen_comm_data): ... here. Make definitions non-static.
15342 * config/arm/arm-cpu-data.h: Regenerated.
15343 * config/arm/arm-cpu-cdata.h: Regenerated.
15345 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15347 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
15348 (cpu_arch_extension): New structure.
15349 (cpu_arch_option, arch_option, cpu_option): New structures.
15350 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
15351 architecture types.
15352 (gen_data): Generate new format data tables.
15353 * config/arm/arm.c (cpu_tune): New structure.
15354 (cpu_option, processors): Delete.
15355 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
15356 (arm_print_hint_for_cpu_option): ... this and ...
15357 (arm_print_hint_for_arch_option): ... this.
15358 (arm_parse_arch_cpu_name): Delete. Replace with ...
15359 (arm_parse_cpu_option_name): ... this and ...
15360 (arm_parse_arch_option_name): ... this.
15361 (arm_unrecognized_feature): Change type of target parameter to
15363 (arm_parse_arch_cpu_features): Delete. Replace with ...
15364 (arm_parse_option_features): ... this.
15365 (arm_configure_build_target): Rework to use new configuration data
15367 (arm_print_tune_info): Rework for new configuration data tables.
15368 * config/arm/arm-cpu-data.h: Regenerated.
15369 * config/arm/arm-cpu.h: Regenerated.
15371 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15373 * Makefile.in (OBJS): Move sbitmap.o from here ...
15374 (OBJS-libcommon): ... to here.
15376 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15378 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
15379 (ISA_ALL_CRYPTO): New macro.
15380 (ISA_ALL_SIMD): New macro
15381 (ISA_ALL_FP): New macro.
15382 * config/arm/arm.c (fpu_bitlist): Update initializer.
15383 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
15385 (arm9e): Add fpu. Add option for nofp
15386 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
15387 (arm926ej-s, arm1026ej-s): Likewise.
15388 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
15389 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
15390 neon-fp16, neon-vfpv4, nofp and nosimd.
15391 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
15392 (cortex-a8): Add fpu. Add option for nofp.
15393 (cortex-a9): Add fpu. Add options for nosimd and nofp.
15394 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
15395 (cortex-r4f): Add fpu.
15396 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
15397 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
15399 (cortex-r8): Likewise.
15400 (cortex-m4): Add fpu. Add option for nofp.
15401 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
15402 (cortex-a17.cortex-a7): Likewise.
15403 (cortex-a32): Add fpu. Add options for crypto and nofp.
15404 (cortex-a35, cortex-a53): Likewise.
15405 (cortex-a57): Add fpu. Add option for crypto.
15406 (cortex-a72, cortex-a73): Likewise.
15407 (exynos-m1): Likewise.
15408 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
15409 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
15410 (cortex-m33): Add fpu. Add option for nofp.
15411 * config/arm/arm-cpu-cdata.h: Regenerated
15412 * config/arm/arm-cpu-data.h: Regenerated.
15414 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15416 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
15417 (armv5te, armv5tej): Likewise.
15418 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
15419 (armv7): Add options fp and vfpv3-d16.
15420 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
15421 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
15423 (armv7ve): Likewise.
15424 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
15425 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
15426 (armv8-a): Add nocrypto option.
15427 (armv8.1-a, armv8.2-a): Likewise.
15428 (armv8-m.main): add options fp, fp.dp and nofp.
15430 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15432 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
15434 (armv8-a+crc): Delete.
15435 (armv8.1-a): Add options simd, crypto and nofp.
15436 (armv8.2-a): Add options fp16, simd, crypto and nofp.
15437 (armv8.2-a+fp16): Delete.
15438 (armv8-m.main): Add option dsp.
15439 (armv8-m.main+dsp): Delete.
15440 (cortex-a8): Add fpu. Add nofp option.
15441 (cortex-a9): Add fpu. Add nofp and nosimd options.
15442 * config/arm/parsecpu.awk (gen_data): Generate option tables and
15443 link to main cpu and architecture data structures.
15444 (gen_comm_data): Only put isa attributes from the main architecture
15446 (option): New statement for architecture and CPU entries.
15447 * arm.c (struct cpu_option): New structure.
15448 (struct processors): Add entry for options.
15449 (arm_unrecognized_feature): New function.
15450 (arm_parse_arch_cpu_name): Ignore any characters after the first
15452 (arm_parse_arch_cpu_feature): New function.
15453 (arm_configure_build_target): Separate out any CPU and architecture
15454 features and parse separately. Don't error out if -mfpu=auto is
15455 used with only an architecture string.
15456 (arm_print_asm_arch_directives): New function.
15457 (arm_file_start): Call it.
15458 * config/arm/arm-cpu-cdata.h: Regenerated.
15459 * config/arm/arm-cpu-data.h: Likewise.
15460 * config/arm/arm-tables.opt: Likewise.
15462 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15464 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
15465 assembler when it is not -mfpu=auto.
15467 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15469 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
15470 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
15471 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
15472 (ASM_CPU_SPEC): Rewrite.
15473 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
15474 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
15475 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
15476 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
15477 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
15478 copied string is NUL-terminated. Also strip any characters prefixed
15480 (arm_rewrite_selected_arch): New function.
15481 (arm_rewrite_march): New function.
15483 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15485 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
15486 (x_arm_cpu_string, x_arm_tune_string): Likewise.
15487 (march, mcpu, mtune): Convert to string-based options.
15488 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
15489 (arm_parse_arch_cpu_name): New function.
15490 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
15491 identify selected architecture or CPU.
15492 (arm_option_save): New function.
15493 (TARGET_OPTION_SAVE): Redefine.
15494 (arm_option_restore): Restore string options.
15495 (arm_option_print): Print string options.
15497 2017-06-16 Martin Sebor <msebor@redhat.com>
15499 PR tree-optimization/80933
15500 PR tree-optimization/80934
15501 * builtins.c (fold_builtin_3): Do not handle bcmp here.
15502 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
15503 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
15504 (gimple_fold_builtin): Call them.
15506 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15508 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
15509 as unlikely; update profile.
15511 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15513 * predict.c (force_edge_cold): Handle declaring edges impossible
15516 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15518 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
15520 (try_unroll_loop_completely): Fix reporting.
15522 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15524 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
15526 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
15529 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
15530 if given a non-constant argument for an intrinsic which requires a
15533 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15535 * profile.c (compare_freqs): New function.
15536 (branch_prob): Sort edge list.
15537 (find_spanning_tree): Assume that the list is priority sorted.
15539 2017-06-16 Richard Biener <rguenther@suse.de>
15541 PR tree-optimization/81090
15542 * passes.def (pass_record_bounds): Remove.
15543 * tree-pass.h (make_pass_record_bounds): Likewise.
15544 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
15545 make_pass_record_bounds): Likewise.
15546 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
15547 not free niter estimates at the beginning but at the end.
15548 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
15550 2017-06-16 Richard Biener <rguenther@suse.de>
15552 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
15553 initializer to workaround ICE in host GCC 4.8.
15555 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15557 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
15559 (clone_inlined_nodes): Update.
15561 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15563 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
15564 prefetch settings, and enable prefetching by default at -O3.
15566 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15568 * config/aarch64/aarch64.c (aarch64_override_options_internal):
15569 Set flag_prefetch_loop_arrays according to tuning data.
15571 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15573 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
15574 New tune structure.
15575 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
15576 [Unrelated to main purpose of the patch] Place the pointer field last
15577 to enable type checking errors when tune structure are wrongly merged.
15578 * config/aarch64/aarch64.c (generic_prefetch_tune,)
15579 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
15580 (thunderx2t99_prefetch_tune): New tune constants.
15581 (tune_params *_tunings): Update all tunings (no functional change).
15582 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
15583 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
15584 from tunings structures.
15586 2017-06-16 Jakub Jelinek <jakub@redhat.com>
15589 * ubsan.c (instrument_null): Add T argument, use it instead
15590 of computing it based on IS_LHS.
15591 (instrument_object_size): Likewise.
15592 (pass_ubsan::execute): Adjust instrument_null and
15593 instrument_object_size callers to pass gimple_get_lhs or
15594 gimple_assign_rhs1 result to it. Use instrument_null instead of
15595 calling get_base_address and instrument_mem_ref. Handle
15596 aggregate call arguments for object-size sanitization.
15598 2017-06-16 Yury Gribov <tetra2005@gmail.com>
15600 PR tree-optimization/81089
15601 * tree-vrp.c (is_masked_range_test): Validate operands of
15604 2017-06-15 Martin Sebor <msebor@redhat.com>
15607 * dumpfile.c (dump_register): Avoid calling memset to initialize
15608 a class with a default ctor.
15609 * gcc.c (struct compiler): Remove const qualification.
15610 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
15611 * hash-table.h: Ditto.
15612 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
15614 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
15615 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
15617 * params.h (struct param_info): Make struct members non-const.
15618 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
15619 with default initialization.
15620 * vec.h (vec_copy_construct, vec_default_construct): New helper
15622 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
15623 with vec_copy_construct.
15624 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
15625 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
15626 * doc/invoke.texi (-Wclass-memaccess): Document.
15628 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15630 * emit-rtl.h (is_leaf): Update comment about local
15631 register allocator.
15633 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15636 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
15637 for a variable to have a section before checking if the section has a
15639 Set section to.persistent if persistent attribute is set.
15640 Warn if .persistent attribute is used on an automatic variable.
15642 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
15644 PR rtl-optimization/80474
15645 * reorg.c (update_block): Do not ignore instructions in a delay slot.
15647 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
15649 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
15652 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
15654 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
15655 (casesi): Emit bounds checking as RTL.
15656 (casesi_internal_mips16_<mode>): Remove bounds checking.
15658 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
15660 * config/xtensa/xtensa.c (xtensa_option_override): Append
15661 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
15662 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
15663 xtensa_doloop_hooks): Define unconditionally.
15664 (xtensa_reorg_loops): Only call reorg_loops in the presence of
15666 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
15667 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
15668 for it in xtensa_option_override.
15669 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
15670 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
15672 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
15674 * doc/cppopts.texi: Document '-' special value to -MF.
15676 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
15678 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
15679 (cortex_a53_fconst): Likewise.
15680 (cortex_a53_fpmul): Likewise.
15681 (cortex_a53_f_load_64): Likewise.
15682 (cortex_a53_f_load_many): Likewise.
15683 (cortex_a53_advsimd_alu): Likewise.
15684 (cortex_a53_advsimd_alu_q): Likewise.
15685 (cortex_a53_advsimd_mul): Likewise.
15686 (cortex_a53_advsimd_mul_q): Likewise.
15687 (fpmac bypass): Add new bypass for fpmac-fpmac case.
15688 Add missing fmul, r2f_cvt and fconst cases.
15690 2017-06-14 Richard Biener <rguenther@suse.de>
15692 PR middle-end/81088
15693 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
15695 (fold_binary_loc): When associating do not treat pre-existing
15696 TREE_OVERFLOW on literal constants as a reason to allow
15697 TREE_OVERFLOW on associated literal constants.
15699 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15701 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
15702 (MASK_FEATURES): New macro.
15703 * config/sparc/sparc.c (sparc_option_override): Remove the special
15704 handling of -mfpu and generalize it to all MASK_FEATURES switches.
15706 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15708 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
15709 a division of 0 if non-call exceptions are enabled.
15711 2017-06-14 Andrew Pinski <apinski@cavium.com>
15712 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15715 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
15716 Improve vector initialization code gen for only variable case.
15718 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15720 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
15722 2017-06-14 Richard Biener <rguenther@suse.de>
15724 PR tree-optimization/81083
15725 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
15728 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15730 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
15731 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
15732 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
15733 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
15734 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
15735 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
15737 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15739 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
15740 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
15742 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15744 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
15746 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15748 * config/rs6000/t-rtems: Don't handle SPE.
15750 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15752 * config/rs6000/t-linux: Don't handle SPE.
15754 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15756 * config/rs6000/eabispe.h: Delete file.
15758 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15760 * config/rs6000/t-spe: Delete file.
15762 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15764 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
15765 (rs6000_legitimate_offset_address_p): Return false for anything in
15766 V2SImode or V2SFmode.
15768 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15770 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
15771 except V2SF and V2SI. Rearrange the vector modes, and add comments.
15772 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
15774 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
15775 (rs6000_legitimate_offset_address_p): Ditto.
15776 (rs6000_emit_move): Ditto.
15777 (rs6000_init_builtins): Remove V4HI_type_node.
15779 2017-06-13 Martin Liska <mliska@suse.cz>
15782 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
15783 (gate_asan): Likewise.
15784 * asan.h (asan_no_sanitize_address_p): Remove the function.
15785 (sanitize_flags_p): New function.
15786 * builtins.def: Fix coding style.
15787 * common.opt: Use renamed enum value.
15788 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
15789 * doc/extend.texi: Document no_sanitize attribute.
15790 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
15791 to SANITIZE_UNDEFINED_NONDEFAULT.
15792 * gcc.c (sanitize_spec_function): Use the renamed enum value.
15793 * gimple-fold.c (optimize_atomic_compare_exchange_p):
15794 Use sanitize_flags_p.
15795 * gimplify.c (gimplify_function_tree): Likewise.
15796 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
15797 * opts.c (parse_no_sanitize_attribute): New function.
15798 (common_handle_option): Use renamed enum value.
15799 * opts.h (parse_no_sanitize_attribute): Declare.
15800 * tree.c (sanitize_flags_p): New function.
15801 * tree.h: Declared here.
15802 * tsan.c: Use sanitize_flags_p.
15803 * ubsan.c (ubsan_expand_null_ifn): Likewise.
15804 (instrument_mem_ref): Likewise.
15805 (instrument_bool_enum_load): Likewise.
15806 (do_ubsan_in_current_function): Remove the function.
15807 (pass_ubsan::execute): Use sanitize_flags_p.
15808 * ubsan.h: Remove do_ubsan_in_current_function
15809 * tree-cfg.c (print_no_sanitize_attr_value): New function.
15810 (dump_function_to_file): Use it here.
15812 2017-06-13 Martin Jambor <mjambor@suse.cz>
15814 PR tree-optimization/80803
15815 PR tree-optimization/81063
15816 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
15817 (propagate_subaccesses_across_link): Enqueue subtree whenever
15818 necessary instead of relying on the caller.
15820 2017-06-13 Martin Jambor <mjambor@suse.cz>
15822 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
15823 that have a first_link.
15824 (sort_and_splice_var_accesses): Do not check first_link before
15826 (subtree_mark_written_and_enqueue): Likewise.
15827 (propagate_all_subaccesses): Likewise and do not stop at first
15828 parent with a first_link.
15830 2017-06-13 Martin Jambor <mjambor@suse.cz>
15832 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
15835 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15837 * match.pd: New pattern.
15839 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15841 * tree-vrp.c (is_masked_range_test): New function.
15842 (register_edge_assert_for): Determine ranges for
15845 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15847 PR tree-optimization/67328
15848 * fold-const.c (maskable_range_p): New function.
15849 (build_range_check): Generate bittests if possible.
15851 2017-06-13 Martin Liska <mliska@suse.cz>
15853 * gimple-pretty-print.c (dump_probability): Add new argument.
15854 (dump_edge_probability): Dump both probability and count.
15855 (dump_gimple_label): Likewise.
15856 (dump_gimple_bb_header): Likewise.
15858 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
15861 * config/avr/avr-devices.c: Fix indentation.
15862 * config/avr/gen-avr-mmcu-specs.c: Dito.
15864 2017-06-13 Richard Biener <rguenther@suse.de>
15866 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
15867 instead get vector type from stmt_info.
15868 (vectorizable_reduction): Adjust. Remove dead code.
15870 2017-06-13 Richard Biener <rguenther@suse.de>
15872 PR middle-end/81065
15873 * fold-const.c (extract_muldiv_1): Remove bogus distribution
15874 case of C * (x * C2 + C3).
15875 (fold_addr_of_array_ref_difference): Properly fold index difference.
15877 2017-06-12 David S. Miller <davem@davemloft.net>
15880 * config/sparc/sparc.md (return expander): Emit frame blockage if
15881 function uses alloca.
15883 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
15885 * combine.c (make_field_assignment): Check len rather than the mode
15886 precision when calling force_to_mode.
15888 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
15890 Support multilibs and devices that see flash in RAM address range.
15893 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
15894 (avr_mcu_t) <flash_pm_offset>: New field.
15895 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
15896 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
15897 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
15898 (AVR_TINY_PM_OFFSET): Remove macro.
15899 * config/avr/avr.opt (-mshort-calls): New option.
15900 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
15901 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
15902 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
15903 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
15904 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
15905 instead of avr_arch->have_jmp_call.
15906 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
15907 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
15908 avr_arch->flash_pm_offset to define.
15909 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
15910 new field flash_pm_offset. Add entry for avrxmega3.
15911 (avr_texinfo): Add entry for avrxmega3.
15912 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
15913 attiny212, attiny214,
15914 attiny412, attiny414, attiny416, attiny417,
15915 attiny814, attiny816, attiny817,
15916 attiny1614, attiny1616, attiny1617,
15917 attiny3214, attiny3216, attiny3217.
15918 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
15919 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
15920 (avr_print_operand_address) [AVR_TINY]: Same.
15921 (avr_asm_init_sections) <readonly_data_section>: Only patch
15922 callback if avr_arch->flash_pm_offset = 0.
15923 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
15924 for rodata if avr_arch->flash_pm_offset != 0.
15925 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
15926 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
15927 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
15928 (m_options): Append opt_rcall.
15929 (m_dirnames): Append dir_rcall.
15930 * config/avr/t-multilib: Regenerate.
15932 * configure.ac [target=avr]: Check whether avrxmega3 default
15933 linker description file works as needed.
15934 * configure: Regenerate.
15935 * doc/avr-mmcu.texi: Regenerate.
15936 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
15937 <__AVR_ARCH__>: Document avrxmega3 and 103.
15938 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
15939 <__AVR_SHORT_CALLS__>: Document it.
15940 <__AVR_PM_BASE_ADDRESS__>: Document it.
15941 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
15942 (AVR Variable Attributes) <progmem>: Document this is
15943 not needed for avrxmega3.
15944 (AVR Named Address Spaces) <__flash>: Dito.
15946 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
15948 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
15950 2017-06-12 Doug Rupp <rupp@adacore.com>
15952 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
15953 Append vxworks-stdint.h to the tm_file list.
15954 * config/vxworks-stdint.h: New file.
15956 2017-06-12 Martin Liska <mliska@suse.cz>
15958 PR tree-optimization/81041
15959 * tree-profile.c (gimple_gen_ic_func_profiler):
15960 Create an extra BB in profile-generate
15961 (gimple_gen_time_profiler): Likewise.
15963 2017-06-12 Jakub Jelinek <jakub@redhat.com>
15965 PR tree-optimization/81003
15966 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
15967 (update_range_test): Use it instead of force_gimple_operand_gsi.
15969 2017-06-12 Richard Biener <rguenther@suse.de>
15971 PR tree-optimization/81053
15972 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
15973 with backedge value not defined in loop. Simplify def stmt
15976 2017-06-11 Tom de Vries <tom@codesourcery.com>
15979 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
15981 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
15982 nvptx_cannot_force_const_mem.
15984 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15986 * opts.c (finish_options): Move test for flag_split_stack after
15987 it has been initialized.
15989 2017-06-11 Jason Merrill <jason@redhat.com>
15991 * tree.h (id_equal): New.
15992 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
15993 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
15994 instead of strcmp of IDENTIFIER_POINTER.
15996 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15998 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
15999 (mark_all_inlined_calls_cdtor): Fix formating.
16000 (inline_transform): Rescale profile before inlining.
16002 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16004 * cgraph.h (cgraph_edge::clone): Update prototype.
16005 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
16006 (cgraph_node::create_clone): Update.
16007 (cgraph_node::create_version_clone): Update.
16008 * tree-inline.c (copy_bb): Update.
16009 (expand_call_inline): Update.
16011 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
16013 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
16014 factored out from ...
16015 (rs6000_emit_prologue): ... here.
16017 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
16019 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
16020 factored out from ...
16021 (rs6000_emit_prologue): ... here.
16023 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16025 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
16027 (handle_missing_profiles): Fix computation of tp_first_run.
16028 (counts_to_freqs): Do not touch freqs when count is 0.
16030 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16032 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
16035 2017-06-10 Tom de Vries <tom@codesourcery.com>
16037 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
16038 attributes): Document signal effective target.
16040 2017-06-10 Tom de Vries <tom@codesourcery.com>
16042 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
16043 Document effective target stack_size.
16045 2017-06-09 David Malcolm <dmalcolm@redhat.com>
16047 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
16048 to the edit_context if they can be auto-applied.
16050 2017-06-9 Ian Lance Taylor <iant@golang.org>
16052 * opts.c (finish_options): If -fsplit-stack, disable implicit
16053 -forder-blocks-and-partition.
16054 * doc/invoke.texi (Optimize Options): Document that when using
16055 -fsplit-stack -forder-blocks-and-partition is not implicitly
16058 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16060 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
16061 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
16062 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
16063 * builtins.def (abort, trap, unreachable): Declare cold.
16064 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
16065 * tree-core.h (ECF_COLD): New.
16066 * tree.c (set_call_expr_flags): Handle ECF_COLD.
16067 (build_common_builtin_nodes): Mark unreachable and abort as cold.
16069 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16071 * predict.c (unlikely_executed_stmt_p): Cleanup.
16073 2017-06-09 Richard Biener <rguenther@suse.de>
16075 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
16076 model if the ref is always written to.
16078 2017-06-09 Tamar Christina <tamar.christina@arm.com>
16080 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
16082 2017-06-09 Tamar Christina <tamar.christina@arm.com>
16084 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
16087 2017-06-09 Tom de Vries <tom@codesourcery.com>
16090 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
16091 "target cannot support label values" when encountering LABEL_REF.
16093 2017-06-09 Martin Liska <mliska@suse.cz>
16095 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
16096 (gimple_gen_ic_func_profiler): Emit direct comparison
16097 of __gcov_indirect_call_callee with NULL.
16098 (gimple_gen_time_profiler): Change probability from
16099 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
16101 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16103 * profile.c (edge_gcov_counts): Turn to pointer.
16104 (compute_branch_probabilities, compute_branch_probabilities): Update.
16105 (branch_prob): Do not clear edge_gcov_count.
16106 * profile.h (edge_gcov_counts): Turn to pointer.
16107 (edge_gcov_count): Update.
16109 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16111 * gimple.h (gimple_check_failed): Mark cold.
16113 2017-06-09 Richard Biener <rguenther@suse.de>
16115 PR tree-optimization/66623
16116 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
16117 refactor check_reduction into two parts, properly computing
16118 whether we have to check reduction validity for outer loop
16121 2017-06-09 Richard Biener <rguenther@suse.de>
16123 PR tree-optimization/79483
16124 * graphite-scop-detection.c (order): New global.
16125 (get_order): Compute bb to order mapping that satisfies code
16126 generation constraints.
16127 (cmp_pbbs): New helper.
16128 (build_scops): Start domwalk at entry block, sort generated
16131 2017-06-09 Richard Biener <rguenther@suse.de>
16133 PR middle-end/81007
16134 * ipa-polymorphic-call.c
16135 (ipa_polymorphic_call_context::restrict_to_inner_class):
16136 Skip FIELD_DECLs with error_mark_node type.
16137 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
16140 2017-06-09 Martin Liska <mliska@suse.cz>
16142 * predict.c (struct branch_predictor): New struct.
16143 (test_prediction_value_range): New test.
16144 (predict_c_tests): New function.
16145 * selftest-run-tests.c (selftest::run_tests): Run the function.
16146 * selftest.h: Declare new tests.
16148 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
16151 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
16152 gen_add3_insn did not fail.
16153 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
16154 r0, construct that number in a temporary reg and add that reg to r0.
16155 If asked to put the result in r0 as well, fail.
16157 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
16159 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16160 for early expansion of vec_eqv.
16162 2017-06-08 Jakub Jelinek <jakub@redhat.com>
16164 PR middle-end/81005
16165 * ubsan.c (instrument_null): Avoid pointless code temporary.
16166 (pass_ubsan::execute): Instrument aggregate arguments of calls.
16168 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
16172 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
16175 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
16178 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16180 * predict.c (unlikely_executed_edge_p): Move ahead.
16181 (probably_never_executed_edge_p): Use it.
16183 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
16185 PR middle-end/79988
16186 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
16187 gimple_call_builtin_p call.
16189 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16191 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
16192 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
16193 rtl_check_failed_type2, rtl_check_failed_code1,
16194 rtl_check_failed_code2, rtl_check_failed_code_mode,
16195 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
16196 rtvec_check_failed_bounds, rtl_check_failed_flag,
16197 _fatal_insn_not_found, _fatal_insn): Likewise.
16198 * tree.h (tree_contains_struct_check_failed,
16199 tree_check_failed, tree_not_check_failed,
16200 tree_class_check_failed, tree_range_check_failed,
16201 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
16202 tree_vec_elt_check_failed, phi_node_elt_check_failed,
16203 tree_operand_check_failed, omp_clause_check_failed,
16204 omp_clause_operand_check_failed, omp_clause_range_check_failed):
16207 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16209 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
16210 flag_branch_probabilities.
16211 * ipa-inline.c (edge_badness): Likewise.
16212 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
16213 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
16214 * predict.c (maybe_hot_frequency_p): Likewise.
16215 (probably_never_executed): Likewise.
16216 * sched-ebb.c (schedule_ebbs): Likewise.
16217 * sched-rgn.c (find_single_block_region): Likewise.
16218 * tracer.c (tail_duplicate): Likewise.
16220 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16222 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
16223 longer requires x_flag_profile_use.
16225 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16227 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
16228 instead of flag_reorder_blocks_and_partition.
16229 * dbxout.c (dbxout_function_end): Likewise.
16230 * dwarf2out.c (gen_subprogram_die): Likewise.
16231 * haifa-sched.c (sched_create_recovery_edges): Likewise.
16232 * hw-doloop.c (reorg_loops): Likewise.
16233 * varasm.c (assemble_start_function,
16234 assemble_end_function): Likewise.
16235 (decide_function_section): Do not check for
16236 flag_reorder_blocks_and_partition.
16238 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
16240 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
16242 (chkp_get_hard_register_fake_addr_expr): Ditto.
16243 (chkp_build_addr_expr): Add check for hard reg case.
16244 (chkp_parse_array_and_component_ref): Ditto.
16245 (chkp_find_bounds_1): Ditto.
16246 (chkp_process_stmt): Don't generate bounds store for
16249 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16251 * predict.c (maybe_hot_bb_p): Do not check profile status.
16252 (maybe_hot_edge_p): Likewise.
16253 (probably_never_executed): Check for zero counts even if profile
16255 (unlikely_executed_edge_p): New function.
16256 (unlikely_executed_stmt_p): New function.
16257 (unlikely_executed_bb_p): New function.
16258 (set_even_probabilities): Use unlikely predicates.
16259 (combine_predictions_for_bb): Likewise.
16260 (predict_paths_for_bb): Likewise.
16261 (predict_paths_leading_to_edge): Likewise.
16262 (determine_unlikely_bbs): New function.
16263 (estimate_bb_frequencies): Use it.
16264 (compute_function_frequency): Use zero counts even if profile is
16266 * profile-count.h: Fix typo.
16268 2017-08-08 Julia Koval <julia.koval@intel.com>
16270 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
16271 _mm512_mask_cvtsepi16_storeu_epi8,
16272 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
16273 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
16274 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
16275 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
16276 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
16277 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
16278 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
16279 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
16280 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
16281 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
16282 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
16283 __builtin_ia32_pmovuswb256mem_mask,
16284 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
16285 __builtin_ia32_pmovwb512mem_mask): New builtins.
16287 2017-08-08 Julia Koval <julia.koval@intel.com>
16289 PR target/73350,80862
16290 * config/i386/subst.md (round): Fix round pattern.
16291 * config/i386/i386.c (ix86_erase_embedded_rounding):
16292 Fix erasing rounding for the fixed pattern.
16294 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16296 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
16298 2017-06-08 Martin Liska <mliska@suse.cz>
16300 PR gcov-profile/80911
16301 * gcov.c (block_info::block_info): New constructor.
16303 2017-06-07 Carl Love <cel@us.ibm.com>
16305 * config/rs6000/rs6000-c: The return type of the following
16306 built-in functions was implemented as int not long long. Fix sign
16307 of return value for the unsigned version of vec_mulo and vec_mule.
16308 vector unsigned long long vec_bperm (vector unsigned long long,
16309 vector unsigned char)
16310 vector signed long long vec_mule (vector signed int,
16312 vector unsigned long long vec_mule (vector unsigned int,
16313 vector unsigned int)
16314 vector signed long long vec_mulo (vector signed int,
16316 vector unsigned long long vec_mulo (vector unsigned int,
16317 vector unsigned int)
16318 * doc/extend.texi: Fix the documentation for the built-in
16321 2017-06-07 Carl Love <cel@us.ibm.com>
16324 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
16327 2017-06-07 Carl Love <cel@us.ibm.com>
16329 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
16330 support, Generate doublehv for signed int/float for BE case only.
16332 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
16334 * doc/invoke.texi (mcx16): Rewrite.
16336 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16338 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
16339 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
16340 *mov<mode>_softfloat, and an anonymous splitter): Use
16341 nonimmediate_operand instead of rs6000_nonimmediate_operand.
16343 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16345 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
16347 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
16348 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
16349 (rs6000_debug_reg_global): Adjust.
16350 (rs6000_init_hard_regno_mode_ok): Adjust.
16351 (rs6000_dbx_register_number): Adjust.
16352 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
16353 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
16354 Remove SPE_ACC and SPEFSCR.
16355 (REG_ALLOC_ORDER): Ditto.
16356 (FRAME_POINTER_REGNUM): Change to 111.
16357 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
16358 (REG_CLASS_NAMES): Ditto.
16359 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
16360 (REGISTER_NAMES): Ditto.
16361 (ADDITIONAL_REG_NAMES): Ditto.
16362 (rs6000_reg_names): Ditto.
16363 * config/rs6000/rs6000.md: Renumber some register number
16366 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16368 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
16370 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
16371 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
16373 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
16374 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
16375 Delete the SPE high registers.
16376 (REG_ALLOC_ORDER): Ditto.
16377 (enum reg_class): Remove SPE_HIGH_REGS.
16378 (REG_CLASS_NAMES): Ditto.
16379 (REG_CLASS_CONTENTS): Delete the SPE high registers.
16380 (REGISTER_NAMES): Ditto.
16381 (rs6000_reg_names): Ditto.
16382 * doc/tm.texi.in: Remove SPE as example.
16383 * doc/tm.texi: Regenerate.
16385 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16387 * config/rs6000/8540.md (ppc8540_brinc): Delete.
16388 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
16389 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
16390 * config/rs6000/rs6000.md (type): Remove "brinc".
16392 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16394 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
16395 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
16396 * config/rs6000/linuxspe.h: Delete file.
16397 * config/rs6000/rs6000.md: Don't include spe.md.
16398 * config/rs6000/spe.h: Delete file.
16399 * config/rs6000/spe.md: Delete file.
16400 * config/rs6000/t-rs6000: Remove spe.md.
16402 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16404 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
16405 (reg_or_none500mem_operand): Delete.
16406 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
16407 instead of reg_or_none500mem_operand.
16409 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16411 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
16412 handling of SPE flags.
16413 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
16415 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16417 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
16419 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
16420 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
16421 paired_divv2sf3): Similar.
16422 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
16423 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
16424 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
16426 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
16427 Rename the paired_* instruction patterns.
16428 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
16430 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
16431 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
16432 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
16433 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
16434 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
16435 PAIRED_VECTOR_MODE.
16436 (struct machine_function): Delete field spe_insn_chain_scanned_p.
16437 (spe_func_has_64bit_regs_p): Delete.
16438 (spe_expand_predicate_builtin): Delete.
16439 (spe_expand_evsel_builtin): Delete.
16440 (TARGET_DWARF_REGISTER_SPAN): Do not define.
16441 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
16442 (invalid_e500_subreg): Delete.
16443 (rs6000_legitimize_address): Always force_reg op2 as well, for
16444 paired single memory accesses.
16445 (rs6000_member_type_forces_blk): Delete.
16446 (rs6000_spe_function_arg): Delete.
16447 (rs6000_expand_unop_builtin): Delete SPE handling.
16448 (rs6000_expand_binop_builtin): Ditto.
16449 (spe_expand_stv_builtin): Delete.
16450 (bdesc_2arg_spe): Delete.
16451 (spe_expand_builtin): Delete.
16452 (spe_expand_predicate_builtin): Delete.
16453 (spe_expand_evsel_builtin): Delete.
16454 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
16455 (spe_init_builtins): Delete.
16456 (spe_func_has_64bit_regs_p): Delete.
16457 (savres_routine_name): Delete "info" parameter. Adjust callers.
16458 (rs6000_emit_stack_reset): Ditto.
16459 (rs6000_dwarf_register_span): Delete.
16460 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
16461 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
16462 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
16464 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
16466 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
16467 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
16468 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
16469 mulv2sf3, divv2sf3): Delete expanders.
16471 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16473 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
16475 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16477 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
16478 * config/rs6000/rs6000.c: Ditto.
16480 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16482 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
16483 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
16484 comparison_operator.
16486 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16488 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
16489 * config/rs6000/rs6000.opt: Ditto.
16490 * config/rs6000/t-rtems: Ditto.
16492 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16494 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
16495 TARGET_E500_SINGLE by 0, simplify.
16496 * config/rs6000/rs6000.c: Ditto.
16497 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
16498 (spe_build_register_parallel): Delete.
16499 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
16500 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
16501 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
16502 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
16503 (E500_CONVERT): Delete.
16504 * config/rs6000/spe.md: Remove many patterns and all define_constants.
16506 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16508 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
16509 * config/rs6000/dfp.md: Ditto.
16510 (negdd2, *negdd2_fpr): Merge.
16511 (absdd2, *absdd2_fpr): Merge.
16512 (negtd2, *negtd2_fpr): Merge.
16513 (abstd2, *abstd2_fpr): Merge.
16514 * config/rs6000/e500.h: Delete file.
16515 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
16516 TARGET_FPRS by 1 and simplify.
16517 * config/rs6000/rs6000-c.c: Ditto.
16518 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
16519 TARGET_DF_SPE by 0.
16520 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
16522 * config/rs6000/rs6000.md: Ditto.
16523 (floatdidf2, *floatdidf2_fpr): Merge.
16524 (move_from_CR_gt_bit): Delete.
16525 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
16526 (E500_CR_IOR_COMPARE): Delete.
16527 (All patterns that require !TARGET_FPRS): Delete.
16528 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
16530 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16532 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
16534 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16536 * graphds.c (add_edge): Intitialize edge's attached data.
16537 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
16538 pointer parameter. Call pointed function on each edge during
16539 graph traversing. Skip traversing the edge when the function
16541 (graphds_dfs, graphds_scc): Ditto.
16542 (for_each_edge): New parameter. Pass the new parameter to callback
16544 * graphds.h (skip_edge_callback): New function pointer type.
16545 (graphds_dfs, graphds_scc): New function pointer parameter.
16546 (graphds_edge_callback, for_each_edge): New parameter.
16548 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16550 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
16551 out code checking if runtime alias check is possible to below ...
16552 Call the new function.
16553 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
16554 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
16556 2017-06-07 Marek Polacek <polacek@redhat.com>
16559 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
16560 TYPE_OVERFLOW_WRAPS check.
16562 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16564 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
16565 if versioning is required.
16566 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
16567 peeling with the check for versioning.
16569 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16571 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
16572 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
16573 Set true to new parameter if new ssa variable is defined.
16574 (vect_gen_vector_loop_niters): Refactor. Set range information
16575 for the new vector loop bound variable.
16576 (vect_do_peeling): Ditto.
16578 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16580 * tree-affine.c (ssa.h): Include header file.
16581 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
16582 has wrapping overflow behavior.
16584 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16586 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
16588 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16590 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
16591 (tree_to_aff_combination): ... here.
16593 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16595 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
16596 reg_pressure model function.
16597 (ivopts_global_cost_for_size): Delete.
16598 (determine_set_costs, iv_ca_recount_cost): Call new model function
16599 ivopts_estimate_reg_pressure.
16601 2017-06-07 Tamar Christina <tamar.christina@arm.com>
16603 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
16604 expensive than udiv. Remove floating point cases from mod.
16606 2017-06-07 Tamar Christina <tamar.christina@arm.com>
16608 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
16609 Increase idiv cost.
16611 2017-06-07 Tamar Christina <tamar.christina@arm.com>
16613 * config/aarch64/aarch64.md
16614 (copysignsf3): Fix mask generation.
16616 2017-06-07 Jakub Jelinek <jakub@redhat.com>
16618 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
16620 (class dump_manager): Add register_dumps method.
16621 * dumpfile.c: Include langhooks.h.
16622 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
16623 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
16624 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
16625 (dump_manager::dump_register): Start with 512 entries instead of 32.
16626 (dump_manager::register_dumps): New method.
16627 * toplev.c (general_init): Instead of invoking register_dumps
16628 langhook, invoke register_dumps method on the dump manager.
16629 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
16632 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
16634 * doc/md.texi: Clarify the restrictions on a define_insn condition.
16635 Say that # requires an associated define_split to exist, and that
16636 the define_split must be suitable for use after register allocation.
16638 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16640 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
16641 (compute_outgoing_frequencies): Also initialize zero counts.
16642 (find_many_sub_basic_blocks): Do not produce uninitialized profile
16643 around loops; preserve more of profile when nothing changes.
16645 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
16647 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
16649 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
16650 * config/arm/arm-cpu-cdata.h: Regenerate.
16651 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
16652 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
16653 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
16654 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
16655 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
16657 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
16659 * config/arm/t-rmprofile: Likewise.
16660 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
16662 2017-06-06 David S. Miller <davem@davemloft.net>
16665 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
16666 blockage if function uses alloca.
16668 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16670 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
16671 New "uid" fields to hold pretty-print IDs of group and ref.
16672 Memory references are now identified as <group_id>:<ref_id>
16673 instead of using [random] addresses.
16674 (dump_mem_details): Simplify, no functional change.
16675 (dump_mem_ref): Simplify and make output more concise.
16676 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
16677 (find_or_create_group): Initialize group uid.
16678 (record_ref): Initialize ref uid. Improve debug output.
16679 (prune_group_by_reuse, should_issue_prefetch_p,)
16680 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
16681 (mark_nontemporal_store, determine_loop_nest_reuse):
16682 Improve debug output.
16684 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16686 * dbgcnt.def (prefetch): New debug counter.
16687 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
16688 (schedule_prefetches): Stop issueing prefetches if debug counter
16691 2017-06-06 Tom de Vries <tom@codesourcery.com>
16693 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
16694 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
16696 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16698 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
16699 Use aarch64_reg_or_zero predicate for operand 4.
16700 (aarch64_compare_and_swap<mode> define_insn_and_split):
16701 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
16702 (aarch64_store_exclusive<mode>): Likewise for operand 2.
16704 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
16706 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
16707 (arm_compute_save_core_reg_mask): This.
16708 (thumb1_compute_save_reg_mask): Rename into ...
16709 (thumb1_compute_save_core_reg_mask): This.
16710 (arm_compute_save_reg0_reg12_mask): Adapt comment.
16711 (arm_compute_frame_layout): Likewise.
16713 2017-06-06 Richard Biener <rguenther@suse.de>
16715 PR tree-optimization/80974
16716 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
16717 keep or clear leaders SSA info.
16719 2017-06-06 Tom de Vries <tom@codesourcery.com>
16721 * config/nvptx/nvptx.c (split_mode_p): New function.
16722 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
16724 2017-06-06 Tom de Vries <tom@codesourcery.com>
16726 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
16728 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16731 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
16734 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16736 * shrink-wrap.c (handle_simple_exit): Update profile.
16737 (try_shrink_wrapping): Upate profile.
16739 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16741 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
16742 (tree_guess_outgoing_edge_probabilities): New.
16743 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
16744 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
16746 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16748 * ipa-split.c (split_function): Initialize return bb profile.
16750 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16752 * profile.c (compute_branch_probabilities): Also initialize
16753 EXIT_BLOCK profile.
16755 2017-06-06 Richard Biener <rguenther@suse.de>
16757 PR tree-optimization/80928
16758 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
16759 (vect_analyze_loop_operations): Properly guard analysis for
16761 (vect_transform_loop): Likewise.
16762 (vect_analyze_loop_2): Also reset SLP type on PHIs.
16763 (vect_model_induction_cost): Do not cost for pure SLP.
16764 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
16765 of induction in inner loop vectorization.
16766 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
16767 (vect_get_and_check_slp_defs): Handle vect_induction_def.
16768 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
16770 (vect_analyze_slp_cost_1): Cost induction.
16771 (vect_detect_hybrid_slp_stmts): Handle PHIs.
16772 (vect_get_slp_vect_defs): Likewise.
16773 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
16774 (vect_transform_stmt): Handle SLP reductions.
16775 * tree-vectorizer.h (vectorizable_induction): Adjust.
16777 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16779 * config/rs6000/rs6000.c (make_resolver_func): Update
16780 init_lowered_empty_function call.
16782 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
16784 * doc/invoke.texi: Document the -fprofile-abs-path option.
16785 * common.opt (fprofile-abs-path): New option.
16786 * gcov-io.h (gcov_write_filename): Declare.
16787 * gcov-io.c (gcov_write_filename): New function.
16788 * coverage.c (coverage_begin_function): Use gcov_write_filename.
16789 * profile.c (output_location): Likewise.
16791 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16793 * shring-wrap.c: Revert accidental commit.
16795 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
16797 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
16799 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16801 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
16803 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
16804 profile in callgraph edge.
16805 * profile-count.h (apply_probability): If THIS is 0, then result is 0
16806 (apply_scale): Likewise.
16807 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
16808 Also scale profile when inlining function with zero profile.
16809 (initialize_cfun): Update exit block profile even when it is zero.
16810 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
16811 when profile is read.
16813 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16815 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
16816 (CLONE_*): New constants to define the processors we can generate
16817 code for with the target_clone attribute.
16818 (rs6000_clone_map): New array to identify which clone processors
16819 the current program is running on.
16820 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
16821 target_clone attribute.
16822 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
16823 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
16824 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
16825 (cpu_expand_builtin): Add support for target_clone attribute.
16826 (rs6000_valid_attribute_p): Allow "default" attribute.
16827 (get_decl_name): New debug function to simplify printing the
16828 current function name in debugging statements.
16829 (rs6000_clone_priority): New functions to support the target_clone
16830 attribute, and be able to generate code to switch between ISA 2.05
16831 through ISA 3.0 (power6 through power9).
16832 (rs6000_compare_version_priority): Likewise.
16833 (rs6000_get_function_versions_dispatcher): Likewise.
16834 (make_resolver_func): Likewise.
16835 (add_condition_to_bb): Likewise.
16836 (dispatch_function_versions): Likewise.
16837 (rs6000_generate_version_dispatcher_body): Likewise.
16838 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
16839 (fusion_gpr_load_p): Fix a spacing issue.
16840 * doc/extend.texi (Common Function Attributes): Document that the
16841 PowerPC supports the target_clone attribute.
16843 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
16845 * config/arm/arm.h: explain F symbol found in description of ARM
16846 register allocation in its legend.
16848 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16850 * config/mips/frame-header-opt.c: Include profile-count.h.
16851 * config/riscv/riscv.c: Include profile-count.h
16853 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16855 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
16857 (sm_set_flag_if_changed): Add bbs field.
16858 (execute_sm_if_changed_flag_set): Pass BBS.
16859 (execute_sm): Update.
16861 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16863 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
16866 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16868 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
16869 (peephole2): New peephole2 to emit the above.
16870 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
16872 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16874 * config/aarch64/aarch64.c (define_peephole2 above
16875 *sub_<shift>_<mode>): New peephole.
16877 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16879 * config/i386/i386.c (make_resolver_func): Update.
16880 * Makefile.in: Add profile-count.h and profile-count.o
16881 * auto-profile.c (afdo_indirect_call): Update to new API.
16882 (afdo_set_bb_count): Update.
16883 (afdo_propagate_edge): Update.
16884 (afdo_propagate_circuit): Update.
16885 (afdo_calculate_branch_prob): Update.
16886 (afdo_annotate_cfg): Update.
16887 * basic-block.h: Include profile-count.h
16888 (struct edge_def): Turn count to profile_count.
16889 (struct basic_block_def): Likewie.
16890 (REG_BR_PROB_BASE): Move to profile-count.h
16891 (RDIV): Move to profile-count.h
16892 * bb-reorder.c (max_entry_count): Turn to profile_count.
16893 (find_traces): Update.
16894 (rotate_loop):Update.
16895 (connect_traces):Update.
16896 (sanitize_hot_paths):Update.
16897 * bt-load.c (migrate_btr_defs): Update.
16898 * cfg.c (RDIV): Remove.
16899 (init_flow): Use alloc_block.
16900 (alloc_block): Uninitialize count.
16901 (unchecked_make_edge): Uninitialize count.
16902 (check_bb_profile): Update.
16903 (dump_edge_info): Update.
16904 (dump_bb_info): Update.
16905 (update_bb_profile_for_threading): Update.
16906 (scale_bbs_frequencies_int): Update.
16907 (scale_bbs_frequencies_gcov_type): Update.
16908 (scale_bbs_frequencies_profile_count): New.
16909 * cfg.h (update_bb_profile_for_threading): Update.
16910 (scale_bbs_frequencies_profile_count): Declare.
16911 * cfgbuild.c (compute_outgoing_frequencies): Update.
16912 (find_many_sub_basic_blocks): Update.
16913 * cfgcleanup.c (try_forward_edges): Update.
16914 (try_crossjump_to_edge): Update.
16915 * cfgexpand.c (expand_gimple_tailcall): Update.
16916 (construct_exit_block): Update.
16917 * cfghooks.c (verify_flow_info): Update.
16918 (dump_bb_for_graph): Update.
16919 (split_edge): Update.
16920 (make_forwarder_block): Update.
16921 (duplicate_block): Update.
16922 (account_profile_record): Update.
16923 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
16924 (get_estimated_loop_iterations): Update.
16925 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
16926 (single_likely_exit): Update.
16927 * cfgloopmanip.c (scale_loop_profile): Update.
16929 (set_zero_probability): Update.
16930 (lv_adjust_loop_entry_edge): Update.
16931 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16932 (purge_dead_edges): Update.
16933 (rtl_account_profile_record): Update.
16934 * cgraph.c (cgraph_node::create): Uninitialize count.
16935 (symbol_table::create_edge): Uninitialize count.
16936 (cgraph_update_edges_for_call_stmt_node): Update.
16937 (cgraph_edge::dump_edge_flags): Update.
16938 (cgraph_node::dump): Update.
16939 (cgraph_edge::maybe_hot_p): Update.
16940 * cgraph.h: Include profile-count.h
16941 (create_clone), create_edge, create_indirect_edge): Update.
16942 (cgraph_node): Turn count to profile_count.
16943 (cgraph_edge0: Likewise.
16944 (make_speculative, clone): Update.
16945 (create_edge): Update.
16946 (init_lowered_empty_function): Update.
16947 * cgraphclones.c (cgraph_edge::clone): Update.
16948 (duplicate_thunk_for_node): Update.
16949 (cgraph_node::create_clone): Update.
16950 * cgraphunit.c (cgraph_node::analyze): Update.
16951 (cgraph_node::expand_thunk): Update.
16952 * final.c (dump_basic_block_info): Update.
16953 * gimple-streamer-in.c (input_bb): Update.
16954 * gimple-streamer-out.c (output_bb): Update.
16955 * graphite.c (print_global_statistics): Update.
16956 (print_graphite_scop_statistics): Update.
16957 * hsa-brig.c: Include basic-block.h.
16958 * hsa-dump.c: Include basic-block.h.
16959 * hsa-gen.c (T sum_slice): Update.
16960 (convert_switch_statements):Update.
16961 * hsa-regalloc.c: Include basic-block.h.
16962 * ipa-chkp.c (chkp_produce_thunks): Update.
16963 * ipa-cp.c (struct caller_statistics): Update.
16964 (init_caller_stats): Update.
16965 (gather_caller_stats): Update.
16966 (ipcp_cloning_candidate_p): Update.
16967 (good_cloning_opportunity_p): Update.
16968 (get_info_about_necessary_edges): Update.
16969 (dump_profile_updates): Update.
16970 (update_profiling_info): Update.
16971 (update_specialized_profile): Update.
16972 (perhaps_add_new_callers): Update.
16973 (decide_about_value): Update.
16974 (ipa_cp_c_finalize): Update.
16975 * ipa-devirt.c (struct odr_type_warn_count): Update.
16976 (struct decl_warn_count): Update.
16977 (struct final_warning_record): Update.
16978 (possible_polymorphic_call_targets): Update.
16979 (ipa_devirt): Update.
16980 * ipa-fnsummary.c (redirect_to_unreachable): Update.
16981 * ipa-icf.c (sem_function::merge): Update.
16982 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
16983 * ipa-inline.c (compute_uninlined_call_time): Update.
16984 (compute_inlined_call_time): Update.
16985 (want_inline_small_function_p): Update.
16986 (want_inline_self_recursive_call_p): Update.
16987 (edge_badness): Update.
16988 (lookup_recursive_calls): Update.
16989 (recursive_inlining): Update.
16990 (inline_small_functions): Update.
16991 (dump_overall_stats): Update.
16992 (dump_inline_stats): Update.
16993 * ipa-profile.c (ipa_profile_generate_summary): Update.
16994 (ipa_propagate_frequency): Update.
16995 (ipa_profile): Update.
16996 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
16997 * ipa-utils.c (ipa_merge_profiles): Update.
16998 * loop-doloop.c (doloop_modify): Update.
16999 * loop-unroll.c (report_unroll): Update.
17000 (unroll_loop_runtime_iterations): Update.
17001 * lto-cgraph.c (lto_output_edge): Update.
17002 (lto_output_node): Update.
17003 (input_node): Update.
17004 (input_edge): Update.
17005 (merge_profile_summaries): Update.
17006 * lto-streamer-in.c (input_cfg): Update.
17007 * lto-streamer-out.c (output_cfg): Update.
17008 * mcf.c (create_fixup_graph): Update.
17009 (adjust_cfg_counts): Update.
17010 (sum_edge_counts): Update.
17011 * modulo-sched.c (sms_schedule): Update.
17012 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
17013 * predict.c (maybe_hot_count_p): Update.
17014 (probably_never_executed): Update.
17015 (dump_prediction): Update.
17016 (combine_predictions_for_bb): Update.
17017 (propagate_freq): Update.
17018 (handle_missing_profiles): Update.
17019 (counts_to_freqs): Update.
17020 (rebuild_frequencies): Update.
17021 (force_edge_cold): Update.
17022 * predict.h: Include profile-count.h
17023 (maybe_hot_count_p, counts_to_freqs): UPdate.
17024 * print-rtl-function.c: Do not include cfg.h
17025 * print-rtl.c: Include basic-block.h
17026 * profile-count.c: New file.
17027 * profile-count.h: New file.
17028 * profile.c (is_edge_inconsistent): Update.
17029 (correct_negative_edge_counts): Update.
17030 (is_inconsistent): Update.
17031 (set_bb_counts): Update.
17032 (read_profile_edge_counts): Update.
17033 (compute_frequency_overlap): Update.
17034 (compute_branch_probabilities): Update; Initialize and deinitialize
17036 (branch_prob): Update.
17037 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
17038 (edge_gcov_count): New.
17039 (bb_gcov_count): New.
17040 * shrink-wrap.c (try_shrink_wrapping): Update.
17041 * tracer.c (better_p): Update.
17042 * trans-mem.c (expand_transaction): Update.
17043 (ipa_tm_insert_irr_call): Update.
17044 (ipa_tm_insert_gettmclone_call): Update.
17045 * tree-call-cdce.c: Update.
17046 * tree-cfg.c (gimple_duplicate_sese_region): Update.
17047 (gimple_duplicate_sese_tail): Update.
17048 (gimple_account_profile_record): Update.
17049 (execute_fixup_cfg): Update.
17050 * tree-inline.c (copy_bb): Update.
17051 (copy_edges_for_bb): Update.
17052 (initialize_cfun): Update.
17053 (freqs_to_counts): Update.
17054 (copy_cfg_body): Update.
17055 (expand_call_inline): Update.
17056 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
17057 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
17058 (try_unroll_loop_completely): Update.
17059 (try_peel_loop): Update.
17060 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17061 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
17062 * tree-ssa-loop-split.c (connect_loops): Update.
17063 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
17064 * tree-ssa-reassoc.c (branch_fixup): Update.
17065 * tree-ssa-tail-merge.c (replace_block_by): Update.
17066 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
17067 (compute_path_counts): Update.
17068 (update_profile): Update.
17069 (recompute_probabilities): Update.
17070 (update_joiner_offpath_counts): Update.
17071 (estimated_freqs_path): Update.
17072 (freqs_to_counts_path): Update.
17073 (clear_counts_path): Update.
17074 (ssa_fix_duplicate_block_edges): Update.
17075 (duplicate_thread_path): Update.
17076 * tree-switch-conversion.c (case_bit_test_cmp): Update.
17077 (struct switch_conv_info): Update.
17078 * tree-tailcall.c (decrease_profile): Update.
17079 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
17080 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17081 * value-prof.c (check_counter): Update.
17082 (gimple_divmod_fixed_value): Update.
17083 (gimple_mod_pow2): Update.
17084 (gimple_mod_subtract): Update.
17085 (gimple_ic_transform): Update.
17086 (gimple_stringop_fixed_value): Update.
17087 * value-prof.h (gimple_ic): Update.
17089 2017-06-02 Carl Love <cel@us.ibm.com>
17091 * config/rs6000/rs6000-c: Add support for built-in functions
17092 vector double vec_doublee (vector signed int);
17093 vector double vec_doublee (vector unsigned int);
17094 vector double vec_doublee (vector float);
17095 vector double vec_doubleh (vector signed int);
17096 vector double vec_doubleh (vector unsigned int);
17097 vector double vec_doubleh (vector float);
17098 vector double vec_doublel (vector signed int);
17099 vector double vec_doublel (vector unsigned int);
17100 vector double vec_doublel (vector float);
17101 vector double vec_doubleo (vector signed int);
17102 vector double vec_doubleo (vector unsigned int);
17103 vector double vec_doubleo (vector float);.
17104 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
17105 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
17107 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
17108 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
17109 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
17111 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
17112 vec_doublel, vec_doubleh.
17113 * doc/extend.texi: Update the built-in documentation file for the
17114 new built-in functions.
17116 2017-06-02 David Malcolm <dmalcolm@redhat.com>
17119 * ipa-inline-analysis.c (free_growth_caches): Set
17120 edge_removal_hook_holder to NULL after removing it.
17122 2017-06-02 Sudakshina Das <sudi.das@arm.com>
17124 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
17125 comparision with zero.
17127 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
17128 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
17129 for early expansion of vec_min and vec_max builtins.
17130 (builtin_function_type): Add min/max unsigned variants to those
17131 identified as having unsigned arguments.
17133 2017-06-02 Olivier Hainque <hainque@adacore.com>
17135 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
17137 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17139 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
17140 Use VALL_F16 iterator rather than VALL.
17142 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17144 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
17145 Emit CBNZ inside loop when doing a strong exchange and comparing
17146 against zero. Generate the CC flags after the loop.
17148 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
17150 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
17151 (dl_section_ref): New.
17152 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
17153 On AIX, append an expression to subtract the size of the
17154 section length to dl_section_ref.
17156 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
17158 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
17159 for early expansion of vector absolute builtins.
17161 2017-06-02 Richard Biener <rguenther@suse.de>
17163 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
17164 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
17166 2017-06-02 Richard Biener <rguenther@suse.de>
17168 PR tree-optimization/80948
17169 * tree-tailcall.c (find_tail_calls): Track stmts to move in
17170 stmt order as well.
17172 2017-06-02 Richard Biener <rguenther@suse.de>
17174 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
17176 * tree-vect-stmts.c (process_use): Do not mark backedge defs
17177 for inductions as relevant.
17179 2017-06-02 Richard Biener <rguenther@suse.de>
17181 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
17182 (vectorizable_induction): ... this. Remove dead code.
17184 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
17186 * builtins. (expand_builtin_alloca): Remove second parameter and
17187 infer its value from the first parameter instead.
17188 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
17190 2017-06-02 Jakub Jelinek <jakub@redhat.com>
17192 PR rtl-optimization/80903
17193 * loop-doloop.c (add_test): Unshare sequence.
17195 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
17197 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
17199 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
17201 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
17203 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
17204 xlogue_layout::get_instance, logue_layout::xlogue_layout,
17205 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
17206 (xlogue_layout::get_stub_rtx): Make static.
17207 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
17208 (xlogue_layout::compute_stub_managed_regs): Rename to...
17209 (xlogue_layout::count_stub_managed_regs): ...this.
17210 (xlogue_layout::is_stub_managed_reg): New function.
17211 (xlogue_layout::m_stub_names): Rename to...
17212 (xlogue_layout::s_stub_names): ...this, make static.
17213 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
17214 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
17215 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
17216 xlogue_layout::s_stub_names): Instantiate statics.
17217 (stub_managed_regs): Remove.
17218 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
17219 (disable_call_ms2sysv_xlogues): Rename to...
17220 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
17221 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
17223 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
17224 change after reload_completed.
17225 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
17227 (ix86_expand_prologue): Likewise.
17228 (ix86_expand_epilogue): Likewise.
17229 (ix86_expand_split_stack_prologue): Likewise.
17230 (ix86_compute_frame_layout): Remove frame parameter ...
17231 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
17232 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
17234 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
17235 (ix86_frame): Move from here ...
17236 * config/i386/i386.h (ix86_frame): ... to here.
17237 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
17238 complete ix86_frame data structure instead. Remove some_ld_name.
17240 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
17242 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
17243 symbols that hold a DECL_VALUE_EXPR.
17245 2017-06-01 Martin Jambor <mjambor@suse.cz>
17247 PR tree-optimization/80898
17248 * tree-sra.c (process_subtree_disqualification): Removed.
17249 (disqualify_candidate): Do not acll
17250 process_subtree_disqualification.
17251 (subtree_mark_written_and_enqueue): New function.
17252 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
17253 RHS has been disqualified and re-queue LHS if necessary. Apart
17254 from that, ignore disqualified RHS.
17256 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17258 * config/s390/s390.c (s390_emit_epilogue): Disable early return
17259 address fetch for z10 or later.
17261 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17263 * config/arc/arc.md (tst_movb): Add guard when splitting.
17265 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17267 * config/arc/arc.c (arc_can_eliminate): Test against
17268 arc_frame_pointer_needed.
17270 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17272 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
17273 to prevent store reordering.
17274 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
17275 (type): Add block type.
17276 (stack_tie): Define special instruction to be used in
17279 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17281 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
17282 constraint. It is not valid for the pattern.
17283 (noncommutative_binary_comparison): Likewise.
17285 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17287 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
17290 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17292 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
17293 be used by the reg-alloc.
17295 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17297 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
17298 reg-alloc when having mul64 or mul32x16 instructions.
17299 (mulsidi3): Likewise.
17300 (umulsidi3): Likewise.
17301 (mulsi32x16): New pattern.
17302 (mulsi64): Likewise.
17303 (mulsidi64): Likewise.
17304 (umulsidi64): Likewise.
17305 (MUL32x16_REG): Define.
17306 (mul64_600): Use MUL32x16_REG.
17307 (mac64_600): Likewise.
17308 (umul64_600): Likewise.
17309 (umac64_600): Likewise.
17311 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17313 * config/arc/arc.md (mulsi3_700): Make it commutative.
17315 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
17317 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
17319 (*sign_extendsidi2_insn): Likewise for movstosw.
17321 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
17323 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
17324 the type of the input discriminant value. Convert the
17325 discriminant value of signedness vary.
17327 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
17329 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
17330 Add to -Wall section.
17332 2017-06-01 Richard Biener <rguenther@suse.de>
17334 PR middle-end/66313
17335 * fold-const.c (fold_plusminus_mult_expr): If the factored
17336 factor may be zero use a wrapping type for the inner operation.
17337 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
17338 and handle moved defs.
17339 (process_assignment): Properly guard the unary op case. Return a
17340 tri-state indicating that moving the stmt before the call may allow
17341 to continue. Pass through to_move.
17342 (find_tail_calls): Handle moving unrelated defs before
17345 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
17348 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
17349 splitter result in the canonical way.
17351 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
17353 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
17354 also for 32bit target. Update insn attributes.
17355 (zero-extendsidi2 splitter): Allow all registers for operand 1.
17357 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
17359 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
17360 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
17361 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
17362 (_mm_maskz_min_ss): New intrinsics.
17364 2017-05-31 Martin Liska <mliska@suse.cz>
17366 * tree-vect-loop.c (vect_create_epilog_for_reduction):
17367 Change comment style to one we normally use.
17368 (vectorizable_reduction): Likewise.
17369 (vectorizable_induction): Likewise.
17370 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
17371 (vectorizable_call): Likewise.
17372 (vectorizable_simd_clone_call): Likewise.
17373 (vectorizable_conversion): Likewise.
17374 (vectorizable_assignment): Likewise.
17375 (vectorizable_shift): Likewise.
17376 (vectorizable_operation): Likewise.
17377 (vectorizable_store): Likewise.
17378 (vectorizable_load): Likewise.
17379 * tree-vectorizer.h: Likewise.
17381 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
17383 * passes.c (emergency_dump_function): New.
17384 * tree-pass.h (emergency_dump_function): Declare.
17385 * plugin.c (plugins_internal_error_function): Remove.
17386 * plugin.h (plugins_internal_error_function): Remove declaration.
17387 * toplev.c (internal_error_function): New static function. Use it...
17388 (general_init): ...here.
17390 2017-05-31 Graham Markall <graham.markall@embecosm.com>
17392 * config/arc/arc.c (arc_print_operand): Handle constant operands.
17393 (arc_rtx_costs): Add costs for new patterns.
17394 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
17395 * config/arc/predicates.md: Add _1_2_3_operand predicate.
17397 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
17399 * tree-ssa-strlen.c (get_next_strinfo): New function.
17400 (get_stridx_plus_constant): Use it.
17401 (zero_length_string): Likewise.
17402 (adjust_related_strinfos): Likewise.
17403 (adjust_last_stmt): Likewise.
17405 2017-05-31 Richard Biener <rguenther@suse.de>
17408 * config/i386/i386.c (ix86_expand_builtin): Remove assert
17409 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
17411 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
17413 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
17414 loop_vinfo argument and use of dependence distance vectors.
17415 Check instead whether the two references differ only in their
17416 initial value and assume that they have the same alignment if the
17417 difference is a multiple of the vector alignment.
17418 (vect_analyze_data_refs_alignment): Update call accordingly.
17420 2017-05-31 Martin Liska <mliska@suse.cz>
17423 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
17425 2017-05-31 Bin Cheng <bin.cheng@arm.com>
17427 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
17428 (create_intersect_range_checks): Move from ...
17429 * tree-data-ref.c (create_intersect_range_checks_index)
17430 (create_intersect_range_checks): ... to here.
17431 (create_runtime_alias_checks): New function factored from ...
17432 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
17433 here. Call above function.
17434 * tree-data-ref.h (create_runtime_alias_checks): New function.
17436 2017-05-31 Bin Cheng <bin.cheng@arm.com>
17438 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
17439 segment length for dr_b and compute it in wide_int.
17441 2017-05-31 Richard Biener <rguenther@suse.de>
17443 PR tree-optimization/80906
17444 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
17445 and pass through iv_map.
17446 (copy_bb_and_scalar_dependences): Adjust.
17447 (translate_pending_phi_nodes): Likewise.
17448 (copy_loop_close_phi_args): Handle code-generating IVs instead
17451 2017-05-30 David Malcolm <dmalcolm@redhat.com>
17453 * diagnostic-color.c (color_dict): Add "type-diff".
17454 (parse_gcc_colors): Update comment.
17455 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
17456 -fdiagnostics-show-template-tree and -fno-elide-type.
17457 (GCC_COLORS): Add type-diff to example.
17459 (-fdiagnostics-show-template-tree): New.
17460 (-fno-elide-type): New.
17461 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
17462 the pp_format_decoder callback. Call any m_format_postprocessor's
17464 (pretty_printer::pretty_printer): Initialize
17465 m_format_postprocessor.
17466 (pretty_printer::~pretty_printer): Delete any
17467 m_format_postprocessor.
17468 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
17469 (class format_postprocessor): New class.
17470 (struct pretty_printer::format_decoder): Document the new parameters.
17471 (struct pretty_printer::m_format_postprocessor): New field.
17472 * tree-diagnostic.c (default_tree_printer): Update for new
17473 bool and const char ** params.
17474 * tree-diagnostic.h (default_tree_printer): Likewise.
17476 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
17478 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
17479 (lwa_operand): Delete rs6000_gen_cell_microcode test.
17480 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
17481 rs6000_gen_cell_microcode code.
17482 (rs6000_final_prescan_insn): Delete.
17483 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
17484 "warn-cell-microcode" entries.
17485 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
17486 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
17487 throughout. Change cc_reg_not_micro_cr0_operand to
17488 cc_reg_not_cr0_operand throughout.
17489 (*extendhi<mode>2_noload): Delete.
17490 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
17491 (mwarn-cell-microcode): Delete.
17492 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
17493 -mgen-cell-microcode and -mwarn-cell-microcode.
17495 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
17498 * config/i386/constraints.md (Yd): New constraint.
17500 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
17501 and (?Yd, r) alternatives. Update insn attributes.
17502 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
17503 and (?*Yd, r) alternatives. Update insn attributes.
17504 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
17506 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
17508 * gimplify.c (gimplify_modify_expr): Don't create a
17509 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
17512 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
17514 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
17516 2017-05-30 Richard Biener <rguenther@suse.de>
17518 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
17519 and reduc_def fields.
17520 (STMT_VINFO_REDUC_TYPE): New define.
17521 (STMT_VINFO_REDUC_DEF): Likewise.
17522 (vect_force_simple_reduction): Adjust prototype.
17523 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
17524 (vect_is_simple_reduction): Remove check_reduction argument.
17525 (vect_force_simple_reduction): Adjust and set
17526 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
17527 (vectorizable_reduction): Do not re-do reduction analysis
17528 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
17529 * tree-parloops.c (gather_scalar_reductions): Adjust.
17531 2017-05-30 Richard Biener <rguenther@suse.de>
17533 PR middle-end/80901
17534 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
17537 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
17539 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
17540 Introduce unknown_misalignment parameter and remove vf.
17541 (vect_peeling_hash_get_lowest_cost):
17542 Pass unknown_misalignment parameter.
17543 (vect_enhance_data_refs_alignment):
17544 Fix unsupportable data ref treatment.
17546 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17548 * tree-vect-data-refs.c (vect_get_data_access_cost):
17549 Workaround for SLP handling.
17550 (vect_enhance_data_refs_alignment):
17551 Compute costs for doing no peeling at all, compare to the best
17552 peeling costs so far and avoid peeling if cheaper.
17554 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17556 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
17557 Return peeling info and set costs to zero for unlimited cost
17559 (vect_enhance_data_refs_alignment): Also inspect all datarefs
17560 with unknown misalignment. Compute and costs for unknown
17561 misalignment, compare them to the costs for known misalignment
17562 and choose the cheapest for peeling.
17564 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17566 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
17567 (vect_get_peeling_costs_all_drs): Create function.
17568 (vect_peeling_hash_get_lowest_cost):
17569 Use vect_get_peeling_costs_all_drs.
17570 (vect_peeling_supportable): Create function.
17571 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
17573 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17575 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
17576 DR_HAS_NEGATIVE_STEP.
17577 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
17578 (vect_enhance_data_refs_alignment): Use.
17579 (vect_duplicate_ssa_name_ptr_info): Use.
17580 * tree-vectorizer.h (dr_misalignment): Use.
17581 (known_alignment_for_access_p): Use.
17583 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17586 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
17588 (has_section_name): New function.
17590 2017-05-30 Martin Liska <mliska@suse.cz>
17593 * auto-profile.c (get_function_decl_from_block): Fix
17596 2017-05-30 Richard Biener <rguenther@suse.de>
17598 PR middle-end/80876
17599 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
17601 2017-05-30 Martin Liska <mliska@suse.cz>
17603 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
17604 * dumpfile.h (struct dump_file_info): Remove ctors.
17606 2017-05-30 Martin Liska <mliska@suse.cz>
17608 * predict.def: Fix GNU coding style.
17610 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
17612 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
17613 Mark 'to' argument with ATTRIBUTE_UNUSED.
17615 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
17617 * config/xtensa/xtensa.c (xtensa_emit_call): Use
17618 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
17619 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
17622 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
17624 * doc/install.texi (Options specification): Restore entry of
17625 --enable-sjlj-exceptions.
17627 2017-05-27 Michael Eager <eager@eagercon.com>
17630 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
17632 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
17634 * config/microblaze/microblaze.h
17635 (FIXED_REGISTERS): Update in macro.
17636 (CALL_USED_REGISTERS): Update in macro.
17638 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
17640 * doc/install.texi: Add links to macOS binary distributions.
17642 2017-05-27 Jakub Jelinek <jakub@redhat.com>
17646 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17648 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
17650 2017-05-26 Martin Liska <mliska@suse.cz>
17652 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
17654 2017-05-26 Martin Liska <mliska@suse.cz>
17656 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
17657 always leading ';; '.
17658 (dump_bb_info): Likewise.
17659 (brief_dump_cfg): Likewise.
17660 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
17661 * dumpfile.c: Remove usage of TDF_VERBOSE.
17662 * dumpfile.h (enum dump_kind): Likewise.
17663 (dump_gimple_bb_header): Do not use TDF_COMMENT.
17664 * print-tree.c (debug_verbose): Remove.
17665 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
17666 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
17667 * tree-diagnostic.c (default_tree_printer): Replace
17668 TDF_DIAGNOSTIC with TDF_SLIM.
17670 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17672 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
17673 in parameter loop, rather than loop_vinfo.
17674 (create_intersect_range_checks): Ditto.
17675 (vect_create_cond_for_alias_checks): Update call to above functions.
17677 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17679 PR tree-optimization/80815
17680 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
17681 for merging runtime alias checks. Handle negative DR_STEPs.
17683 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17685 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
17687 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
17688 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
17689 out code pruning runtime alias checks.
17690 * tree-data-ref.c (prune_runtime_alias_test_list): New function
17691 factored out from above.
17692 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
17694 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
17696 (prune_runtime_alias_test_list): New decalaration.
17698 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17700 * tree-vect-data-refs.c (compare_tree): Rename and move ...
17701 * tree-data-ref.c (data_ref_compare_tree): ... to here.
17702 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
17703 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
17704 (operator==, comp_dr_with_seg_len_pair): Ditto.
17705 (vect_prune_runtime_alias_test_list): Ditto.
17707 2017-05-26 Martin Liska <mliska@suse.cz>
17710 * params.def: Bound partial-inlining-entry-probability param.
17712 2017-05-26 Marek Polacek <polacek@redhat.com>
17715 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
17718 2017-05-26 Richard Biener <rguenther@suse.de>
17720 PR tree-optimization/80842
17721 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
17724 2017-05-26 Richard Biener <rguenther@suse.de>
17726 PR tree-optimization/80844
17727 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
17729 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
17731 * doc/md.texi (Machine Constraints): Update x86 family
17732 machine constraints section to match 'config/i386/constraints.md'.
17734 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
17736 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
17738 2017-05-25 Nathan Sidwell <nathan@acm.org>
17740 * doc/invoke.texi (--enable-languages): Update documentation.
17742 2017-05-25 Martin Liska <mliska@suse.cz>
17744 * dumpfile.c: Add TDF_FOLDING.
17745 * dumpfile.h (enum dump_kind): Likewise.
17746 * genmatch.c (dt_simplify::gen_1): Use it.
17748 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17750 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
17752 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17754 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
17755 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
17757 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17759 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
17760 * match.pd (X == C): Rewrite it here.
17761 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
17762 with_certain_nonzero_bits2): New predicates.
17763 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
17765 2017-05-24 Nathan Sidwell <nathan@acm.org>
17767 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
17770 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
17773 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
17775 * config/powerpcspe: New port. Files are copied from the rs6000
17776 port, with "rs6000" in filenames replaced by "powerpcspe".
17778 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
17780 PR rtl-optimization/80754
17781 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
17783 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
17785 * config/sparc/sparc.md (length): Return the correct value for -mflat
17786 sibcalls to match output_sibcall.
17788 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
17792 * config/rs6000/rs6000.c (struct machine_function): Add new field
17794 (rs6000_get_separate_components): Init that field, use it.
17795 (rs6000_components_for_bb): Use the field.
17797 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
17799 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
17801 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
17803 PR middle-end/80823
17804 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
17806 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17809 * config/s390/s390.c (s390_check_qrst_address): Check incoming
17810 address against address_operand predicate.
17811 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
17813 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
17815 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
17816 parameters passed indirectly.
17818 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
17820 * config/i386/i386.md (*movdi_internal): Remove SSE4
17821 alternative 18 (?r, *v). Update insn attributes.
17822 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
17823 Update insn attributes.
17824 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
17825 Update insn attributes.
17826 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
17827 alternative 1 (r, v). Remove isa attribute.
17828 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
17829 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
17830 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
17832 2017-05-23 Tom de Vries <tom@codesourcery.com>
17834 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
17837 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17839 * cgraphunit.c (symbol_table::process_new_functions): Update.
17840 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
17841 (inline_generate_summary): Rename to ...
17842 (ipa_fn_summary_generate): ... this one.
17843 (inline_read_summary): Rename to ...
17844 (ipa_fn_summary_read): ... this one.
17845 (inline_write_summary): Rename to ...
17846 (ipa_fn_summary_write): ... this one.
17847 (inline_free_summary): Rename to ...
17848 (ipa_free_fn_summary): ... this one.
17849 (pass_data_local_fn_summary, pass_local_fn_summary,
17850 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
17851 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
17852 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
17853 make_pass_ipa_fn_summary): New.
17854 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
17855 inline_write_summary, inline_free_summary): Remove.
17856 (ipa_free_fn_summary) : New.
17857 * ipa-inline.c (ipa_inline): Update.
17858 (pass_ipa_inline): Do not generate summaries.
17859 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
17861 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
17862 and add pass_ipa_fn_summary.
17863 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
17865 (make_pass_inline_parameters): Remove.
17867 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
17869 * omp-low.c (struct omp_context): Remove "default_kind" member.
17872 * omp-offload.c (execute_oacc_device_lower): Remove the
17873 parallelism dimensions function attributes for unparallelized
17874 OpenACC kernels constructs.
17876 2017-05-23 Martin Liska <mliska@suse.cz>
17878 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
17880 (cgraph_edge::make_speculative): Likewise.
17881 (cgraph_edge::resolve_speculation): Likewise.
17882 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
17883 (cgraph_node::dump): Likewise.
17884 * cgraph.h: Likewise.
17885 * cgraphunit.c (analyze_functions): Likewise.
17886 (symbol_table::compile): Likewise.
17887 * ipa-cp.c (print_all_lattices): Likewise.
17888 (determine_versionability): Likewise.
17889 (initialize_node_lattices): Likewise.
17890 (ipcp_verify_propagated_values): Likewise.
17891 (estimate_local_effects): Likewise.
17892 (update_profiling_info): Likewise.
17893 (create_specialized_node): Likewise.
17894 (perhaps_add_new_callers): Likewise.
17895 (decide_about_value): Likewise.
17896 (decide_whether_version_node): Likewise.
17897 (identify_dead_nodes): Likewise.
17898 (ipcp_store_bits_results): Likewise.
17899 * ipa-devirt.c (dump_targets): Likewise.
17900 (ipa_devirt): Likewise.
17901 * ipa-icf.c (sem_item::dump): Likewise.
17902 (sem_function::equals): Likewise.
17903 (sem_variable::equals): Likewise.
17904 (sem_item_optimizer::read_section): Likewise.
17905 (sem_item_optimizer::execute): Likewise.
17906 (congruence_class::dump): Likewise.
17907 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
17908 (dump_inline_summary): Likewise.
17909 (estimate_node_size_and_time): Likewise.
17910 (inline_analyze_function): Likewise.
17911 * ipa-inline-transform.c (inline_call): Likewise.
17912 * ipa-inline.c (report_inline_failed_reason): Likewise.
17913 (want_early_inline_function_p): Likewise.
17914 (edge_badness): Likewise.
17915 (update_edge_key): Likewise.
17916 (inline_small_functions): Likewise.
17917 * ipa-profile.c (ipa_profile): Likewise.
17918 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
17919 (ipa_make_edge_direct_to_target): Likewise.
17920 (remove_described_reference): Likewise.
17921 (ipa_impossible_devirt_target): Likewise.
17922 (propagate_controlled_uses): Likewise.
17923 (ipa_print_node_params): Likewise.
17924 (ipcp_transform_function): Likewise.
17925 * ipa-pure-const.c (pure_const_read_summary): Likewise.
17926 (propagate_pure_const): Likewise.
17927 * ipa-reference.c (generate_summary): Likewise.
17928 (read_write_all_from_decl): Likewise.
17929 (propagate): Likewise.
17930 (ipa_reference_read_optimization_summary): Likewise.
17931 * ipa-utils.c (ipa_merge_profiles): Likewise.
17932 * ipa.c (walk_polymorphic_call_targets): Likewise.
17933 (symbol_table::remove_unreachable_nodes): Likewise.
17934 (ipa_single_use): Likewise.
17935 * passes.c (execute_todo): Likewise.
17936 * predict.c (drop_profile): Likewise.
17937 * symtab.c (symtab_node::get_dump_name): New function.
17938 (symtab_node::dump_name): Likewise.
17939 (symtab_node::dump_asm_name): Likewise.
17940 (symtab_node::dump_references): Likewise.
17941 (symtab_node::dump_referring): Likewise.
17942 (symtab_node::dump_base): Likewise.
17943 (symtab_node::debug_symtab): Likewise.
17944 * tree-sra.c (convert_callers_for_node): Likewise.
17945 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
17946 * value-prof.c (init_node_map): Likewise.
17948 2017-05-23 Martin Liska <mliska@suse.cz>
17950 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
17951 and symtab_node::debug_symtab to symbol_table::debug.
17952 * cgraphunit.c (analyze_functions): Use the renamed function.
17953 (symbol_table::compile): Likewise.
17954 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
17955 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
17956 * passes.c (execute_todo): Likewise.
17957 * symtab.c (symbol_table::dump): New function.
17958 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
17960 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17962 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
17963 that nonconst implies exec.
17965 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17967 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
17968 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
17969 (inline_edge_summary_vec): Turn into ...
17970 (ipa_call_summaries): ... this one.
17971 (redirect_to_unreachable, edge_set_predicate,
17972 evaluate_properties_for_edge, inline_summary_alloc,
17973 reset_ipa_call_summary, reset_inline_summary,
17974 inline_summary_t::duplicate): Update.
17975 (inline_edge_duplication_hook): Turn to ...
17976 (ipa_call_summary_t::duplicate): ... this one.
17977 (inline_edge_removal_hook): Turn to ...
17978 (ipa_call_summary_t::remove): ... this one.
17979 (dump_inline_edge_summary): Turn to ...
17980 (dump_ipa_call_summary): ... this one.
17981 (estimate_function_body_sizes): Update.
17982 (inline_update_callee_summaries): Update.
17983 (remap_edge_change_prob): Update.
17984 (remap_edge_summaries): Update.
17985 (inline_merge_summary): Update.
17986 (do_estimate_edge_time): Update.
17987 (inline_generate_summary): Update.
17988 (inline_read_section): Update.
17989 (inline_read_summary): Update.
17990 (inline_free_summary): Update.
17991 * ipa-inline.c (can_inline_edge_p): Update.
17992 (compute_inlined_call_time): Update.
17993 (want_inline_small_function_p): Update.
17994 (edge_badness): Update.
17995 (early_inliner): Update.
17996 * ipa-inline.h (inline_edge_summary): Turn to ...
17997 (ipa_call_summary): ... this one.
17998 (ipa_call_summary_t): New class.
17999 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
18000 (ipa_call_summaries): New.
18001 (inline_edge_summary): Remove.
18002 (estimate_edge_growth): Update.
18003 * ipa-profile.c (ipa_propagate_frequency_1): Update.
18004 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
18005 * ipa-split.c (execute_split_functions): Update.
18006 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
18008 2017-05-23 Tom de Vries <tom@codesourcery.com>
18010 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
18011 attributes): Document rdrand effective target.
18013 2017-05-23 Tom de Vries <tom@codesourcery.com>
18015 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
18016 attributes): Sort alphabetically.
18018 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
18020 * config/avr/genmultilib.awk: Use gsub instead of gensub.
18022 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
18025 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
18026 V2DF/V2DI splat into two separate patterns, one that handles
18027 registers, and the other that only handles memory. Drop support
18028 for splatting from a GPR on ISA 2.07 and then splitting the
18029 splat into direct move and splat.
18030 (vsx_splat_<mode>_reg): Likewise.
18031 (vsx_splat_<mode>_mem): Likewise.
18033 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
18035 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
18037 2017-05-22 Jakub Jelinek <jakub@redhat.com>
18039 PR middle-end/80809
18040 * omp-low.c (finish_taskreg_remap): New function.
18041 (finish_taskreg_scan): If unit size of ctx->record_type
18042 is non-constant, unshare the size expression and replace
18043 decls in it with possible outer var refs.
18045 PR middle-end/80809
18046 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
18047 GOVD_SHARED rather than GOVD_PRIVATE with it.
18048 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
18049 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
18051 PR middle-end/80853
18052 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
18053 as last argument to build_outer_var_ref for pointer bases of array
18054 section reductions.
18056 2017-05-19 Martin Sebor <msebor@redhat.com>
18058 * print-tree.c (print_node): Print DECL_READ_P flag.
18060 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18062 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
18063 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
18064 * cgraph.c: Likewise.
18065 * cgraphunit.c: Likewise.
18066 * gengtype.c: Likewise.
18067 * ipa-cp.c: Likewise.
18068 * ipa-devirt.c: Likewise.
18069 * ipa-icf.c: Likewise.
18070 * ipa-predicate.c: Likewise.
18071 * ipa-profile.c: Likewise.
18072 * ipa-prop.c: Likewise.
18073 * ipa-split.c: Likewise.
18075 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
18076 edge_predicate_pool, dump_inline_hints,
18077 inline_summary::account_size_time, redirect_to_unreachable,
18078 edge_set_predicate, set_hint_predicate,
18079 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
18080 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
18081 inline_summary_t::remove, remap_hint_predicate_after_duplication,
18082 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
18083 ipa_call_summary_t::remove, initialize_growth_caches,
18084 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
18085 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
18086 mark_modified, unmodified_parm_1, unmodified_parm,
18087 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
18088 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
18089 compute_bb_predicates, will_be_nonconstant_expr_predicate,
18090 will_be_nonconstant_predicate, record_modified_bb_info,
18091 get_minimal_bb, record_modified, param_change_prob,
18092 phi_result_unknown_predicate, predicate_for_phi_result,
18093 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
18094 estimate_function_body_sizes, compute_inline_parameters,
18095 compute_inline_parameters_for_curren, pass_data_inline_parameters,
18096 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
18097 inline_update_callee_summaries, remap_edge_change_prob,
18098 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
18099 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
18100 inline_analyze_function, inline_summary_t::insert,
18101 inline_generate_summary, read_ipa_call_summary, inline_read_section,
18102 inline_read_summary, write_ipa_call_summary, inline_write_summary,
18103 inline_free_summary): Move to ipa-fnsummary.h
18104 (predicate_t): Remove.
18105 * ipa-fnsummary.c: New file.
18106 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
18107 (enum inline_hints_vals, inline_hints, agg_position_info,
18108 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
18109 inline_summaries, ipa_call_summary, ipa_call_summary_t,
18110 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
18111 dump_inline_summary, dump_inline_hints, inline_generate_summary,
18112 inline_read_summary, inline_write_summary, inline_free_summary,
18113 inline_analyze_function, initialize_inline_failed,
18114 inline_merge_summary, inline_update_overall_summary,
18115 compute_inline_parameters): Move to ipa-fnsummary.h
18116 * ipa-fnsummary.h: New file.
18117 * ipa-inline-transform.h: Include ipa-inline.h.
18118 * ipa-inline.c: LIkewise.
18120 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18122 * ipa-inline.c (edge_badness): Use inlined_time instead of
18123 inline_summaries->get.
18125 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18127 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
18129 2017-05-22 Nathan Sidwell <nathan@acm.org>
18131 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
18132 (fdump-lang): Document 'raw' option.
18133 * dumpfile.h (TDI_tu): Delete.
18134 * dumpfile.c (dump_files): Remove translation-unit.
18135 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
18137 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
18139 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
18140 command option from $(AWK) call.
18141 * config/avr/genmultilib.awk: Simplify and rewrite so that it
18142 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
18143 [FORMAT]: Remove handling of variable.
18144 * config/avr/t-multilib: Regenerate.
18146 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18148 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
18150 (dump_inline_summary): Do not print self_time.
18151 (estimate_function_body_sizes): Do not set self_time.
18152 (compute_inline_parameters): Likewise.
18153 (inline_read_section, inline_write_summary): Do not stream self_time.
18154 * ipa-inline.h (inline_summary): Drop self_time.
18156 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18158 * ipa-inline-analysis.c (account_size_time): Rename to ...
18159 (inline_summary::account_size_time): ... this one.
18160 (reset_ipa_call_summary): Turn to ...
18161 (ipa_call_summary::reset): ... this one.
18162 (reset_inline_summary): Turn to ...
18163 (inline_summary::reset): ... this one.
18164 (inline_summary_t::remove): Update.
18165 (inline_summary_t::duplicate): Update.
18166 (ipa_call_summary_t::remove): Update.
18167 (dump_inline_summary): Update.
18168 (estimate_function_body_sizes): Update.
18169 (compute_inline_parameters): Update.
18170 (estimate_node_size_and_time): Update.
18171 (inline_merge_summary): Update.
18172 (inline_update_overall_summary): Update.
18173 (inline_read_section): Update.
18174 (inline_write_summary): Update.
18175 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
18176 add account_size_time and reset member functions.
18177 (ipa_call_summary): Add reset function.
18178 * ipa-predicate.h (predicate::operator &): Constify.
18180 2017-05-22 Richard Biener <rguenther@suse.de>
18182 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
18184 2017-05-19 Jason Merrill <jason@redhat.com>
18186 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
18188 2017-05-19 Marek Polacek <polacek@redhat.com>
18191 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
18192 TYPE_OVERFLOW_WRAPS checks.
18194 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
18196 * tree-core.h (enum omp_clause_default_kind): Add
18197 "OMP_CLAUSE_DEFAULT_PRESENT".
18198 * tree-pretty-print.c (dump_omp_clause): Handle it.
18199 * gimplify.c (enum gimplify_omp_var_data): Add
18200 "GOVD_MAP_FORCE_PRESENT".
18201 (gimplify_adjust_omp_clauses_1): Map it to
18202 "GOMP_MAP_FORCE_PRESENT".
18203 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
18205 * gimplify.c (oacc_default_clause): Clarify.
18207 2017-05-19 Nathan Sidwell <nathan@acm.org>
18209 LANG_HOOK_REGISTER_DUMPS
18210 * toplev.c (general_init): Call register dump lang hook.
18211 * doc/invoke.texi: Document -fdump-lang option family.
18212 * dumpfile.c (dump_files): Remove class dump here.
18213 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
18214 * dumpfile.h (tree_dump_index): Remove TDI_class.
18215 * langhooks-def.h (lhd_register_dumps): Declare.
18216 (LANG_HOOKS_REGISTER_DUMPS): Define.
18217 (LANG_HOOKS_INITIALIZER): Add it.
18218 * langhooks.c (lhd_register_dumps): Define.
18219 * langhooks.h (struct lang_hooks): Add register_dumps.
18221 2017-05-19 Nathan Sidwell <nathan@acm.org>
18223 * context.h (context::set_passes): New.
18224 * context.c (context::context): Do not create pass manager.
18225 * toplev.c (general_init): Create pass manager here.
18227 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
18229 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
18230 use this splitter if two add or or instructions would also work for
18231 the constant we want to generate.
18233 2017-05-19 Richard Biener <rguenther@suse.de>
18236 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
18237 predicate evaluation.
18239 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
18241 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
18243 * ipa-inline.c (want_inline_small_function_p): Do not cast to
18246 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
18248 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
18249 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
18250 (inline_edge_summary_vec): Turn into ...
18251 (ipa_call_summaries): ... this one.
18252 (redirect_to_unreachable, edge_set_predicate,
18253 evaluate_properties_for_edge, inline_summary_alloc,
18254 reset_ipa_call_summary, reset_inline_summary,
18255 inline_summary_t::duplicate): Update.
18256 (inline_edge_duplication_hook): Turn to ...
18257 (ipa_call_summary_t::duplicate): ... this one.
18258 (inline_edge_removal_hook): Turn to ...
18259 (ipa_call_summary_t::remove): ... this one.
18260 (dump_inline_edge_summary): Turn to ...
18261 (dump_ipa_call_summary): ... this one.
18262 (estimate_function_body_sizes): Update.
18263 (inline_update_callee_summaries): Update.
18264 (remap_edge_change_prob): Update.
18265 (remap_edge_summaries): Update.
18266 (inline_merge_summary): Update.
18267 (do_estimate_edge_time): Update.
18268 (inline_generate_summary): Update.
18269 (inline_read_section): Update.
18270 (inline_read_summary): Update.
18271 (inline_free_summary): Update.
18272 * ipa-inline.c (can_inline_edge_p): Update.
18273 (compute_inlined_call_time): Update.
18274 (want_inline_small_function_p): Update.
18275 (edge_badness): Update.
18276 (early_inliner): Update.
18277 * ipa-inline.h (inline_edge_summary): Turn to ...
18278 (ipa_call_summary): ... this one.
18279 (ipa_call_summary_t): New class.
18280 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
18281 (ipa_call_summaries): New.
18282 (inline_edge_summary): Remove.
18283 (estimate_edge_growth): Update.
18284 * ipa-profile.c (ipa_propagate_frequency_1): Update.
18285 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
18286 * ipa-split.c (execute_split_functions): Update.
18287 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
18289 2017-05-19 Richard Biener <rguenther@suse.de>
18291 PR middle-end/80764
18292 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
18294 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
18296 * config/rs6000/rs6000.c (struct machine_function): Add field
18297 fpr_is_wrapped_separately.
18298 (rs6000_get_separate_components): Use 64 components. Handle the
18299 new FPR components.
18300 (rs6000_components_for_bb): Handle the FPR components.
18301 (rs6000_emit_prologue_components): Handle the FPR components.
18302 (rs6000_emit_epilogue_components): Handle the FPR components.
18303 (rs6000_set_handled_components): Handle the FPR components.
18304 (rs6000_emit_prologue): Don't output prologue code for those FPRs
18305 that are already separately shrink-wrapped.
18306 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
18307 that are already separately shrink-wrapped.
18309 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
18312 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
18315 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
18316 (define_peephole2 for Altivec d-form load): Add peepholes to catch
18317 cases where the register allocator uses a move and an offsettable
18318 memory operation to/from a FPR register on ISA 2.06/2.07.
18319 (define_peephole2 for Altivec d-form store): Likewise.
18321 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
18324 * config/i386/mmx.md (*mov<mode>_internal): Enable
18325 alternatives 11, 12, 13 and 14 also for 32bit targets.
18326 Remove alternatives 15, 16, 17 and 18.
18327 * config/i386/sse.md (vec_concatv2di): Change
18328 alternative (!x, *y) to (x, ?!*Yn).
18330 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
18332 * dumpfile.h (enum dump_kind): Remove stray comma.
18334 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
18336 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
18337 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
18338 predicate::num_conditions
18339 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
18340 (CHANGED): turn into predicate::changed.
18341 (agg_position_info): Move to ipa-predicate.h
18342 (add_condition, predicate::add_clause, predicate::operator &=,
18343 predicate::or_with, predicate::evaluate, predicate::probability,
18344 dump_condition, dump_clause, predicate::dump,
18345 predicate::remap_after_duplication, predicate::remap_after_inlining,
18346 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
18347 (evaluate_conditions_for_known_args): Update.
18348 (set_cond_stmt_execution_predicate): Update.
18349 * ipa-inline.h: Include ipa-predicate.h
18350 (condition, inline_param_summary, conditions, agg_position_info,
18351 predicate): Move to ipa-predicate.h
18352 * ipa-predicate.c: New file.
18353 * ipa-predicate.h: New file.
18355 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
18357 * final.c (leaf_function_p): Check we are not in a sequence.
18359 2017-05-18 Martin Liska <mliska@suse.cz>
18361 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
18362 * dumpfile.c (dump_register): Use new enum dump_kind.
18363 (get_dump_file_name): Likewise.
18364 (dump_enable_all): Likewise.
18365 (dump_switch_p_1): Likewise.
18366 (enable_rtl_dump_file): Remove usage of TDF_RTL.
18367 * dumpfile.h (enum dump_kind): New enum type.
18368 (struct dump_file_info): Create constructor and
18369 format fields and comments.
18370 * passes.c (pass_manager::register_one_dump_file):
18372 * statistics.c (statistics_early_init): Likewise.
18373 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
18374 TDF_TREE with TDF_SLIM.
18375 (gather_memory_references_ref): Likewise.
18377 2017-05-18 Martin Liska <mliska@suse.cz>
18379 * vec.h (struct vnull): Use it.
18381 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
18383 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
18384 (true_predicate, false_predicate, true_predicate_p,
18385 false_predicate_p): Remove.
18386 (single_cond_predicate, not_inlined_predicate): Turn to member function
18388 (add_condition): Update.
18389 (add_clause): Turn to...
18390 (predicate::add_clause): ... this one; update; allow passing NULL
18392 (and_predicates): Turn to ...
18393 (predicate::operator &=): ... this one.
18394 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
18395 (or_predicates): Turn to ...
18396 (predicate::or_with): ... this one.
18397 (evaluate_predicate): Turn to ...
18398 (predicate::evaluate): ... this one.
18399 (predicate_probability): Turn to ...
18400 (predicate::probability): ... this one.
18401 (dump_condition): Update.
18402 (dump_predicate): Turn to ...
18403 (predicate::dump): ... this one.
18404 (account_size_time): Update.
18405 (edge_set_predicate): Update.
18406 (set_hint_predicate): UPdate.
18407 (evaluate_conditions_for_known_args): Update.
18408 (evaluate_properties_for_edge): Update.
18409 (remap_predicate_after_duplication): Turn to...
18410 (predicate::remap_after_duplication): ... this one.
18411 (remap_hint_predicate_after_duplication): Update.
18412 (inline_summary_t::duplicate): UPdate.
18413 (dump_inline_edge_summary): Update.
18414 (dump_inline_summary): Update.
18415 (set_cond_stmt_execution_predicate): Update.
18416 (set_switch_stmt_execution_predicate): Update.
18417 (compute_bb_predicates): Update.
18418 (will_be_nonconstant_expr_predicate): Update.
18419 (will_be_nonconstant_predicate): Update.
18420 (phi_result_unknown_predicate): Update.
18421 (predicate_for_phi_result): Update.
18422 (array_index_predicate): Update.
18423 (estimate_function_body_sizes): Update.
18424 (estimate_node_size_and_time): Update.
18425 (estimate_ipcp_clone_size_and_time): Update.
18426 (remap_predicate): Rename to ...
18427 (predicate::remap_after_inlining): ... this one.
18428 (remap_hint_predicate): Update.
18429 (inline_merge_summary): Update.
18430 (inline_update_overall_summary): Update.
18431 (estimate_size_after_inlining): Update.
18432 (read_predicate): Rename to ...
18433 (predicate::stream_in): ... this one.
18434 (read_inline_edge_summary): Update.
18435 (write_predicate): Rename to ...
18436 (predicate::stream_out): ... this one.
18437 (write_inline_edge_summary): Update.
18438 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
18439 (clause_t): Turn to uint32_t
18440 (predicate): Turn to class; implement constructor and operators
18442 (size_time_entry): Update.
18443 (inline_summary): Update.
18444 (inline_edge_summary): Update.
18446 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
18448 * fold-const.c (fold_binary_loc): Move transformation...
18449 * match.pd (C - X CMP X): ... here.
18451 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
18453 * config/sparc/sparc.c (sparc_option_override): Set function
18454 alignment for -mcpu=niagara7 to 64 to match the I$ line.
18455 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
18457 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
18459 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
18461 2017-05-18 Marek Polacek <polacek@redhat.com>
18464 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
18465 (pass_ubsan::execute): Call gimple_assign_single_p instead of
18466 gimple_assign_load_p.
18468 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
18470 PR middle-end/80692
18471 * real.c (do_compare): Give decimal_do_compare preference over
18472 comparing just the signs.
18474 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
18476 * doc/md.texi (Canonicalization of Instructions): Describe the
18477 canonical form of instructions that inherently set a condition
18480 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
18482 PR middle-end/80775
18483 * tree-cfg.c: Move deletion of unreachable case statements to after
18484 the merging of consecutive case labels.
18486 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
18488 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
18489 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
18490 restoring of callee-saved registers.
18492 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
18494 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
18495 * config/visium/visium.c (single_set_and_flags): Likewise.
18496 * config/visium/visium.md (Substitutions): Likewise.
18498 2017-05-17 Martin Liska <mliska@suse.cz>
18500 * cfg.c: Introduce dump_flags_t type and
18501 use it instead of int type.
18503 * cfghooks.c: Likewise.
18504 * cfghooks.h (struct cfg_hooks): Likewise.
18505 * cfgrtl.c: Likewise.
18506 * cfgrtl.h: Likewise.
18507 * cgraph.c (cgraph_node::get_body): Likewise.
18508 * coretypes.h: Likewise.
18509 * domwalk.c: Likewise.
18510 * domwalk.h: Likewise.
18511 * dumpfile.c (struct dump_option_value_info): Likewise.
18512 (dump_enable_all): Likewise.
18513 (dump_switch_p_1): Likewise.
18514 (opt_info_switch_p): Likewise.
18515 * dumpfile.h (enum tree_dump_index): Likewise.
18516 (struct dump_file_info): Likewise.
18517 * genemit.c: Likewise.
18518 * generic-match-head.c: Likewise.
18519 * gengtype.c (open_base_files): Likewise.
18520 * gimple-pretty-print.c: Likewise.
18521 * gimple-pretty-print.h: Likewise.
18522 * graph.c (print_graph_cfg): Likewise.
18523 * graphite-scop-detection.c (dot_all_sese): Likewise.
18524 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
18525 * loop-unroll.c (report_unroll): Likewise.
18526 * passes.c (pass_manager::register_one_dump_file): Likewise.
18527 * print-tree.c: Likewise.
18528 * statistics.c: Likewise.
18529 * tree-cfg.c: Likewise.
18530 * tree-cfg.h: Likewise.
18531 * tree-dfa.c: Likewise.
18532 * tree-dfa.h: Likewise.
18533 * tree-dump.c (dump_function): Likewise.
18534 * tree-dump.h (struct dump_info): Likewise.
18535 * tree-pretty-print.c: Likewise.
18536 * tree-pretty-print.h: Likewise.
18537 * tree-ssa-live.c: Likewise.
18538 * tree-ssa-live.h: Likewise.
18539 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
18540 * tree-vect-loop.c: Likewise.
18541 * tree-vect-slp.c: Likewise.
18543 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
18544 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18546 PR tree-optimization/80457
18547 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
18548 of all arguments to a statement as scalar_to_vec operations.
18549 (vectorizable_call): Adjust call to vect_model_simple_cost for
18551 (vectorizable_conversion): Likewise.
18552 (vectorizable_assignment): Likewise.
18553 (vectorizable_shift): Likewise.
18554 (vectorizable_operation): Likewise.
18555 (vectorizable_comparison): Likewise.
18556 (vect_is_simple_cond): Record the def types for operands.
18557 (vectorizable_condition): Likewise, call vect_model_simple_cost.
18558 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
18559 for statement argument count.
18561 2017-05-16 Carl Love <cel@us.ibm.com>
18563 * config/rs6000/rs6000-c: Add support for built-in functions
18564 vector unsigned long long vec_bperm (vector unsigned long long,
18565 vector unsigned char)
18566 vector signed long long vec_mule (vector signed int,
18568 vector unsigned long long vec_mule (vector unsigned int,
18569 vector unsigned int)
18570 vector signed long long vec_mulo (vector signed int,
18572 vector unsigned long long vec_mulo (vector unsigned int,
18573 vector unsigned int)
18574 vector signed char vec_sldw (vector signed char,
18575 vector signed char,
18577 vector unsigned char vec_sldw (vector unsigned char,
18578 vector unsigned char,
18580 vector signed short vec_sldw (vector signed short,
18581 vector signed short,
18583 vector unsigned short vec_sldw (vector unsigned short,
18584 vector unsigned short,
18586 vector signed int vec_sldw (vector signed int,
18589 vector unsigned int vec_sldw (vector unsigned int,
18590 vector unsigned int,
18592 vector signed long long vec_sldw (vector signed long long,
18593 vector signed long long,
18595 vector unsigned long long vec_sldw (vector unsigned long long,
18596 vector unsigned long long,
18598 * config/rs6000/rs6000-c: Add support for built-in functions
18599 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
18600 * config/rs6000/altivec.h: Add defintion for vec_sldw.
18601 * doc/extend.texi: Update the built-in documentation for the
18602 new built-in functions.
18604 2017-05-16 Marek Polacek <polacek@redhat.com>
18608 * tree.c (save_expr): Don't fold the expression.
18610 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
18612 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
18613 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
18614 and (?*y,m). Update insn attributes.
18616 2017-05-16 Martin Liska <mliska@suse.cz>
18618 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
18619 flags argument of print_gimple_stmt, print_gimple_expr,
18620 print_generic_stmt and print_generic_expr.
18621 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
18622 * coretypes.h: Likewise.
18623 * except.c (dump_eh_tree): Likewise.
18624 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
18625 * gimple-pretty-print.h: Likewise.
18626 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
18627 (backprop::push_to_worklist): Likewise.
18628 (backprop::pop_from_worklist): Likewise.
18629 (backprop::process_use): Likewise.
18630 (backprop::intersect_uses): Likewise.
18631 (note_replacement): Likewise.
18632 * gimple-ssa-store-merging.c
18633 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
18634 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
18635 (pass_store_merging::execute): Likewise.
18636 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
18637 (ssa_base_cand_dump_callback): Likewise.
18638 (dump_incr_vec): Likewise.
18639 (replace_refs): Likewise.
18640 (replace_mult_candidate): Likewise.
18641 (create_add_on_incoming_edge): Likewise.
18642 (create_phi_basis): Likewise.
18643 (insert_initializers): Likewise.
18644 (all_phi_incrs_profitable): Likewise.
18645 (introduce_cast_before_cand): Likewise.
18646 (replace_one_candidate): Likewise.
18647 * gimplify.c (gimplify_expr): Likewise.
18648 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
18649 (set_rename): Likewise.
18650 (rename_uses): Likewise.
18651 (copy_loop_phi_nodes): Likewise.
18652 (add_close_phis_to_merge_points): Likewise.
18653 (copy_loop_close_phi_args): Likewise.
18654 (copy_cond_phi_args): Likewise.
18655 (graphite_copy_stmts_from_block): Likewise.
18656 (translate_pending_phi_nodes): Likewise.
18657 * graphite-poly.c (print_pdr): Likewise.
18658 (dump_gbb_cases): Likewise.
18659 (dump_gbb_conditions): Likewise.
18660 (print_scop_params): Likewise.
18661 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
18662 (build_cross_bb_scalars_use): Likewise.
18663 (gather_bbs::before_dom_children): Likewise.
18664 * hsa-dump.c (dump_hsa_immed): Likewise.
18665 * ipa-cp.c (print_ipcp_constant_value): Likewise.
18666 (get_replacement_map): Likewise.
18667 * ipa-inline-analysis.c (dump_condition): Likewise.
18668 (estimate_function_body_sizes): Likewise.
18669 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
18670 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
18671 * ipa-prop.c (ipa_dump_param): Likewise.
18672 (ipa_print_node_jump_functions_for_edge): Likewise.
18673 (ipa_modify_call_arguments): Likewise.
18674 (ipa_modify_expr): Likewise.
18675 (ipa_dump_param_adjustments): Likewise.
18676 (ipa_dump_agg_replacement_values): Likewise.
18677 (ipcp_modif_dom_walker::before_dom_children): Likewise.
18678 * ipa-pure-const.c (check_stmt): Likewise.
18679 (pass_nothrow::execute): Likewise.
18680 * ipa-split.c (execute_split_functions): Likewise.
18681 * omp-offload.c (dump_oacc_loop_part): Likewise.
18682 (dump_oacc_loop): Likewise.
18683 * trans-mem.c (tm_log_emit): Likewise.
18684 (tm_memopt_accumulate_memops): Likewise.
18685 (dump_tm_memopt_set): Likewise.
18686 (dump_tm_memopt_transform): Likewise.
18687 * tree-cfg.c (gimple_verify_flow_info): Likewise.
18688 (print_loop): Likewise.
18689 * tree-chkp-opt.c (chkp_print_addr): Likewise.
18690 (chkp_gather_checks_info): Likewise.
18691 (chkp_get_check_result): Likewise.
18692 (chkp_remove_check_if_pass): Likewise.
18693 (chkp_use_outer_bounds_if_possible): Likewise.
18694 (chkp_reduce_bounds_lifetime): Likewise.
18695 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
18696 (chkp_mark_completed_bounds): Likewise.
18697 (chkp_register_incomplete_bounds): Likewise.
18698 (chkp_mark_invalid_bounds): Likewise.
18699 (chkp_maybe_copy_and_register_bounds): Likewise.
18700 (chkp_build_returned_bound): Likewise.
18701 (chkp_get_bound_for_parm): Likewise.
18702 (chkp_build_bndldx): Likewise.
18703 (chkp_get_bounds_by_definition): Likewise.
18704 (chkp_generate_extern_var_bounds): Likewise.
18705 (chkp_get_bounds_for_decl_addr): Likewise.
18706 * tree-chrec.c (chrec_apply): Likewise.
18707 * tree-data-ref.c (dump_data_reference): Likewise.
18708 (dump_subscript): Likewise.
18709 (dump_data_dependence_relation): Likewise.
18710 (analyze_overlapping_iterations): Likewise.
18711 * tree-inline.c (expand_call_inline): Likewise.
18712 (tree_function_versioning): Likewise.
18713 * tree-into-ssa.c (dump_defs_stack): Likewise.
18714 (dump_currdefs): Likewise.
18715 (dump_names_replaced_by): Likewise.
18716 (dump_update_ssa): Likewise.
18717 (update_ssa): Likewise.
18718 * tree-object-size.c (pass_object_sizes::execute): Likewise.
18719 * tree-parloops.c (build_new_reduction): Likewise.
18720 (try_create_reduction_list): Likewise.
18721 (ref_conflicts_with_region): Likewise.
18722 (oacc_entry_exit_ok_1): Likewise.
18723 (oacc_entry_exit_single_gang): Likewise.
18724 * tree-pretty-print.h: Likewise.
18725 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
18726 (get_scalar_evolution): Likewise.
18727 (add_to_evolution): Likewise.
18728 (get_loop_exit_condition): Likewise.
18729 (analyze_evolution_in_loop): Likewise.
18730 (analyze_initial_condition): Likewise.
18731 (analyze_scalar_evolution): Likewise.
18732 (instantiate_scev): Likewise.
18733 (number_of_latch_executions): Likewise.
18734 (gather_chrec_stats): Likewise.
18735 (final_value_replacement_loop): Likewise.
18736 (scev_const_prop): Likewise.
18737 * tree-sra.c (dump_access): Likewise.
18738 (disqualify_candidate): Likewise.
18739 (create_access): Likewise.
18740 (reject): Likewise.
18741 (maybe_add_sra_candidate): Likewise.
18742 (create_access_replacement): Likewise.
18743 (analyze_access_subtree): Likewise.
18744 (analyze_all_variable_accesses): Likewise.
18745 (sra_modify_assign): Likewise.
18746 (initialize_constant_pool_replacements): Likewise.
18747 (find_param_candidates): Likewise.
18748 (decide_one_param_reduction): Likewise.
18749 (replace_removed_params_ssa_names): Likewise.
18750 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
18751 * tree-ssa-copy.c (dump_copy_of): Likewise.
18752 (copy_prop_visit_cond_stmt): Likewise.
18753 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
18754 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
18755 (record_equivalences_from_stmt): Likewise.
18756 * tree-ssa-dse.c (compute_trims): Likewise.
18757 (delete_dead_call): Likewise.
18758 (delete_dead_assignment): Likewise.
18759 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
18760 (forward_propagate_into_cond): Likewise.
18761 (pass_forwprop::execute): Likewise.
18762 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
18763 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
18765 (move_computations_worker): Likewise.
18766 (execute_sm): Likewise.
18767 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
18768 (remove_exits_and_undefined_stmts): Likewise.
18769 (remove_redundant_iv_tests): Likewise.
18770 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
18771 (adjust_iv_update_pos): Likewise.
18772 * tree-ssa-math-opts.c (bswap_replace): Likewise.
18773 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
18774 (value_replacement): Likewise.
18775 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
18776 * tree-ssa-pre.c (print_pre_expr): Likewise.
18777 (get_representative_for): Likewise.
18778 (create_expression_by_pieces): Likewise.
18779 (insert_into_preds_of_block): Likewise.
18780 (eliminate_insert): Likewise.
18781 (eliminate_dom_walker::before_dom_children): Likewise.
18782 (eliminate): Likewise.
18783 (remove_dead_inserted_code): Likewise.
18784 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
18785 * tree-ssa-reassoc.c (get_rank): Likewise.
18786 (eliminate_duplicate_pair): Likewise.
18787 (eliminate_plus_minus_pair): Likewise.
18788 (eliminate_not_pairs): Likewise.
18789 (undistribute_ops_list): Likewise.
18790 (eliminate_redundant_comparison): Likewise.
18791 (update_range_test): Likewise.
18792 (optimize_range_tests_var_bound): Likewise.
18793 (optimize_vec_cond_expr): Likewise.
18794 (rewrite_expr_tree): Likewise.
18795 (rewrite_expr_tree_parallel): Likewise.
18796 (linearize_expr): Likewise.
18797 (break_up_subtract): Likewise.
18798 (linearize_expr_tree): Likewise.
18799 (attempt_builtin_powi): Likewise.
18800 (attempt_builtin_copysign): Likewise.
18801 (transform_stmt_to_copy): Likewise.
18802 (transform_stmt_to_multiply): Likewise.
18803 (dump_ops_vector): Likewise.
18804 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
18805 (print_scc): Likewise.
18806 (set_ssa_val_to): Likewise.
18807 (visit_reference_op_store): Likewise.
18808 (visit_use): Likewise.
18809 (sccvn_dom_walker::before_dom_children): Likewise.
18810 (run_scc_vn): Likewise.
18811 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
18813 (expr_hash_elt::print): Likewise.
18814 (const_and_copies::pop_to_marker): Likewise.
18815 (const_and_copies::record_const_or_copy_raw): Likewise.
18816 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
18817 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
18818 (dump_predicates): Likewise.
18819 (find_uninit_use): Likewise.
18820 (warn_uninitialized_phi): Likewise.
18821 (pass_late_warn_uninitialized::execute): Likewise.
18822 * tree-ssa.c (verify_vssa): Likewise.
18823 (verify_ssa): Likewise.
18824 (maybe_optimize_var): Likewise.
18825 * tree-vrp.c (dump_value_range): Likewise.
18826 (dump_all_value_ranges): Likewise.
18827 (dump_asserts_for): Likewise.
18828 (register_edge_assert_for_2): Likewise.
18829 (vrp_visit_cond_stmt): Likewise.
18830 (vrp_visit_switch_stmt): Likewise.
18831 (vrp_visit_stmt): Likewise.
18832 (vrp_visit_phi_node): Likewise.
18833 (simplify_cond_using_ranges_1): Likewise.
18834 (fold_predicate_in): Likewise.
18835 (evrp_dom_walker::before_dom_children): Likewise.
18836 (evrp_dom_walker::push_value_range): Likewise.
18837 (evrp_dom_walker::pop_value_range): Likewise.
18838 (execute_early_vrp): Likewise.
18840 2017-05-16 Richard Biener <rguenther@suse.de>
18842 * dwarf2out.c (loc_list_from_tree_1): Do not create
18843 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
18845 2017-05-16 Richard Biener <rguenther@suse.de>
18847 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
18849 (note_variable_value_in_expr): If we resolved the decl ref
18850 do not push to the stack.
18852 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
18854 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
18855 operations in fast-math mode.
18856 (vaddq_f16): Likewise.
18857 (vmul_f16): Likewise.
18858 (vmulq_f16): Likewise.
18859 (vsub_f16): Likewise.
18860 (vsubq_f16): Likewise.
18861 * config/arm/neon.md (add<mode>3): New.
18863 (fma:<VH:mode>3): New. Also remove outdated comment.
18866 2017-05-16 Martin Liska <mliska@suse.cz>
18870 * ipa-devirt.c (warn_types_mismatch): Fix typo.
18871 (odr_types_equivalent_p): Likewise.
18873 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
18875 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
18877 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
18880 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
18881 non-interunit SSE move alternatives with '?'.
18882 (zero-extendsidi peephole2): New peephole to skip intermediate
18883 general register in SSE zero-extend sequence.
18885 2017-05-15 Jeff Law <law@redhat.com>
18887 * reorg.c (relax_delay_slots): Create a new variable to hold
18888 the temporary target rather than clobbering TARGET_LABEL.
18890 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
18891 missing argument to extract_bit_field call.
18892 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
18894 2017-05-15 Martin Liska <mliska@suse.cz>
18897 * gcc.c (process_command): Do not allow empty argument of -o option.
18899 2017-05-15 Renlin Li <renlin.li@arm.com>
18901 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
18902 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
18903 * config/aarch64/constraints.md (Usf): Add long call check.
18904 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
18905 (call_value): Likewise.
18906 (sibcall): Likewise.
18907 (sibcall_value): Likewise.
18909 (call_value_insn): New.
18910 (sibcall_insn): Update rtx pattern.
18911 (sibcall_value_insn): Likewise.
18912 (call_internal): Remove.
18913 (call_value_internal): Likewise.
18914 (sibcall_internal): Likewise.
18915 (sibcall_value_internal): Likewise.
18916 (call_reg): Likewise.
18917 (call_symbol): Likewise.
18918 (call_value_reg): Likewise.
18919 (call_value_symbol): Likewise.
18921 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
18924 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
18926 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
18928 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
18929 compatible with CCGOCmode and with CCZmode.
18931 2017-05-14 Martin Sebor <msebor@redhat.com>
18933 PR middle-end/77671
18934 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
18935 (gimple_fold_builtin_snprintf): Same.
18936 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
18937 (gimple_fold_builtin_snprintf): Same.
18938 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
18939 of character types.
18940 (is_call_safe): New function.
18941 (try_substitute_return_value): Call it.
18942 (try_simplify_call): New function.
18943 (pass_sprintf_length::handle_gimple_call): Call it.
18945 2017-05-14 Martin Sebor <msebor@redhat.com>
18947 PR middle-end/80669
18948 * builtins.c (expand_builtin_stpncpy): Simplify.
18950 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
18952 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
18953 * config/i386/i386.h
18954 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
18955 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
18956 (struct machine_function): Add new members call_ms2sysv,
18957 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
18958 (struct machine_frame_state): New fields sp_realigned and
18959 sp_realigned_offset.
18960 * config/i386/i386.c
18961 (enum xlogue_stub): New enum.
18962 (enum xlogue_stub_sets): New enum.
18963 (class xlogue_layout): New class.
18964 (struct ix86_frame): New fields stack_realign_allocate_offset,
18965 stack_realign_offset and outlined_save_offset. Modify comments to
18966 detail stack layout when using out-of-line stubs.
18967 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
18968 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
18969 -mcall-ms2sysv-xlogues.
18970 (stub_managed_regs): New static variable.
18971 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
18972 registers managed by out-of-line stub.
18973 (disable_call_ms2sysv_xlogues): New function.
18974 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
18975 m->call_ms2sysv when appropriate and compute frame layout for
18977 (sp_valid_at, fp_valid_at): New inline functions.
18978 (choose_basereg): New function.
18979 (choose_baseaddr): Add align parameter, use choose_basereg and modify
18981 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
18982 Use align parameter of choose_baseaddr to generated aligned SSE movs
18984 (pro_epilogue_adjust_stack): Modify to track
18985 machine_frame_state::sp_realigned.
18986 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
18987 (ix86_nsaved_sseregs): Likewise.
18988 (ix86_emit_save_regs): Likewise.
18989 (ix86_emit_save_regs_using_mov): Likewise.
18990 (ix86_emit_save_sse_regs_using_mov): Likewise.
18991 (get_scratch_register_on_entry): Likewise.
18992 (gen_frame_set): New function.
18993 (gen_frame_load): Likewise.
18994 (gen_frame_store): Likewise.
18995 (emit_outlined_ms2sysv_save): Likewise.
18996 (emit_outlined_ms2sysv_restore): Likewise.
18997 (ix86_expand_prologue): Modify stack re-alignment code and call
18998 emit_outlined_ms2sysv_save when appropriate.
18999 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
19000 parameter rtx_insn *insn, which allows the function to be used to only
19001 generate the notes.
19002 (ix86_expand_epilogue): Modify validity checks of frame and stack
19003 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
19004 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
19005 * config/i386/predicates.md
19006 (save_multiple): New predicate.
19007 (restore_multiple): Likewise.
19008 * config/i386/sse.md
19009 (save_multiple<mode>): New pattern.
19010 (save_multiple_realign<mode>): Likewise.
19011 (restore_multiple<mode>): Likewise.
19012 (restore_multiple_and_return<mode>): Likewise.
19013 (restore_multiple_leave_return<mode>): Likewise.
19014 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
19016 2017-05-14 Julia Koval <julia.koval@intel.com>
19018 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
19019 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
19020 (__builtin_ia32_xsetbv): New builtins.
19021 * config/i386/i386.c (ix86_expand_special_args_builtin):
19023 (ix86_expand_builtin): Special expand for new intrinsics.
19024 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
19025 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
19026 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
19028 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19030 * cfganal.c (inverted_post_order_compute): Change argument type
19032 * cfganal.h (inverted_post_order_compute): Adjust prototype.
19033 * df-core.c (rest_of_handle_df_initialize): Adjust.
19034 (rest_of_handle_df_finish): Likewise.
19035 (df_analyze_1): Likewise.
19036 (df_analyze): Likewise.
19037 (loop_inverted_post_order_compute): Change argument to be a vec *.
19038 (df_analyze_loop): Adjust.
19039 (df_get_n_blocks): Likewise.
19040 (df_get_postorder): Likewise.
19041 * df.h (struct df_d): Change field to be a vec.
19042 * lcm.c (compute_laterin): Adjust.
19043 (compute_available): Likewise.
19044 * lra-lives.c (lra_create_live_ranges_1): Likewise.
19045 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
19046 * tree-ssa-pre.c (compute_antic): Likewise.
19048 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19050 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
19051 (depth_first_search::depth_first_search): Change structure init
19052 function to this constructor.
19053 (depth_first_search::add_bb): Rename function to this member.
19054 (depth_first_search::execute): Likewise.
19055 (flow_dfs_compute_reverse_finish): Adjust.
19057 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19059 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
19060 (longest_simple_path): Likewise.
19061 * shrink-wrap.c (spread_components): Likewise.
19062 (disqualify_problematic_components): Likewise.
19063 (emit_common_heads_for_components): Likewise.
19064 (emit_common_tails_for_components): Likewise.
19065 (insert_prologue_epilogue_for_components): Likewise.
19067 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19069 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
19072 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19074 * df-core.c (df_set_blocks): Start using auto_bitmap.
19075 (df_compact_blocks): Likewise.
19076 * df-problems.c (df_rd_confluence_n): Likewise.
19077 * df-scan.c (df_insn_rescan_all): Likewise.
19078 (df_process_deferred_rescans): Likewise.
19079 (df_update_entry_block_defs): Likewise.
19080 (df_update_exit_block_uses): Likewise.
19081 (df_entry_block_bitmap_verify): Likewise.
19082 (df_exit_block_bitmap_verify): Likewise.
19083 (df_scan_verify): Likewise.
19084 * lra-constraints.c (lra_constraints): Likewise.
19085 (undo_optional_reloads): Likewise.
19086 (lra_undo_inheritance): Likewise.
19087 * lra-remat.c (calculate_gen_cands): Likewise.
19088 (do_remat): Likewise.
19089 * lra-spills.c (assign_spill_hard_regs): Likewise.
19090 (spill_pseudos): Likewise.
19091 * tree-ssa-pre.c (bitmap_set_and): Likewise.
19092 (bitmap_set_subtract_values): Likewise.
19094 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19096 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
19097 management with auto_bitmap.
19098 (fix_inter_tick): Likewise.
19099 (fix_recovery_deps): Likewise.
19100 * ira.c (add_store_equivs): Likewise.
19101 (find_moveable_pseudos): Likewise.
19102 (split_live_ranges_for_shrink_wrap): Likewise.
19103 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
19104 (rtx_reuse_manager::seen_def_p): Likewise.
19105 (rtx_reuse_manager::set_seen_def): Likewise.
19106 * print-rtl.h (class rtx_reuse_manager): Likewise.
19108 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19110 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
19112 (migrate_btr_def): Likewise.
19113 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
19114 * df-core.c (loop_post_order_compute): Likewise.
19115 (loop_inverted_post_order_compute): Likewise.
19116 * hsa-common.h: Likewise.
19117 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
19118 * init-regs.c (initialize_uninitialized_regs): Likewise.
19119 * ipa-inline.c (resolve_noninline_speculation): Likewise.
19120 (inline_small_functions): Likewise.
19121 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
19122 * ira.c (combine_and_move_insns): Likewise.
19123 (build_insn_chain): Likewise.
19124 * loop-invariant.c (find_invariants): Likewise.
19125 * lower-subreg.c (propagate_pseudo_copies): Likewise.
19126 * predict.c (tree_predict_by_opcode): Likewise.
19127 (predict_paths_leading_to): Likewise.
19128 (predict_paths_leading_to_edge): Likewise.
19129 (estimate_loops_at_level): Likewise.
19130 (estimate_loops): Likewise.
19131 * shrink-wrap.c (try_shrink_wrapping): Likewise.
19132 (spread_components): Likewise.
19133 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
19134 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
19135 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
19136 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
19137 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
19138 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
19139 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
19140 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
19141 (mark_threaded_blocks): Likewise.
19142 (thread_through_all_blocks): Likewise.
19143 * tree-ssa.c (verify_ssa): Likewise.
19144 (execute_update_addresses_taken): Likewise.
19145 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
19147 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19149 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
19151 (post_order_compute): Likewise.
19152 (inverted_post_order_compute): Likewise.
19153 (pre_and_rev_post_order_compute_fn): Likewise.
19155 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19157 * genrecog.c (int_set::int_set): Explicitly construct our
19158 auto_vec base class.
19159 * vec.h (auto_vec::auto_vec): New constructor.
19161 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19163 * bitmap.h (class auto_bitmap): New constructor taking
19164 bitmap_obstack * argument.
19166 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19168 * bitmap.h (class auto_bitmap): Change type of m_bits to
19169 bitmap_head, and adjust ctor / dtor and member operators.
19171 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
19173 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
19174 when returned register mode doesn't match original mode.
19176 2017-05-12 Jeff Law <law@redhat.com>
19177 Jakub Jelinek <jakub@redhat.com>
19179 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
19180 we look for cc setter after the compare-elim changes.
19181 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
19182 within the vector to match what compare-elim now expects.
19183 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
19184 (xorsi3_flags, one_cmplsi2_flags): Likewise.
19186 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
19187 after the compare-elim changes.
19188 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
19189 the vector to match what compare-elim now expects.
19190 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
19191 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
19192 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
19193 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
19194 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
19196 * config/visium/visium.c (single_set_and_flags): Fix where
19197 we look for cc setter after the compare-elim changes.
19198 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
19199 with the vector to match what compare-elim now expects.
19200 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
19201 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
19202 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
19203 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
19204 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
19205 (neg<mode>2_insn_set_overflow): Likewise.
19207 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
19209 PR middle-end/79794
19210 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
19211 maybe_expand_insn call, set ops[0].target. If still set after call,
19212 set alt_rtl. Add extra arg to recursive calls.
19213 (extract_bit_field): Add alt_rtl argument. Pass to
19215 * expmed.h (extract_bit_field): Fix prototype.
19216 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
19217 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
19218 to extract_bit_field_calls.
19219 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
19220 Pass alt_rtl to extract_bit_field calls.
19221 * calls.c (store_unaligned_arguments_into_psuedos)
19222 load_register_parameters): Pass extra NULL to extract_bit_field calls.
19223 * optabs.c (maybe_legitimize_operand): Clear op->target when call
19225 * optabs.h (struct expand_operand): Add target bitfield.
19227 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
19229 * compare-elim.c (try_eliminate_compare): Canonicalize
19230 operation with embedded compare to
19231 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
19232 (set (reg) (operation)].
19234 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
19236 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
19239 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
19240 cost of adding a carry flag for ADC instruction.
19241 [case MINUS]: Ignore the cost of subtracting a carry flag
19242 for SBB instruction.
19244 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
19246 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
19248 * config/rs6000/bmiintrin.h: New file.
19249 * config/rs6000/bmi2intrin.h: New file.
19250 * config/rs6000/x86intrin.h: New file.
19252 2017-05-12 Jeff Law <law@redhat.com>
19254 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
19257 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
19259 PR middle-end/80707
19260 * tree-cfg.c: Remove cfg edges of unreachable case statements.
19262 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19264 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
19265 early expansion of vector divide builtins.
19266 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
19267 builtins identified as having unsigned arguments.
19269 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19271 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
19272 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
19273 expansion of vector logical operations (and, andc, or, xor,
19276 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19278 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
19279 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
19281 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19283 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
19284 early GIMPLE expansion of vector multiplies.
19286 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19288 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
19289 TARGET_HAVE_MOVT conditional.
19290 (movt splitter): Likewise.
19292 2017-05-12 Richard Biener <rguenther@suse.de>
19294 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
19295 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
19296 Fold all stmts not inplace.
19298 2017-05-12 Richard Biener <rguenther@suse.de>
19300 PR tree-optimization/80713
19301 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
19302 inserted_exprs bit for not removed stmts.
19304 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
19306 PR middle-end/69921
19307 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
19308 parallelized" attribute for parallelized OpenACC kernels.
19309 * omp-offload.c (execute_oacc_device_lower): Use it.
19311 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
19312 Set "oacc kernels" attribute.
19313 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
19314 parameter. Adjust all users.
19315 (oacc_fn_attrib_kernels_p): Remove function.
19316 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
19317 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
19318 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
19319 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
19320 assert "oacc kernels" attribute is set.
19322 2017-05-11 Carl Love <cel@us.ibm.com>
19324 * config/rs6000/rs6000-c: Add support for built-in functions
19325 vector unsigned char vec_popcnt (vector signed char)
19326 vector unsigned char vec_popcnt (vector unsigned char)
19327 vector unsigned short vec_popcnt (vector signed short)
19328 vector unsigned short vec_popcnt (vector unsigned short)
19329 vector unsigned int vec_popcnt (vector signed int)
19330 vector unsigned int vec_popcnt (vector unsigned int)
19331 vector unsigned long long vec_popcnt (vector signed long long)
19332 vector unsigned long long vec_popcnt (vector unsigned long long)
19333 vector signed long long vec_slo (vector signed long long,
19334 vector signed char)
19335 vector signed long long vec_slo (vector signed long long,
19336 vector unsigned char)
19337 vector unsigned long long vec_slo (vector unsigned long long,
19338 vector signed char)
19339 vector unsigned long long vec_slo (vector unsigned long long,
19340 vector unsigned char)
19341 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
19342 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
19343 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
19344 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
19345 * doc/extend.texi: Update the built-in documentation file for the
19346 new built-in functions.
19348 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
19350 * attribs.h (sorted_attr_string): Move machine independent
19351 functions for target clone support from the i386 port to common
19352 code. Rename ix86_function_versions to common_function_versions.
19353 Rename make_name to make_unique_name.
19354 (common_function_versions): Likewise.
19355 (make_unique_name): Likewise.
19356 (make_dispatcher_decl): Likewise.
19357 (is_function_default_version): Likewise.
19358 * attribs.c (attr_strcmp): Likewise.
19359 (sorted_attr_string): Likewise.
19360 (common_function_versions): Likewise.
19361 (make_unique_name): Likewise.
19362 (make_dispatcher_decl): Likewise.
19363 (is_function_default_version): Likewise.
19364 * config/i386/i386.c (attr_strcmp): Likewise.
19365 (sorted_attr_string): Likewise.
19366 (ix86_function_versions): Likewise.
19367 (make_name): Likewise.
19368 (make_dispatcher_decl): Likewise.
19369 (is_function_default_version): Likewise.
19370 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
19372 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19375 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
19376 Account for direct move costs for vec_construct of integer
19379 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
19382 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
19383 (UNSPEC_STX_ATOMIC): Ditto.
19384 (loaddi_via_sse): New insn.
19385 (storedi_via_sse): Ditto.
19386 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
19387 Update corresponding peephole2 patterns.
19388 (atomic_storedi_fpu): Ditto.
19390 2017-05-11 Julia Koval <julia.koval@intel.com>
19392 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
19393 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
19395 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
19396 (__builtin_ia32_rsqrt14ss_mask): New builtins.
19397 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
19399 2017-05-11 Nathan Sidwell <nathan@acm.org>
19401 * graphite-poly.c: Include dumpfile.h.
19403 * dumpfle.h (dump_function): Declare here ...
19404 * tree-dump.h (dump_function): ... not here.
19405 * dumpfile.c: #include tree-cfg.h.
19406 (dump_function): Move here from ...
19407 * tree-dump.c (dump_function): ... here.
19408 * gimplify.c: #include splay-tree.h, not tree-dump.h.
19409 * graphite-poly.c: Don't include tree-dump.h.
19410 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
19411 * print-tree.c: Likewise.
19412 * stor-layout.c: Likewise.
19413 * tree-nested.c: Likewise.
19415 * dumpfile.c (dump_start): Use TDF_FLAGS.
19416 (dump_enable_all): Fix TDF_KIND check thinko.
19418 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
19420 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19421 array entries to represent two legal parameterizations of the
19422 overloaded __builtin_cmpb function, as represented by the
19423 P6_OV_BUILTIN_CMPB constant.
19424 (altivec_resolve_overloaded_builtin): Add special case handling
19425 for the __builtin_cmpb function, as represented by the
19426 P6_OV_BUILTIN_CMPB constant.
19427 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
19428 (BU_P6_64BIT_2): New macro.
19429 (BU_P6_OVERLOAD_2): New macro
19430 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
19431 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
19432 (CMPB): Add overload support to represent both 32-bit and 64-bit
19433 compare-bytes function.
19434 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
19435 support for TARGET_CMPB.
19436 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
19437 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19438 documentation of the __builtin_cmpb overloaded built-in function.
19440 2017-05-11 Richard Biener <rguenther@suse.de>
19442 PR tree-optimization/80705
19443 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
19444 bases are not vectorizable.
19446 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19448 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
19449 when counting register pressure.
19451 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19453 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
19454 (iv_ca_more_deps): Renamed to ...
19455 (iv_ca_compare_deps): ... this.
19456 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
19458 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19460 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
19462 (determine_group_iv_costs): ... here.
19463 (find_inv_vars_cb): Record inv var if it's not recorded before.
19465 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19467 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
19468 (get_shiftadd_cost): Ditto.
19470 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19472 * tree-ssa-address.c: Include header file.
19473 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
19475 (add_to_parts): Refactor.
19476 (addr_to_parts): New parameter. Update use of move_hint_to_base.
19477 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
19480 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19482 PR tree-optimization/53090
19483 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
19485 (extract_cond_operands): Detect condition with IV on both sides
19486 and return COMP_IV_EXPR_2.
19487 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
19488 (rewrite_use_compare): Simplify by removing call to function
19489 extract_cond_operands.
19491 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19493 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
19494 (extract_cond_operands): Detect condition comparing against non-
19495 invariant bound and return appropriate enum value.
19496 (find_interesting_uses_cond): Update use of extract_cond_operands.
19497 Handle its return value accordingly.
19498 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
19500 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19502 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
19503 nonlinear iv_use computation in loop invariant sensitive way.
19505 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19507 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
19508 (find_iv_candidates): Call relate_compare_use_with_all_cands.
19510 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19512 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
19513 (dump_cand): Support iv_cand.inv_exprs.
19514 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
19516 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
19519 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19521 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
19523 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
19524 as local function. Include necessary header files.
19525 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
19527 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19529 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
19531 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19533 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
19534 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
19535 RSHIFT_EXPR and BIT_NOT_EXPR.
19537 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19539 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
19540 (adjust_setup_cost): New parameter supporting round up adjustment.
19541 (struct address_cost_data): Delete.
19542 (force_expr_to_var_cost): Don't bound cost with spill_cost.
19543 (split_address_cost, ptr_difference_cost): Delete.
19544 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
19545 (struct ainc_cost_data): New struct.
19546 (get_address_cost_ainc): New function.
19547 (get_address_cost, get_computation_cost): Reimplement.
19548 (determine_group_iv_cost_address): Record inv_expr for all uses of
19550 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
19551 (iv_ca_has_deps): Reimplemented to ...
19552 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
19554 (iv_ca_extend): Call iv_ca_more_deps.
19556 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19558 * tree-ssa-address.c (struct mem_address): Move to header file.
19559 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
19560 * tree-ssa-address.h (struct mem_address): Move from C file.
19561 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
19563 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19565 * tree-affine.h (aff_combination_type): New interface.
19566 (aff_combination_zero_p): Remove static.
19567 (aff_combination_const_p): New interface.
19568 (aff_combination_singleton_var_p): New interfaces.
19570 2017-05-11 Richard Biener <rguenther@suse.de>
19572 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
19573 Skip unreachable blocks and destinations.
19574 (eliminate): Move stmt removal and fixup ...
19575 (fini_eliminate): ... here. Skip inserted exprs.
19576 (pass_pre::execute): Move fini_pre after fini_eliminate.
19577 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
19578 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
19579 PRE to get rid of dead code that has invalid SSA form and
19580 split critical edges again.
19582 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19584 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19586 2017-05-11 Richard Biener <rguenther@suse.de>
19588 * passes.c (execute_function_todo): Verify loops if they are
19589 said to be up-to-date.
19590 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
19591 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
19593 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
19596 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
19597 handle calling assemble_external ourself.
19600 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
19601 modes with zero size. Enhance comment.
19603 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19605 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
19606 built-ins for vec_xl and vec_xst with short and char pointer
19609 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
19611 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
19612 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
19613 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
19614 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
19615 (_mm_maskz_min_round_ss): New intrinsics.
19616 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
19617 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
19618 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
19619 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
19620 (__builtin_ia32_minss_mask_round): New builtins.
19621 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19622 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
19623 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
19625 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
19626 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
19628 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
19631 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
19633 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
19634 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
19635 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
19636 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
19637 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
19638 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
19639 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
19640 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19641 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
19642 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
19643 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
19644 (__builtin_ia32_mulss_mask_round): New builtins.
19645 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19646 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
19647 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
19649 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
19650 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
19652 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
19655 2017-05-10 Julia Koval <julia.koval@intel.com>
19657 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
19658 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
19659 (_mm256_setr_m128i): New intrinsics.
19661 2017-05-10 Julia Koval <julia.koval@intel.com>
19663 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
19664 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
19665 (_mm_maskz_rcp14_ss): New intrinsics.
19666 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
19667 (__builtin_ia32_rcp14ss_mask): New builtins.
19668 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
19670 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
19672 PR tree-optimization/51513
19673 * tree-cfg.c (gimple_seq_unreachable_p): New function.
19674 (assert_unreachable_fallthru_edge_p): Use it.
19675 (group_case_labels_stmt): Likewise.
19676 * tree-cfg.h: Prototype it.
19677 * stmt.c: Include cfghooks.h and tree-cfg.h.
19678 (emit_case_dispatch_table) <gap_label>: New local variable.
19679 Use it to fill dispatch table gaps.
19680 Test for default_label before updating probabilities.
19681 (expand_case) <default_label>: Remove unneeded initialization.
19682 Test for unreachable default case statement and remove its edge.
19683 Set default_label accordingly.
19684 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
19686 2017-05-10 Carl Love <cel@us.ibm.com>
19688 * config/rs6000/rs6000-c: Add support for built-in functions
19689 vector signed char vec_neg (vector signed char)
19690 vector signed short int vec_neg (vector short int)
19691 vector signed int vec_neg (vector signed int)
19692 vector signed long long vec_neg (vector signed long long)
19693 vector float vec_neg (vector float)
19694 vector double vec_neg (vector double)
19695 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
19697 * config/rs6000/altivec.h: Add define for vec_neg
19698 * doc/extend.texi: Update the built-in documentation for the
19699 new built-in functions.
19701 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19703 PR tree-optimization/77644
19704 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
19706 2017-05-10 Nathan Sidwell <nathan@acm.org>
19708 * dumpfile.h (TDI_lang_all): New.
19709 (TDF_KIND): New. Renumber others
19710 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
19712 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
19714 (get_dump_file_name): Adjust suffix generation.
19715 (dump_enable_all): Use TDF_KIND.
19716 * doc/invoke.texi (-fdump-lang-all): Document.
19718 * dumpfile.h: Tabify.
19720 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
19723 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
19724 Move member access before delete.
19726 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
19728 * tree-inline.c (expand_call_inline): Split block at stmt
19731 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
19734 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
19735 are now unused after splitting mov{sf,sd}_hardfloat.
19736 (f32_lr2): Likewise.
19737 (f32_lm): Likewise.
19738 (f32_lm2): Likewise.
19739 (f32_li): Likewise.
19740 (f32_li2): Likewise.
19741 (f32_lv): Likewise.
19742 (f32_sr): Likewise.
19743 (f32_sr2): Likewise.
19744 (f32_sm): Likewise.
19745 (f32_sm2): Likewise.
19746 (f32_si): Likewise.
19747 (f32_si2): Likewise.
19748 (f32_sv): Likewise.
19749 (f32_dm): Likewise.
19750 (f32_vsx): Likewise.
19751 (f32_av): Likewise.
19752 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
19753 For movsf, order stores so the VSX stores occur before the GPR
19754 store which encourages the register allocator to use a traditional
19755 FPR instead of a GPR. For movsd, order the stores so that the GPR
19756 store comes before the VSX stores to allow the power6 to work.
19757 This is due to the power6 not having a 32-bit integer store
19758 instruction from a FPR.
19759 (movsf_hardfloat): Likewise.
19760 (movsd_hardfloat): Likewise.
19762 2017-05-09 Martin Sebor <msebor@redhat.com>
19764 PR translation/80280
19765 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
19768 PR translation/80280
19769 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
19770 data member added in r247778.
19771 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
19773 2017-05-09 Nathan Sidwell <nathan@acm.org>
19775 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
19777 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
19780 2017-05-09 Marek Polacek <polacek@redhat.com>
19782 * doc/invoke.texi: Fix typo.
19784 2017-05-09 Richard Biener <rguenther@suse.de>
19786 * tree-vrp.c (vrp_val_is_max): Adjust comment.
19787 (vrp_val_is_min): Likewise.
19788 (set_value_range_to_value): Likewise.
19789 (set_value_range_to_nonnegative): Likewise.
19790 (gimple_assign_nonzero_p): Likewise.
19791 (gimple_stmt_nonzero_p): Likewise.
19792 (vrp_int_const_binop): Likewise. Remove unreachable case.
19793 (adjust_range_with_scev): Adjust comments.
19794 (compare_range_with_value): Likewise.
19795 (extract_range_from_phi_node): Likewise.
19796 (test_for_singularity): Likewise.
19798 2017-05-09 Richard Biener <rguenther@suse.de>
19800 * tree-vrp.c (get_single_symbol): Add assert that we don't
19801 get overflowed constants as invariant part.
19802 (compare_values_warnv): Add comment before the TREE_NO_WARNING
19803 checks. Use wi::cmp instead of recursing for integer constants.
19804 (compare_values): Just ignore whether we assumed undefined
19805 overflow instead of failing the compare.
19806 (extract_range_for_var_from_comparison_expr): Add comment before the
19807 TREE_NO_WARNING sets.
19808 (test_for_singularity): Likewise.
19809 (extract_range_from_comparison): Do not disable optimization
19810 when we assumed undefined overflow.
19811 (extract_range_basic): Remove init of unused var.
19813 2017-05-09 Richard Biener <rguenther@suse.de>
19815 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
19816 (extract_range_from_multiplicative_op_1): Adjust.
19817 (extract_range_from_binary_expr_1): Use int_const_binop.
19819 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
19822 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
19823 rs6000_store_data_bypass_p in seven define_bypass directives and
19824 in several comments.
19825 * config/rs6000/rs6000-protos.h: Add prototype for
19826 rs6000_store_data_bypass_p function.
19827 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
19828 function implements slightly different (rs6000-specific) semantics
19829 than store_data_bypass_p, returning false rather than aborting
19830 with assertion error when arguments do not satisfy the
19831 requirements of store data bypass.
19832 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
19833 rs6000_store_data_bypass_p.
19835 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
19837 * config/xtensa/xtensa-protos.h
19838 (xtensa_initial_elimination_offset): New declaration.
19839 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
19840 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
19841 macro definition, add case for FRAME_POINTER_REGNUM when
19842 FRAME_GROWS_DOWNWARD.
19843 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
19844 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
19845 xtensa_initial_elimination_offset.
19847 2017-05-08 Nathan Sidwell <nathan@acm.org>
19849 * doc/invoke.texi: Alphabetize -fdump options.
19851 2017-05-08 Martin Sebor <msebor@redhat.com>
19853 PR translation/80280
19854 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
19856 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
19858 * target.def (compute_frame_layout): New optional target hook.
19859 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
19860 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
19861 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
19863 * reload1.c (verify_initial_elim_offsets): Likewise.
19864 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
19865 (use_simple_return_p): Call arm_compute_frame_layout if needed.
19866 (arm_get_frame_offsets): Split up into this ...
19867 (arm_compute_frame_layout): ... and this function.
19869 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
19871 * config/aarch64/constraints.md (Usa): New constraint.
19872 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
19874 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
19876 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
19877 with_multilib_list after it has been checked.
19879 2017-05-08 Richard Biener <rguenther@suse.de>
19881 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
19882 (bitmap_set_subtract_values): Likewise.
19884 2017-05-08 Richard Biener <rguenther@suse.de>
19886 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
19887 (gimple_assign_nonzero): ... this and remove strict_overflow_p
19889 (gimple_stmt_nonzero_warnv_p): Rename to ...
19890 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
19892 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
19893 (extract_range_basic): Adjust, do not disable propagation on
19894 strict overflow sensitive simplification.
19895 (vrp_visit_cond_stmt): Likewise.
19897 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
19899 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
19900 body size unconditionally.
19902 2017-05-07 Jeff Law <law@redhat.com>
19905 2017-05-06 Jeff Law <law@redhat.com>
19906 PR tree-optimization/78496
19907 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
19910 PR tree-optimization/78496
19911 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
19912 (simplify_stmt_using_ranges): Call it.
19913 (vrp_dom_walker::before_dom_children): Extract equivalences
19914 from an ASSERT_EXPR with an equality comparison against a
19917 2017-05-06 Jeff Law <law@redhat.com>
19919 PR tree-optimization/78496
19920 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
19923 PR tree-optimization/78496
19924 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
19925 (simplify_stmt_using_ranges): Call it.
19926 (vrp_dom_walker::before_dom_children): Extract equivalences
19927 from an ASSERT_EXPR with an equality comparison against a
19930 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
19932 * lra-constraints.c (lra_copy_reg_equiv): New function.
19933 (split_reg): Use it to copy equivalence information from the
19934 original register to the spill register.
19936 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
19938 PR rtl-optimization/75964
19939 * simplify-rtx.c (simplify_const_relational_operation): Remove
19940 invalid handling of comparisons of integer ABS.
19942 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
19944 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
19945 initialize to zero.
19946 (init_regs): Remove declaration.
19947 (function_arg_advance_32): Initialize error_p as boolean variable.
19949 2017-05-05 Nathan Sidwell <nathan@acm.org>
19951 * store-motion.c (remove_reachable_equiv_notes): Reformat long
19952 lines. Use for (;;).
19954 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19956 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
19957 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
19958 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
19959 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
19960 VF=2 that require versioning.
19962 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19964 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
19967 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19969 * diagnostic.h (diagnostic_override_option_index): Convert from
19970 macro to inline function.
19972 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19974 * diagnostic.c (last_module_changed_p): New function.
19975 (set_last_module): New function.
19976 (diagnostic_report_current_module): Convert macro usage to
19977 the above functions.
19978 * diagnostic.h (diagnostic_context::last_module): Strengthen
19979 from const line_map * to const line_map_ordinary *.
19980 (diagnostic_last_module_changed): Delete macro.
19981 (diagnostic_set_last_module): Delete macro.
19983 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19985 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
19986 with diagnostic_report_diagnostic.
19987 (diagnostic_n_impl_richloc): Likewise.
19988 * diagnostic.h (report_diagnostic): Delete macro.
19989 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
19990 with diagnostic_report_diagnostic.
19991 * substring-locations.c (format_warning_va): Likewise.
19993 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19995 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
19996 save/restor of format_spec. Move option-printing code to...
19997 (print_option_information): ...this new function, and
19998 reimplement by simply printing to the pretty_printer,
19999 rather than appending to the format string.
20001 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20003 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
20004 handling logic into...
20005 (update_effective_level_from_pragmas): ...this new function.
20007 2017-05-04 Andrew Waterman <andrew@sifive.com>
20009 * config/riscv/riscv.opt (mstrict-align): New option.
20010 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
20011 (SLOW_UNALIGNED_ACCESS): Define.
20012 (riscv_slow_unaligned_access): Declare.
20013 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
20015 (riscv_slow_unaligned_access): New variable.
20016 (rocket_tune_info): Set slow_unaligned_access to true.
20017 (optimize_size_tune_info): Set slow_unaligned_access to false.
20018 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
20019 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
20020 (riscv_option_override): Set riscv_slow_unaligned_access.
20021 * doc/invoke.texi: Add -mstrict-align to RISC-V.
20023 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
20025 * config/riscv/riscv.md: Unify indentation.
20027 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20032 * config/rs6000/rs6000.md (u code attribute): Add FIX and
20034 (extendsi<mode>2): Add support for doing sign extension via
20035 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
20036 don't have ISA 3.0 instructions.
20037 (extendsi<mode>2 splitter): Likewise.
20038 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
20039 generate the normal insns since SImode can now go in vector
20040 registers. Disallow the special UNSPECs needed for previous
20041 machines to hide SImode being used. Add new insns
20042 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
20043 (fix_trunc<mode>si2_stfiwx): Likewise.
20044 (fix_trunc<mode>si2_internal): Likewise.
20045 (fixuns_trunc<mode>si2): Likewise.
20046 (fixuns_trunc<mode>si2_stfiwx): Likewise.
20047 (fctiw<u>z_<mode>_smallint): Likewise.
20048 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
20049 of floating point to 32-bit integer from doing a direct move to
20050 the GPR registers to do a store.
20051 (fctiwz_<mode>): Break long line.
20053 2017-05-05 Bin Cheng <bin.cheng@arm.com>
20055 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
20056 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
20057 (addr_list, addr_offset_valid_p): New.
20058 (split_address_groups): Check offset validity with above function.
20059 (gt-tree-ssa-loop-ivopts.h): Include header file.
20061 2017-05-05 Nathan Sidwell <nathan@acm.org>
20063 * config.gcc (arm*-*-*): Add missing 'fi'.
20065 2017-05-05 Steve Ellcey <sellcey@cavium.com>
20067 * doc/invoke.texi (-fopt-info): Explicitly say order of options
20068 included in -fopt-info does not matter.
20069 * doc/optinfo.texi (-fopt-info): Fix description of default
20070 behavour. Explicitly say order of options included in -fopt-info
20073 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
20075 * config.gcc: Allow combinations of aprofile and rmprofile values for
20076 --with-multilib-list.
20077 * config/arm/t-multilib: New file.
20078 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
20079 variables. Remove setting of ISA and floating-point ABI in
20080 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
20081 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
20082 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
20083 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
20085 * config/arm/t-rmprofile: Likewise except for the matches changes.
20086 * doc/install.texi (--with-multilib-list): Document the combination of
20087 aprofile and rmprofile values and warn about pitfalls in doing that.
20089 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
20091 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
20092 (movdi_aarch64): Likewise.
20094 2017-05-05 Jakub Jelinek <jakub@redhat.com>
20096 PR tree-optimization/80632
20097 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
20099 (build_arrays): Initialize it for virtual phis.
20100 (fix_phi_nodes): Use it for virtual phis.
20102 PR tree-optimization/80558
20103 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
20104 [x, y] op z into [x op, y op z] for op & or | if conditions
20107 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
20108 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20111 * config/arm/arm.md (use_literal_pool): Remove.
20112 (64-bit immediate split): No longer takes cost into consideration
20113 if arm_disable_literal_pool is enabled.
20114 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
20115 used when arm_disable_literal_pool is enabled.
20116 (arm_max_const_double_inline_cost): Remove use of
20117 arm_disable_literal_pool.
20118 (push_minipool_fix): Add assert.
20119 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
20120 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
20121 (no_literal_pool_sf_immediate): New.
20123 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20125 PR tree-optimization/80613
20126 * tree-ssa-dce.c (propagate_necessity): Remove cases for
20127 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
20129 2017-05-05 Richard Biener <rguenther@suse.de>
20131 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
20133 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
20135 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
20136 of this flag from insn conditions due to removal from r247495.
20138 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
20140 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
20142 (arm_early_store_addr_dep_ptr): Likewise.
20143 * config/arm/aarch-common-protos.h
20144 (arm_early_load_addr_dep_ptr): Add prototype.
20145 (arm_early_store_addr_dep_ptr): Likewise.
20146 * config/arm/cortex-a53.md: Add new bypasses.
20148 2017-05-05 Jakub Jelinek <jakub@redhat.com>
20150 * tree.c (next_type_uid): Change type to unsigned.
20151 (type_hash_canon): Decrement back next_type_uid if
20152 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
20153 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
20156 2017-05-04 Martin Sebor <msebor@redhat.com>
20158 * builtins.c: Fix a trivial typo in a comment.
20160 PR middle-end/79234
20161 * builtins.c (check_sizes): Adjust to handle reading past the end.
20162 Avoid printing excessive upper bound of ranges. Use %E to print
20163 tree nodes instead of converting them to %wu.
20164 (expand_builtin_memchr): New function.
20165 (compute_dest_size): Rename...
20166 (compute_objsize): ...to this.
20167 (expand_builtin_memcpy): Adjust.
20168 (expand_builtin_mempcpy): Adjust.
20169 (expand_builtin_strcat): Adjust.
20170 (expand_builtin_strcpy): Adjust.
20171 (check_strncat_sizes): Adjust.
20172 (expand_builtin_strncat): Adjust.
20173 (expand_builtin_strncpy): Adjust and simplify.
20174 (expand_builtin_memset): Adjust.
20175 (expand_builtin_bzero): Adjust.
20176 (expand_builtin_memcmp): Adjust.
20177 (expand_builtin): Handle memcmp.
20178 (maybe_emit_chk_warning): Check strncat just once.
20180 2017-05-04 Martin Sebor <msebor@redhat.com>
20182 PR preprocessor/79214
20183 PR middle-end/79222
20184 PR middle-end/79223
20185 * builtins.c (check_sizes): Add inlining context and issue
20186 warnings even when -Wno-system-headers is set.
20187 (check_strncat_sizes): Same.
20188 (expand_builtin_strncat): Same.
20189 (expand_builtin_memmove): New function.
20190 (expand_builtin_stpncpy): Same.
20191 (expand_builtin): Handle memmove and stpncpy.
20193 2017-05-04 Bin Cheng <bin.cheng@arm.com>
20195 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
20196 which is not used any more.
20198 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20200 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
20202 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20204 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
20205 (cortexa53_tunings): Likewise.
20206 (cortexa57_tunings): Likewise.
20207 (cortexa72_tunings): Likewise.
20208 (cortexa73_tunings): Likewise.
20210 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20212 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
20213 Set loop alignment to 8.
20215 2017-05-04 Martin Sebor <msebor@redhat.com>
20217 PR translation/80280
20218 * builtins.c (expand_builtin_object_size): Add missing quoting to
20219 %D and like directives.
20220 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
20221 (hsa_type_for_tree_type): Same.
20222 (verify_function_arguments): Same.
20223 * symtab.c (symbol_table::change_decl_assembler_name): Same.
20224 * varasm.c (get_section): Same.
20227 2017-05-04 Martin Sebor <msebor@redhat.com>
20229 PR translation/80280
20230 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
20232 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20234 * config/aarch64/aarch64.c (generic_addrcost_table):
20235 Change HI/TI mode setting.
20237 2017-05-04 Martin Jambor <mjambor@suse.cz>
20239 PR tree-optimization/80622
20240 * tree-sra.c (comes_initialized_p): New function.
20241 (build_accesses_from_assign): Only set write lazily when
20242 comes_initialized_p is false.
20243 (analyze_access_subtree): Use comes_initialized_p.
20244 (propagate_subaccesses_across_link): Assert !comes_initialized_p
20245 instead of testing for PARM_DECL.
20247 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20249 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
20250 constraint on operand 0 to allow more general addressing modes.
20251 Adjust output template.
20252 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
20254 * config/aarch64/aarch64-protos.h
20255 (aarch64_address_valid_for_prefetch_p): Declare prototype.
20256 * config/aarch64/constraints.md (Dp): New address constraint.
20257 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
20260 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
20262 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
20263 update use of estimate_ipcp_clone_size_and_time.
20264 (estimate_local_effects): Update use of
20265 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
20266 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
20267 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
20268 Return nonspecialized time.
20270 2017-05-04 Richard Biener <rguenther@suse.de>
20272 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
20273 for the last VUSE which def dominates the PHI. Directly call
20275 (get_continuation_for_phi_1): Remove.
20277 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
20279 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
20280 to explain the use of truncating division. Cap the number of
20281 iterations to the maximum given by nb_iterations_upper_bound,
20284 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20286 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
20287 * configure: Regenerate.
20288 * config.in: Regenerate.
20289 * config/i386/driver-mingw32.c: new file.
20290 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
20291 * config.host: Link driver-mingw32.o on MinGW host.
20292 * doc/install.texi: Document new --enable-mingw-wildcard configure
20295 2017-05-04 Marek Polacek <polacek@redhat.com>
20297 PR tree-optimization/80612
20298 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
20300 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20301 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
20303 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
20304 (movt splitter): Likewise.
20305 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
20306 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
20307 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
20308 block for Thumb-1 with MOVT.
20309 (thumb2_legitimate_address_p): Move code block ...
20310 (can_avoid_literal_pool_for_label_p): ... into this new function.
20311 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
20313 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
20314 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
20315 "M-profile targets with the MOVT instruction".
20317 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20319 * config/arm/arm-builtins.c (arm_init_builtins): Rename
20320 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
20321 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
20323 2017-05-04 Martin Liska <mliska@suse.cz>
20325 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
20326 variable cond_code.
20328 2017-05-04 Richard Biener <rguenther@suse.de>
20330 * tree.c (array_at_struct_end_p): Handle arrays at struct
20331 end with flexarrays more conservatively. Refactor and treat
20332 arrays of arrays or aggregates more strict. Fix
20333 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
20334 * tree.c (array_at_struct_end_p): Adjust prototype.
20335 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
20336 * gimple-fold.c (get_range_strlen): Likewise.
20337 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
20339 2017-05-04 Richard Biener <rguenther@suse.de>
20341 PR tree-optimization/31130
20342 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
20344 (supports_overflow_infinity): Likewise.
20345 (is_negative_overflow_infinity): Likewise.
20346 (is_positive_overflow_infinity): Likewise.
20347 (is_overflow_infinity): Likewise.
20348 (stmt_overflow_infinity): Likewise.
20349 (overflow_infinity_range_p): Likewise.
20350 (usable_range_p): Remove as always returning true.
20351 (make_overflow_infinity): Remove.
20352 (negative_overflow_infinity): Likewise.
20353 (positive_overflow_infinity): Likewise.
20354 (avoid_overflow_infinity): Likewise.
20355 (set_value_range): Adjust accordingly.
20356 (set_value_range_to_nonnegative): Likewise, remove now unused
20357 overflow_infinity arg.
20358 (vrp_operand_equal_p): Adjust.
20359 (update_value_range): Likewise.
20360 (range_int_cst_singleton_p): Likewise.
20361 (operand_less_p): Likewise.
20362 (compare_values_warnv): Likewise.
20363 (extract_range_for_var_from_comparison_expr): Likewise.
20364 (vrp_int_const_binop): Likewise.
20365 (zero_nonzero_bits_from_vr): Likewise.
20366 (extract_range_from_multiplicative_op_1): Likewise.
20367 (extract_range_from_binary_expr_1): Likewise.
20368 (extract_range_from_unary_expr): Likewise.
20369 (extract_range_from_comparison): Likewise.
20370 (extract_range_basic): Likewise.
20371 (adjust_range_with_scev): Likewise.
20372 (compare_ranges): Likewise.
20373 (compare_range_with_value): Likewise.
20374 (dump_value_range): Likewise.
20375 (test_for_singularity): Likewise, remove strict_overflow_p parameter
20377 (simplify_cond_using_ranges): Adjust.
20379 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
20381 * brig-builtins.def: Added a builtin for class_f64.
20382 * builtin-types.def: Added a builtin type needed by class_f64.
20384 2017-05-03 Jason Merrill <jason@redhat.com>
20386 * timevar.def: Add TV_CONSTEXPR.
20388 2017-05-03 David Malcolm <dmalcolm@redhat.com>
20390 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
20392 2017-05-03 Martin Jambor <mjambor@suse.cz>
20394 * ipa-prop.c (ipa_update_after_lto_read): Removed.
20395 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
20396 * ipa-cp.c (ipcp_propagate_stage): Do not call
20397 ipa_update_after_lto_read.
20398 * ipa-inline.c (ipa_inline): Likewise.
20400 2017-05-03 Martin Jambor <mjambor@suse.cz>
20402 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
20403 tag. Added a default constructor and a destructor.
20404 (ipa_edge_args_sum_t): New class;
20405 (ipa_edge_args_sum): Declare.
20406 (ipa_edge_args_vector): Remove declaration.
20407 (IPA_EDGE_REF): Use ipa_edge_args_sum.
20408 (ipa_free_edge_args_substructures): Remove declaration.
20409 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
20410 (ipa_edge_args_info_available_for_edge_p): Likewise.
20411 * ipa-prop.c (ipa_edge_args_vector): Removed.
20412 (edge_removal_hook_holder): Likewise.
20413 (edge_duplication_hook_holder): Likewise.
20414 (ipa_edge_args_sum): New variable.
20415 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
20416 ipa_edge_args_vector.
20417 (ipa_free_edge_args_substructures): Likewise.
20418 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
20419 ipa_edge_args_vector.
20420 (ipa_edge_removal_hook): Turned into method
20421 ipa_edge_args_sum_t::remove.
20422 (ipa_edge_duplication_hook): Turned into method
20423 ipa_edge_args_sum_t::duplicate.
20424 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
20425 registering edge hooks.
20426 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
20427 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
20428 ipa_edge_args_sum instead of ipa_edge_args_vector.
20429 * ipa-profile.c (ipa_profile): Likewise.
20431 2017-05-03 Martin Jambor <mjambor@suse.cz>
20433 * symbol-summary.h (function_summary): New method exists.
20434 (function_summary::symtab_removal): Deallocate through release.
20435 (call_summary): New class.
20436 (gt_ggc_mx): New overload.
20437 (gt_pch_nx): Likewise.
20438 (gt_pch_nx): Likewise.
20440 2017-05-03 Jeff Law <law@redhat.com>
20442 PR tree-optimization/78496
20443 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
20444 from simplify_cond_using_ranges. Split off code to walk
20445 backwards through casts into ...
20446 (simplify_cond_using_ranges_2): New function.
20447 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
20448 (execute_vrp): After identifying jump threads, call
20449 simplify_cond_using_ranges_2.
20451 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
20454 * ipa-inline.h (inline_summary): Add ctor.
20455 (create_ggc): Do not use ggc_cleared_alloc.
20457 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
20458 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20460 * gcc.c (handle_braces): Support escaping in switch matching
20462 * doc/invoke.texi (Spec Files): Document it.
20463 Remove superfluous @code markup in items.
20465 2017-05-03 David Malcolm <dmalcolm@redhat.com>
20467 * diagnostic-show-locus.c (struct column_range): New struct.
20468 (get_affected_columns): New function.
20469 (get_printed_columns): New function.
20470 (struct correction): New struct.
20471 (correction::ensure_capacity): New function.
20472 (correction::ensure_terminated): New function.
20473 (struct line_corrections): New struct.
20474 (line_corrections::~line_corrections): New dtor.
20475 (line_corrections::add_hint): New function.
20476 (layout::print_trailing_fixits): Reimplement in terms of the new
20478 (selftest::test_overlapped_fixit_printing): New function.
20479 (selftest::diagnostic_show_locus_c_tests): Call it.
20481 2017-05-03 Nathan Sidwell <nathan@acm.org>
20483 Canonicalize canonical type hashing
20484 * tree.h (type_hash_canon_hash): Declare.
20485 * tree.c (type_hash_list, attribute_hash_list): Move into
20486 type_hash_canon_hash.
20487 (build_type_attribute_qual_variant): Break out hash code calc into
20488 type_hash_canon_hash.
20489 (type_hash_canon_hash): New. Generic type hash computation.
20490 (build_range_type_1, build_array_type_1, build_function_type,
20491 build_method_type_directly, build_offset_type, build_complex_type,
20492 make_vector_type): Call it.
20494 2017-05-03 Richard Biener <rguenther@suse.de>
20496 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
20497 When all DRs have unknown misaligned do not always peel
20498 when there is a store but apply the same costing model as if
20499 there were only loads.
20501 2017-05-03 Richard Biener <rguenther@suse.de>
20504 PR tree-optimization/80492
20505 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
20506 compare_base_decls returning dont-know properly.
20508 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
20510 * config/arm/iterators.md (CCSI): New mode iterator.
20511 (arch): New mode attribute.
20512 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
20513 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
20514 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
20515 code iterator for success result mode.
20516 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
20517 the corresponding new insn generators.
20519 2017-05-03 Bin Cheng <bin.cheng@arm.com>
20522 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20523 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
20525 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
20527 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
20528 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
20529 (DDR_A): Wrap DDR argument in brackets.
20530 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
20531 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
20532 (DDR_REVERSED_P): Likewise.
20534 2017-05-03 Jakub Jelinek <jakub@redhat.com>
20536 PR tree-optimization/79472
20537 * tree-switch-conversion.c (struct switch_conv_info): Add
20538 contiguous_range and default_case_nonstandard fields.
20539 (collect_switch_conv_info): Compute contiguous_range and
20540 default_case_nonstandard fields, don't clear final_bb if
20541 contiguous_range and only the default case doesn't have the required
20543 (check_all_empty_except_final): Set default_case_nonstandard instead
20544 of failing if contiguous_range and the default case doesn't have empty
20546 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
20547 and only the default case doesn't have the required constants. Skip
20549 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
20550 if default_case_nonstandard.
20551 (build_constructors): Build constant 1 just once. Assert that default
20552 values aren't inserted in between cases if contiguous_range. Skip
20554 (build_arrays): Skip virtual phis.
20555 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
20556 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
20557 Handle virtual phis.
20558 (gen_inbound_check): Handle default_case_nonstandard case.
20559 (process_switch): Adjust check_final_bb caller. Call
20560 gather_default_values with the first non-default case instead of
20561 default case if default_case_nonstandard.
20563 2017-05-02 Nathan Sidwell <nathan@acm.org>
20565 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
20566 check. Fix formatting.
20568 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
20570 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
20571 errors when comparing specialized and unspecialized times.
20573 2017-05-02 David Malcolm <dmalcolm@redhat.com>
20575 * diagnostic-show-locus.c
20576 (layout::should_print_annotation_line_p): Make private.
20577 (layout::print_annotation_line): Make private.
20578 (layout::annotation_line_showed_range_p): Make private.
20579 (layout::show_ruler): Make private.
20580 (layout::print_source_line): Make private. Pass in line and
20581 line_width, rather than calling location_get_source_line. Drop
20583 (layout::print_leading_fixits): New method.
20584 (layout::print_any_fixits): Rename to...
20585 (layout::print_trailing_fixits): ...this, and make private.
20586 Don't print newline fixits.
20587 (diagnostic_show_locus): Move logic for printing one row into...
20588 (layout::print_line): ...this new function. Move the
20589 location_get_source_line call and error-handling from
20590 print_source_line to here. Call print_leading_fixits, and rename
20591 print_any_fixits to print_trailing_fixits.
20592 (selftest::test_fixit_insert_containing_newline): Update now that
20593 newlines are partially supported.
20594 (selftest::test_fixit_insert_containing_newline_2): New test.
20595 (selftest::test_fixit_replace_containing_newline): Update comments.
20596 (selftest::diagnostic_show_locus_c_tests): Call the new test.
20597 * edit-context.c (class added_line): New class.
20598 (class edited_line): Describe newline handling in comment.
20599 (edited_line::actually_edited_p): New method.
20600 (edited_line::print_content): Delete redundant decl.
20601 (edited_line::m_predecessors): New field.
20602 (edited_file::print_content): Call edited_line::print_content.
20603 (edited_file::print_diff): Update to support newlines.
20604 (edited_file::print_diff_hunk): Likewise.
20605 (edited_file::print_run_of_changed_lines): New function.
20606 (edited_file::print_diff_line): Convert to...
20607 (print_diff_line): ...this.
20608 (edited_file::get_effective_line_count): New function.
20609 (edited_line::edited_line): Initialize new field m_predecessors.
20610 (edited_line::~edited_line): Clean up m_predecessors.
20611 (edited_line::apply_fixit): Handle newlines.
20612 (edited_line::get_effective_line_count): New function.
20613 (edited_line::print_content): New function.
20614 (edited_line::print_diff_lines): New function.
20615 (selftest::test_applying_fixits_insert_containing_newline): New
20617 (selftest::test_applying_fixits_replace_containing_newline): New
20619 (selftest::insert_line): New function.
20620 (selftest::test_applying_fixits_multiple_lines): Add example of
20622 (selftest::edit_context_c_tests): Call the new tests.
20624 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20626 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
20627 parameter cand. Update dump information.
20628 (get_computation_cost): Update uses.
20630 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20632 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
20633 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
20634 (get_computation_at, rewrite_use_address): Update use of
20635 get_computation_aff.
20637 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20639 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
20640 (get_computation): Delete.
20641 (get_computation_cost): Implement like get_computation_cost_at.
20642 Use get_computation_at.
20643 (get_computation_cost_at): Delete.
20644 (rewrite_use_nonlinear_expr): Use get_computation_at.
20645 (rewrite_use_compare, remove_unused_ivs): Ditto.
20647 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20649 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
20651 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20653 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
20654 (ivopts_global_cost_for_size): Rename parameter and update uses.
20655 (iv_ca_recount_cost): Update uses.
20656 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
20657 candidates seperately in n_invs and n_cands.
20658 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
20660 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20662 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
20663 (find_inv_vars_cb): New.
20664 (find_depends): Renamed to ...
20665 (find_inv_vars): ... this.
20666 (add_candidate_1, force_var_cost): Call find_inv_vars.
20667 (split_address_cost, determine_group_iv_cost_cond): Ditto.
20669 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20671 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
20672 inv_vars. Add inv_exprs.
20673 (struct iv_cand): Rename depends_on to inv_vars.
20674 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
20675 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
20676 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
20677 (dump_cand): Dump inv_vars.
20678 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
20679 (record_invariant, find_depends, add_candidate_1): Ditto.
20680 (set_group_iv_cost, force_var_cost): Ditto.
20681 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
20682 (get_computation_cost_at, get_computation_cost): Ditto.
20683 (determine_group_iv_cost_generic): Ditto.
20684 (determine_group_iv_cost_address): Ditto.
20685 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
20686 (determine_group_iv_costs): Ditto.
20687 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
20688 (iv_ca_set_remove_invariants): Renamed to ...
20689 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
20690 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
20691 (iv_ca_set_add_invariants): Renamed to ...
20692 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
20693 (iv_ca_set_cp): Use iv_ca_set_add_invs.
20694 (iv_ca_has_deps): Support inv_vars and inv_exprs.
20695 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
20696 (create_new_ivs): Remove useless dump.
20698 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20700 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
20702 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
20703 (iv_ca_set_no_cp, create_new_iv): Ditto.
20705 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20707 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
20709 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
20711 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
20712 function tree_check2.
20714 2017-05-02 Martin Liska <mliska@suse.cz>
20716 * doc/gcov.texi: Add missing preposition.
20717 * gcov.c (function_info::function_info): Properly fill up
20718 all member variables.
20720 2017-05-02 Tamar Christina <tamar.christina@arm.com>
20722 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
20724 2017-05-02 Tamar Christina <tamar.christina@arm.com>
20726 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
20728 2017-05-02 Martin Liska <mliska@suse.cz>
20731 * lto-streamer-in.c (lto_read_tree_1): Remove
20732 LTO_STREAMER_DEBUG.
20733 * lto-streamer.c (struct tree_hash_entry): Likewise.
20734 (struct tree_entry_hasher): Likewise.
20735 (tree_entry_hasher::hash): Likewise.
20736 (tree_entry_hasher::equal): Likewise.
20737 (lto_streamer_init): Likewise.
20738 (lto_orig_address_map): Likewise.
20739 (lto_orig_address_get): Likewise.
20740 (lto_orig_address_remove): Likewise.
20741 * lto-streamer.h: Likewise.
20742 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
20743 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
20745 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
20747 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
20748 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
20749 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
20750 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
20751 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
20752 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
20753 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
20754 (mm_maskz_sub_ss): New intrinsics.
20755 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20756 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
20757 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
20758 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
20759 (__builtin_ia32_subss_mask_round): New builtins.
20760 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20761 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20762 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
20764 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
20765 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
20767 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
20770 2017-05-02 Martin Jambor <mjambor@suse.cz>
20772 PR tree-optimization/78687
20773 * tree-sra.c (access): New field parent.
20774 (process_subtree_disqualification): New function.
20775 (disqualify_candidate): Call it.
20776 (build_accesses_from_assign): Reset write flag if creating an
20778 (build_access_subtree): Fill in parent field and also prpagate
20779 down grp_write flag.
20780 (create_artificial_child_access): New parameter set_grp_write, set
20781 grp_write to its value.
20782 (propagate_subaccesses_across_link): Also propagate grp_write flag
20784 (propagate_all_subaccesses): Push the closest parent back to work
20785 queue if add_access_to_work_queue returned true.
20787 2017-05-02 Richard Biener <rguenther@suse.de>
20789 * common.opt (fstrict-overflow): Alias negative to fwrapv.
20790 * doc/invoke.texi (fstrict-overflow): Remove all traces of
20791 -fstrict-overflow documentation.
20792 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
20793 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
20794 flag_strict_overflow.
20795 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
20796 * lto-opts.c (lto_write_options): Do not stream it.
20797 * lto-wrapper.c (merge_and_complain): Do not handle it.
20798 * opts.c (default_options_table): Do not set -fstrict-overflow.
20799 (finish_options): Likewise do not clear it when sanitizing.
20800 * simplify-rtx.c (simplify_const_relational_operation): Do not
20801 test flag_strict_overflow.
20803 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
20805 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
20806 using enabled attribute.
20807 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
20808 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
20809 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
20810 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
20811 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
20812 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
20813 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
20814 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
20815 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
20816 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
20818 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
20820 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
20822 2017-05-02 Richard Biener <rguenther@suse.de>
20824 PR tree-optimization/80591
20826 2017-04-10 Richard Biener <rguenther@suse.de>
20828 * tree-ssa-structalias.c (find_func_aliases): Properly handle
20831 2017-05-02 Richard Biener <rguenther@suse.de>
20833 PR tree-optimization/80549
20834 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
20835 (cleanup_tree_cfg_noloop): Create forwarders to known loop
20836 headers if they do not have a preheader.
20838 2017-05-02 Martin Liska <mliska@suse.cz>
20841 * common.opt: Fix typo.
20842 * doc/invoke.texi: Likewise.
20844 2017-05-01 Jan Beulich <jbeulich@suse.com>
20846 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
20847 swapping, add (x,x,m,x,n) alternative.
20849 2017-05-01 Nathan Sidwell <nathan@acm.org>
20851 * calls.c (combine_pending_stack_adjustment_and_call): Remove
20852 unnecessary unadjusted_alignment check.
20854 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
20857 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
20859 * gimplify.c (gimplify_cilk_detach): New function.
20860 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
20861 * tree-core.h: Document EXPR_CILK_SPAWN.
20862 * tree.h (EXPR_CILK_SPAWN): Define.
20864 2017-05-01 David Malcolm <dmalcolm@redhat.com>
20866 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
20867 to use new fixit_hint representation, using the "replace" logic.
20868 (get_line_span_for_fixit_hint): Likewise.
20869 (layout::print_any_fixits): Likewise.
20870 (selftest::test_one_liner_many_fixits): Rename to...
20871 (selftest::test_one_liner_many_fixits_1): ...this, and update
20872 comment and expected output to reflect that the multiple fix-it
20873 hints are now consolidated into one insertion.
20874 (selftest::test_one_liner_many_fixits_2): New test.
20875 (selftest::test_diagnostic_show_locus_one_liner): Update for
20877 (selftest::test_fixit_consolidation): Update for fix-it API
20879 * diagnostic.c (print_parseable_fixits): Likewise.
20880 * edit-context.c (edited_line::m_line_events): Convert from
20881 auto_vec <line_event *> to auto_vec <line_event>.
20882 (class line_event): Convert from abstract base class to a concrete
20883 class, taking over the role of replace_event.
20884 (class insert_event): Delete.
20885 (class replace_event): Rename to class line_event. Convert to
20887 (edit_context::add_fixits): Reimplement.
20888 (edit_context::apply_insert): Delete.
20889 (edit_context::apply_replace): Rename to...
20890 (edit_context::apply_fixit): ...this. Convert to half-open range.
20891 (edited_file::apply_insert): Delete.
20892 (edited_file::apply_replace): Rename to...
20893 (edited_file::apply_fixit): ...this.
20894 (edited_line::~edited_line): Drop deletion of events.
20895 (edited_line::apply_insert): Delete.
20896 (edited_line::apply_replace): Rename to...
20897 (edited_line::apply_fixit): ...this. Convert to half-open range.
20898 Update for change to type of m_line_events.
20899 * edit-context.h (edit_context::apply_insert): Delete.
20900 (edit_context::apply_replace): Rename to...
20901 (edit_context::apply_fixit): ...this.
20903 2017-05-01 Martin Sebor <msebor@redhat.com>
20905 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
20908 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
20911 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
20912 __get_cpuid_max returns 0.
20913 (__get_cpuid_count): Ditto.
20915 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
20917 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
20918 replacement expression is another instance of one of its arguments.
20920 2017-05-01 Jakub Jelinek <jakub@redhat.com>
20923 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
20924 check for stack push/pop autoinc.
20925 * config/i386/i386.c (ix86_agi_dependent): Return false
20926 if the only reason why modified_in_p returned true is that
20927 addr is SP based and set_insn is a push or pop.
20929 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
20931 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
20934 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
20937 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
20938 (account_size_time): Use two predicates - exec_pred and
20940 (evaluate_conditions_for_known_args): Compute both clause and
20942 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
20943 (inline_summary_t::duplicate): Update.
20944 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
20946 (compute_inline_parameters): Likewise.
20947 (estimate_edge_size_and_time): Update caluclation of time.
20948 (estimate_node_size_and_time): Compute both time and nonspecialized
20950 (estimate_ipcp_clone_size_and_time): Update.
20951 (inline_merge_summary): Update.
20952 (do_estimate_edge_time): Update.
20953 (do_estimate_edge_size): Update.
20954 (do_estimate_edge_hints): Update.
20955 (inline_read_section, inline_write_summary): Stream both new predicates.
20956 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
20958 (compute_inlined_call_time): Cleanup.
20959 (big_speedup_p): Update.
20960 (edge_badness): Update.
20961 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
20962 (size_time_entry): Replace predicate by exec_predicate and
20963 nonconst_predicate.
20964 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
20965 (estimate_edge_time): Return also nonspec_time.
20966 (reset_edge_growth_cache): Update.
20968 2017-04-29 Jakub Jelinek <jakub@redhat.com>
20970 PR rtl-optimization/80491
20971 * ifcvt.c (noce_process_if_block): When looking for x setter
20972 with missing else_bb, don't check only the insn right before
20973 cond_earliest, but look for the last insn that x is modified in
20974 within the same bb.
20976 PR rtl-optimization/80491
20977 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
20979 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
20981 PR tree-optimization/80487
20982 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
20984 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20986 PR tree-optimization/79697
20987 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
20988 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
20989 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
20991 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
20992 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
20994 2017-04-28 Martin Sebor <msebor@redhat.com>
20996 PR tree-optimization/80523
20997 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
20998 (init_target_to_host_charmap, target_to_host, target_strtol10): New
21000 (maybe_warn, format_directive, parse_directive): Use new functions.
21001 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
21003 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
21005 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
21007 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
21009 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
21010 target_header_dir): Set correctly.
21011 * configure: Regenerated.
21012 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21013 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21014 instead of SYSTEM_HEADER_DIR.
21016 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
21018 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
21019 (estimate_local_effects): Likewise.
21020 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
21021 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
21022 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
21023 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
21024 do_estimate_edge_time, estimate_edge_time): Likewise.
21025 * ipa-inline-analysis.c (estimate_node_size_and_time,
21026 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
21027 (estimate_time_after_inlining): Remove.
21029 2017-04-28 Martin Liska <mliska@suse.cz>
21031 * doc/gcov.texi: Enhance documentation of gcov.
21033 2017-04-28 Martin Liska <mliska@suse.cz>
21035 * doc/gcov.texi: Sort options in alphabetic order.
21036 * doc/gcov-dump.texi: Likewise.
21037 * doc/gcov-tool.texi: Likewise.
21038 * gcov.c (print_usage): Likewise.
21039 * gcov-dump.c (print_usage): Likewise.
21040 * gcov-tool.c (print_merge_usage_message): Likewise.
21041 (print_rewrite_usage_message): Likewise.
21042 (print_overlap_usage_message): Likewise.
21044 2017-04-28 Martin Liska <mliska@suse.cz>
21046 PR gcov-profile/53915
21047 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
21049 2017-04-28 Martin Liska <mliska@suse.cz>
21051 PR gcov-profile/79891
21052 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
21053 is marked by compiler as living on a line.
21054 (get_cycles_count): Remove usage of the union.
21055 (output_intermediate_file): Likewise.
21056 (find_source): Fix GNU coding style.
21057 (accumulate_line_counts): Remove old non-all block mode.
21058 (output_lines): Remove usage of the union.
21059 * profile.c (output_location): Include all BBs, even if
21060 belonging to a same line (and file) as a previous BB.
21062 2017-04-28 Martin Liska <mliska@suse.cz>
21064 * gcov.c (process_args): Handle new argument 'w'.
21065 (read_graph_file): Assign ID to BBs.
21066 (output_branch_count): Display BB # if verbose flag is set.
21067 (output_lines): Likewise for arcs.
21068 (print_usage): Add '--verbose' option help.
21069 * doc/gcov.texi: Document --verbose (-w) option.
21071 2017-04-28 Martin Liska <mliska@suse.cz>
21073 * gcov.c (struct block_location_info): New struct.
21074 (process_file): Fill up the new structure.
21075 (read_graph_file): Replace usage of encoding by the newly added
21077 (add_line_counts): Likewise.
21078 (accumulate_line_counts): Remove usage of the union.
21079 (function_info::function_info): New function.
21080 (function_info::~function_info): Likewise.
21081 (process_file): Call delete instead of release_function.
21082 (release_function): Release the function.
21083 (release_structures): Call delete instead of release_function.
21084 (solve_flow_graph): Replace usage of num_blocks.
21085 (find_exception_blocks): Likewise.
21086 (output_lines): Fix GNU coding style.
21088 2017-04-28 Martin Liska <mliska@suse.cz>
21091 * coverage.c (coverage_remove_note_file): New function.
21092 * coverage.h: Declare the function.
21093 * toplev.c (finalize): Clean if an error has been seen.
21095 2017-04-28 Martin Liska <mliska@suse.cz>
21097 PR gcov-profile/80031
21098 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
21099 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
21100 * gcov.c (read_graph_file): Read just number of blocks.
21101 * profile.c (branch_prob): Do not stream 0 flags per a basic
21104 2017-04-28 Martin Liska <mliska@suse.cz>
21106 * gcov-dump.c (tag_*): Add new argument to declarations.
21107 (dump_gcov_file): Likewise.
21108 (tag_blocks): Add and use new argument depth.
21109 (tag_arcs): Likewise.
21110 (tag_lines): Likewise.
21111 (tag_counters): Likewise.
21112 (tag_summary): Likewise.
21113 (dump_working_sets): Use depth to do a proper indentation.
21115 2017-04-28 Jakub Jelinek <jakub@redhat.com>
21118 * cgraph.h (symtab_node::debug_symtab): No longer inline.
21119 * symtab.c (symtab_node::debug_symtab): Move definition here.
21121 2017-04-28 Richard Biener <rguenther@suse.de>
21123 * lto-streamer.h (LTO_major_version): Bump to 7.
21125 2017-04-28 Richard Biener <rguenther@suse.de>
21127 * tree-vrp.c (assert_info): New struct.
21128 (add_assert_info): New helper.
21129 (register_edge_assert_for_2): Refactor to add asserts to a vector
21131 (register_edge_assert_for_1): Likewise.
21132 (register_edge_assert_for): Likewise.
21133 (finish_register_edge_assert_for): New helper actually registering
21134 asserts where live on edge.
21135 (find_conditional_asserts): Adjust.
21136 (find_switch_asserts): Likewise.
21137 (evrp_dom_walker::try_find_new_range): Generalize.
21138 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
21140 2017-04-27 Marek Polacek <polacek@redhat.com>
21143 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
21144 arg10 and arg11 to itype.
21146 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
21148 * doc/extend.texi (Object Size Checking): Improve grammar.
21150 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
21153 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
21154 that the logic for permitting reciprocal estimates matches that
21157 2017-04-27 Jakub Jelinek <jakub@redhat.com>
21160 * tree.c (type_cache_hasher::equal): Only compare
21161 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
21162 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
21163 non-aggregate element types.
21164 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
21165 about the flag on ARRAY_TYPEs in the comment, formatting fix.
21167 2017-04-27 Richard Biener <rguenther@suse.de>
21169 PR middle-end/80533
21170 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
21171 stripping ARRAY_REFs from MEM_EXPR make sure we're not
21172 keeping a reference to a trailing array.
21174 2017-04-27 Richard Biener <rguenther@suse.de>
21176 PR middle-end/80539
21177 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
21178 being in loop-closed SSA form conservatively.
21179 (chrec_fold_multiply_poly_poly): Likewise.
21181 2017-04-27 Tamar Christina <tamar.christina@arm.com>
21183 PR middle-end/79665
21184 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
21185 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
21187 2017-04-27 Jakub Jelinek <jakub@redhat.com>
21190 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
21191 (aarch64_function_arg_alignment): Return unsigned int again, but still
21192 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
21193 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
21194 Don't emit -Wpsabi note.
21195 (aarch64_function_arg_boundary): Likewise.
21196 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
21199 2017-04-26 Nathan Sidwell <nathan@acm.org>
21201 * tree.h (crc32_unsigned_n): Declare.
21202 (crc32_unsigned, crc32_unsigned): Make inline.
21203 * tree.c (crc32_unsigned_bits): Replace with ...
21204 (crc32_unsigned_n): ... this.
21205 (crc32_unsigned, crc32_byte): Remove.
21206 (crc32_string): Remove unnecessary braces.
21208 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
21210 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
21211 * ipa-inline-analysis.c (MAX_TIME): Remove.
21212 (account_size_time): Use sreal for time.
21213 (dump_inline_summary): Update.
21214 (estimate_function_body_sizes): Update.
21215 (estimate_edge_size_and_time): Update.
21216 (estimate_calls_size_and_time): Update.
21217 (estimate_node_size_and_time): Update.
21218 (inline_merge_summary): Update.
21219 (inline_update_overall_summary): Update.
21220 (estimate_time_after_inlining): Update.
21221 (inline_read_section): Update.
21222 (inline_write_summary): Update.
21223 * ipa-inline.c (compute_uninlined_call_time): Update.
21224 (compute_inlined_call_time): Update.
21225 (recursive_inlining): Update.
21226 (inline_small_functions): Update.
21227 (dump_overall_stats): Update.
21228 * ipa-inline.h: Include sreal.h.
21229 (size_time_entry): Turn time to sreal.
21230 (inline_summary): Turn self_time nad time to sreal.
21232 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
21234 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
21236 (sreal::stream_out, sreal::stream_in): New.
21237 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
21239 2017-04-25 Jakub Jelinek <jakub@redhat.com>
21241 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
21244 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
21247 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
21249 (dimode_scalar_chain::compute_convert_gain): Ditto.
21250 (dimode_scalar_chain::make_vector_copies): Ditto.
21251 (dimode_scalar_chain::convert_reg): Ditto.
21252 (dimode_scalar_chain::convert_insn): Ditto.
21253 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
21254 (VI248_AVX512BW_1): New mode iterator.
21255 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
21256 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
21259 2017-04-25 Martin Sebor <msebor@redhat.com>
21261 PR tree-optimization/80497
21262 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
21263 constants are representable in HOST_WIDE_INT.
21264 (parse_directive): Ditto.
21266 2017-04-25 Martin Sebor <msebor@redhat.com>
21269 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
21270 (new_zero_array): Adjust signature.
21271 (dom_info::dom_init): Used unsigned rather that size_t.
21272 (dom_info::dom_info): Same.
21274 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21275 Jakub Jelinek <jakub@redhat.com>
21278 * config/arm/arm.c: Include gimple.h.
21279 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
21280 returns negative, increment ncrn only if it returned positive.
21281 (arm_needs_doubleword_align): Return int instead of bool,
21282 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
21283 members, but if there is any such non-FIELD_DECL
21284 > PARM_BOUNDARY aligned decl, return -1 instead of false.
21285 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
21286 returns negative, increment nregs only if it returned positive.
21287 (arm_setup_incoming_varargs): Likewise.
21288 (arm_function_arg_boundary): Emit -Wpsabi note if
21289 arm_needs_doubleword_align returns negative, return
21290 DOUBLEWORD_ALIGNMENT only if it returned positive.
21292 2017-04-25 Marek Polacek <polacek@redhat.com>
21295 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
21296 first argument to type.
21298 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
21301 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
21302 type checks to test for compatibility instead of equality.
21304 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21305 Jakub Jelinek <jakub@redhat.com>
21308 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
21310 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
21311 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
21312 the alignment computation, but return their maximum in warn_alignment.
21313 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
21314 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
21316 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
21317 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
21320 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21322 * config/arc/simdext.md (dmpyh): Fix typo.
21324 2017-04-25 Richard Biener <rguenther@suse.de>
21326 PR tree-optimization/80492
21327 * alias.c (compare_base_decls): Handle registers with asm
21328 specification conservatively.
21329 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
21330 compare_base_decls returning dont-know properly.
21332 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21334 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
21335 (legitimate_offset_address_p): New function.
21336 (arc_legitimate_address_p): Use above function.
21338 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21340 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
21342 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21344 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
21345 ACCH registers whenever they are available.
21347 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21349 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
21350 double regs fix when not used.
21352 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21354 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
21356 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
21357 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
21359 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21361 * config/arc/arc.c (arc_output_addsi): Check for h-register class
21362 when emitting short ADD instructions.
21364 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21366 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
21368 (cmpsi_cc_c_insn): Likewise.
21369 (cbranchsi4_scratch): Compute proper instruction length using
21370 compact_hreg_operand.
21371 * config/arc/predicates.md (compact_hreg_operand): New predicate.
21373 2017-04-25 Richard Biener <rguenther@suse.de>
21375 PR middle-end/80509
21376 * passes.c (pass_manager::pass_manager): Initialize
21377 m_name_to_pass_map.
21379 2017-04-25 Richard Biener <rguenther@suse.de>
21381 PR tree-optimization/79201
21382 * tree-ssa-sink.c (statement_sink_location): Handle calls.
21384 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21387 * config/s390/vector.md: Split MEM->GPR vector moves for
21388 non-s_operand addresses.
21390 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21393 * config/s390/predicates.md (reload_const_wide_int_operand): New
21395 * config/s390/s390.md ("movti"): Remove d/P alternative.
21396 ("movti_bigconst"): New pattern definition.
21398 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
21401 * s390-protos.h (s390_expand_cs_hqi): Removed.
21402 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
21403 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
21404 modes as well as CCZ1mode and CCZmode.
21405 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
21406 signature of s390_emit_compare_and_swap.
21407 (s390_expand_cs_hqi): Likewise, make static.
21408 (s390_expand_cs_tdsi): Generate an explicit compare before trying
21409 compare-and-swap, in some cases.
21410 (s390_expand_cs): Wrapper function.
21411 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
21413 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
21414 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
21415 patterns for small and large integers. Forbid symref memory operands.
21416 Move expander to s390.c. Require cc register.
21417 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
21418 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
21419 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
21420 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
21421 symref memory operands. Remove CC mode and call s390_match_ccmode
21423 ("atomic_exchange<mode>"): Allow and implement all integer modes.
21425 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
21427 * config/s390/s390.md (define_peephole2): New peephole to help
21428 combining the load-and-test pattern with volatile memory.
21430 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
21432 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
21433 with CCZmode for TARGET_Z196.
21435 2017-04-25 Jakub Jelinek <jakub@redhat.com>
21437 PR rtl-optimization/80501
21438 * combine.c (make_compound_operation_int): Set subreg_code to SET
21439 even for AND with mask of the sign bit of mode.
21441 PR rtl-optimization/80500
21442 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
21443 sum's initial value.
21445 2017-04-25 Julian Brown <julian@codesourcery.com>
21446 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21448 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
21450 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
21452 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
21454 2017-04-25 Julian Brown <julian@codesourcery.com>
21455 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21457 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
21458 (thunderx2t99_sha): New Reservation.
21460 2017-04-25 Julian Brown <julian@codesourcery.com>
21461 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21463 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
21464 type for 1-element load.
21466 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
21468 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
21470 2017-04-24 Martin Jambor <mjambor@suse.cz>
21472 PR tree-optimization/80293
21473 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
21474 char arrays not totally scalarizable if it is false.
21475 (analyze_all_variable_accesses): Pass correct value in the new
21476 parameter. Add a statistics counter.
21478 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
21480 PR middle-end/79931
21481 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
21483 2017-04-24 Richard Biener <rguenther@suse.de>
21485 PR tree-optimization/80494
21486 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
21487 out for complex types.
21489 2017-04-24 Richard Biener <rguenther@suse.de>
21491 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
21492 * tree-ssa-sccvn.c (print_scc): Print SCC size.
21493 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
21494 (DFS): Adjust and never fail.
21495 (sccvn_dom_walker::fail): Remove.
21496 (sccvn_dom_walker::before_dom_children): Adjust.
21497 (run_scc_vn): Likewise and never fail.
21498 * tree-ssa-pre.c (pass_pre::execute): Adjust.
21499 (pass_fre::execute): Likewise.
21501 2017-04-24 Richard Biener <rguenther@suse.de>
21503 PR tree-optimization/79725
21504 * tree-ssa-sink.c (statement_sink_location): Return whether
21505 failure reason was zero uses. Move that check later.
21506 (sink_code_in_bb): Deal with zero uses by removing the stmt
21509 2017-04-24 Richard Biener <rguenther@suse.de>
21512 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
21513 pointer-based references.
21515 2017-04-24 Richard Biener <rguenther@suse.de>
21518 * pass_manager.h (pass_manager::operator new): Remove.
21519 (pass_manager::operator delete): Likewise.
21520 * passes.c (pass_manager::operator new): Remove.
21521 (pass_manager::operator delete): Likewise.
21522 (pass_manager::pass_manager): Zero individual pass members.
21524 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
21527 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
21528 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
21529 Check "XEXP (src, 1)" operand here.
21530 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
21531 Check "XEXP (src, 1)" operand here.
21532 (dimode_scalar_chain::make_vector_copies): Detect count register
21533 of a shift instruction. Zero extend count register from QImode
21534 to DImode to satisfy vector shift pattern count operand predicate.
21535 Substitute vector shift count operand with a DImode copy.
21536 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
21539 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
21541 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
21542 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
21543 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
21544 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
21545 (UNSPEC_NOREX_MEM): Remove definition.
21547 2017-04-21 Richard Biener <rguenther@suse.de>
21549 PR tree-optimization/79547
21550 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
21551 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
21552 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
21553 without any constraints.
21555 2017-04-21 Richard Biener <rguenther@suse.de>
21557 PR tree-optimization/78847
21558 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
21560 2017-04-21 Richard Biener <rguenther@suse.de>
21562 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
21563 (build_distinct_type_copy): Likewise.
21564 (build_variant_type_copy): Likewise.
21565 * tree.c (build_qualified_type): Pass down mem-stat info.
21566 (build_distinct_type_copy): Likewise.
21567 (build_variant_type_copy): Likewise.
21569 2017-04-21 Richard Biener <rguenther@suse.de>
21571 PR tree-optimization/80237
21572 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
21574 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
21575 for a simplified result.
21577 2016-04-21 Richard Biener <rguenther@suse.de>
21579 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
21580 sth as strict as a simple_iv but a chrec without symbols and an
21581 operand defined in the loop we are peeling (and not some subloop).
21582 (propagate_constants_for_unrolling): Propagate all constants.
21584 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
21587 * config/i386/i386.c (print_reg): Remove assert for disalowed
21588 regno values, call output_operand_lossage instead.
21590 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
21593 * config/i386/constraints.md (Yc): New register constraint.
21594 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
21595 Use Yc constraint for alternative 2 of operand 0. Remove
21596 preferred_for_speed attribute.
21598 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
21600 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
21601 lastprivate clauses in SIMT case.
21603 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
21605 * doc/invoke.texi (-Wextra-semi): Document new warning option.
21607 2017-04-20 Richard Biener <rguenther@suse.de>
21609 PR tree-optimization/57796
21610 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
21611 as N scalar stores.
21612 (vect_model_load_cost): Cost gathers as N scalar loads.
21614 2017-04-20 Richard Biener <rguenther@suse.de>
21616 * ggc-page.c (ggc_allocated_p): Rename to ...
21617 (safe_lookup_page_table_entry): ... this and return the lookup
21619 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
21621 2017-04-20 Richard Biener <rguenther@suse.de>
21623 PR tree-optimization/80453
21624 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
21625 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
21626 from the conditions.
21627 (vn_phi_eq): Pass them down.
21628 (vn_phi_lookup): Record them.
21629 (vn_phi_insert): Likewise.
21631 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
21633 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
21634 uninitialized variable warning to avoid buffer overrun.
21636 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
21639 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
21640 is suppressed for '{ 0 }' in C.
21642 2017-04-20 Jakub Jelinek <jakub@redhat.com>
21644 * BASE-VER: Set to 8.0.0.
21646 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
21648 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
21649 priority .init_array and .fini_array section with SECTION_NOTYPE
21652 2017-04-20 Jakub Jelinek <jakub@redhat.com>
21654 PR middle-end/80423
21655 * tree.h (build_array_type): Add typeless_storage default argument.
21656 * tree.c (type_cache_hasher::equal): Also compare
21657 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
21658 (build_array_type): Add typeless_storage argument, set
21659 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
21661 (build_nonshared_array_type): Adjust build_array_type_1 caller.
21662 (build_array_type): Likewise. Add typeless_storage argument.
21664 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
21665 Jakub Jelinek <jakub@redhat.com>
21667 PR tree-optimization/80426
21668 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
21669 operation on symbolic operands, also compute the overflow for the
21670 invariant part when the operation degenerates into a negation.
21672 2017-04-19 Jakub Jelinek <jakub@redhat.com>
21675 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
21676 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
21679 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
21681 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
21684 * config/avr/avr.c (tree.h): Include it.
21685 (cgraph.h): Include it.
21686 (avr_encode_section_info): Don't warn for uninitialized progmem
21687 variable if it's just an alias.
21689 2017-04-19 Richard Biener <rguenther@suse.de>
21692 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
21693 when needed by AutoPGO.
21695 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
21698 * doc/lto.texi: Remove an extra 'that'.
21700 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
21702 PR rtl-optimization/80429
21703 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
21704 are only used in debug insns.
21706 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
21707 Vladimir Makarov <vmakarov@redhat.com>
21709 * config/sparc/predicates.md (input_operand): Add comment. Return
21710 true for any memory operand when LRA is in progress.
21711 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
21713 2017-04-18 Jeff Law <law@redhat.com>
21716 * mips.md ({return,simple_return}_internal): Do not overwrite
21719 2017-04-18 Jakub Jelinek <jakub@redhat.com>
21721 PR tree-optimization/80443
21722 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
21723 instead of adding 1, subtract -1 and similarly instead of subtracting
21726 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
21728 PR rtl-optimization/80357
21729 * haifa-sched.c (tmp_bitmap): New variable.
21730 (model_recompute): Handle duplicate use records.
21731 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
21732 (free_global_sched_pressure_data): Free it.
21734 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
21737 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
21738 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21739 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21740 instead of SYSTEM_HEADER_DIR.
21742 2017-04-18 Jeff Law <law@redhat.com>
21744 PR middle-end/80422
21745 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
21746 predecessors after walking up the insn chain.
21748 2017-04-18 Jakub Jelinek <jakub@redhat.com>
21751 * dwarf2out.c (modified_type_die): Try harder not to emit internal
21752 sizetype type into debug info.
21754 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
21757 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
21758 unneeded test for TARGET_UPPER_REGS_SF.
21759 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
21761 2017-04-18 Jakub Jelinek <jakub@redhat.com>
21764 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
21765 instead of gsi_after_labels.
21767 2017-04-18 Jeff Law <law@redhat.com>
21769 * regcprop.c (maybe_mode_change): Avoid creating copies of the
21773 2017-04-13 Jeff Law <law@redhat.com>
21774 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
21775 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
21777 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
21780 * config/avr/avr.c (intl.h): Include it.
21781 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
21783 2017-04-18 Martin Liska <mliska@suse.cz>
21785 PR gcov-profile/78783
21786 * gcov-tool.c (gcov_output_files): Validate that destination
21787 file is either removed by the tool or by a user.
21789 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
21790 Guy Benyei <guybe@mellanox.com>
21792 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
21793 block, and do not negate it, the stored id is already negative.
21795 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
21797 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
21799 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
21802 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
21803 masks of options that should be turned off if the VSX vector
21804 options are turned off.
21805 (OTHER_P8_VECTOR_MASKS): Likewise.
21806 (OTHER_VSX_VECTOR_MASKS): Likewise.
21807 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
21808 rs6000_disable_incompatible_switches to validate no type switches
21810 (rs6000_incompatible_switch): New function to disallow turning on
21811 other vector options if -mno-vsx, -mno-power8-vector, or
21812 -mno-power9-vector are specified.
21814 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21816 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
21818 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21820 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
21821 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
21822 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
21823 (ARG_POINTER_CFA_OFFSET): Likewise.
21825 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21827 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
21828 conditions to take advantage of various optimizations.
21830 2017-04-13 Jeff Law <law@redhat.com>
21832 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
21833 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
21834 (zero_extendsidi2_dext): Likewise.
21836 2017-04-13 Jakub Jelinek <jakub@redhat.com>
21839 * fold-const.c (fold_ternary_loc): Revert
21840 use op0 instead of fold_convert_loc (loc, type, arg0) part of
21843 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
21845 PR rtl-optimization/80343
21846 * lra-remat.c (update_scratch_ops): Assign original hard reg to
21847 new scratch pseudo.
21849 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
21852 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
21853 to ubsan_encode_value.
21855 2017-04-13 Jeff Law <law@redhat.com>
21857 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
21858 appearing in DEBUG_INSNs.
21860 2017-04-13 Martin Liska <mliska@suse.cz>
21862 PR gcov-profile/80413
21863 * gcov-io.c (gcov_write_string): Copy to buffer just when
21864 allocated size is greater than zero.
21866 2017-04-13 Jakub Jelinek <jakub@redhat.com>
21869 * dwarf2out.c (decls_for_scope): Ignore declarations of
21870 current_function_decl in BLOCK_NONLOCALIZED_VARS.
21872 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
21875 * ipa-visibility.c (non_local_p): Fix typos.
21876 (localize_node): When localizing symbol in same comdat group,
21877 dissolve the group only when we know external symbols are going
21879 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
21881 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21883 PR tree-optimization/79390
21884 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
21885 order does not result in usable sequence, retry with reversed operand
21891 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
21892 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
21893 op0 instead of fold_convert_loc (loc, type, arg0).
21895 2017-04-12 Jeff Law <law@redhat.com>
21897 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
21898 has a delay slot in the generated code.
21900 * config/cris/cris.md (cris_preferred_reload_class): Return
21901 GENNONACR_REGS rather than GENERAL_REGS.
21903 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21906 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
21907 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
21908 signedness of the result type.
21910 2017-04-12 Richard Biener <rguenther@suse.de>
21911 Jeff Law <law@redhat.com>
21913 PR tree-optimization/80359
21914 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
21915 trim stores to TARGET_MEM_REFs.
21917 2017-04-12 Richard Biener <rguenther@suse.de>
21919 PR tree-optimization/79390
21920 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
21921 threading case even more.
21923 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
21926 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
21927 for quad_address_p for TImode, instead of just not indexed_address.
21929 2017-04-12 Richard Biener <rguenther@suse.de>
21930 Bernd Edlinger <bernd.edlinger@hotmail.de>
21932 PR middle-end/79671
21933 * alias.c (component_uses_parent_alias_set_from): Handle
21934 TYPE_TYPELESS_STORAGE.
21935 (get_alias_set): Likewise.
21936 * tree-core.h (tree_type_common): Add typeless_storage flag.
21937 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
21938 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
21939 for types containing members with TYPE_TYPELESS_STORAGE.
21940 (place_field): Likewise.
21941 (layout_type): Likewise for ARRAY_TYPE.
21942 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
21943 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
21944 TYPE_TYPELESS_STORAGE.
21945 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
21947 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21950 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
21951 first argument to type.
21953 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21957 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
21958 CONST0_RTX (mode) rather than const0_rtx where appropriate.
21959 (rs6000_expand_binop_builtin): Likewise.
21960 (rs6000_expand_ternop_builtin): Likewise; also add missing
21961 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
21962 vshasigma built-ins.
21963 * doc/extend.texi: Document that vec_xxpermdi's third argument
21964 must be a constant.
21966 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
21968 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
21969 Use shift_const cost parameter when calculating gain of STV shifts.
21971 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
21973 PR rtl-optimization/70478
21974 * lra-constraints.c (process_alt_operands): Check memory for
21975 disfavoring memory insn operand.
21977 2017-04-11 Jakub Jelinek <jakub@redhat.com>
21979 PR middle-end/80100
21980 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
21981 left shift in unsigned HOST_WIDE_INT type.
21983 PR rtl-optimization/80385
21984 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
21985 (not (neg X)) into (plus X -1) for complex or non-integral modes.
21988 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
21989 if they have any depend clauses.
21991 2017-04-11 Martin Liska <mliska@suse.cz>
21994 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
21995 * ipa-split.c (split_function): Create a local comdat symbol
21996 if caller is in a comdat group.
21998 2017-04-11 Martin Liska <mliska@suse.cz>
22001 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
22004 2017-04-11 Martin Sebor <msebor@redhat.com>
22006 PR middle-end/80364
22007 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
22008 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
22010 (directive::set_width, directive::set_precision, format_character):
22012 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
22015 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
22018 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
22019 conflict, set target->arch_name instead of target->cpu_name.
22021 2017-04-11 Richard Biener <rguenther@suse.de>
22023 PR tree-optimization/80374
22024 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
22025 build_zero_cst, remove fold_convertible_p check again.
22027 2017-04-11 Martin Liska <mliska@suse.cz>
22030 * ubsan.c (instrument_object_size): Do not instrument register
22033 2017-04-11 Jakub Jelinek <jakub@redhat.com>
22036 * config/i386/i386-builtin-types.def
22037 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
22038 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
22039 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
22040 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
22041 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
22042 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
22043 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
22044 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
22045 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
22046 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
22047 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
22048 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
22049 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
22050 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
22051 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
22052 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
22053 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
22054 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
22055 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
22056 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
22057 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
22058 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
22059 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
22060 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
22061 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
22062 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
22063 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
22064 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
22065 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
22066 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
22067 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
22068 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
22069 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
22070 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
22071 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
22072 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
22073 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
22074 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
22075 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
22076 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
22077 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
22078 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
22079 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
22080 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
22081 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
22082 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
22084 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
22085 flag to second_arg_count, handle 4 argument function type _COUNT
22086 aliases, handle second_arg_count on second argument rather than last.
22088 2017-04-10 Jeff Law <law@redhat.com>
22090 PR tree-optimization/80374
22091 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
22092 record anything if we can not convert integer_zero_node to the
22095 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
22098 * config/rs6000/rs6000.c (rs6000_option_override_internal):
22099 Enhance special handling given to the TARGET_P9_MINMAX option in
22100 relation to certain other options.
22102 2017-04-10 Bin Cheng <bin.cheng@arm.com>
22104 PR tree-optimization/80153
22105 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
22106 remove POINTER_PLUS_EXPR's base part directly, rather than through
22109 2017-04-10 Richard Biener <rguenther@suse.de>
22110 Bin Cheng <bin.cheng@arm.com>
22112 PR tree-optimization/80153
22113 * tree-affine.c (aff_combination_to_tree): Get base pointer from
22114 the first element of pointer type aff_tree. Build result expr in
22116 (add_elt_to_tree): Convert to type unconditionally. Remove other
22117 fold_convert calls.
22118 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
22119 (rewrite_use_nonlinear_expr): Check invariant using iv information.
22121 2017-04-10 Richard Biener <rguenther@suse.de>
22123 * tree-ssa-structalias.c (find_func_aliases): Properly handle
22126 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
22128 PR rtl-optimization/70478
22129 * lra-constraints.c (curr_small_class_check): New.
22130 (update_and_check_small_class_inputs): New.
22131 (process_alt_operands): Update curr_small_class_check. Disfavor
22132 alternative insn memory operands. Check available regs for small
22135 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
22138 * config/mips/mips.opt (-mvirt): Update description.
22139 * doc/invoke.texi (-mvirt): Likewise.
22141 2017-04-10 Richard Biener <rguenther@suse.de>
22143 PR middle-end/80362
22144 * fold-const.c (fold_binary_loc): Look at unstripped ops when
22145 looking for NEGATE_EXPR in -A / -B to A / B folding.
22147 2017-04-10 Martin Liska <mliska@suse.cz>
22149 PR gcov-profile/80224
22150 * gcov.c (print_usage): Fix usage string.
22151 (get_gcov_intermediate_filename): Remove.
22152 (output_gcov_file): Use both for normal and intermediate format.
22153 (generate_results): Do not initialize special file for
22154 intermediate format.
22156 2017-04-10 Richard Biener <rguenther@suse.de>
22158 PR tree-optimization/80304
22159 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
22162 2017-04-10 Nathan Sidwell <nathan@acm.org>
22165 * config/rs6000/rs6000.c (rs6000_vector_type): New.
22166 (rs6000_init_builtins): Use it.
22168 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22170 * config/arm/arm.md (<mrc>): Add mode to SET source.
22171 (<mrrc>): Likewise.
22173 2017-04-10 Richard Biener <rguenther@suse.de>
22175 PR middle-end/80344
22176 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
22178 2017-04-10 Jakub Jelinek <jakub@redhat.com>
22181 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
22182 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
22183 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
22184 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
22185 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
22186 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
22187 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
22188 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
22189 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
22190 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
22191 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
22192 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
22193 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
22194 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
22195 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
22196 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
22197 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
22198 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
22199 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
22200 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
22201 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
22202 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
22203 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
22205 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
22207 PR rtl-optimization/70478
22208 * lra-constraints.c: Reverse the last patch.
22210 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
22212 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
22213 Add comment for WCHAR_T.
22215 2017-04-08 Martin Liska <mliska@suse.cz>
22218 2017-04-07 Martin Liska <mliska@suse.cz>
22221 * ipa-split.c (split_function): Add function part to a same comdat
22224 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22227 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
22229 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
22231 * rs6000/rs6000.c (vec_load_pendulum): Rename...
22232 (vec_pairing): ...to this.
22233 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
22234 (rs6000_sched_init): Adjust for name change.
22235 (struct rs6000_sched_context): Likewise.
22236 (rs6000_init_sched_context): Likewise.
22237 (rs6000_set_sched_context): Likewise.
22239 2017-04-07 Jakub Jelinek <jakub@redhat.com>
22245 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
22247 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
22248 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
22249 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
22251 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
22253 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
22255 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
22257 PR rtl-optimization/70703
22258 * ira-color.c (update_conflict_hard_regno_costs): Use
22259 int64_t instead of HOST_WIDE_INT.
22261 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
22263 PR rtl-optimization/70478
22264 * lra-constraints.c (process_alt_operands): Disfavor alternative
22265 insn memory operands.
22267 2017-04-07 Jeff Law <law@redhat.com>
22269 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
22270 CALL and NOTE_INSN_CALL_ARG_LOCATION.
22272 2017-04-07 Martin Liska <mliska@suse.cz>
22275 * config/aarch64/aarch64.c (aarch64_process_target_attr):
22276 Show error message instead of an ICE.
22278 2017-04-07 Martin Liska <mliska@suse.cz>
22281 * ipa-split.c (split_function): Add function part to a same comdat
22284 2017-04-07 Richard Biener <rguenther@suse.de>
22286 PR middle-end/80341
22287 * tree.c (get_unwidened): Also handle ! for_type case for
22289 * convert.c (do_narrow): Split out from ...
22290 (convert_to_integer_1): ... here. Do not pass final truncation
22291 type to get_unwidened for TRUNC_DIV_EXPR.
22293 2017-04-07 Richard Biener <rguenther@suse.de>
22295 * tree-affine.c (wide_int_ext_for_comb): Take type rather
22297 (aff_combination_const): Adjust.
22298 (aff_combination_scale): Likewise.
22299 (aff_combination_add_elt): Likewise.
22300 (aff_combination_add_cst): Likewise.
22301 (aff_combination_convert): Likewise.
22302 (add_elt_to_tree): Likewise. Remove unused argument.
22303 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
22305 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
22307 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
22309 * config/arm/arm.c (arm_default_short_enums): Use
22310 ARM_DEFAULT_SHORT_ENUMS.
22311 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
22313 2017-04-06 Jakub Jelinek <jakub@redhat.com>
22316 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
22317 members with redundant out-of-class redeclaration.
22319 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
22322 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
22323 * config/i386/i386.md (*zero_extendsidi2):
22324 Add (?*x,*x) and (?*v,*v) alternatives.
22326 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
22329 * config/i386/i386.c (ix86_expand_builtin)
22330 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
22331 mode from insn data. Convert operands to insn operand mode.
22332 Copy operands that don't satisfy insn predicate to a register.
22334 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
22336 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
22339 2017-04-06 Richard Biener <rguenther@suse.de>
22341 PR tree-optimization/80334
22342 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
22343 preserve alignment of accesses.
22345 2017-04-06 Richard Biener <rguenther@suse.de>
22347 PR tree-optimization/80262
22348 * tree-sra.c (build_ref_for_offset): Preserve address-space
22350 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
22351 Drop useless address-space information on MEM_REF offsets.
22353 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
22355 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
22357 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
22359 PR rtl-optimization/70703
22360 * ira-color.c (update_conflict_hard_regno_costs): Use
22361 HOST_WIDE_INT instead of long.
22363 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
22366 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
22367 not defined for x86_64 target. Add -mmmx target option when __SSE2__
22369 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
22370 for x86_64 target. Handle -m3dnowa option.
22372 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
22374 PR rtl-optimization/70703
22375 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
22376 (update_conflict_hard_regno_costs): Use long instead of unsigned
22377 arithmetic for cost calculation.
22379 2017-04-05 Jakub Jelinek <jakub@redhat.com>
22380 Bernd Edlinger <bernd.edlinger@hotmail.de>
22383 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
22386 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
22389 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
22390 ptr_mode with Pmode throughout.
22391 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
22392 into probe_stack_range and use DImode.
22394 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
22397 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
22398 call_eh_return is true.
22400 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22402 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
22403 Initialize last_match_fntype_index.
22405 2017-04-05 Jakub Jelinek <jakub@redhat.com>
22408 * tree-nvr.c: Include internal-fn.h.
22409 (pass_return_slot::execute): Ignore internal calls without
22412 2017-04-04 Jakub Jelinek <jakub@redhat.com>
22413 Richard Biener <rguenther@suse.de>
22416 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
22417 captures used multiple times, except for the last use.
22418 * generic-match-head.c: Include gimplify.h.
22420 2017-04-04 Jakub Jelinek <jakub@redhat.com>
22422 PR tree-optimization/79390
22423 * target.h (struct noce_if_info): Declare.
22424 * targhooks.h (default_noce_conversion_profitable_p): Declare.
22425 * target.def (noce_conversion_profitable_p): New target hook.
22426 * ifcvt.h (struct noce_if_info): New type, moved from ...
22427 * ifcvt.c (struct noce_if_info): ... here.
22428 (noce_conversion_profitable_p): Renamed to ...
22429 (default_noce_conversion_profitable_p): ... this. No longer
22431 (noce_try_store_flag_constants, noce_try_addcc,
22432 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
22433 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
22434 instead of noce_conversion_profitable_p.
22435 * config/i386/i386.c: Include ifcvt.h.
22436 (ix86_option_override_internal): Don't override
22437 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
22438 (ix86_noce_conversion_profitable_p): New function.
22439 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
22440 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
22441 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
22442 * doc/tm.texi: Regenerated.
22444 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22446 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
22449 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22452 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
22453 instructions for small multiply cores.
22455 2017-04-04 Jeff Law <law@redhat.com>
22457 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
22459 (mips_expand_vec_perm_const): Initialize elements in orig_perm
22460 that are not set by the loop over the elements.
22462 2017-04-04 Jakub Jelinek <jakub@redhat.com>
22465 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
22466 int mode, convert_modes it to mode as unsigned, otherwise use
22467 lowpart_subreg to mode rather than SImode.
22468 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
22469 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
22470 Use DImode instead of SImode for the shift count operand.
22471 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
22474 2017-04-04 Richard Biener <rguenther@suse.de>
22476 PR middle-end/80281
22477 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
22478 arithmetic done for the negate or the plus. Simplify.
22479 (A - (-B) -> A + B): Likewise.
22480 * fold-const.c (split_tree): Make sure to not negate pointers.
22482 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
22484 PR rtl-optimization/60818
22485 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
22486 a compare of comparisons with the thing compared if this results
22487 in a different machine mode.
22489 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
22491 * alias.c (base_alias_check): Fix typo in comment.
22492 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
22493 * cgraphunit.c (symbol_table::compile): Likewise.
22494 * collect2.c (maybe_run_lto_and_relink): Likewise.
22495 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
22496 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
22497 * config/avr/avr.c (avr_map_op_t): Likewise.
22498 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
22499 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
22500 * config/epiphany/epiphany.md (movcc): Likewise.
22501 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
22502 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
22504 * config/mips/mips.c (mips_save_restore_reg): Likewise.
22505 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
22506 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
22507 * config/sh/sh.c (sh_rtx_costs): Likewise.
22508 * fold-const.c (fold_truth_andor): Likewise.
22509 * genautomata.c (collapse_flag): Likewise.
22510 * gengtype.h (struct type::u::s): Likewise.
22511 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
22512 * input.c (FORMAT_AMOUNT): Likewise.
22513 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
22514 (known_aggs_to_agg_replacement_list): Likewise.
22515 * ipa-inline-analysis.c: Likewise.
22516 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
22517 * ipa-polymorphic-call.c
22518 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
22519 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
22520 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
22522 * modulo-sched.c (apply_reg_moves): Likewise.
22523 * omp-expand.c (build_omp_regions_1): Likewise.
22524 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
22525 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
22526 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
22527 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
22528 * value-prof.c: Likewise.
22529 * var-tracking.c (val_reset): Likewise.
22531 2017-04-03 Richard Biener <rguenther@suse.de>
22533 PR tree-optimization/80275
22534 * fold-const.c (split_address_to_core_and_offset): Handle
22537 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
22539 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
22540 descriptors is at least equal to that of functions.
22542 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
22544 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
22546 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
22549 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
22550 (mov<IMOD4:mode>): New expander.
22551 (*mov<IMOD4:mode>_internal): New insn and split pattern.
22553 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
22555 PR rtl-optimization/79405
22556 * fwprop.c (propagations_left): New variable.
22557 (forward_propagate_into): Decrement it.
22558 (fwprop_init): Initialize it.
22559 (fw_prop): If the variable has reached zero, stop propagating.
22560 (fwprop_addr): Ditto.
22562 2017-03-31 Jakub Jelinek <jakub@redhat.com>
22565 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
22566 a FUNCTION_DECL, pass it as decl instead of origin to
22569 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
22571 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
22574 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
22577 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
22578 TARGET_VSX_SMALL_INTEGER.
22580 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22582 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
22583 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
22585 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
22587 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
22588 extraction from odd-numbered MSA register.
22590 2017-03-31 Jakub Jelinek <jakub@redhat.com>
22592 PR middle-end/80173
22593 * expmed.c (store_bit_field_1): Don't attempt to create
22594 a word subreg out of hard registers wider than word if they
22595 have HARD_REGNO_NREGS of 1 for their mode.
22597 PR middle-end/80163
22598 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
22599 conversions to integer types wider than word and pointer.
22602 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
22603 (rtx_equal_for_cselib_p): Pass 0 to it.
22604 * cselib.c (cselib_hasher::equal): Likewise.
22605 (rtx_equal_for_cselib_1): Add depth argument. If depth
22606 is 128, don't look up VALUE locs and punt. Increment
22607 depth in recursive calls when walking VALUE locs.
22609 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
22611 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
22612 (make_gcov_file_name): Use the canonical path name for generating
22614 (read_line): Fix handling of files with ascii null bytes.
22616 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
22618 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
22619 to initialise a vector register instead
22620 of using a const_int.
22622 2017-03-30 Jakub Jelinek <jakub@redhat.com>
22624 PR translation/80189
22625 * gimplify.c (omp_default_clause): Use %qs instead of %s in
22626 diagnostic messages.
22628 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
22631 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
22632 (dfp_diex_<mode>): Update mode of operand 1.
22633 * doc/extend.texi (dxex, dxexq): Document change to return type.
22634 (diex, diexq): Document change to argument type.
22636 2017-03-30 Martin Jambor <mjambor@suse.cz>
22639 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
22640 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
22641 it reflects the signature changes performed at the callee side.
22642 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
22643 to cgraph_build_function_type_skip_args.
22644 (build_function_decl_skip_args): Adjust call to the above function.
22646 2017-03-30 Jakub Jelinek <jakub@redhat.com>
22649 * config/i386/sse.md
22650 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
22651 register as dest whenever it is a MEM not rtx_equal_p to the
22652 corresponding dup operand, and when forcing into reg move the
22653 reg into the memory afterwards.
22654 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
22655 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
22656 for the force_reg mode.
22657 (avx512vl_vextractf128<mode>): Use register as dest either
22658 always when a MEM, or when it is a MEM not rtx_equal_p to the
22659 corresponding dup operand, or even not when it is a CONST_VECTOR
22660 depending on the mode and lo vs. hi.
22661 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
22663 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
22664 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
22665 Likewise. Require that operands[2] is even.
22666 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
22667 Remove extraneous parens. Require that operands[2] is a multiple
22669 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
22670 operands[0] is a MEM if <mask_applied>, the predicates/constraints
22671 disallow memory then.
22673 2017-03-30 Richard Biener <rguenther@suse.de>
22675 PR tree-optimization/77498
22676 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
22677 to non-constants over backedges.
22679 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
22681 PR rtl-optimization/80233
22682 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
22683 as last_combined_insn. Do not test for BARRIER_P separately.
22685 2017-03-29 Andreas Schwab <schwab@suse.de>
22688 * calls.c (prepare_call_address): Convert funexp to Pmode before
22689 copying to temp reg.
22691 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22693 PR tree-optimization/80158
22694 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
22695 Handle possible future case of more than one alternate
22697 (replace_rhs_if_not_dup): Likewise.
22698 (replace_one_candidate): Likewise.
22700 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
22702 PR rtl-optimization/80193
22703 * ira.c (ira): Do not check allocation for LRA.
22705 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
22707 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
22708 (nvptx_output_simt_exit): Declare.
22709 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
22710 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
22711 (init_softstack_frame): Move initialization of crtl->is_leaf to...
22712 (nvptx_declare_function_name): ...here. Emit declaration of local
22713 memory space buffer for omp_simt_enter insn.
22714 (nvptx_output_unisimt_switch): New.
22715 (nvptx_output_softstack_switch): New.
22716 (nvptx_output_simt_enter): New.
22717 (nvptx_output_simt_exit): New.
22718 * config/nvptx/nvptx.h (struct machine_function): New fields
22719 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
22720 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
22721 (UNSPECV_SIMT_EXIT): Ditto.
22722 (omp_simt_enter_insn): New insn.
22723 (omp_simt_enter): New expansion.
22724 (omp_simt_exit): New insn.
22725 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
22727 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
22728 (expand_GOMP_SIMT_ENTER_ALLOC): New.
22729 (expand_GOMP_SIMT_EXIT): New.
22730 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
22731 (GOMP_SIMT_ENTER_ALLOC): Ditto.
22732 (GOMP_SIMT_EXIT): Ditto.
22733 * target-insns.def (omp_simt_enter): New insn.
22734 (omp_simt_exit): Ditto.
22735 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
22737 (lower_rec_simd_input_clauses): Implement SIMT privatization.
22738 (lower_rec_input_clauses): Likewise.
22739 (lower_lastprivate_clauses): Handle SIMT privatization.
22741 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
22742 (ompdevlow_adjust_simt_enter): New.
22743 (find_simtpriv_var_op): New.
22744 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
22745 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
22747 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
22748 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
22749 (copy_decl_for_dup_finish): Ditto.
22751 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
22753 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
22756 * config/i386/i386.c (ix86_option_override_internal): Always
22757 allow -mpreferred-stack-boundary=3 for 64-bit targets.
22759 2017-03-28 Bin Cheng <bin.cheng@arm.com>
22761 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
22763 2017-03-28 Bin Cheng <bin.cheng@arm.com>
22765 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
22766 mark new edge's irreducible flag accordign to it.
22767 (vect_do_peeling): Check loop preheader edge's irreducible flag
22768 and pass it to function slpeel_add_loop_guard.
22770 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
22772 PR tree-optimization/80218
22773 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
22774 Update block frequencies and counts.
22776 2017-03-28 Richard Biener <rguenther@suse.de>
22778 PR tree-optimization/78644
22779 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
22780 of a simplification result we may not use it at all.
22782 2017-03-28 Richard Biener <rguenther@suse.de>
22785 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
22786 without arguments, generate default definition of a SSA name.
22788 2017-03-28 Richard Biener <rguenther@suse.de>
22790 PR middle-end/80222
22791 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
22792 TYPE_REF_CAN_ALIAS_ALL references.
22793 * fold-const.c (fold_indirect_ref_1): Likewise.
22795 2017-03-28 Martin Liska <mliska@suse.cz>
22798 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
22799 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
22801 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
22802 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
22804 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
22805 (EXTRA_SPECS): Define.
22806 (SUBTARGET_EXTRA_SPECS): Likewise.
22807 (SUBTARGET_CPP_SPEC): Likewise.
22808 * config/arc/elf.h (EXTRA_SPECS): Renamed to
22809 SUBTARGET_EXTRA_SPECS.
22810 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
22812 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
22814 * config/arc/simdext.md (vst64_insn): Update pattern.
22815 (vld32wh_insn): Likewise.
22816 (vld32wl_insn): Likewise.
22817 (vld64_insn): Likewise.
22818 (vld32_insn): Likewise.
22820 2017-03-28 Marek Polacek <polacek@redhat.com>
22823 * fold-const.c (fold_comparison): Use protected_set_expr_location
22824 instead of SET_EXPR_LOCATION.
22826 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
22828 * tree.c (add_expr): Avoid name lookup warning.
22830 2017-03-27 Jeff Law <law@redhat.com>
22832 PR tree-optimization/80216
22833 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
22834 function name. Limit recursion depth.
22835 (record_temporary_equivalences): Corresponding changes.
22837 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
22839 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
22842 2017-03-27 Jakub Jelinek <jakub@redhat.com>
22845 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
22847 * cfgcleanup.c (reg_note_cfa_p): New array.
22848 (insns_have_identical_cfa_notes): New function.
22849 (old_insns_match_p): Don't cross-jump in between /f
22850 and non-/f instructions. If both i1 and i2 are frame related,
22851 verify all CFA notes, their order and content.
22853 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
22856 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
22857 HImode and SImode with zero extend to DImode to one insn.
22858 (bswap<mode>2_extenddi): Likewise.
22859 (bswapsi2_extenddi): Likewise.
22860 (bswaphi2_extendsi): Likewise.
22861 (bswaphi2): Combine bswap HImode and SImode into one insn.
22862 Separate memory insns from swapping register.
22863 (bswapsi2): Likewise.
22864 (bswap<mode>2): Likewise.
22865 (bswaphi2_internal): Delete, no longer used.
22866 (bswapsi2_internal): Likewise.
22867 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
22868 store, and gpr<-gpr swap insns.
22869 (bswap<mode>2_store): Likewise.
22870 (bswaphi2_reg): Register only splitter, combine with the splitter.
22871 (bswaphi2 splitter): Likewise.
22872 (bswapsi2_reg): Likewise.
22873 (bswapsi2 splitter): Likewise.
22874 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
22875 the insns into load, store, and register/register insns.
22876 (bswapdi2_ldbrx): Likewise.
22877 (bswapdi2_load): Likewise.
22878 (bswapdi2_store): Likewise.
22879 (bswapdi2_reg): Likewise.
22881 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
22883 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
22884 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
22886 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
22889 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
22891 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
22892 special handling for target option conflicts between dform
22893 options (-mpower9-dform, -mpower9-dform-vector,
22894 -mpower9-dform-scalar) and -mno-direct-move.
22896 2017-03-27 Richard Biener <rguenther@suse.de>
22898 PR tree-optimization/80181
22899 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
22901 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22903 * config/arc/predicates.md (move_double_src_operand): Replace the
22904 call to move_double_src_operand with a call to address_operand.
22906 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22908 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
22909 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
22910 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
22912 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22914 * config/arc/predicates.md (long_immediate_loadstore_operand):
22915 Consider scaled addresses cases.
22917 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22919 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
22920 restored when in interrupt.
22921 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
22922 doesn't have delay slot.
22924 2017-03-27 Richard Biener <rguenther@suse.de>
22927 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
22928 inlined thunk clones.
22930 2017-03-27 Jakub Jelinek <jakub@redhat.com>
22933 * asan.c (instrument_derefs): Copy over last operand from
22934 original COMPONENT_REF to the new COMPONENT_REF with
22935 DECL_BIT_FIELD_REPRESENTATIVE.
22936 * ubsan.c (instrument_object_size): Likewise.
22938 2017-03-27 Richard Biener <rguenther@suse.de>
22940 PR tree-optimization/80170
22941 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
22942 sure DR/SCEV didnt fold in constants we do not see when looking
22943 at the reference base alignment.
22945 2017-03-27 Richard Biener <rguenther@suse.de>
22947 PR middle-end/80171
22948 * gimple-fold.c (fold_ctor_reference): Properly guard against
22949 NULL return value from canonicalize_constructor_val.
22951 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
22954 * config/i386/i386.c (ix86_expand_builtin)
22955 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
22956 flags reg setting and flags reg using instructions.
22957 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
22958 clobbering instructions to zero extend op2.
22960 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
22962 * doc/install.texi (Configuration) <--with-aix-soname>:
22963 Update link to AIX ld.
22965 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
22967 PR rtl-optimization/80160
22968 PR rtl-optimization/80159
22969 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
22970 reg_alternate_class into account.
22972 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
22975 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
22976 to consider in curr_insn_transform.
22978 2017-03-24 Jakub Jelinek <jakub@redhat.com>
22980 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
22981 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
22982 and emit_mode_inner.
22984 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22986 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
22987 argument to the overloaded builtin variants. Use the new flag to
22988 deprecate certain builtin variants.
22989 * config/s390/s390-builtin-types.def: Add new builtin types.
22990 * config/s390/s390-builtins.h: Support new flags field for
22991 overloaded builtins.
22992 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
22993 (s390_macro_to_expand): Enable vector float data type.
22994 (s390_cpu_cpp_builtins_internal): Indicate support of the new
22995 builtins by incrementing the __VEC__ version number.
22996 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
22998 (s390_resolve_overloaded_builtin): Emit error messages depending
22999 on the builtin flags.
23000 * config/s390/s390.c (s390_expand_builtin): Support additional
23001 flags argument. Change error message to match the messages
23002 emitted in s390-c.c.
23003 * config/s390/s390.md: New UNSPEC_* constants.
23004 (op_type): Add new instruction types.
23005 * config/s390/vecintrin.h: Add new builtins and test data class
23007 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
23008 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
23009 (VEC_INEXACT, VEC_NOINEXACT): New constants.
23010 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
23011 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
23012 ("vec_mergel<mode>"): V_HW -> VEC_HW.
23014 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
23015 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
23016 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
23017 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
23019 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
23020 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
23021 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
23022 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
23024 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
23025 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
23026 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
23027 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
23028 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
23029 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
23030 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
23032 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
23033 ("vec_scatter_element<V_HW_4:mode>_DI")
23034 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
23035 ("vec_fpint<mode>", "vflls")
23036 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
23037 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
23038 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
23039 ("*vec_cmphe<mode>_cc"): ... these.
23041 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
23042 mode constant instead of magic value.
23044 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23046 * config/s390/s390.c (s390_expand_vec_compare): Support other
23047 vector floating point modes than just V2DF.
23048 (s390_expand_vcond): Likewise.
23049 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
23050 (s390_cannot_change_mode_class): Prevent mode changes between TF
23051 and V1TF in vector registers.
23052 * config/s390/s390.md (DF, SF): New mode attributes.
23053 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
23054 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
23055 SFmode support for VRs.
23056 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
23058 (VFT, VF_HW): New mode iterators.
23059 (vw, sdx): New mode attributes.
23060 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
23061 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
23062 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
23063 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
23064 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
23065 also the new vector floating point modes. Renaming to ...
23067 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
23068 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
23069 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
23070 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
23071 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
23072 ("vec_unordered<mode>"): ... these.
23074 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
23075 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
23076 ("*vec_extendv2df"): New insn definitions.
23078 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23080 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
23081 ("mulditi3_2", "*muldi3_sign"): New patterns.
23082 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
23083 rename the pattern definition.
23085 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23087 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
23089 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
23091 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23093 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
23094 instruction if possible.
23095 * config/s390/vector.md (vec_halfnumelts): New mode
23097 ("*vec_vllezlf<mode>"): New pattern.
23099 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23101 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
23102 ("popcountv4si2", "popcountv2di2"): Rename to ...
23103 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
23104 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
23106 ("popcount<mode>2_vxe"): New pattern.
23108 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23110 * common/config/s390/s390-common.c (processor_flags_table): Add
23112 * config.gcc: Add arch12.
23113 * config/s390/driver-native.c (s390_host_detect_local_cpu):
23114 Default to arch12 for unknown CPU model numbers.
23115 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
23116 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
23117 PROCESSOR_max sanity check.
23118 * config/s390/s390-opts.h (enum processor_type): Add
23120 * config/s390/s390.c (processor_table): Add arch12.
23121 (s390_expand_builtin): Add check for B_VXE flag.
23122 (s390_issue_rate): Add PROCESSOR_ARCH12.
23123 (s390_get_sched_attrmask): Likewise.
23124 (s390_get_unit_mask): Likewise.
23125 (s390_sched_score): Enable z13 scheduling for arch12.
23126 (s390_sched_reorder): Likewise.
23127 (s390_sched_variable_issue): Likewise.
23128 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
23130 (s390_tune_attr): Use z13 scheduling also for arch12.
23131 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
23132 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
23133 (TARGET_VXE_P): New macros.
23134 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
23135 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
23136 * config/s390/s390.opt: Add arch12 as processor_type.
23138 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23140 * config/s390/s390.md
23141 ("fixuns_truncdddi2", "fixuns_trunctddi2")
23142 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
23143 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
23145 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
23146 Rename expanders to ...
23148 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
23149 ("fixuns_truncdddi2_emu"): ... these.
23151 ("fixuns_trunc<mode>si2_emu"): New expander.
23153 ("*fixuns_truncdfdi2_z13"): Rename to ...
23154 ("*fixuns_truncdfdi2_vx"): ... this.
23156 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23158 * config/s390/2964.md: Remove the single element vector compare
23159 instructions which are no longer used.
23160 * config/s390/s390.c (s390_select_ccmode): Remove handling of
23162 (s390_canonicalize_comparison): Remove handling of DFmode
23164 (s390_expand_vec_compare_scalar): Remove function.
23165 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
23166 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
23168 ("*cmp<mode>_ccs"): Add wfcdb instruction.
23170 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23172 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
23174 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
23175 will anyway by matched by mov<mode>_64dfp.
23177 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23179 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
23180 vlef/vstef. Add missing operand to vleif.
23182 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23184 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
23185 pair for all vector types with 64 bit elements.
23186 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
23187 * config/s390/vector.md (V_HW_64): ... here.
23188 (V_128_NOSINGLE): New mode iterator.
23189 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
23190 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
23191 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
23192 ("*vec_load_pairv2di"): Change to ...
23193 ("*vec_load_pair<mode>"): ... this one.
23195 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23197 * config/s390/constraints.md: Add comments.
23198 (jKK): Reject element sizes > 8 bytes.
23199 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
23201 * config/s390/s390.md: Add the s_operand checks formerly in
23202 s390_split_ok_p to various splitters where they are still
23204 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
23205 for 128 bit vectors. Plus two splitters.
23207 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23209 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
23212 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23215 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
23216 error if the boundary argument is not constant.
23218 2017-03-24 Jakub Jelinek <jakub@redhat.com>
23220 PR rtl-optimization/80112
23221 * loop-doloop.c (doloop_condition_get): Don't check condition
23222 if cmp isn't SET with IF_THEN_ELSE src.
23224 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23226 PR tree-optimization/80158
23227 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
23228 replacing a candidate statement, also replace it for the
23229 candidate's alternate interpretation.
23230 (replace_rhs_if_not_dup): Likewise.
23231 (replace_one_candidate): Likewise.
23233 2017-03-24 Richard Biener <rguenther@suse.de>
23235 PR tree-optimization/80167
23236 * graphite-isl-ast-to-gimple.c
23237 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
23239 (translate_isl_ast_to_gimple::get_rename): Likewise.
23241 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
23243 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
23244 handling of certain combinations of target options, including the
23245 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
23246 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
23248 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23251 * config/arm/arm.md (*load_multiple): Add reload_completed to
23252 matching condition.
23254 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23255 Richard Biener <rguenth@suse.de>
23257 PR tree-optimization/79908
23258 PR tree-optimization/80136
23259 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
23260 been cast away, gimplify_and_add suffices.
23262 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
23264 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
23266 2017-03-23 Richard Biener <rguenther@suse.de>
23268 PR tree-optimization/80032
23269 * gimplify.c (gimple_push_cleanup): Forced unconditional
23270 cleanups still have to go to the conditional_cleanups
23273 2017-03-22 Jakub Jelinek <jakub@redhat.com>
23275 PR tree-optimization/80072
23276 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
23278 (next_operand_entry_id): Change type to unsigned int.
23279 (sort_by_operand_rank): Make sure to return the right return value
23280 even if unsigned fields are bigger than INT_MAX.
23281 (struct oecount): Change cnt and id type to unsigned int.
23282 (oecount_hasher::equal): Formatting fix.
23283 (oecount_cmp): Make sure to return the right return value
23284 even if unsigned fields are bigger than INT_MAX.
23285 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
23288 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
23289 TREE_READONLY on result if writing it more than once.
23292 * doc/invoke.texi (-fsanitize=thread): Document that with
23293 -fnon-call-exceptions atomics are not able to throw
23297 * tsan.c: Include tree-eh.h.
23298 (instrument_builtin_call): Call maybe_clean_eh_stmt or
23299 maybe_clean_or_replace_eh_stmt where needed.
23300 (instrument_memory_accesses): Add cfg_changed argument.
23301 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
23302 if it returned true.
23303 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
23305 PR rtl-optimization/63191
23306 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
23307 wrapper function, moved the whole old content into ...
23308 (ix86_delegitimize_address_1): ... this. New inline function.
23309 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
23310 true as last argument instead of ix86_delegitimize_address.
23312 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
23314 * config/aarch64/aarch64.c (generic_branch_cost): Copy
23315 cortexa57_branch_cost.
23317 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
23319 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
23321 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23324 * doc/md.texi (Constraints): Document wA constraint.
23325 * config/rs6000/constraints.md (wA): New.
23326 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
23327 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
23328 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
23329 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
23331 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
23334 * gimplify.c (is_oacc_declared): New function.
23335 (oacc_default_clause): Use it to set default flags for acc declared
23336 variables inside parallel regions.
23337 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
23338 declared variables.
23339 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
23340 declare attribute to any decl as necessary.
23342 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
23345 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
23346 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
23347 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
23348 (arm_arch_lpae): This.
23349 * config/arm/arm.c (arm_arch7ve): Rename into ...
23350 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
23351 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
23354 2017-03-22 Martin Liska <mliska@suse.cz>
23357 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
23358 error message instead of an ICE.
23360 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23362 * doc/extend.texi (6.11 Additional Floating Types): Revise.
23364 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
23366 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
23368 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
23371 2017-03-21 Martin Sebor <msebor@redhat.com>
23373 * doc/extend.texi: Use "cannot" instead of "can't."
23374 * doc/hostconfig.texi: Same.
23375 * doc/install.texi: Same.
23376 * doc/invoke.texi: Same.
23377 * doc/loop.texi: Same.
23378 * doc/md.texi: Same.
23379 * doc/objc.texi: Same.
23380 * doc/rtl.texi: Same.
23381 * doc/tm.texi: Same.
23382 * doc/tm.texi.in: Same.
23383 * doc/trouble.texi: Same.
23385 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
23388 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
23389 (collect_checksum_attributes): Set it.
23390 (die_checksum_ordered): Use it.
23392 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23394 PR tree-optimization/79908
23395 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
23396 change: For a VA_ARG whose LHS has been cast away, use
23397 force_gimple_operand to construct the side effects.
23399 2017-03-21 David Malcolm <dmalcolm@redhat.com>
23401 PR translation/80001
23402 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
23403 more amenable to translation.
23404 (oacc_loop_auto_partitions): Likewise.
23406 2017-03-21 Marek Polacek <polacek@redhat.com>
23407 Martin Sebor <msebor@redhat.com>
23409 PR tree-optimization/80109
23410 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
23411 on INTEGRAL_TYPE_P.
23413 2017-03-21 Jakub Jelinek <jakub@redhat.com>
23414 Segher Boessenkool <segher@kernel.crashing.org>
23417 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
23418 check reg_used_between_p between insn and one of succ or succ2
23419 depending on if succ is artificial insn not inserted into insn
23422 2017-03-21 Martin Liska <mliska@suse.cz>
23424 PR gcov-profile/80081
23425 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
23426 * doc/gcc.texi: Include gcov-dump stuff.
23427 * doc/gcov-dump.texi: New file.
23429 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
23431 PR rtl-optimization/79150
23432 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
23433 conditional jump, if the jump is the last insn of the loop.
23435 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23436 Richard Biener <rguenth@suse.de>
23438 PR tree-optimization/79908
23439 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
23440 been cast away, use force_gimple_operand to construct the side
23443 2017-03-21 Martin Liska <mliska@suse.cz>
23445 PR libfortran/79956
23446 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
23449 2017-03-21 Brad Spengler <spender@grsecurity.net>
23452 * plugin.c (htab_hash_plugin): New function.
23453 (add_new_plugin): Use it and adjust.
23454 (parse_plugin_arg_opt): Adjust.
23455 (init_one_plugin): Likewise.
23457 2017-03-21 Richard Biener <rguenther@suse.de>
23459 PR tree-optimization/80032
23460 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
23461 if set force the cleanup to happen unconditionally.
23462 (gimplify_target_expr): Push inserted clobbers with force_uncond
23463 to avoid them being removed by control-dependent DCE.
23465 2017-03-21 Richard Biener <rguenther@suse.de>
23467 PR tree-optimization/80122
23468 * tree-inline.c (copy_bb): Do not expans va-arg packs or
23469 va_arg_pack_len when the inlined call stmt requires pack
23471 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
23473 2017-03-21 Jakub Jelinek <jakub@redhat.com>
23476 * tsan.c (instrument_builtin_call): If the memory model argument
23477 is not a constant, assume it is valid.
23480 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
23483 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
23485 PR rtl-optimization/79910
23486 * combine.c (can_combine_p): Do not allow combining an I0 or I1
23487 if its dest is used by an insn before I2 (other than the combined
23488 insns themselves, which are properly handled already).
23490 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
23493 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
23495 * combine.c (record_used_regs): New static function.
23496 (try_combine): Handle situations where there is an additional
23497 instruction between I2 and I3 which needs to have a LOG_LINK
23501 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
23503 * combine.c (try_combine): Delete redundant i1 test. Call
23504 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
23506 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23509 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
23510 alternatives 13/14.
23512 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23514 PR tree-optimization/80054
23515 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
23516 the optimization if a PHI or any of its arguments is not dominated
23517 by the candidate's basis. Use gphi* rather than gimple* as
23519 (replace_profitable_candidates): Clean up a gimple* variable that
23520 should be a gphi* variable.
23522 2017-03-20 Martin Sebor <msebor@redhat.com>
23525 * doc/extend.texi (attribute constructor): Document present limitation.
23527 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
23530 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
23531 __POWER9_VECTOR__ #ifdef control, change template definition to
23532 use Power9-specific built-in function.
23533 (vec_any_eq): Likewise.
23534 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
23535 to control outcomes from this test.
23536 (vector_ae_<mode>p): For VEC_F modes, likewise.
23538 2017-03-20 Ian Lance Taylor <iant@google.com>
23540 * config/i386/i386.c (ix86_function_regparm): Save an extra
23541 register for -fsplit-stack with DECL_STATIC_CHAIN.
23543 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
23546 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
23547 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
23549 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
23551 * config/riscv/riscv.c (riscv_print_operand): Use "fence
23553 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
23556 2017-03-20 Marek Polacek <polacek@redhat.com>
23559 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
23561 2017-03-20 Richard Biener <rguenther@suse.de>
23563 PR tree-optimization/80113
23564 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
23565 allocate extra SSA name for PHI def.
23566 (add_close_phis_to_outer_loops): Likewise.
23567 (add_close_phis_to_merge_points): Likewise.
23568 (copy_loop_close_phi_args): Likewise.
23569 (copy_cond_phi_nodes): Likewise.
23571 2017-03-20 Martin Liska <mliska@suse.cz>
23573 PR middle-end/79753
23574 * tree-chkp.c (chkp_build_returned_bound): Do not build
23575 returned bounds for a LHS that's not a BOUNDED_P type.
23577 2017-03-20 Martin Liska <mliska@suse.cz>
23581 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
23582 COMPLEX_CST and VECTOR_CST.
23584 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23587 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
23588 target operand. A new splitter adds the clobber statement in case
23589 the target operand is dead anyway.
23591 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
23593 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
23594 to age-old versions of binutils and glibc.
23596 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
23598 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
23600 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
23602 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
23604 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
23606 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
23607 requirement for binutils 2.13.
23609 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
23611 * combine.c (try_combine): Delete redundant i1 test. Call
23612 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
23614 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
23616 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
23617 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
23619 <riscv64-*-elf>: Re-arrange section
23620 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
23621 <riscv32-*-linux>: Likewise.
23622 <riscv64-*-elf>: Likewise
23623 <riscv64-*-linux>: Likewise.
23625 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
23628 * aarch64.opt(verbose-cost-dump): Fix typo.
23630 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
23633 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
23634 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
23636 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
23638 * reload.c (find_reloads): When reloading a nonoffsettable address,
23639 use RELOAD_OTHER for it and its address reloads.
23641 PR rtl-optimization/79910
23642 * combine.c (record_used_regs): New static function.
23643 (try_combine): Handle situations where there is an additional
23644 instruction between I2 and I3 which needs to have a LOG_LINK
23647 2017-03-17 Jeff Law <law@redhat.com>
23649 PR tree-optimization/71437
23650 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
23651 conditional in the hash table first.
23652 (vrp_dom_walker::before_dom_children): Extract condition from
23653 ASSERT_EXPR. Record condition, its inverion and any implied
23654 conditions as well.
23656 2017-03-17 Marek Polacek <polacek@redhat.com>
23657 Markus Trippelsdorf <markus@trippelsdorf.de>
23659 PR tree-optimization/80079
23660 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
23663 2017-03-17 Richard Biener <rguenther@suse.de>
23665 PR middle-end/80075
23666 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
23667 Properly verify the LHS before the RHS possibly claims to be
23669 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
23672 2017-03-17 Martin Jambor <mjambor@suse.cz>
23674 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
23675 (List of -O2 options): Likewise.
23676 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
23679 2017-03-17 Tom de Vries <tom@codesourcery.com>
23681 * gcov-dump.c (print_usage): Print bug_report_url.
23683 2017-03-17 Richard Biener <rguenther@suse.de>
23685 PR middle-end/80050
23686 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
23687 (parser::peek): Likewise.
23689 2017-03-17 Richard Biener <rguenther@suse.de>
23691 PR tree-optimization/80048
23692 * sese.c (free_sese_info): Properly release rename_map and
23693 copied_bb_map elements.
23695 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
23697 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
23698 Add linked-list forward and backlinks. Insert on
23699 construction, remove on destruction.
23700 (class pass_store_merging): Add m_stores_head field.
23701 (pass_store_merging::terminate_and_process_all_chains):
23702 Iterate over m_stores_head list.
23703 (pass_store_merging::terminate_all_aliasing_chains):
23705 (pass_store_merging::execute): Check for debug stmts first.
23706 Push new chains onto the m_stores_head stack.
23708 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
23711 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
23712 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
23713 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
23715 2017-03-16 Jeff Law <law@redhat.com>
23717 PR tree-optimization/71437
23718 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
23719 member function. Implementation moved into after_dom_children
23720 member function and into the threader's thread_outgoing_edges
23722 (dom_opt_dom_walker::after_dom_children): Simplify by moving
23723 some code into new thread_outgoing_edges.
23724 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
23725 definition. Simplify marker handling (do it here). Assume we always
23726 have the available expression and the const/copies tables.
23727 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
23729 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
23730 * tree-vrp.c (equiv_stack): No longer file scoped.
23731 (vrp_dom_walker): New class.
23732 (vrp_dom_walker::before_dom_children): New member function.
23733 (vrp_dom_walker::after_dom_children): Likewise.
23734 (identify_jump_threads): Setup domwalker. Use it rather than
23735 walking edges in a random order by hand. Simplify setup/finalization.
23736 (finalize_jump_threads): Remove.
23737 (vrp_finalize): Do not call identify_jump_threads here.
23738 (execute_vrp): Do it here instead and call thread_through_all_blocks
23741 PR tree-optimization/71437
23742 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
23744 (simplify_stmt_for_jump_threading): Add basic_block argument. All
23746 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
23747 (dom_opt_dom_walker::thread_across_edge): Remove
23748 handle_dominating_asserts argument. All callers changed.
23749 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
23750 changes. Remove calls to lhs_of_dominating_assert. Other
23751 uses of handle_dominating_asserts turn into unconditional code
23752 (simplify_control_stmt_condition_1): Likewise.
23753 (simplify_control_stmt_condition): Likewise.
23754 (thread_through_normal_block, thread_across_edge): Likewise.
23755 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
23756 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
23757 object if it is not an SSA_NAME.
23758 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
23759 before calling into the VRP specific simplifiers.
23760 (identify_jump_threads): Remove handle_dominating_asserts
23763 2017-03-16 Jakub Jelinek <jakub@redhat.com>
23766 * tree-diagnostic.c (default_tree_printer): No longer static.
23767 * tree-diagnostic.h (default_tree_printer): New prototype.
23769 2017-03-16 Tamar Christina <tamar.christina@arm.com>
23771 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
23772 Change ins into fmov.
23774 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23776 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
23777 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
23778 Use h_con constraint for operand 1.
23779 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
23780 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
23782 2017-03-15 Jeff Law <law@redhat.com>
23784 PR tree-optimization/71437
23785 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
23786 (record_temporary_equivalences): Use it.
23788 PR tree-optimization/71437
23789 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
23790 tree-ssa-scopedtables.
23791 (lookup_avail_expr, build_and_record_new_cond): Likewise.
23792 (record_conditions, record_cond, vuse_eq): Likewise.
23793 (record_edge_info): Adjust to API tweak of record_conditions.
23794 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
23795 (record_temporary_equivalences, optimize_stmt): Likewise.
23796 (eliminate_redundant_computations): Likewise.
23797 (record_equivalences_from_stmt): Likewise.
23798 * tree-ssa-scopedtables.c: Include options.h and params.h.
23799 (vuse_eq): New function, moved from tree-ssa-dom.c
23800 (build_and_record_new_cond): Likewise.
23801 (record_conditions): Likewise. Accept vector of conditions rather
23802 than edge_equivalence structure for first argument.
23803 for the first argument.
23804 (avail_exprs_stack::lookup_avail_expr): New member function, moved
23805 from tree-ssa-dom.c.
23806 (avail_exprs_stack::record_cond): Likewise.
23807 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
23808 from tree-ssa-dom.c.
23809 (avail_exprs_stack): Add new member functions lookup_avail_expr
23811 (record_conditions): Declare.
23813 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
23816 * lra-constraints.c (process_alt_operands): Increase reject for
23817 reloading an input/output operand.
23819 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
23822 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
23823 insns to convert from signed/unsigned char/short to IEEE 128-bit
23825 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
23827 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
23830 * config/i386/i386.c (ix86_vector_duplicate_value): Create
23831 subreg of inner mode for values already in registers.
23833 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
23835 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
23836 iteration reg is used after the loop.
23838 2017-03-14 Martin Sebor <msebor@redhat.com>
23840 PR tree-optimization/79800
23841 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
23842 precision in negative-positive range.
23843 (format_floating): Call non-const overload with adjusted precision.
23845 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
23848 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
23851 2017-03-14 Martin Sebor <msebor@redhat.com>
23853 PR middle-end/80020
23854 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
23855 * builtins.def (aligned_alloc): Use it.
23858 * Makefile.in (GTFILES): Add calls.c.
23859 * calls.c: Include "gt-calls.h".
23861 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
23863 PR rtl-optimization/79728
23864 * regs.h (struct target_regs): New field
23865 x_contains_allocatable_regs_of_mode.
23866 (contains_allocatable_regs_of_mode): New macro.
23867 * reginfo.c (init_reg_sets_1): Initialize it, and change
23868 contains_reg_of_mode so it includes global regs as well.
23869 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
23870 rather than contains_regs_of_mode.
23872 2017-03-14 Martin Liska <mliska@suse.cz>
23874 * doc/invoke.texi: Document options that can't be combined with
23875 -fcheck-pointer-bounds.
23877 2017-03-14 Martin Liska <mliska@suse.cz>
23879 PR middle-end/79831
23880 * doc/invoke.texi (-Wchkp): Document the option.
23882 2017-03-14 Martin Liska <mliska@suse.cz>
23884 * Makefile.in: Install gcov-dump.
23886 2017-03-14 Martin Liska <mliska@suse.cz>
23888 * multiple_target.c (expand_target_clones): Bail out for
23889 an invalid attribute.
23891 2017-03-14 Richard Biener <rguenther@suse.de>
23893 * alias.c (struct alias_set_entry): Pack properly.
23894 * cfgloop.h (struct loop): Likewise.
23895 * cse.c (struct set): Likewise.
23896 * ipa-utils.c (struct searchc_env): Likewise.
23897 * loop-invariant.c (struct invariant): Likewise.
23898 * lra-remat.c (struct cand): Likewise.
23899 * recog.c (struct change_t): Likewise.
23900 * rtl.h (struct address_info): Likewise.
23901 * symbol-summary.h (function_summary): Likewise.
23902 * tree-loop-distribution.c (struct partition): Likewise.
23903 * tree-object-size.c (struct object_size_info): Likewise.
23904 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
23905 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
23906 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
23907 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
23908 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
23909 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
23910 (struct _stmt_vec_info): Likewise.
23912 2017-03-14 Martin Liska <mliska@suse.cz>
23915 * multiple_target.c (create_dispatcher_calls): Check that
23916 a target can create a function dispatcher.
23918 2017-03-14 Martin Liska <mliska@suse.cz>
23921 * multiple_target.c (expand_target_clones): Drop local.local
23922 flag for default implementation.
23924 2017-03-14 Richard Biener <rguenther@suse.de>
23926 PR tree-optimization/80030
23927 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
23929 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
23931 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
23932 gcc_fallthrough() instead of __attribute__((fallthrough));
23934 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
23936 * doc/gcc.texi: Remove "up" link to (DIR).
23937 * doc/gccint.texi: Ditto.
23939 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
23941 * doc/install.texi (Specific) <avr>: Remove reference to
23944 2017-03-13 Jeff Law <law@redhat.com>
23946 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
23947 attribute rather than comments.
23949 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
23950 match_scratch operand is highest.
23952 2017-03-13 Martin Liska <mliska@suse.cz>
23954 PR middle-end/78339
23955 * ipa-pure-const.c (warn_function_noreturn): If the declarations
23956 is a CHKP clone, use original declaration.
23958 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23960 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
23961 (arc_conditional_register_usage): Use a different allocation order
23962 when optimizing for size.
23963 * common/config/arc/arc-common.c (arc_option_optimization_table):
23964 Section anchors default on when optimizing for size.
23966 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23968 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
23970 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23972 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
23973 * config/arc/arc.md (cpu_facility): Add cd variant.
23974 (*movqi_insn): Add code density variant.
23975 (*movhi_insn): Likewise.
23976 (*movqi_insn): Likewise.
23977 (*addsi3_mixed): Likewise.
23978 (subsi3_insn): Likewise.
23980 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23982 * config/arc/arc.md (movsi_cond_exec): Update constraint.
23984 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23986 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
23987 expressions with MINUS and UNARY ops.
23989 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23992 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
23994 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
23995 between vec_select and vector argument.
23996 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
23997 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
23998 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
23999 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
24000 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
24001 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
24003 2017-03-13 Richard Biener <rguenther@suse.de>
24006 * params.def (vect-max-peeling-for-alignment): Fix typo.
24008 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
24010 * doc/install.texi (Specific) <mips-*-*>: Remove description of
24011 issue that only occurred with binutils below 2.18.
24013 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
24015 * doc/install.texi (Specific) <cris-axis-elf>: No longer
24016 refer to binutils 2.11/2.12 minimum.
24018 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
24020 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
24021 ftp.kernel.org and simplify binutils requirement.
24023 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
24025 * doc/invoke.texi (Warning Options): Fix spelling of link-time
24027 (Optimize Options): Ditto. Also remove redundancy.
24029 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24031 PR translation/79848
24032 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
24034 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
24035 to G_ to avoid double translation.
24037 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24039 PR translation/79923
24040 * auto-profile.c (get_combined_location): Convert leading
24041 character of diagnostics to lower case and remove trailing period.
24042 (read_profile): Likewise for various diagnostics.
24043 * config/arm/arm.c (arm_option_override): Remove trailing period
24044 from various diagnostics.
24045 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
24046 (msp430_expand_delay_cycles): Likewise.
24048 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24051 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
24052 full command-line argument, rather than just "str".
24053 (aarch64_validate_march): Likewise.
24054 (aarch64_validate_mtune): Likewise.
24056 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
24058 PR rtl-optimization/78911
24059 * lra-assigns.c (must_not_spill_p): New function.
24060 (spill_for): Use it.
24062 2017-03-10 Jakub Jelinek <jakub@redhat.com>
24064 PR tree-optimization/79981
24065 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
24066 ATOMIC_COMPARE_EXCHANGE ifn result.
24067 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
24068 IFN_ATOMIC_COMPARE_EXCHANGE.
24070 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24073 * opts.c (parse_sanitizer_options): Add missing question mark to
24074 "did you mean" message.
24076 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24078 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
24080 (VMULEUH_UNS): Likewise.
24081 (VMULOUB_UNS): Likewise.
24082 (VMULOUH_UNS): Likewise.
24083 * config/rs6000/rs6000.c (builtin_function_type): Remove
24084 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
24086 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24089 * read-rtl-function.c (function_reader::read_rtx_operand): Update
24090 x with result of extra_parsing_for_operand_code_0.
24091 (function_reader::extra_parsing_for_operand_code_0): Convert
24092 return type from void to rtx, returning x. When reading
24093 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
24094 larger size containing struct block_symbol.
24096 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
24098 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
24099 -mfloat128-hardware without -m64.
24101 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
24104 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
24105 entries to the case statement that marks unsigned arguments to
24106 overloaded functions.
24108 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
24110 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
24111 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
24113 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
24116 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
24117 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
24119 2017-03-10 Martin Liska <mliska@suse.cz>
24123 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
24124 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
24127 2017-03-10 Olivier Hainque <hainque@adacore.com>
24129 * tree-switch-conversion (array_value_type): Start by resetting
24130 candidate type to it's main variant.
24132 2017-03-10 Jakub Jelinek <jakub@redhat.com>
24134 PR rtl-optimization/79909
24135 * combine.c (try_combine): Use simplify_replace_rtx on individual
24136 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
24137 of the whole CALL_INSN_FUNCTION_USAGE.
24139 PR tree-optimization/79972
24140 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
24141 get_range_info on SSA_NAMEs. Formatting fixes.
24143 2017-03-10 Richard Biener <rguenther@suse.de>
24144 Jakub Jelinek <jakub@redhat.com>
24146 PR tree-optimization/77975
24147 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
24148 edge to be constant.
24149 (get_val_for): For constant x return it. Formatting fix.
24150 (loop_niter_by_eval): Avoid pointless looping if the next iteration
24151 would use the same bases as the current one.
24153 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24155 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
24156 instead of vec_select for V1TImode.
24157 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
24159 (VSX_LE_128): Add V1TI to this mode iterator.
24160 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
24161 (*vsx_le_perm_store_<mode>): Likewise.
24162 (pre-reload splitter for VSX stores): Likewise.
24163 (post-reload splitter for VSX stores): Likewise.
24164 (*vsx_xxpermdi2_le_<mode>): Likewise.
24165 (*vsx_lxvd2x2_le_<mode>): Likewise.
24166 (*vsx_stxvd2x2_le_<mode>): Likewise.
24168 2017-03-09 Michael Eager <eager@eagercon.com>
24170 Correct failures with --enable-checking=yes,rtl.
24172 * config/microblaze/microblaze.c (microblaze_expand_shift):
24173 Replace GET_CODE test with CONST_INT_P and INTVAL test with
24174 test for const0_rtx.
24175 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
24176 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
24178 2017-03-09 Richard Biener <rguenther@suse.de>
24180 PR tree-optimization/79977
24181 * graphite-scop-detection.c (scop_detection::merge_sese):
24182 Handle the case of extra exits to blocks dominating the entry.
24184 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
24186 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
24189 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
24191 PR rtl-optimization/79949
24192 * lra-constraints.c (process_alt_operands): Check memory when
24193 trying to predict a cycle. Print about the overall increase.
24195 2017-03-09 Richard Biener <rguenther@suse.de>
24197 PR middle-end/79971
24198 * gimple-expr.c (useless_type_conversion_p): Preserve
24199 TYPE_SATURATING for fixed-point types.
24201 2017-03-09 Richard Biener <rguenther@suse.de>
24204 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
24205 alignment of BLKmode params.
24207 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24210 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
24211 (VALL_NO_V2Q): Likewise.
24213 * config/aarch64/aarch64-simd.md
24214 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
24216 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
24217 VALL_NO_V2Q mode iterator.
24218 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
24220 2017-03-09 Martin Liska <mliska@suse.cz>
24222 PR tree-optimization/79631
24223 * tree-chkp-opt.c (chkp_is_constant_addr): Call
24224 tree_int_cst_sign_bit just for INTEGER constants.
24226 2017-03-09 Martin Liska <mliska@suse.cz>
24230 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
24233 2017-03-09 Marek Polacek <polacek@redhat.com>
24236 * tree.c (inchash::add_expr): Handle TREE_VEC.
24238 2017-03-09 Martin Liska <mliska@suse.cz>
24241 (chkp_narrow_size_and_offset): New function.
24242 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
24243 (void chkp_parse_bit_field_ref): New function.
24244 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
24245 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
24247 2017-03-09 Martin Liska <mliska@suse.cz>
24250 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
24251 (chkp_find_bounds_1): Remove gcc_unreachable.
24253 2017-03-09 Jakub Jelinek <jakub@redhat.com>
24256 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
24257 BUILT_IN_SYNC*, determine the access type from the size suffix and
24258 always build a MEM_REF with that type. Handle forgotten
24259 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
24262 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
24263 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
24264 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
24265 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
24266 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
24267 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
24268 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
24269 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
24270 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
24271 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
24272 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
24273 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
24274 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
24275 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
24276 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
24277 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
24278 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
24279 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
24280 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
24281 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
24282 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
24283 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
24284 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
24285 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
24286 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
24287 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
24288 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
24289 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
24290 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
24291 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
24292 definitions outside of __OPTIMIZE__ guarded section.
24295 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
24296 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
24297 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
24298 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
24301 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24303 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
24304 ("vfenez<mode>"): Add missing constraints.
24306 2017-03-08 Martin Sebor <msebor@redhat.com>
24309 * config/nds32/nds32.c (nds32_option_override):
24310 Fix misspelled diagnostic.
24312 2017-03-08 Jakub Jelinek <jakub@redhat.com>
24315 * gimplify.c (gimplify_omp_for): Replace index var in outer
24316 taskloop statement with an artificial variable and add
24317 OMP_CLAUSE_PRIVATE clause for it.
24319 2017-03-08 Richard Biener <rguenther@suse.de>
24321 PR tree-optimization/79955
24322 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
24323 for accesses that are completely outside of the variable.
24325 2017-03-08 Andrew Haley <aph@redhat.com>
24327 PR tree-optimization/79943
24328 * tree-ssa-loop-split.c (compute_new_first_bound): When
24329 calculating the new upper bound, (END-BEG) should be added, not
24332 2017-03-08 Jakub Jelinek <jakub@redhat.com>
24334 * config/avr/avr.md (setmemhi): Make sure match_dup
24335 operand number comes before match_scratch.
24337 2017-03-08 Richard Biener <rguenther@suse.de>
24339 PR tree-optimization/79920
24340 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
24341 with ncopies == 1 to ...
24342 (vect_transform_slp_perm_load): ... here. Properly compute
24343 all element loads by iterating VF times over the group. Do
24344 not handle ncopies (computed in a broken way) in
24345 vect_create_mask_and_perm.
24347 2017-03-08 Jakub Jelinek <jakub@redhat.com>
24350 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
24351 is a uniform vector, use uniform_vector_p return value instead of
24352 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
24354 2017-03-07 Marek Polacek <polacek@redhat.com>
24356 PR middle-end/79809
24357 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
24358 (alloca_call_type): Likewise.
24360 2017-03-07 Martin Liska <mliska@suse.cz>
24362 * gcov.c (process_args): Put comment to correct location.
24364 2017-03-07 Martin Liska <mliska@suse.cz>
24366 PR middle-end/68270
24367 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
24368 Use array_at_struct_end_p instead of DECL_CHAIN (field).
24369 (chkp_narrow_bounds_for_field): Likewise.
24370 (chkp_parse_array_and_component_ref): Pass one more argument to
24373 2017-03-07 Richard Biener <rguenther@suse.de>
24375 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
24378 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
24380 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
24381 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
24383 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24386 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
24387 to end of description.
24388 (PARAM_MAX_STORES_TO_MERGE): Likewise.
24390 2017-03-07 Jakub Jelinek <jakub@redhat.com>
24392 PR rtl-optimization/79901
24393 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
24395 (*avx512f_<code><mode>3<mask_name>): ... this.
24396 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
24397 iterator instead of VI8_AVX2_AVX512BW.
24399 PR rtl-optimization/79901
24400 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
24401 min/max expander, expand it using expand_vec_cond_expr.
24404 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
24407 2017-03-06 Jakub Jelinek <jakub@redhat.com>
24410 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
24411 to void * for PCH reasons.
24412 * dwarf2out.c (output_loc_operands, output_die): Cast
24413 v.val_vec.array to unsigned char *.
24415 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
24418 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
24421 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
24423 PR rtl-optimization/79571
24424 * lra-constraints.c (process_alt_operands): Calculate static
24425 reject and subtract it from overall when only addresses will be
24428 2017-03-06 Julia Koval <julia.koval@intel.com>
24431 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
24432 incoming stack boundary to 128 for 64-bit targets.
24434 2017-03-06 Richard Biener <rguenther@suse.de>
24436 PR tree-optimization/79894
24437 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
24438 to NULL after folding it.
24440 2017-03-06 Richard Biener <rguenther@suse.de>
24442 PR tree-optimization/79824
24443 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
24444 check disabling peeling for gaps.
24446 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
24448 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
24449 attributes): Document gettimeofday.
24451 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
24453 * config/s390/s390.c (s390_option_override_internal): Set
24454 PARAM_MIN_VECT_LOOP_BOUND
24456 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
24458 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
24459 * config/s390/s390.md: Likewise.
24461 2017-03-06 Jakub Jelinek <jakub@redhat.com>
24464 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
24465 (<avx2_avx512>_perm<mode>): Rename to ...
24466 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
24468 (<avx512>_perm<mode>_mask): Rename to ...
24469 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
24471 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
24472 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
24473 instead of VI8F_256_512.
24474 (avx512f_perm<mode>): New define_expand.
24475 (avx512f_perm<mode>_mask): Likewise.
24476 (avx512f_perm<mode>_1<mask_name>): New define_insn.
24477 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
24479 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
24481 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
24482 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
24484 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
24486 2017-03-06 Martin Liska <mliska@suse.cz>
24489 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
24490 when having a SSA NAME w/o VAR_DECL assigned to it.
24492 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
24494 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
24495 msa_dpsub_<su>_d): Fix MODE for vec_select.
24497 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
24499 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
24501 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
24503 2017-03-06 Richard Biener <rguenther@suse.de>
24505 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
24506 * plugin.c (register_plugin_info): Likewise.
24507 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
24509 2017-03-05 Jakub Jelinek <jakub@redhat.com>
24511 * config/i386/sse.md (sse_storehps, sse_storelps,
24512 avx_<castmode><avxsizesuffix>_<castmode>,
24513 avx512f_<castmode><avxsizesuffix>_<castmode>,
24514 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
24515 in condition that at least one operand is not a MEM.
24517 2017-03-03 Jakub Jelinek <jakub@redhat.com>
24519 PR middle-end/79805
24520 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
24521 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
24523 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
24524 gimple_call_nothrow_p flag based on whether original builtin can throw.
24525 If it can, emit following stmts on the fallthrough edge.
24526 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
24527 don't create new bb if inserting just debug stmts on the edge, try to
24528 insert them on the fallthru bb or just reset debug stmts.
24530 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
24533 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
24534 restore recog_data (including the operand rtxes inside it) around
24535 the call to get_insn_template.
24537 2017-03-03 Martin Sebor <msebor@redhat.com>
24539 PR tree-optimization/79699
24540 * context.c (context::~context): Free MPFR caches to avoid
24541 a memory leak on program exit.
24543 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24545 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
24546 Use wide_int::ulow () instead of .elt (0).
24548 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
24550 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
24551 (*pushxf): Limit oF constraint to 32bit targets and add oC
24552 constraint for 64bit targets.
24553 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
24554 (*pushdf): Change rmF constraint to rmC.
24556 2017-03-03 Martin Liska <mliska@suse.cz>
24558 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
24559 Remove unused variable.
24561 2017-03-03 Jakub Jelinek <jakub@redhat.com>
24564 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
24565 is a memory operand, increase num_memory.
24566 (ix86_expand_args_builtin): Likewise.
24568 2017-03-03 Jan Hubicka <jh@suse.cz>
24571 * ipa-devirt.c (maybe_record_node): Properly handle
24572 __cxa_pure_virtual visibility.
24574 2017-03-03 Martin Liska <mliska@suse.cz>
24576 PR tree-optimization/79803
24577 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
24579 (pass_loop_prefetch::execute): Disabled optimization if an
24580 assumption about L1 cache size is not met.
24582 2017-03-03 Martin Liska <mliska@suse.cz>
24584 PR rtl-optimization/79574
24585 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
24586 (hash_scan_set): Likewise.
24587 (dump_hash_table): Likewise.
24588 (hoist_code): Likewise.
24590 2017-03-03 Richard Biener <rguenther@suse.de>
24592 * fixed-value.c (fixed_from_string): Restore use of elt (1)
24593 in place of uhigh ().
24594 (fixed_convert_from_real): Likewise.
24596 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
24599 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
24601 2017-03-03 Richard Biener <rguenther@suse.de>
24603 PR middle-end/79818
24604 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
24605 TYPE_OVERFLOW_UNDEFINED check.
24607 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24609 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
24611 (vector_ae_<mode>_p): Likewise.
24612 (vector_nez_<mode>_p): Likewise.
24613 (vector_ne_v2di_p): Likewise.
24614 (vector_ae_v2di_p): Likewise.
24615 (vector_ne_<mode>_p): Likewise.
24616 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
24618 (vsx_tsqrt<mode>2_fe): Likewise.
24620 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
24623 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
24625 2017-03-02 Jakub Jelinek <jakub@redhat.com>
24627 PR rtl-optimization/79780
24628 * cprop.c (one_cprop_pass): When second and further conditional trap
24629 in a single basic block is turned into an unconditional trap, turn it
24630 into a deleted note to avoid RTL verification failures.
24632 2017-03-02 Richard Biener <rguenther@suse.de>
24634 * fold-const.c (const_binop): Use ulow () instead of elt (0).
24636 2017-03-02 Richard Biener <rguenther@suse.de>
24638 PR tree-optimization/79345
24640 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
24641 param and abort the walk, returning -1 if it is hit.
24642 (walk_aliased_vdefs): Take a limit param and pass it on.
24643 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
24644 defaulting to 0 and return a signed int.
24645 * tree-ssa-uninit.c (struct check_defs_data): New struct.
24646 (check_defs): New helper.
24647 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
24648 about uninitialized memory.
24649 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
24650 bogus uninitialized warning.
24651 (fixed_convert_from_real): Likewise.
24653 2017-03-02 Bin Cheng <bin.cheng@arm.com>
24655 PR tree-optimization/66768
24656 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
24657 iv_use if base object can't be determined.
24659 2017-03-02 Jakub Jelinek <jakub@redhat.com>
24661 PR tree-optimization/79345
24662 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
24663 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
24664 (get_pattern_stats): Initialize it.
24665 * genemit.c (gen_expand): Verify match_scratch numbers come after
24666 match_operand/match_dup numbers.
24667 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
24668 match_scratch numbers.
24669 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
24671 * config/s390/s390.md (trunctdsd2): Likewise.
24673 2017-03-02 Richard Biener <rguenther@suse.de>
24675 * wide-int.h (wide_int_storage::operator=): Implement in terms
24678 2017-03-02 Richard Biener <rguenther@suse.de>
24680 PR tree-optimization/79777
24681 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
24682 the to insert expression to sth existing.
24684 2017-03-01 Martin Sebor <msebor@redhat.com>
24686 PR middle-end/79692
24687 * gimple-ssa-sprintf.c
24688 (directive::known_width_and_precision): New function.
24689 (format_integer): Use it.
24690 (get_mpfr_format_length): Consider the full range of precision
24691 when computing %g output with the # flag. Set the likely byte
24692 count to 3 rather than 1 when precision is indeterminate.
24693 (format_floating): Correct the lower bound of precision.
24695 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24697 * doc/invoke.texi: Document default code model for 64-bit Linux.
24699 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24702 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
24703 udiv rather than div since input pattern is unsigned.
24705 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
24707 * config/i386/i386.c (print_reg): Warn for values of
24708 unsupported size in integer register.
24710 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
24713 * config/rs6000/predicates.md (current_file_function_operand): Do
24714 not allow self calls to be local if the function is replaceable.
24716 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
24719 * config/rs6000/altivec.h (vec_ctz and others): Change the
24720 preprocessor macro that controls conditional compilation from
24721 _ARCH_PWR9 to __POWER9_VECTOR__.
24722 (vec_all_ne): Change parameterization of __altivec_scalar_pred
24723 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
24724 control (instead of _ARCH_PWR9 control) so that template
24725 definition uses power9-specific function.
24726 (vec_any_eq): Likewise.
24727 (vec_all_ne): Change macro definition to use a power9-specific
24728 expansion under #ifdef __POWER9_VECTOR__ control (instead of
24729 _ARCH_PWR9 control).
24730 (vec_any_eq) Likewise.
24731 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
24732 expansion for CMPNEF to remove support for xvcmpnesp instruction.
24733 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
24734 support for xvcmpnedp instruction.
24735 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
24736 macro expansion so that Power9 implementation of vec_all_ne does
24737 not use the AltiVec predicate framework.
24738 (VCMPNEH_P): Likewise.
24739 (VCMPNEW_P): Likewise.
24740 (VCMPNED_P): Likewise.
24741 (VCMPNEFP_P): Likewise.
24742 (VCMPNEDP_P): Likewise.
24743 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
24744 implementation of vec_any_eq to not use AltiVec predicate
24746 (VCMPAEH_P): Likewise.
24747 (VCMPAEW_P): Likewise.
24748 (VCMPAED_P): Likewise.
24749 (VCMPAEFP_P): Likewise.
24750 (VCMPAEDP_P): Likewise.
24751 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
24752 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
24753 not use the AltiVec predicate framework.
24754 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
24755 of vec_any_eq to not use AltiVec predicate framework.
24756 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
24757 support for predefined __POWER9_VECTOR__ macro to indicate that
24758 Power9 instruction selection is enabled.
24759 (altivec_overloaded_builtins): Remove extraneous
24760 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
24761 function argument types RS6000_BTI_bool_V16QI and
24762 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
24763 entry for overloaded function argument types RS6000_BTI_bool_V4SI
24764 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
24765 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
24766 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
24767 Power9 for implementations of vec_cmpne. Change the signature for
24768 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
24769 (representing vec_all_ne) to remove the previously described first
24770 argument of type RS6000_BTI_INTSI, as this was an artifact of
24771 reliance on the AltiVec predicate framework, which is no longer
24772 used in the implementation of these functions. Add
24773 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
24774 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
24775 since, unlike the AltiVec predicate framework implementation, we
24776 do not share function descriptors between vec_alle and vec_anyeq.
24777 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
24778 set of modes that receive special treatment even when
24779 TARGET_P9_VECTOR is true. The special treatment emits code that
24780 does not depend on Power9 instructions.
24781 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
24782 define_expand to not rely on AltiVec predicate framework.
24783 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
24785 (vector_ne_v2di_p): Change this define_expand to not rely on
24786 AltiVec predicate framework.
24787 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
24789 (vector_ne_<mode>_p): Change this define_expand to not rely on
24790 AltiVec predicate framework.
24791 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
24793 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
24794 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
24795 define_insn pattern.
24796 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
24797 define_insn pattern because the xvcmpne<VSs>. instruction is not
24799 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
24800 instruction is not supported.
24802 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24804 * config/nvptx/nvptx.c: Include intl.h.
24806 2017-03-01 Martin Jambor <mjambor@suse.cz>
24809 * ipa-prop.h (ipa_bits): Removed field known.
24810 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
24811 to pointers. Adjusted their comments to warn about their sharing.
24812 (ipcp_transformation_summary): Change bits to a vector of pointers.
24813 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
24814 (ipa_get_ipa_bits_for_value): Declare.
24815 * tree-vrp.h (value_range): Mark as GTY((for_user)).
24816 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
24817 (ipa_bits_hash_table): Likewise.
24818 (ipa_vr_ggc_hash_traits): Likewise.
24819 (ipa_vr_hash_table): Likewise.
24820 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
24821 being pointers and vr_known being removed.
24822 (ipa_set_jf_unknown): Likewise.
24823 (ipa_get_ipa_bits_for_value): New function.
24824 (ipa_set_jfunc_bits): Likewise.
24825 (ipa_get_value_range): New overloaded functions.
24826 (ipa_set_jfunc_vr): Likewise.
24827 (ipa_compute_jump_functions_for_edge): Use the above functions to
24828 construct bits and vr parts of jump functions.
24829 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
24830 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
24832 (ipcp_grow_transformations_if_necessary): Also allocate
24833 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
24835 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
24836 them. Fix too long lines.
24837 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
24838 vr_known being removed.
24839 (ipa_read_jump_function): Use new setter functions to construct bits
24840 and vr parts of jump functions or set them to NULL.
24841 (write_ipcp_transformation_info): Adjust for bits being pointers.
24842 (read_ipcp_transformation_info): Likewise.
24843 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
24845 Include gt-ipa-prop.h.
24846 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
24848 (ipcp_store_bits_results): Likewise.
24849 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
24850 Do not write to existing jump functions but use a temporary instead.
24852 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24855 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
24856 attempt to use its first operand as BIT_FIELD_REF base.
24858 2017-03-01 Richard Biener <rguenther@suse.de>
24860 PR middle-end/79721
24861 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
24862 interpolating formula in wrapping arithmetic.
24863 (chrec_apply): Convert chrec_evaluate return value to wanted type.
24865 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24867 PR tree-optimization/79734
24868 * tree-vect-generic.c (expand_vector_condition): Optimize
24869 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
24870 Handle VEC_COND_EXPR where comparison has different inner width from
24871 type's inner width.
24873 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
24875 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
24876 markup, and similar issues. Remove @opindex entries for things
24877 that aren't options. Add missing -mmpy-option entries.
24879 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24881 PR tree-optimization/79737
24882 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
24883 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
24884 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
24885 instead of byte_size. Formatting fix.
24886 (shift_bytes_in_array_right): Formatting fix.
24888 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
24891 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
24892 condition on optimize for the leaf function test.
24894 2017-02-28 Martin Liska <mliska@suse.cz>
24897 * read-rtl-function.c (function_reader::handle_unknown_directive):
24898 Bail out when one uses -flto.
24900 2017-02-28 Martin Liska <mliska@suse.cz>
24902 * common.opt: Replace space with tabular for options of <number>
24904 * config/i386/i386.opt: Show <number> value for
24905 -mlarge-data-threshold.
24906 * opts.c (print_filtered_help): Do not display number in hexadecimal
24909 2017-02-28 Martin Liska <mliska@suse.cz>
24911 * common.opt: Fix --help=option -Q for options which are of
24914 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
24916 * config/i386/i386.c (print_reg): Error out for values
24917 of 8-bit size in invalid integer register.
24919 2017-02-28 Martin Sebor <msebor@redhat.com>
24921 PR tree-optimization/79691
24922 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
24924 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24927 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
24928 gcc_unreachable with output_operand_lossage.
24930 2017-02-28 Richard Biener <rguenther@suse.de>
24932 PR tree-optimization/79740
24933 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
24935 (visit_nary_op): Insert the nary into the hashtable if we
24936 pattern-matched sth.
24937 * tree-ssa-pre.c (eliminate_insert): Robustify.
24939 2017-02-28 Richard Biener <rguenther@suse.de>
24941 PR middle-end/79731
24942 * fold-const.c (decode_field_reference): Reject out-of-bound
24945 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24947 * config/i386/i386.c: Include intl.h.
24948 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
24949 instead of just cond ? "..." : "...".
24950 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
24951 * coverage.c (read_counts_file): Likewise.
24952 * omp-offload.c: Include intl.h.
24953 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
24954 of just cond ? "..." : "...".
24955 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
24956 of just cond ? "..." : "...".
24958 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
24961 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
24963 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
24964 'tune for' CPU name.
24965 * config/arm/arm-cpu-data.h: Regenerated.
24967 2017-02-28 Richard Biener <rguenther@suse.de>
24969 PR tree-optimization/79732
24970 * tree-inline.c (expand_call_inline): Do not shadow var.
24972 2017-02-28 Richard Biener <rguenther@suse.de>
24974 PR tree-optimization/79723
24975 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
24976 address-space properly.
24978 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
24980 * doc/optinfo.texi (Optimization groups): Fix option used for
24982 * doc/invoke.texi (-fopt-info): Document "omp".
24983 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
24984 (OPTGROUP_ALL): Add OPTGROUP_OMP.
24985 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
24986 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
24987 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
24989 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
24991 * dumpfile.c (optgroup_options): Instead of "openmp", associate
24992 OPTGROUP_OMP with "omp".
24994 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
24997 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
24998 for arithmetic shift of unsigned V2DI.
25000 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
25002 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
25003 arc/linux.h headers.
25004 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
25005 (LINK_SPEC): Likewise.
25006 (ARC_TLS_EXTRA_START_SPEC): Likewise.
25007 (EXTRA_SPECS): Likewise.
25008 (STARTFILE_SPEC): Likewise.
25009 (ENDFILE_SPEC): Likewise.
25010 (LIB_SPEC): Likewise.
25011 (TARGET_SDATA_DEFAULT): Likewise.
25012 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
25013 (MULTILIB_DEFAULTS): Likewise.
25014 (DWARF2_UNWIND_INFO): Likewise.
25015 * config/arc/big.h: New file.
25016 * config/arc/elf.h: Likewise.
25017 * config/arc/linux.h: Likewise.
25018 * config/arc/t-uClibc: Remove.
25020 2017-02-27 Bin Cheng <bin.cheng@arm.com>
25022 PR tree-optimization/77536
25023 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
25024 (tree_transform_and_unroll_loop): Use above function to compute the
25025 estimated niter of unrolled loop and use it when scaling profile.
25026 Also use count info rather than frequency if it's non-zero.
25027 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
25028 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
25029 (vect_transform_loop): Call above function.
25031 2017-02-27 Richard Biener <rguenther@suse.de>
25033 PR tree-optimization/45397
25034 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
25035 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
25036 (visit_nary_op): Add pattern matching for CSEing sign-changed
25037 or truncated operations with wider ones.
25039 2017-02-27 Richard Biener <rguenther@suse.de>
25041 PR tree-optimization/79690
25042 * tree-vect-stmts.c (vectorizable_store): Use vector type
25043 built from the DR with address-space.
25045 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
25047 * doc/invoke.texi (Optimize Options): Refine the description
25048 of asan-use-after-return.
25050 2017-02-25 Alan Modra <amodra@gmail.com>
25052 PR rtl-optimization/79584
25053 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
25054 base, not ad->base_term, the reg within base. Remove assertion
25055 that ad->base == ad->base_term. Replace gen_int_mode using
25056 bogus mode with const0_rtx.
25058 2017-02-25 Jakub Jelinek <jakub@redhat.com>
25060 PR middle-end/79396
25061 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
25062 FMA_EXPR like tcc_binary or tcc_unary.
25064 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
25067 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
25069 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
25070 (output_loc_operands): Handle DW_OP_call_ref and
25071 DW_OP_GNU_variable_value.
25072 (struct variable_value_struct): New type.
25073 (struct variable_value_hasher): Likewise.
25074 (variable_value_hash): New variable.
25075 (string_types): Remove.
25076 (copy_loc_descr): New function.
25077 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
25078 (prepend_loc_descr_to_each): New function.
25079 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
25080 instead of add_loc_descr_to_each if the first argument is single
25081 location list and the second has multiple.
25082 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
25083 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
25084 when looking for variable value which doesn't have other location info.
25085 (loc_list_from_tree): Formatting fix.
25086 (gen_array_type_die): Simplify DW_AT_string_length handling.
25087 (adjust_string_types): Remove.
25088 (gen_subprogram_die): Don't call adjust_string_types nor test/set
25089 string_types. Call resolve_variable_values.
25090 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
25091 (resolve_addr_in_expr): Likewise. Add A argument.
25092 (copy_deref_exprloc): Remove deref argument. Adjust for the
25093 original expression being DW_OP_GNU_variable_value with optionally
25094 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
25095 optionally after it.
25096 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
25097 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
25098 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
25099 (variable_value_hasher::hash, variable_value_hasher::equal): New
25101 (resolve_variable_value_in_expr, resolve_variable_value,
25102 resolve_variable_values, note_variable_value_in_expr,
25103 note_variable_value): New functions.
25104 (dwarf2out_early_finish): Call note_variable_value on all toplevel
25107 2017-02-24 Jakub Jelinek <jakub@redhat.com>
25110 * opts.h (handle_generated_option): Add GENERATED_P argument.
25111 * opts-common.c (handle_option): Adjust function comment.
25112 (handle_generated_option): Add GENERATED_P argument, pass it to
25114 (control_warning_option): Pass false to handle_generated_option
25116 * opts.c (maybe_default_option): Pass true to handle_generated_option
25118 * optc-gen.awk: Likewise.
25120 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25122 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
25123 a REG, look at the REG it is a SUBREG of.
25124 (splitter for cmpeqsi_t): Ditto.
25126 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25128 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
25129 the special USEs with the pattern of the insn, not the insn itself.
25131 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
25134 * doc/invoke.texi: Document -mload-store-pairs.
25136 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25137 Sandra Loosemore <sandra@codesourcery.com>
25139 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
25140 argument isn't a CONST_INT.
25141 (nios2_alternate_compare_const): Assert op is a CONST_INT.
25142 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
25143 (nios2_validate_compare): Bypass alternate compare logic if *op2
25144 is not a CONST_INT.
25145 (ldstwm_operation_p): Return false if first_base is not a REG or
25146 if first_offset is not a CONST_INT.
25148 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25150 * config/cris/cris.md: Use correct operand in a define_peephole2.
25152 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25154 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
25156 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25158 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
25159 this_insn if it is an INSN or JUMP_INSN.
25160 (force_offsettable): Look at base, not at addr.
25161 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
25162 on things that aren't necessarily CONST_INTs.
25164 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
25166 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
25167 -mfpmath=sse is the default also for x86-32 targets with SSE2
25168 instruction set when @option{-ffast-math} is enabled
25170 2017-02-24 Jeff Law <law@redhat.com>
25172 PR rtl-optimizatoin/79286
25173 * ira.c (update_equiv_regs): Drop may_trap_p exception to
25176 2017-02-24 Richard Biener <rguenther@suse.de>
25178 PR tree-optimization/79389
25179 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
25182 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
25184 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
25185 function comment to reflect reality.
25186 (loop_exits_before_overflow): Fix typo in function description.
25188 2017-02-24 Richard Biener <rguenther@suse.de>
25190 PR tree-optimization/79389
25191 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
25192 properly that a threading opportunity exists. Detect conditional
25193 copy/constant propagation opportunities.
25195 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
25197 * config/visium/visium.md (type): Add trap.
25198 (b): New mode attribute.
25199 (*btst): Rename into...
25200 (*btst<mode>): ...this and adjust.
25201 (*cbranchsi4_btst_insn): Rename into...
25202 (*cbranch<mode>4_btst_insn): ...this and adjust.
25203 (trap): New define_insn.
25205 2017-02-23 Jakub Jelinek <jakub@redhat.com>
25207 PR tree-optimization/79389
25208 * ifcvt.c (struct noce_if_info): Add rev_cond field.
25209 (noce_reversed_cond_code): New function.
25210 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
25211 reversed_comparison_code. Formatting fix.
25212 (noce_try_store_flag): Test rev_cond != NULL in addition to
25213 reversed_comparison_code.
25214 (noce_try_store_flag_constants): Likewise.
25215 (noce_try_store_flag_mask): Likewise.
25216 (noce_try_addcc): Use rev_cond if non-NULL instead of
25217 reversed_comparison_code.
25218 (noce_try_cmove_arith): Likewise. Formatting fixes.
25219 (noce_try_minmax, noce_try_abs): Clear rev_cond.
25220 (noce_find_if_block): Initialize rev_cond.
25221 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
25222 instead of false as last argument never attempt to reverse it
25225 2017-02-23 Bin Cheng <bin.cheng@arm.com>
25227 PR tree-optimization/79663
25228 * tree-predcom.c (combine_chains): Process refs in reverse order
25229 only for ZERO length chains, and add explaining comment.
25231 2017-02-23 Jeff Law <law@redhat.com>
25233 PR tree-optimization/79578
25234 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
25235 in call to operand_equal_p.
25237 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
25240 * config/i386/cpuid.h: Fix another undefined behavior.
25242 2017-02-23 Richard Biener <rguenther@suse.de>
25244 PR tree-optimization/79683
25245 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
25246 vector types for data-refs.
25248 2017-02-23 Martin Liska <mliska@suse.cz>
25250 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
25252 2017-02-23 Jakub Jelinek <jakub@redhat.com>
25254 PR middle-end/79665
25255 * internal-fn.c (get_range_pos_neg): Moved to ...
25256 * tree.c (get_range_pos_neg): ... here. No longer static.
25257 * tree.h (get_range_pos_neg): New prototype.
25258 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
25259 are known to be in between 0 and signed maximum inclusive, try to
25260 expand both unsigned and signed divmod and use the cheaper one from
25263 2017-02-22 Jeff Law <law@redhat.com>
25265 PR tree-optimization/79578
25266 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
25267 to compare base operands.
25269 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
25272 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
25273 gpc_reg_operand instead of fpr_reg_operand.
25275 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
25277 * config/mips/mips.c (mips_return_in_memory): Force FP
25278 vector types to be returned in memory for o32 ABI.
25280 2017-02-22 Jakub Jelinek <jakub@redhat.com>
25282 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
25283 instead of DW_TAG_member for static data member declarations and don't
25284 set no_linkage_name for static inline data members.
25285 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
25288 2017-02-22 Martin Liska <mliska@suse.cz>
25290 * doc/invoke.texi: Replace inequality signs with square brackets
25293 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25295 PR tree-optimization/68644
25296 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
25298 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
25301 * lra-constraints.c (simplify_operand_subreg): Handle
25302 WORD_REGISTER_OPERATIONS targets.
25304 2017-02-22 Jakub Jelinek <jakub@redhat.com>
25307 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
25308 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
25309 elimination by swapping fld*.
25311 2017-02-22 Richard Biener <rguenther@suse.de>
25313 PR tree-optimization/79673
25314 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
25315 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
25316 irrelevant address-space qualifiers and avoiding a
25317 ADDR_SPACE_CONVERT_EXPR from fold_convert.
25319 2017-02-22 Richard Biener <rguenther@suse.de>
25321 PR tree-optimization/79666
25322 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
25323 to not symbolically negate if that may introduce undefined
25326 2017-02-22 Martin Liska <mliska@suse.cz>
25329 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
25330 * data-streamer-out.c (streamer_write_gcov_count_stream):
25332 * value-prof.c (stream_out_histogram_value): Make assert more
25333 precise based on type of counter.
25335 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
25338 * config/i386/i386.md (standard_x87sse_constant_load splitter):
25339 Use nonimmediate_operand instead of memory_operand for operand 1.
25340 (float-extend standard_x87sse_constant_load splitter): Ditto.
25342 2017-02-21 Jeff Law <law@redhat.com>
25344 PR tree-optimization/79621
25345 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
25346 blocks with edges to themselves.
25348 2017-02-21 Jakub Jelinek <jakub@redhat.com>
25351 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
25352 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
25353 Use gimple_call_builtin_p.
25356 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
25357 on temporarily removed DEBUG_INSNs.
25359 PR tree-optimization/79649
25360 * tree-loop-distribution.c (classify_partition): Give up on
25361 non-generic address space loads/stores.
25363 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
25365 * doc/loop.texi (Loop manipulation): Remove nonexistent
25366 tree_ssa_loop_version from the documentation.
25367 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
25369 2017-02-21 Jakub Jelinek <jakub@redhat.com>
25372 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
25373 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
25374 * config/rs6000/rs6000.c: Include except.h.
25375 (rs6000_expand_split_stack_prologue): Call
25376 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
25378 2017-02-21 Martin Jambor <mjambor@suse.cz>
25381 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
25382 have been analyzed.
25384 2017-02-21 Martin Jambor <mjambor@suse.cz>
25386 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
25387 for backward compatibility only.
25388 * doc/invoke.texi (Option Summary): Remove all references to
25389 -fipa-cp-alignment.
25391 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
25395 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25397 * lra-constraints.c (curr_insn_transform): Handle
25398 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
25400 2017-02-21 Martin Liska <mliska@suse.cz>
25402 * config/i386/i386.opt: Replace -masm-dialect with -masm.
25404 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
25406 PR translation/79638
25407 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
25409 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
25412 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
25413 (arm_function_ok_for_sibcall): Return false for an indirect call by
25414 descriptor if all the argument registers are used.
25415 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
25416 alignment of the function.
25418 2017-02-21 Jakub Jelinek <jakub@redhat.com>
25420 PR tree-optimization/61441
25421 * simplify-rtx.c (simplify_const_unary_operation): For
25422 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
25423 the sNaN unmodified.
25425 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
25427 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25428 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25429 instead of SYSTEM_HEADER_DIR.
25431 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
25432 Martin Liška <mliska@suse.cz>
25434 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
25435 Fix typos and grammar, use active voice, and clarify.
25437 2017-02-20 Marek Polacek <polacek@redhat.com>
25439 PR middle-end/79537
25440 * gimplify.c (gimplify_expr): Handle unused *&&L;.
25443 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
25445 2017-02-20 Jakub Jelinek <jakub@redhat.com>
25448 * config/i386/i386.c (ix86_expand_builtin): Handle
25449 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
25450 ix86_builtins_isa[fcode].isa as a requirement of those
25451 flags and any other flag in the bitmask.
25452 (ix86_init_mmx_sse_builtins): Use 0 instead of
25453 ~OPTION_MASK_ISA_64BIT as mask.
25454 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
25455 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
25456 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
25457 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
25459 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25462 * lra-constraints.c (split_reg): Check requested split mode
25463 is supported by the register.
25465 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25467 * lra-constraints.c (simplify_operand_subreg): Remove early
25470 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25473 * lra-constraints.c (curr_insn_transform): Tighten condition
25474 for converting SUBREG reloads from OP_OUT to OP_INOUT.
25476 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25479 * lra-constraints.c (curr_insn_transform): Handle
25480 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
25482 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
25485 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
25487 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
25489 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
25492 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
25495 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25497 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
25500 2017-02-18 Jakub Jelinek <jakub@redhat.com>
25502 * final.c (last_columnnum, override_columnnum): New variables.
25503 (final_start_function): Set last_columnnum, pass it to begin_prologue
25504 hook and pass 0 to dwarf2out_begin_prologue.
25505 (final_scan_insn): Update override_columnnum. Pass last_columnnum
25506 to source_line debug hook.
25507 (notice_source_line): Compute last_columnnum and for debug_column_info
25508 return true on column changes.
25509 * debug.h (struct gcc_debug_hooks): Add column argument to
25510 source_line and begin_prologue hooks.
25511 (debug_nothing_int_charstar_int_bool): Remove prototype.
25512 (debug_nothing_int_int_charstar,
25513 debug_nothing_int_int_charstar_int_bool): New prototypes.
25514 (dwarf2out_begin_prologue): Add column argument.
25515 * debug.c (do_nothing_debug_hooks): Adjust source_line and
25516 begin_prologue hooks.
25517 (debug_nothing_int_charstar_int_bool): Remove.
25518 (debug_nothing_int_int_charstar,
25519 debug_nothing_int_int_charstar_int_bool): New functions.
25520 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
25521 through to dwarf2out_source_line.
25522 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
25523 (dwarf2out_source_line): Add column argument, emit it if requested.
25524 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
25526 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
25527 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
25528 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
25529 through to dwarf2out_begin_prologue.
25530 (vmsdbgout_source_line): Add column argument, pass it through to
25531 dwarf2out_source_line.
25532 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
25533 dbxout_source_line caller.
25534 (dbxout_source_line): Add column argument.
25536 * common.opt (gno-column-info, gcolumn-info): New options.
25537 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
25538 (check_die): Also test for multiple DW_AT_decl_column attributes.
25539 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
25540 DW_AT_decl_column if requested.
25541 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
25543 (gen_variable_die): Likewise.
25544 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
25545 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
25548 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
25549 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
25550 (ix86_handle_option): Handle OPT_m3dnowa.
25551 * doc/invoke.texi (-m3dnowa): Document.
25552 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
25553 -m3dnowa instead of -m3dnow -march=athlon.
25556 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
25557 instead of gcc_assert for K, r and R code checks. Formatting fixes.
25559 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25562 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
25563 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
25564 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
25565 generator for vsx_xxpermdi_<mode>_be.
25566 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
25567 force big-endian semantics.
25568 (vsx_xxpermdi_<mode>_be): New define_expand with same
25569 implementation as previous version of vsx_xxpermdi_<mode>.
25571 2017-02-17 Jakub Jelinek <jakub@redhat.com>
25573 PR tree-optimization/79327
25574 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
25575 variable, its initialization and use.
25577 2017-02-17 Julia Koval <julia.koval@intel.com>
25579 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
25580 (OPTION_MASK_ISA_PKU_UNSET): New.
25581 (ix86_handle_option): Handle -mrdpid.
25582 * config/i386/cpuid.h (bit_RDPID): New.
25583 * config/i386/driver-i386.c (host_detect_local_cpu):
25584 Detect RDPID feature.
25585 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
25586 * config/i386/i386-c.c (ix86_target_macros_internal):
25588 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
25589 (ix86_valid_target_attribute_inner_p): Add "rdpid".
25590 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
25591 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
25592 * config/i386/i386.md (define_insn "rdpid"): New.
25593 * config/i386/i386.opt Add -mrdpid.
25594 * config/i386/immintrin.h (_rdpid_u32): New.
25596 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
25598 PR rtl-optimization/79541
25599 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
25600 instead of transforming it into USE.
25602 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
25604 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
25605 If HONOR_SNANS (SFmode) force the input to a register.
25606 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
25607 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
25608 an frsp or similar insn.
25610 2017-02-17 Martin Liska <mliska@suse.cz>
25612 PR rtl-optimization/79577
25613 * params.def (selsched-max-sched-times): Increase minimum to 1.
25615 2017-02-17 Martin Liska <mliska@suse.cz>
25617 PR rtl-optimization/79574
25618 * gcse.c (want_to_gcse_p): Prevent integer overflow.
25620 2017-02-17 Martin Liska <mliska@suse.cz>
25622 PR tree-optimization/79529
25623 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
25624 ssa_defined_default_def_p to handle cases which are implicitly
25626 * tree-ssa.c (ssa_defined_default_def_p): New function.
25627 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
25628 which are implicitly defined.
25629 * tree-ssa.h (ssa_defined_default_def_p): Declare.
25631 2017-02-17 Richard Biener <rguenther@suse.de>
25633 PR middle-end/79576
25634 * params.def (max-ssa-name-query-depth): Limit to 10.
25636 2017-02-17 Richard Biener <rguenther@suse.de>
25638 PR tree-optimization/79552
25639 * tree-ssa-structalias.c (visit_loadstore): Properly verify
25642 2017-02-17 Richard Biener <rguenther@suse.de>
25645 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
25647 2017-02-17 Marek Polacek <polacek@redhat.com>
25649 PR middle-end/79536
25650 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
25651 (fold_negate_expr): New wrapper.
25653 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
25655 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
25656 Correct terminology and de-emphasize pre-standard behavior.
25658 2017-02-16 Alan Modra <amodra@gmail.com>
25660 PR rtl-optimization/79286
25661 * ira.c (def_dominates_uses): New function.
25662 (update_equiv_regs): Don't create an equivalence for insns that
25663 may trap where the register def does not dominate the use.
25665 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
25667 PR rtl-optimization/78127
25668 * lra.c (lra): Call lra_eliminate before finish the loop after
25671 2017-02-16 Richard Biener <rguenther@suse.de>
25673 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
25675 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
25676 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
25677 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
25678 (isl_val_int_from_wi): New function.
25679 (extract_affine_gmp): Rename to ...
25680 (extract_affine_wi): ... this, take a widest_int.
25681 (extract_affine_int): Just wrap extract_affine_wi.
25682 (add_param_constraints): Use isl_val_int_from_wi.
25683 (add_loop_constraints): Likewise, and extract_affine_wi.
25685 2017-02-15 Jeff Law <law@redhat.com>
25687 PR middle-end/79521
25688 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
25689 ira_init_register_move_cost_if_necessary.
25691 2017-02-15 Martin Sebor <msebor@redhat.com>
25693 PR middle-end/32003
25694 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
25695 removed in a prior commit.
25697 2017-02-15 Bin Cheng <bin.cheng@arm.com>
25699 PR tree-optimization/79347
25700 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
25701 counters during peeling.
25703 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
25705 * Makefile.in (site.exp): Remove "set ISLVER".
25707 2017-02-15 Jakub Jelinek <jakub@redhat.com>
25710 * real.c (real_from_integer): Call real_convert even for decimal.
25712 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
25715 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
25717 2017-02-14 Andrew Pinski <apinski@cavium.com>
25719 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
25720 cores and change the partno/implementer to be correct.
25721 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
25722 the 'B" as the implementer.
25723 * config/aarch64/aarch64-tune.md: Regenerate.
25725 2017-02-14 Carl Love <cel@us.ibm.com>
25727 * config/rs6000/rs6000.c: Add case statement entry to make the
25728 xvcvuxdsp built-in argument unsigned.
25729 * config/rs6000/vsx.md: Fix the source and return operand types so they
25730 match the instruction definitions from the ISA document. Fix typo
25731 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
25734 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
25737 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
25738 member early_clobber_alts.
25739 * lra-lives.c (reg_early_clobber_p): New.
25740 (process_bb_lives): Use it.
25741 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
25742 (debug_operand_data): Initialize early_clobber_alts.
25743 (setup_operand_alternative): Set up early_clobber_alts.
25744 (collect_non_operand_hard_regs): Ditto. Pass early clobber
25745 alternatives to new_insn_reg.
25746 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
25748 (lra_update_insn_regno_info): Pass the new arg.
25750 2017-02-14 Jakub Jelinek <jakub@redhat.com>
25752 PR middle-end/79505
25753 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
25754 (new_oacc_loop_raw): Don't clear already cleared fields.
25757 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
25758 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
25759 _mm512_prefetch_i64gather_ps): New inline functions and macros.
25761 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
25764 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
25766 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
25769 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
25770 the extra instruction to the right place to store 128-bit constant
25773 2017-02-14 Martin Sebor <msebor@redhat.com>
25775 PR middle-end/79448
25776 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
25777 warning for strings of unknown length.
25779 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
25781 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
25783 2017-02-14 Jeff Law <law@redhat.com>
25786 * ira-costs.c (scan_one_insn): Initialize register move costs
25787 for pseudos seen in USE/CLOBBER insns.
25789 PR tree-optimization/79095
25790 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
25791 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
25792 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
25793 if the operands are known to be not equal, then the resulting range
25795 (intersect_ranges): If the new range is ~[0,0] and the old range is
25796 wide, then prefer ~[0,0].
25797 * tree-vrp.c (overflow_comparison_p_1): New function.
25798 (overflow_comparison_p): New function.
25799 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
25800 if NAME is used in an overflow test.
25801 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
25802 overflow check that can be expressed as an equality test, then adjust
25803 ops to be that equality test.
25805 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25807 * config/s390/s390-builtin-types.def: Remove flags argument.
25808 * config/s390/s390.c (s390_init_builtins): Likewise.
25810 2017-02-14 Martin Liska <mliska@suse.cz>
25812 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
25813 vector. Fix trailing white spaces.
25815 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
25817 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
25820 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25822 PR rtl-optimization/68664
25823 * config/arm/arm.c (arm_sched_can_speculate_insn):
25824 New function. Declare prototype.
25825 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
25827 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25829 PR rtl-optimization/68664
25830 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
25832 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
25834 2017-02-14 Amit Pawar <amit.pawar@amd.com>
25836 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
25837 max skip bytes for function, loop and jump.
25839 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25841 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
25842 ABS_EXPR for gimple dump.
25844 2017-02-14 Jakub Jelinek <jakub@redhat.com>
25847 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
25849 PR tree-optimization/79408
25850 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
25851 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
25852 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
25853 also if rhs1 is INTEGER_CST.
25855 2017-02-14 Richard Biener <rguenther@suse.de>
25857 PR middle-end/79432
25858 * tree-into-ssa.c (insert_phi_nodes): When the function can
25859 have abnormal edges rewrite SSA names with broken use-def
25860 dominance out of SSA and register them for PHI insertion.
25862 2017-02-13 Martin Sebor <msebor@redhat.com>
25864 PR middle-end/79496
25865 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
25866 clearing info.nowrite flag when snprintf size argument is a range.
25868 2017-02-13 Jakub Jelinek <jakub@redhat.com>
25870 * cprop.c (cprop_jump): Add missing space in string literal.
25871 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
25872 (get_constraint_for_component_ref): Likewise.
25873 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
25874 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
25875 * lra-constraints.c (process_alt_operands): Likewise.
25876 * ipa-inline.c (inline_small_functions): Likewise.
25877 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
25878 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
25879 * trans-mem.c (diagnose_tm_1_op): Likewise.
25880 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
25881 (grid_parallel_clauses_gridifiable): Likewise.
25883 * config/nvptx/mkoffload.c (process): Add space in between
25886 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
25887 "MOD4_SSE_REGS" and "ALL_REGS".
25889 * spellcheck.c (test_data): Add , in between "foo" and "food".
25891 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25894 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
25895 boundary crossing check and subsequent code generation agree.
25897 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25899 * config/aarch64/aarch64.c (has_memory_op): Delete.
25900 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
25903 2017-02-13 Jakub Jelinek <jakub@redhat.com>
25905 PR rtl-optimization/79388
25906 PR rtl-optimization/79450
25907 * combine.c (distribute_notes): When removing TEM_INSN for which
25908 corresponding dest has last value recorded, invalidate that last
25911 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25913 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
25914 of explicit '@'. Add missing assembly comment marker on branch costs
25917 2017-02-13 Nathan Sidwell <nathan@acm.org>
25919 * gengtype-lex.l (<in_struct>): Add '/'.
25921 2017-02-13 Martin Liska <mliska@suse.cz>
25924 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
25926 2017-02-13 Richard Biener <rguenther@suse.de>
25928 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
25930 * configure: Re-generate.
25931 * config.in: Likewise.
25932 * graphite-dependences.c: Simplify as if
25933 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
25934 * graphite-isl-ast-to-gimple.c: Likewise.
25935 * graphite-optimize-isl.c: Likewise.
25936 * graphite-poly.c: Likewise.
25937 * graphite-sese-to-poly.c: Likewise.
25938 * graphite.h: Likewise.
25939 * toplev.c: Include isl/version.h and use isl_version () for
25940 printing the ISL version.
25941 * doc/install.texi: Update ISL requirement.
25943 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
25945 * doc/standards.texi (Standards): Update reference to
25948 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
25950 * doc/extend.texi (Named Address Spaces): sourceware.org now
25952 * doc/install.texi (Binaries): Ditto.
25955 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
25957 * doc/cpp.texi: Replace "stringify"/"stringification" with C
25958 standard terminology "stringize"/"stringizing" throughout.
25959 * doc/cppinternals.texi: Likewise.
25961 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
25963 * doc/extend.texi: Fix some spelling mistakes and typos.
25964 * doc/invoke.texi: Likewise.
25966 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
25969 * params.def (inline-min-speedup) Change from 10 to 8.
25971 2017-02-11 Jakub Jelinek <jakub@redhat.com>
25973 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
25976 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
25979 * ipa-inline-analysis.c (get_minimal_bb): New function.
25980 (record_modified): Use it.
25981 (remap_edge_change_prob): Handle also ancestor functions.
25983 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
25985 * doc/contrib.texi (Contributors): Remove broken link into
25986 the Mauve CVS repository.
25988 2017-02-11 Jakub Jelinek <jakub@redhat.com>
25990 PR middle-end/79454
25991 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
25992 result computation whenever lhs doesn't have vector mode, not
25993 just when it has BLKmode.
25995 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
25997 * doc/makefile.texi (profiledbootstrap): Refer to the
25998 installation instructions only in textual form.
26000 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26003 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
26005 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
26007 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
26008 (Specific): Update mingw-w64 reference.
26010 (Specific): Remove broken link to Renesas RX processor.
26012 2017-02-10 Richard Biener <rguenther@suse.de>
26014 * toplev.c (process_options): Do not mention obsolete graphite
26015 options when printing sorry message about missing graphite support.
26016 Mention -floop-nest-optimize.
26018 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
26020 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
26021 (vtst_p16): Likewise.
26022 (vtstq_p8): Likewise.
26023 (vtstq_p16): Likewise.
26025 (vtstq_p64): Likewise.
26026 * config/arm/arm_neon.h (vgetq_lane_p64): New.
26027 (vset_lane_p64): New.
26028 (vsetq_lane_p64): New.
26030 2017-02-10 Jakub Jelinek <jakub@redhat.com>
26032 PR tree-optimization/79411
26033 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
26034 stmt operands are SSA_NAMEs used in abnormal phis.
26035 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
26038 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
26041 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
26044 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
26046 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
26048 2017-02-09 Jakub Jelinek <jakub@redhat.com>
26050 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
26054 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
26055 not arbitrary TREE_CONSTANT.
26058 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
26059 "omp declare target link" attribute unless is_global_var.
26060 * omp-offload.c (find_link_var_op): Likewise.
26062 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
26063 Chung-Lin Tang <cltang@codesourcery.com>
26065 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
26067 (gimplify_adjust_omp_clauses): Don't delete TILE.
26068 (gimplify_omp_for): Deal with TILE.
26069 * internal-fn.c (expand_GOACC_TILE): New function.
26070 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
26072 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
26073 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
26075 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
26076 avoid DIV for outermost collapse var.
26077 (expand_oacc_for): Insert tile element loop as needed. Adjust.
26078 Remove out of date comments, fix whitespace.
26079 * omp-general.c (omp_extract_for_data): Deal with tiling.
26080 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
26081 adjust OLF_DIM_BASE value.
26082 (struct omp_for_data): Add tiling field.
26083 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
26084 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
26085 for auto loops. Remove default auto determining, moved to
26086 oacc_loop_fixed_partitions.
26087 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
26088 stmts, add e_mask field.
26089 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
26090 (oacc_thread_numbers): Use oacc_dim_call.
26091 (oacc_xform_tile): New.
26092 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
26093 (finish_oacc_loop): Adjust for ifns vector.
26094 (oacc_loop_discover_walk): Append loop abstraction sites to list,
26095 add case for GOACC_TILE fns.
26096 (oacc_loop_xform_loop): Delete.
26097 (oacc_loop_process): Iterate over call list directly, and add
26098 handling for GOACC_TILE fns.
26099 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
26101 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
26102 vector partitioning to outer loops. Assign 2 partitions to loops
26103 when available. Add TILE handling.
26104 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
26105 (execite_oacc_device_lower): Process GOACC_TILE fns,
26106 ignore unknown specs.
26107 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
26108 * tree.c (omp_clause_num_ops): Adjust TILE ops.
26109 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
26111 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
26113 * configure.ac (ACX_BUGURL): Update.
26114 * configure: Regenerate.
26116 2017-02-09 Richard Biener <rguenther@suse.de>
26118 PR tree-optimization/69823
26119 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
26120 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
26122 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
26124 * config/arc/arc-c.def: Add __NPS400__ definition.
26125 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
26126 (TARGET_NPS400): Define.
26128 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
26130 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
26132 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
26133 pointer, arch_info.
26134 (arc_cpu_types): Fill the arch_info field with a pointer into the
26135 arc_arch_types table.
26136 (arc_selected_cpu): Declare.
26137 * config/arc/arc.c (arc_selected_cpu): Make global.
26138 (arc_selected_arch): Delete.
26139 (arc_base_cpu): Delete.
26140 (arc_override_options): Remove references to deleted variables,
26141 update access to arch information.
26142 (ARC_OPT): Update access to arch information.
26143 (ARC_OPTX): Likewise.
26144 * config/arc/arc.h (arc_base_cpu): Remove declaration.
26145 (TARGET_ARC600): Update access to arch information.
26146 (TARGET_ARC601): Likewise.
26147 (TARGET_ARC700): Likewise.
26148 (TARGET_EM): Likewise.
26149 (TARGET_HS): Likewise.
26150 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
26153 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
26156 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
26157 condition/operands for integer GE/LE/GEU/LEU operations.
26159 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
26161 PR translation/79397
26162 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
26165 2017-02-08 Martin Jambor <mjambor@suse.cz>
26168 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
26169 whether allocation happened.
26170 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
26171 nothing was allocated.
26173 2017-02-08 Jakub Jelinek <jakub@redhat.com>
26175 PR tree-optimization/79408
26176 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
26177 constant, but SSA_NAME with a known integer range, use the minimum
26178 of that range instead of op1 to determine if modulo can be replaced
26179 with its first operand.
26181 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26183 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
26185 2017-02-08 Richard Biener <rguenther@suse.de>
26187 PR tree-optimization/71824
26188 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
26189 Check all loops contained in the merged region.
26191 2017-02-07 Andrew Pinski <apinski@cavium.com>
26193 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
26195 2017-02-07 Andrew Pinski <apinski@cavium.com>
26197 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
26198 (thunderxt88): Likewise.
26199 (thunderxt81): Disable LSE and change v8.1 to v8.
26200 (thunderxt83): Likewise.
26202 2017-02-07 Jakub Jelinek <jakub@redhat.com>
26203 Richard Biener <rguenther@suse.de>
26205 PR middle-end/79399
26206 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
26207 type from int to size_t.
26208 * ira-costs.c (struct_costs_size): Change type from int to size_t.
26210 2017-02-07 Jakub Jelinek <jakub@redhat.com>
26212 PR rtl-optimization/79386
26213 * cprop.c (bypass_conditional_jumps): Initialize
26214 bypass_last_basic_block already before splitting bbs after
26215 unconditional traps...
26216 (bypass_conditional_jumps): ... rather than here.
26219 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
26220 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
26221 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
26222 fix -masm=intel patterns.
26224 2017-02-07 Richard Biener <rguenther@suse.de>
26226 PR tree-optimization/79256
26227 PR middle-end/79278
26228 * builtins.c (get_object_alignment_2): Use min_align_of_type
26229 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
26230 and ADJUST_FIELD_ALIGN.
26232 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
26234 * doc/tm.texi: Regenerate.
26235 * stor-layout.c (layout_decl): Adjust.
26236 (update_alignment_for_field): Likewise.
26237 (place_field): Likewise.
26238 (min_align_of_type): Likewise.
26239 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
26240 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
26241 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
26242 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
26243 * config/frv/frv.c (frv_adjust_field_align): Likewise.
26244 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
26245 * config/i386/i386.c (x86_field_alignment): Likewise.
26246 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
26247 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
26248 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
26249 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
26250 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
26251 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
26255 2017-01-30 Richard Biener <rguenther@suse.de>
26257 PR tree-optimization/79256
26258 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
26259 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
26262 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
26264 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
26265 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
26266 builtins to SImode and emit a zero-extend, if necessary.
26268 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
26270 * docs/invoke.texi (RISC-V Options): Alphabetize.
26272 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
26274 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
26277 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
26279 * config/riscv/riscv.c: New file.
26280 * common/config/riscv/riscv-common.c: Likewise.
26281 * config.gcc: Likewise.
26282 * config/riscv/constraints.md: Likewise.
26283 * config/riscv/elf.h: Likewise.
26284 * config/riscv/generic.md: Likewise.
26285 * config/riscv/linux.h: Likewise.
26286 * config/riscv/multilib-generator: Likewise.
26287 * config/riscv/peephole.md: Likewise.
26288 * config/riscv/pic.md: Likewise.
26289 * config/riscv/predicates.md: Likewise.
26290 * config/riscv/riscv-builtins.c: Likewise.
26291 * config/riscv/riscv-c.c: Likewise.
26292 * config/riscv/riscv-ftypes.def: Likewise.
26293 * config/riscv/riscv-modes.def: Likewise.
26294 * config/riscv/riscv-opts.h: Likewise.
26295 * config/riscv/riscv-protos.h: Likewise.
26296 * config/riscv/riscv.h: Likewise.
26297 * config/riscv/riscv.md: Likewise.
26298 * config/riscv/riscv.opt: Likewise.
26299 * config/riscv/sync.md: Likewise.
26300 * config/riscv/t-elf-multilib: Likewise.
26301 * config/riscv/t-linux: Likewise.
26302 * config/riscv/t-linux-multilib: Likewise.
26303 * config/riscv/t-riscv: Likewise.
26304 * configure.ac: Likewise.
26305 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
26306 Waterman as RISC-V maintainers.
26307 * doc/install.texi: Add RISC-V entries.
26308 * doc/invoke.texi: Add RISC-V options section.
26309 * doc/md.texi: Add RISC-V constraints section.
26310 * configure: Regenerated.
26312 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
26315 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
26316 false values to be constant vectors with all 0 or all 1 bits set.
26317 (vcondu<mode><mode>): Likewise.
26318 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
26320 (fpmask_comparison_operator): Update comment.
26321 (vecint_comparison_operator): New predicate.
26322 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
26323 vector conditionals when the true and false values are constant
26324 vectors with all 0 bits or all 1 bits set.
26326 2017-02-06 Martin Sebor <msebor@redhat.com>
26328 PR tree-optimization/79376
26329 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
26331 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
26333 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
26334 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
26335 to simplify split condition.
26337 2017-02-06 Jakub Jelinek <jakub@redhat.com>
26339 * omp-expand.c (oxpand_omp_atomic_fetch_op,
26340 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
26343 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
26345 PR rtl-optimization/68664
26346 * target.def (can_speculate_insn): New hook.
26347 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
26348 * doc/tm.texi: Regenerate.
26349 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
26350 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
26351 (rs6000_sched_can_speculate_insn): New function.
26353 2017-02-06 Jakub Jelinek <jakub@redhat.com>
26355 PR tree-optimization/79284
26356 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
26357 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
26358 vectorizable_mask_load_store, vectorizable_operation,
26359 vect_is_simple_cond, get_same_sized_vectype): Use it instead
26360 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
26361 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
26362 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
26363 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
26364 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
26365 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
26366 is_gimple_assign (stmt). Replace another such test with
26367 is_gimple_assign (stmt).
26369 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
26372 * config/avr/avr.c (rtl-iter.h): Include it.
26373 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
26374 (avr_legitimate_combined_insn): ...and implementation.
26376 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
26378 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
26379 * config/s390/s390.c (s390_const_operand_ok)
26380 (s390_canonicalize_comparison, s390_extract_part)
26381 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
26382 (s390_contiguous_bitmask_p, s390_rtx_costs)
26383 (legitimize_pic_address): Likewise.
26384 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
26385 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
26386 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
26387 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
26388 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
26390 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
26392 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
26393 REGNO($0) == REGNO($1).
26395 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26397 * config/s390/linux.h(SIZE_TYPE): Add comment.
26399 2017-02-06 Julian Brown <julian@codesourcery.com>
26400 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26401 Virendra Pathak <virendra.pathak@broadcom.com>
26403 * config/aarch64/aarch64-cores.def: Change the scheduler
26405 * config/aarch64/aarch64.md: Include thunderx2t99.md.
26406 * config/aarch64/thunderx2t99.md: New file.
26408 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
26410 * doc/standards.texi (Go Language): Update link to language
26413 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
26415 * tree-eh.c (lower_resx): Sanitize profile.
26416 (cleanup_empty_eh_move_lp): Likewise.
26418 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
26421 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
26423 * cfgloopmanip.h (loop_version): Update prototype.
26424 * modulo-sched.c (sms_schedule): Update call of loop_version.
26425 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
26426 * tree-parloops.c (gen_parallel_loop): Likewise.
26427 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
26428 * tree-ssa-loop-split.c (split_loop): Likewise.
26429 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
26430 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
26432 2017-02-05 Martin Liska <mliska@suse.cz>
26435 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
26437 (print_operand_address): Initialize a struct to zero.
26439 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
26441 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
26442 garbage collector only in textual form.
26444 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
26446 * doc/extend.texi (x86 specific memory model extensions for
26447 transactional memory): Simplify a phrase.
26449 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
26452 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
26453 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
26454 (atomic_storedi_1): Likewise.
26456 2017-02-04 Jakub Jelinek <jakub@redhat.com>
26458 PR tree-optimization/79338
26459 * tree-parloops.c (gather_scalar_reductions): Don't call
26460 vect_analyze_loop_form for loop->inner before destroying loop's
26463 2017-02-03 Martin Sebor <msebor@redhat.com>
26465 PR tree-optimization/79327
26466 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
26467 when precision has resulted in leading zeros.
26468 (format_integer): Adjust the likely counter to assume an unknown
26469 argument that may be zero is non-zero.
26471 2017-02-03 Jason Merrill <jason@redhat.com>
26474 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
26475 avoid copying non-taken branch.
26477 2017-02-03 Jakub Jelinek <jakub@redhat.com>
26479 PR tree-optimization/79340
26480 * tree-vect-loop.c (vectorizable_reduction): Release
26481 vec_defs elements after safe_splicing them into other vectors.
26484 PR tree-optimization/79327
26485 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
26486 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
26488 (format_integer): Use wide_int_to_tree instead of build_int_cst
26489 + to_?hwi. If argmin is NULL, just set argmin and argmax to
26490 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
26491 of shortest and longest sequence.
26493 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
26495 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
26496 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
26498 2017-02-03 Walter Lee <walt@tilera.com>
26501 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
26502 after initial stackframe link reg save.
26503 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
26505 2017-02-03 Jakub Jelinek <jakub@redhat.com>
26508 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
26509 wu for stxssp alternative.
26511 2017-02-03 Martin Sebor <msebor@redhat.com>
26513 PR tree-optimization/79352
26514 * gimple-fold.c (get_range_strlen): Add argument.
26515 (get_range_strlen): Change return type to bool.
26516 (get_maxval_strlen): Pass in a dummy argument.
26517 * gimple-fold.h (get_range_strlen): Change return type to bool.
26518 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
26519 * tree.h (array_at_struct_end_p): Add argument.
26520 * tree.c (array_at_struct_end_p): Handle it.
26522 2017-02-03 Martin Liska <mliska@suse.cz>
26525 * multiple_target.c (create_dispatcher_calls): Redirect edge
26526 from a caller of a dispatcher.
26527 (expand_target_clones): Make the clones local.
26528 (ipa_target_clone): Do both target clones and resolvers.
26529 (ipa_dispatcher_calls): Remove the pass.
26530 (pass_dispatcher_calls::gate): Likewise.
26531 (make_pass_dispatcher_calls): Likewise.
26532 * passes.def (pass_target_clone): Put as very first IPA early
26535 2017-02-03 Martin Liska <mliska@suse.cz>
26537 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
26538 in case of a function with ifunc attribute.
26540 2017-02-03 Martin Liska <mliska@suse.cz>
26542 * cgraph.c (cgraph_node::dump): Dump function version info.
26543 * symtab.c (symtab_node::dump_base): Add missing new line.
26545 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
26547 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
26548 (ifcombine_ifandif): Use it.
26550 2017-02-03 Martin Liska <mliska@suse.cz>
26552 * doc/invoke.texi: Document default value for
26553 use-after-scope-direct-emission-threshold.
26555 2017-02-03 Martin Liska <mliska@suse.cz>
26557 PR tree-optimization/79339
26558 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
26559 (format_floating): Likewise.
26561 2017-02-03 Martin Liska <mliska@suse.cz>
26564 * ipa-prop.c (ipa_node_params_t::insert): Remove current
26566 (ipa_node_params_t::remove): Likewise.
26567 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
26568 initialization from removed ipa_node_params_t::insert.
26569 (ipa_node_params::~ipa_node_params): Move from removed
26570 ipa_node_params_t::release.
26571 * symbol-summary.h (symbol_summary::m_released): New member.
26572 Do not release a summary twice. Do not allow to call finalizer
26573 for types of a summary that live in GGC memory.
26575 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26577 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
26580 2017-02-02 Martin Sebor <msebor@redhat.com>
26582 PR middle-end/79275
26583 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
26584 (format_string): Tighten up the range of output for non-constant
26585 strings and correct the expected range for wide non-constant strings.
26587 2017-02-02 Martin Sebor <msebor@redhat.com>
26589 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
26591 PR middle-end/32003
26592 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
26594 (-fdump-tree-@var): Add to index and document how to come up
26595 with pass-specific option and dump file names.
26596 (-fdump-passes): Clarify where to look for output.
26598 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
26600 PR middle-end/77445
26601 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
26602 statistics of the analyzed path; allow threading for speed when
26603 any of BBs along the path are optimized for speed.
26605 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
26607 PR middle-end/78468
26608 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
26609 settings of the virtual registers.
26612 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
26614 * explow.c (get_dynamic_stack_size): Take known alignment of stack
26615 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
26618 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26620 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
26621 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
26623 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26625 * config/s390/s390.md: Add missing comments with the expanded
26627 * config/s390/vector.md: Likewise.
26628 * config/s390/vx-builtins.md: Likewise.
26630 2017-02-02 Jakub Jelinek <jakub@redhat.com>
26633 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
26634 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
26635 conditions on a single line.
26637 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26639 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
26640 __S390_VX__ to __VX__.
26642 2017-02-01 Andrew Pinski <apinski@cavium.com>
26644 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
26645 stmt_info to record_stmt_cost.
26646 (vect_get_known_peeling_cost): Pass stmt_info if known to
26648 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
26649 cpu_vector_cost field into
26650 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
26651 field into vec_int_stmt_cost and vec_fp_stmt_cost.
26652 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
26653 splitting of scalar_stmt_cost and vec_stmt_cost.
26654 (thunderx_vector_cost): Likewise.
26655 (cortexa57_vector_cost): LIkewise.
26656 (exynosm1_vector_cost): Likewise.
26657 (xgene1_vector_cost): Likewise.
26658 (thunderx2t99_vector_cost): Improve after the splitting of the two
26660 (aarch64_builtin_vectorization_cost): Update for the splitting of
26661 scalar_stmt_cost and vec_stmt_cost.
26663 2017-02-01 Torvald Riegel <triegel@redhat.com>
26664 Richard Henderson <rth@redhat.com>
26666 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
26667 conditional on existance of a fast atomic load.
26668 * optabs-query.c (can_atomic_load_p): New function.
26669 * optabs-query.h (can_atomic_load_p): Declare it.
26670 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
26671 no fast atomic load is available for the particular size of access.
26672 (expand_atomic_compare_and_swap): Likewise.
26673 (expand_atomic_load): Likewise.
26674 (expand_atomic_store): Likewise.
26675 (expand_atomic_fetch_op): Likewise.
26676 * testsuite/lib/target-supports.exp
26677 (check_effective_target_sync_int_128): Remove x86 because it provides
26678 no fast atomic load.
26679 (check_effective_target_sync_int_128_runtime): Likewise.
26681 2017-02-01 Richard Biener <rguenther@suse.de>
26683 * graphite.c: Include tree-vectorizer.h for find_loop_location.
26684 (graphite_transform_loops): Provide opt-info for optimized nests.
26685 * tree-parloop.c (parallelize_loops): Provide opt-info for
26686 parallelized loops.
26688 2017-02-01 Richard Biener <rguenther@suse.de>
26690 PR middle-end/79315
26691 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
26692 was not set before.
26694 2017-02-01 Richard Biener <rguenther@suse.de>
26696 PR tree-optimization/71824
26697 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
26698 Verify the loops are valid in the merged SESE region.
26699 (scop_detection::can_represent_loop_1): Check analyzing the
26700 evolution of the number of iterations in the region succeeds.
26702 2017-01-31 Ian Lance Taylor <iant@golang.org>
26704 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
26705 REG_ARGS_SIZE note to 32-bit push insns and call insn.
26707 2017-01-31 David Malcolm <dmalcolm@redhat.com>
26709 PR preprocessor/79210
26710 * input.c (get_substring_ranges_for_loc): Replace line_width
26711 assertion with error-handling.
26713 2017-01-31 Richard Biener <rguenther@suse.de>
26715 PR tree-optimization/77318
26716 * graphite-sese-to-poly.c (extract_affine): Fix assert.
26717 (create_pw_aff_from_tree): Take loop parameter.
26718 (add_condition_to_pbb): Pass loop of the condition to
26719 create_pw_aff_from_tree.
26721 2017-01-31 Jakub Jelinek <jakub@redhat.com>
26723 * config/s390/s390.c (s390_asan_shadow_offset): New function.
26724 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
26726 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
26730 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
26732 (convert_int_to_float128): Likewise.
26733 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
26734 (convert_int_to_float128): Likewise.
26735 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
26736 (UNSPEC_IEEE128_CONVERT): Likewise.
26737 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
26738 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
26739 Use local variables for IBM extended format.
26740 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
26741 (fix_trunc<mode>si2_fprs): Likewise.
26742 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
26743 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
26744 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
26745 to know that we can now have integers of all sizes in vector
26747 (fix<uns>_<mode>di2_hw): Likewise.
26748 (float<uns>_<mode>si2_hw): Likewise.
26749 (fix_<mode>si2_hw): Likewise.
26750 (fixuns_<mode>si2_hw): Likewise.
26751 (float<uns>_<mode>di2_hw): Likewise.
26752 (float_<mode>di2_hw): Likewise.
26753 (float_<mode>si2_hw): Likewise.
26754 (floatuns_<mode>di2_hw): Likewise.
26755 (floatuns_<mode>si2_hw): Likewise.
26756 (xscvqp<su>wz_<mode>): Delete, no longer used.
26757 (xscvqp<su>dz_<mode>): Likewise.
26758 (xscv<su>dqp_<mode>): Likewise.
26759 (ieee128_mfvsrd_64bit): Likewise.
26760 (ieee128_mfvsrd_32bit): Likewise.
26761 (ieee128_mfvsrwz): Likewise.
26762 (ieee128_mtvsrw): Likewise.
26763 (ieee128_mtvsrd_64bit): Likewise.
26764 (ieee128_mtvsrd_32bit): Likewise.
26766 2017-01-31 Martin Liska <mliska@suse.cz>
26769 * ipa-prop.c (ipa_free_all_node_params): Call release method
26770 instead of ~sumbol_summary to not to trigger double times
26773 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
26775 PR tree-optimization/71691
26776 * bitmap.h (class auto_bitmap): New.
26777 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
26778 is_maybe_undefined instead of ssa_undefined_value_p.
26780 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26782 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
26783 __S390_ARCH_LEVEL__ to __ARCH__.
26785 2017-01-31 Jakub Jelinek <jakub@redhat.com>
26787 PR tree-optimization/79267
26788 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
26789 if should_remove_lhs_p is true.
26791 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
26794 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
26795 (add_alignment_attribute): New.
26796 (base_type_die): Add alignment attribute.
26797 (subrange_type_die): Likewise.
26798 (modified_type_die): Likewise.
26799 (gen_array_type_die): Likewise.
26800 (gen_descr_array_type_die: Likewise.
26801 (gen_enumeration_type_die): Likewise.
26802 (gen_subprogram_die): Likewise.
26803 (gen_variable_die): Likewise.
26804 (gen_field_die): Likewise.
26805 (gen_ptr_to_mbr_type_die): Likewise.
26806 (gen_struct_or_union_type_die): Likewise.
26807 (gen_subroutine_type_die): Likewise.
26808 (gen_typedef_die): Likewise.
26809 (base_type_cmp): Compare alignment attribute.
26811 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26814 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
26815 (setb_unsigned) New pattern for setb with CCUNS.
26816 * config/rs6000/rs6000.c (expand_block_compare): Use a different
26817 subfc./subfe sequence to avoid overflow problems. Generate a
26818 shorter sequence with cmpld/setb for power9.
26819 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
26820 for generating subfc. instruction.
26821 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
26822 now uses this instruction.
26824 2017-01-30 Ian Lance Taylor <iant@google.com>
26827 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
26828 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
26830 2017-01-30 Martin Sebor <msebor@redhat.com>
26832 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
26833 Move constant to the right of a relational operator.
26834 (get_mpfr_format_length, format_character, format_string): Ditto.
26835 (should_warn_p, maybe_warn): Same.
26837 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
26839 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
26842 * asan.c (get_translation_unit_decl): Remove function.
26843 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
26845 2017-01-30 Martin Liska <mliska@suse.cz>
26847 PR gcov-profile/79259
26848 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
26849 -fprofile-generate.
26851 2017-01-30 Martin Liska <mliska@suse.cz>
26854 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
26855 Initialize variables with NULL value.
26857 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
26860 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
26862 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
26864 2017-01-30 Richard Biener <rguenther@suse.de>
26866 PR tree-optimization/79276
26867 * tree-vrp.c (process_assert_insertions): Properly adjust common
26868 when removing a duplicate.
26870 * gcc.dg/torture/pr79276.c: New testcase.
26872 2017-01-30 Richard Biener <rguenther@suse.de>
26874 PR tree-optimization/79256
26875 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
26876 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
26878 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
26880 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
26883 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
26884 ("*r<noxa>sbg_<mode>_sll_bitmask")
26885 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
26886 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
26887 Use contiguous_bitmask_nowrap_operand.
26889 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26892 * config/rs6000/altivec.h (vec_xl): Revise #define.
26893 (vec_xst): Likewise.
26895 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
26897 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
26899 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
26901 PR rtl-optimization/79194
26902 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
26903 traps before call to bypass_conditional_jumps.
26905 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
26907 PR tree-optimization/71374
26908 * lra-constraints.c (check_conflict_input_operands): New.
26909 (match_reload): Use it.
26911 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
26914 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
26915 account to calculate conflict_set.
26917 2017-01-27 Bin Cheng <bin.cheng@arm.com>
26919 PR rtl-optimization/78559
26920 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
26921 other_insn in combine.
26923 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26925 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
26926 uint16_type_node for BT_UINT16.
26928 2017-01-27 David Malcolm <dmalcolm@redhat.com>
26930 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
26931 "RTL Tests" to menu.
26932 (GIMPLE Tests): New node.
26933 (RTL Tests): New node.
26935 2017-01-27 Richard Biener <rguenther@suse.de>
26937 PR tree-optimization/79245
26938 * tree-loop-distribution.c (distribute_loop): Apply cost
26939 modeling also to detected patterns.
26941 2017-01-27 Richard Biener <rguenther@suse.de>
26943 PR tree-optimization/71433
26944 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
26945 (compare_assert_loc): New function.
26946 (process_assert_insertions): Sort and optimize assert locations
26947 to remove duplicates and push down identical assertions on
26948 edges to their destination block.
26950 2017-01-27 Richard Biener <rguenther@suse.de>
26952 PR tree-optimization/79244
26953 * tree-vrp.c (remove_range_assertions): Forcefully propagate
26954 out SSA names even if abnormal.
26956 2017-01-27 Jakub Jelinek <jakub@redhat.com>
26958 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
26959 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
26960 instead of MPFR_RNDN.
26962 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
26965 * arm.c (arm_option_override): Don't call build_target_option_node
26966 until after doing all option overrides.
26967 (arm_valid_target_attribute_tree): Likewise.
26969 2017-01-27 Martin Liska <mliska@suse.cz>
26971 * doc/invoke.texi (-fprofile-arcs): Document profiling support
26972 for {cd}tors and C++ {cd}tors.
26974 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
26976 * config/s390/s390.md ("*setmem_long_and")
26977 ("*setmem_long_and_31z"): Use zero_extend instead of and.
26979 2017-01-26 Martin Sebor <msebor@redhat.com>
26981 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
26984 2017-01-26 Martin Sebor <msebor@redhat.com>
26986 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
26987 HAVE_DFmode before using XFmode or DFmode.
26988 (parse_directive): Avoid using the z length modifier to avoid
26989 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
26991 PR middle-end/78703
26992 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
26993 to accept adjustment as an array.
26994 (get_int_range): New function.
26995 (struct directive): Make width and prec arrays.
26996 (directive::set_width, directive::set_precision): Call get_int_range.
26997 (format_integer, format_floating): Handle width and precision ranges.
26998 (format_string, parse_directive): Same.
27000 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27003 * dwarf2out.c (generate_skeleton_bottom_up): For children with
27004 comdat_type_p set, just clone them, but keep the children in the
27008 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
27009 which have direct callers with -fvar-tracking-assignments enabled
27011 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
27012 inside of type units.
27014 2017-01-26 Martin Sebor <msebor@redhat.com>
27016 PR middle-end/78703
27017 * gimple-ssa-sprintf.c (struct result_range): Add likely and
27019 (struct format_result): Replace number_chars, number_chars_min,
27020 and number_chars_max with a single member of struct result_range.
27022 (format_result::operator+=): Adjust.
27023 (struct fmtresult): Remove bounded. Handle likely and unlikely
27025 (fmtresult::adjust_for_width_or_precision): New function.
27026 (fmtresult:type_max_digits): New function.
27027 (bytes_remaining): Handle likely and unlikely counters.
27028 (min_bytes_remaining): Remove.
27029 (format_percent): Simplify.
27030 (format_integer, format_floating): Set likely and unlikely counters.
27031 (get_string_length, format_character, format_string): Same.
27032 (format_plain, should_warn_p): New function.
27033 (maybe_warn): Call should_warn_p. Update diagnostic messages
27034 and handle those for all directives, including plain strings.
27035 (format_directive): Handle likely and unlikely counters.
27036 Remove unnecessary quoting from diagnostics. Add an informational
27038 (add_bytes): Remove.
27039 (pass_sprintf_length::compute_format_length): Simplify.
27040 (try_substitute_return_value): Handle likely and unlikely counters.
27042 2017-01-26 Carl Love <cel@us.ibm.com>
27044 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
27045 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
27047 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
27050 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
27051 endianess for subregs into account.
27052 * lra-constraints.c (lra_constraints): Do risky transformations
27053 always on the first iteration.
27054 * lra-lives.c (check_pseudos_live_through_calls): Add arg
27055 last_call_used_reg_set.
27056 (process_bb_lives): Define and use last_call_used_reg_set.
27057 * lra.c (lra): Always continue after lra_constraints on the first
27060 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
27062 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
27064 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
27066 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27068 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
27069 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
27070 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
27071 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
27072 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
27073 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
27074 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
27075 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
27076 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
27078 2017-01-26 Marek Polacek <polacek@redhat.com>
27081 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
27082 for the third operand.
27084 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27086 PR middle-end/79236
27087 * omp-low.c (struct omp_context): Add simt_stmt field.
27088 (scan_omp_for): Return omp_context *.
27089 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
27090 context to the _simt_ SIMD stmt.
27091 (lower_omp_for): For combined SIMD with sibling _simt_
27092 SIMD, make sure to use the same decls in _looptemp_
27093 clauses as in the sibling.
27095 2017-01-26 David Sherwood <david.sherwood@arm.com>
27097 PR middle-end/79212
27098 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
27101 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27104 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
27105 emit fld b; fld a; if possible.
27107 * brig-builtins.def: Update copyright years.
27108 * config/arm/arm_acle_builtins.def: Update copyright years.
27110 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
27113 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
27114 constraint instead of o for the stxsd instruction.
27116 2017-01-25 Carl Love <cel@us.ibm.com>
27118 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
27119 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
27121 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
27123 * doc/invoke.texi (C++ Dialect Options): Fix typo.
27125 2017-01-25 Richard Biener <rguenther@suse.de>
27127 PR tree-optimization/69264
27128 * target.def (vector_alignment_reachable): Improve documentation.
27129 * doc/tm.texi: Regenerate.
27130 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
27132 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
27133 earlier changes with respect to TYPE_USER_ALIGN.
27134 (vector_alignment_reachable_p): Likewise. Improve dumping.
27136 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27139 * config/arm/arm.md (xordi3): Force constant operand into a register
27142 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27144 * doc/invoke.texi (-fstore-merging): Correct default optimization
27145 levels at which it is enabled.
27146 (-O): Move -fstore-merging from list to...
27149 2017-01-25 Richard Biener <rguenther@suse.de>
27152 * omp-expand.c: Include debug.h.
27153 (expand_omp_taskreg): Make sure to generate early debug before
27154 outlining anything from a function.
27155 (expand_omp_target): Likewise.
27156 (grid_expand_target_grid_body): Likewise.
27158 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
27161 * asan.c (get_translation_unit_decl): New function.
27162 (asan_add_global): Extract modules file name from globals
27163 TRANSLATION_UNIT_DECL name.
27165 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
27168 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
27169 for long calls with APCS frame and VFP.
27171 2017-01-24 David Malcolm <dmalcolm@redhat.com>
27173 * cfg.c (original_copy_tables_initialized_p): New function.
27174 * cfg.h (original_copy_tables_initialized_p): New decl.
27175 * cfgrtl.c (relink_block_chain): Guard the call to
27176 free_original_copy_tables with a call to
27177 original_copy_tables_initialized_p.
27178 * cgraph.h (symtab_node::native_rtl_p): New decl.
27179 * cgraphunit.c (symtab_node::native_rtl_p): New function.
27180 (symtab_node::needed_p): Don't assert for early assembly output
27181 for __RTL functions.
27182 (cgraph_node::finalize_function): Set "force_output" for __RTL
27184 (cgraph_node::analyze): Bail out early for __RTL functions.
27185 (analyze_functions): Update assertion to support __RTL functions.
27186 (cgraph_node::expand): Bail out early for __RTL functions.
27187 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
27189 * function.h (struct function): Update comment for field
27191 * gimple-expr.c: Include "tree-pass.h".
27192 (gimple_has_body_p): Return false for __RTL functions.
27193 * Makefile.in (OBJS): Add run-rtl-passes.o.
27194 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
27196 (gcc::pass_manager::get_clean_slate): New accessor.
27197 * passes.c: Include "insn-addr.h".
27198 (should_skip_pass_p): Add logging. Update logic for running
27199 "expand" to be compatible with both __GIMPLE and __RTL. Guard
27200 property-provider override so it is only done for gimple passes.
27202 (skip_pass): New function.
27203 (execute_one_pass): Call skip_pass when skipping passes.
27204 * read-md.c (md_reader::read_char): Support filtering
27205 the input to a subset of line numbers.
27206 (md_reader::md_reader): Initialize fields
27207 m_first_line and m_last_line.
27208 (md_reader::read_file_fragment): New function.
27209 * read-md.h (md_reader::read_file_fragment): New decl.
27210 (md_reader::m_first_line): New field.
27211 (md_reader::m_last_line): New field.
27212 * read-rtl-function.c (function_reader::create_function): Only
27213 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
27214 curr_properties. Set DECL_INITIAL to a dummy block.
27215 (read_rtl_function_body_from_file_range): New function.
27216 * read-rtl-function.h (read_rtl_function_body_from_file_range):
27218 * run-rtl-passes.c: New file.
27219 * run-rtl-passes.h: New file.
27221 2017-01-24 Jeff Law <law@redhat.com>
27223 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
27226 2017-01-24 Bin Cheng <bin.cheng@arm.com>
27228 PR tree-optimization/79159
27229 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
27230 (record_nonwrapping_iv): Improve boundary using above function if no
27231 value range information.
27233 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
27234 Martin Jambor <mjambor@suse.cz>
27236 * brig-builtins.def: New file.
27237 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
27238 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
27239 (DEF_HSAIL_SAT_BUILTIN): Likewise.
27240 (DEF_HSAIL_INTR_BUILTIN): Likewise.
27241 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
27242 * builtin-types.def (BT_INT8): New.
27243 (BT_INT16): Likewise.
27244 (BT_UINT8): Likewise.
27245 (BT_UINT16): Likewise.
27246 (BT_FN_ULONG): Likewise.
27247 (BT_FN_UINT_INT): Likewise.
27248 (BT_FN_UINT_ULONG): Likewise.
27249 (BT_FN_UINT_LONG): Likewise.
27250 (BT_FN_UINT_PTR): Likewise.
27251 (BT_FN_ULONG_PTR): Likewise.
27252 (BT_FN_INT8_FLOAT): Likewise.
27253 (BT_FN_INT16_FLOAT): Likewise.
27254 (BT_FN_UINT32_FLOAT): Likewise.
27255 (BT_FN_UINT16_FLOAT): Likewise.
27256 (BT_FN_UINT8_FLOAT): Likewise.
27257 (BT_FN_UINT64_FLOAT): Likewise.
27258 (BT_FN_UINT16_UINT32): Likewise.
27259 (BT_FN_UINT32_UINT16): Likewise.
27260 (BT_FN_UINT16_UINT16_UINT16): Likewise.
27261 (BT_FN_INT_PTR_INT): Likewise.
27262 (BT_FN_UINT_PTR_UINT): Likewise.
27263 (BT_FN_LONG_PTR_LONG): Likewise.
27264 (BT_FN_ULONG_PTR_ULONG): Likewise.
27265 (BT_FN_VOID_UINT64_UINT64): Likewise.
27266 (BT_FN_UINT8_UINT8_UINT8): Likewise.
27267 (BT_FN_INT8_INT8_INT8): Likewise.
27268 (BT_FN_INT16_INT16_INT16): Likewise.
27269 (BT_FN_INT_INT_INT): Likewise.
27270 (BT_FN_UINT_FLOAT_UINT): Likewise.
27271 (BT_FN_FLOAT_UINT_UINT): Likewise.
27272 (BT_FN_ULONG_UINT_UINT): Likewise.
27273 (BT_FN_ULONG_UINT_PTR): Likewise.
27274 (BT_FN_ULONG_ULONG_ULONG): Likewise.
27275 (BT_FN_UINT_UINT_UINT): Likewise.
27276 (BT_FN_VOID_UINT_PTR): Likewise.
27277 (BT_FN_UINT_UINT_PTR: Likewise.
27278 (BT_FN_UINT32_UINT64_PTR): Likewise.
27279 (BT_FN_INT_INT_UINT_UINT): Likewise.
27280 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
27281 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
27282 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
27283 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
27284 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
27285 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
27286 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
27287 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
27288 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
27289 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
27290 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
27291 * doc/frontends.texi: List BRIG FE.
27292 * doc/install.texi (Testing): Add BRIG tesring requirements.
27293 * doc/invoke.texi (Overall Options): Mention BRIG.
27294 * doc/standards.texi (Standards): Doucment BRIG HSA version.
27296 2017-01-24 Richard Biener <rguenther@suse.de>
27298 PR translation/79208
27299 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
27301 2017-01-24 Martin Jambor <mjambor@suse.cz>
27304 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
27305 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
27306 and known_contexts.
27308 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
27310 PR middle-end/79123
27311 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
27312 casts from signed to unsigned really don't have a range.
27314 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
27316 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
27317 GMP_RNDx for compatiblity.
27319 2017-01-24 Martin Liska <mliska@suse.cz>
27322 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
27323 that would prevent us to call alloca with -1 as argument.
27325 2017-01-24 Jakub Jelinek <jakub@redhat.com>
27327 * dwarf2out.c (output_compilation_unit_header, output_file_names):
27328 Avoid -Wformat-security warning.
27330 2017-01-23 Andrew Pinski <apinski@cavium.com>
27332 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
27335 2017-01-23 Martin Sebor <msebor@redhat.com>
27337 PR middle-end/78703
27338 * gimple-ssa-sprintf.c (warn_level): New global.
27339 (format_integer): Use it here and throughout the rest of the file.
27340 Use the same switch to compute sign as base.
27341 (maybe_warn): New function.
27342 (format_directive): Factor out warnings into maybe_warn.
27343 Add debugging output. Use warn_level.
27344 (add_bytes): Use warn_level.
27345 (pass_sprintf_length::compute_format_length): Add debugging output.
27346 (try_substitute_return_value): Same.
27347 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
27349 PR middle-end/78703
27350 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
27351 (struct fmtresult, format_integer, format_floating): Adjust.
27352 (fmtresult::fmtresult): Set max correctly in two argument ctor.
27353 (get_string_length, format_string,format_directive): Same.
27354 (pass_sprintf_length::compute_format_length): Same.
27355 (try_substitute_return_value): Simplify slightly.
27357 PR middle-end/78703
27358 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
27359 (fmtresult::operator+=): Outlined.
27360 (struct fmtresult): Add ctors.
27361 (struct conversion_spec): Rename...
27362 (struct directive): ...to this. Add and remove data members.
27363 (directive::set_width, directive::set_precision): New functions.
27364 (format_percent): Use fmtresult ctor.
27365 (get_width_and_precision): Remove.
27366 (format_integer): Make naming changes. Avoid computing width and
27368 (format_floating): Same. Adjust indentation.
27369 (format_character, format_none): New functions.
27370 (format_string): Moved character handling to format_character.
27371 (format_directive): Remove arguments, change return type.
27372 (parse_directive): New function.
27373 (pass_sprintf_length::compute_format_length): Move directive
27374 parsing to parse_directive.
27376 2017-01-23 Jakub Jelinek <jakub@redhat.com>
27378 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
27379 (assign_assembler_name_if_needed): ... this.
27380 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
27381 (assign_assembler_name_if_needed): ... this.
27382 (free_lang_data_in_cgraph): Adjust callers.
27383 * cgraphunit.c (cgraph_node::analyze): Likewise.
27384 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
27387 2017-01-23 Richard Biener <rguenther@suse.de>
27389 PR tree-optimization/79088
27390 PR tree-optimization/79188
27391 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
27392 resetting loop bounds after last path deletion. Reset loop
27393 bounds of the target loop, make code match the comments.
27394 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
27395 Make sure loops need no fixups.
27397 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
27399 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
27400 exponent support with double type for first argument.
27401 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
27402 type returned by __builtin_vec_extract_sig,
27403 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
27404 functions from "vector int" to "vector unsigned int" or from
27405 "vector long long int" to "vector unsigned long long int".
27406 Changed type returned by __builtin_vec_extract_exp,
27407 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
27408 functions from "vector int" to "vector unsigned int" or from
27409 "vector long long int" to "vector unsigned long long int".
27410 Changed return type of __builtin_vec_test_data_class,
27411 __builtin_vec_test_data_class_sp, and
27412 __builtin_vec_test_data_class_dp from "vector int" to
27413 "vector bool int" or from "vector long long int" to "vector bool
27414 long long int" and changed second argument type from "unsigned
27415 int" to "int". Added new overloaded function forms "vector float
27416 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
27417 "vector float __builtin_vec_insert_exp_sp (vector float, vector
27418 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
27419 double, vector unsigned long long int)" and "vector double
27420 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
27421 long int)". Changed return type of
27422 __builtin_scalar_test_data_class and
27423 __builtin_scalar_test_data_class_sp and
27424 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
27425 int" and changed second argument from "unsigned int" to "int".
27426 Changed type returned by __builtin_scalar_test_neg,
27427 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
27428 from "int" to "bool int". Added new overloaded function form
27429 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
27430 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
27431 exponent double-precision with floating point first argument.
27432 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
27433 documentation of scalar_test_data_class, scalar_test_neg,
27434 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
27435 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
27436 vec_test_data_class built-in functions to reflect refinements in
27437 their type signatures.
27439 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
27441 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
27443 (aarch64_elf_asm_destructor): Likewise.
27445 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
27447 PR rtl-optimization/78634
27448 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
27449 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
27450 * ifcvt.c (noce_try_cmove): Add missing cost check.
27452 PR rtl-optimization/71724
27453 * combine.c (if_then_else_cond): Look for situations where it is
27454 beneficial to undo the work of one of the recursive calls.
27456 2017-01-23 Bin Cheng <bin.cheng@arm.com>
27458 PR tree-optimization/70754
27459 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
27460 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
27461 combined stmt before it if not NULL.
27462 (combine_chains): Process refs reversely and compute dominance point
27465 2017-01-23 Martin Liska <mliska@suse.cz>
27467 PR tree-optimization/79196
27468 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
27469 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
27471 (strlen_optimize_stmt): Call the renamed function.
27473 2017-01-23 Michael Matz <matz@suse.de>
27475 PR tree-optimization/78384
27476 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
27478 2017-01-23 Richard Biener <rguenther@suse.de>
27480 PR tree-optimization/79186
27481 * tree-vrp.c (register_new_assert_for): Make sure we've seen
27482 both incoming edges before moving an assert.
27484 2017-01-23 Martin Jambor <mjambor@suse.cz>
27486 * ipa-prop.c (load_from_param_1): Removed.
27487 (load_from_unmodified_param): Bits from load_from_param_1 put back
27489 (load_from_param): Removed.
27490 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
27491 with stmt. Reverted back to use of load_from_unmodified_param.
27493 2017-01-23 Martin Jambor <mjambor@suse.cz>
27496 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
27497 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
27498 field a pointer to garbage collected vector, mark lattices and
27499 ipcp_orig_node with GTY((skip)).
27500 (ipa_get_param_count): Adjust to descriptors being a pointer.
27501 (ipa_get_param): Likewise.
27502 (ipa_get_type): Likewise.
27503 (ipa_get_param_move_cost): Likewise.
27504 (ipa_set_param_used): Likewise.
27505 (ipa_get_controlled_uses): Likewise.
27506 (ipa_set_controlled_uses): Likewise.
27507 (ipa_is_param_used): Likewise.
27508 (ipa_node_params_t): Move into garbage collector. New methods insert
27510 (ipa_node_params_sum): Annotate wth GTY(()).
27511 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
27513 (ipa_load_from_parm_agg): Adjust declaration.
27514 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
27515 * ipa-profile.c (ipa_profile): Likewise.
27516 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
27517 (ipa_populate_param_decls): Make descriptors parameter garbage
27519 (ipa_dump_param): Adjust to descriptors being a pointer.
27520 (ipa_alloc_node_params): Likewise.
27521 (ipa_initialize_node_params): Likewise.
27522 (load_from_param_1): Make descriptors parameter garbage collected.
27523 (load_from_unmodified_param): Likewise.
27524 (load_from_param): Likewise.
27525 (ipa_load_from_parm_agg): Likewise.
27526 (ipa_node_params::~ipa_node_params): Removed.
27527 (ipa_free_all_node_params): Remove call to delete operator.
27528 (ipa_node_params_t::insert): New.
27529 (ipa_node_params_t::remove): Likewise.
27530 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
27531 copy known_csts and known_contexts vectors.
27532 (ipa_read_node_info): Adjust to descriptors being a pointer.
27533 (ipcp_modif_dom_walker): Make m_descriptors field garbage
27535 (ipcp_transform_function): Make descriptors variable garbage
27538 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
27540 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
27541 * config/i386/avx512dqintrin.h: Ditto.
27542 * config/i386/avx512fintrin.h: Ditto.
27543 * config/i386/i386.c: Handle new builtins.
27544 * config/i386/i386-builtin.def: Add new builtins.
27545 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
27546 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
27548 2017-01-23 Jakub Jelinek <jakub@redhat.com>
27549 Martin Liska <mliska@suse.cz>
27551 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
27552 * asan.c (asan_expand_poison_ifn): Support stores and use
27553 appropriate ASAN report function.
27554 * internal-fn.c (expand_ASAN_POISON_USE): New function.
27555 * internal-fn.def (ASAN_POISON_USE): Declare.
27556 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
27557 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
27558 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
27559 ASAN_POISON calls w/o LHS.
27560 * tree-ssa.c (execute_update_addresses_taken): Create clobber
27561 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
27562 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
27563 * gimplify.c (asan_poison_variables): Add attribute
27564 use_after_scope_memory to variables that really needs to live
27566 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
27567 having the attribute.
27569 2017-01-23 Martin Liska <mliska@suse.cz>
27571 * asan.c (create_asan_shadow_var): New function.
27572 (asan_expand_poison_ifn): Likewise.
27573 * asan.h (asan_expand_poison_ifn): New declaration.
27574 * internal-fn.c (expand_ASAN_POISON): Likewise.
27575 * internal-fn.def (ASAN_POISON): New builtin.
27576 * sanopt.c (pass_sanopt::execute): Expand
27577 asan_expand_poison_ifn.
27578 * tree-inline.c (copy_decl_for_dup_finish): Make function
27580 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
27581 * tree-ssa.c (is_asan_mark_p): New function.
27582 (execute_update_addresses_taken): Rewrite local variables
27583 (identified just by use-after-scope as addressable) into SSA.
27585 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
27587 * doc/install.texi (Specific): opensource.apple.com uses https
27588 now. Remove trailing slash.
27590 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
27592 * README.Portability: Remove note on an Irix compatibility issue.
27594 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
27596 * gcov.c (INCLUDE_ALGORITHM): Define.
27597 (INCLUDE_VECTOR): Define.
27598 No longer include <vector> and <algorithm> directly.
27600 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
27602 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
27604 * doc/invoke.texi (Code Gen Options): Ditto.
27606 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
27609 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
27611 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
27613 rtl-optimization/79125
27614 * cprop.c (local_cprop_pass): Handle cases where we make an
27615 unconditional trap.
27617 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
27621 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
27622 read from, for big endian.
27624 2017-01-20 Jiong Wang <jiong.wang@arm.com>
27626 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
27627 register pauth builtins for LP64 only.
27629 2017-01-20 Marek Polacek <polacek@redhat.com>
27632 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
27635 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
27637 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
27639 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
27640 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
27641 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
27643 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27646 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
27647 in big-endian mode when they are not a single duplicated value.
27649 2017-01-20 Richard Biener <rguenther@suse.de>
27651 * BASE-VER: Bump to 7.0.1.
27653 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
27655 * omp-low.c (omplow_simd_context): New struct. Use it...
27656 (lower_rec_simd_input_clauses): ...here and...
27657 (lower_rec_input_clauses): ...here to hold common data. Adjust all
27658 references to idx, lane, max_vf, is_simt.
27660 2017-01-20 Graham Markall <graham.markall@embecosm.com>
27662 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
27665 2017-01-20 Martin Jambor <mjambor@suse.cz>
27667 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
27668 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
27670 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
27671 (GTFILES): Rename hsa.c to hsa-common.c.
27672 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
27673 * hsa-dump.c: Likewise.
27674 * hsa-gen.c: Likewise.
27675 * hsa-regalloc.c: Likewise.
27676 * ipa-hsa.c: Likewise.
27677 * omp-expand.c: Likewise.
27678 * omp-low.c: Likewise.
27679 * toplev.c: Likewise.
27681 2017-01-20 Marek Polacek <polacek@redhat.com>
27684 * doc/invoke.texi: Document -Wduplicated-branches.
27685 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
27686 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
27687 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
27688 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
27689 return 0 only when not OEP_LEXICOGRAPHIC.
27690 (fold_build_cleanup_point_expr): Use the expression
27691 location when building CLEANUP_POINT_EXPR.
27692 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
27693 * tree.c (add_expr): Handle error_mark_node.
27695 2017-01-20 Martin Liska <mliska@suse.cz>
27698 * tree-profile.c (init_ic_make_global_vars): Do not call
27700 (gimple_init_gcov_profiler): Likewise.
27702 2017-01-20 Martin Liska <mliska@suse.cz>
27705 * cgraph.h (maybe_create_reference): Remove argument and
27707 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
27709 * ipa-cp.c (create_specialized_node): Likewise.
27710 * symtab.c (symtab_node::maybe_create_reference): Handle
27711 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
27713 2017-01-20 Martin Liska <mliska@suse.cz>
27715 * read-rtl-function.c (function_reader::create_function): Use
27716 build_decl instread of build_decl_stat.
27718 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
27720 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
27721 * config/i386/avx512dqintrin.h: Ditto.
27722 * config/i386/avx512fintrin.h: Ditto.
27723 * config/i386/i386-builtin-types.def: Add new types.
27724 * config/i386/i386.c: Handle new types.
27725 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
27726 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
27727 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
27728 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
27729 (__builtin_ia32_kshiftridi): New.
27730 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
27732 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
27736 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
27737 define to rs6000_init_stack_protect_guard.
27738 (rs6000_init_stack_protect_guard): New function.
27740 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
27741 Yunqiang Su <yunqiang.su@imgtec.com>
27743 * config.gcc (supported_defaults): Add madd4.
27744 (with_madd4): Add validation.
27745 (all_defaults): Add madd4.
27746 * config/mips/mips.opt (mmadd4): New option.
27747 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
27749 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
27751 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
27752 (ISA_HAS_FUSED_MADD4): Likewise.
27753 * doc/invoke.texi (-mmadd4): Document the new option.
27754 * doc/install.texi (--with-madd4): Document the new option.
27756 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27758 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
27759 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
27760 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
27761 (aarch64_init_pauth_hint_builtins): New.
27762 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
27763 (aarch64_expand_builtin): Expand new builtins.
27765 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27767 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
27768 * combine-stack-adj.c (no_unhandled_cfa): Handle
27769 REG_CFA_TOGGLE_RA_MANGLE.
27770 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
27771 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
27772 info for return address signing.
27773 (aarch64_expand_epilogue): Likewise.
27775 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27777 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
27778 * config/aarch64/aarch64-protos.h
27779 (aarch64_return_address_signing_enabled): New declaration.
27780 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
27782 (aarch64_expand_prologue): Sign return address before it's pushed onto
27784 (aarch64_expand_epilogue): Authenticate return address fetched from
27786 (aarch64_override_options): Sanity check for ILP32 and ISA level.
27787 (aarch64_attributes): New function attributes for "sign-return-address".
27788 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
27789 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
27790 ("*do_return"): Generate combined instructions according to key index.
27791 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
27792 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
27794 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
27795 * config/aarch64/aarch64.opt (msign-return-address=): New.
27796 * doc/extend.texi (AArch64 Function Attributes): Documents
27797 "sign-return-address=".
27798 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
27800 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
27802 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
27803 overall option summary.
27805 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27807 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
27808 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
27809 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
27810 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
27812 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
27814 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
27815 -mpower9-minmax by default for -mcpu=power9.
27816 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
27817 128-bit floating point.
27819 2017-01-20 Alan Modra <amodra@gmail.com>
27821 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
27822 optimizing for size.
27824 2017-01-20 Alan Modra <amodra@gmail.com>
27827 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
27828 for strcmp and strncmp from corresponding builtin decl.
27830 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
27832 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
27833 instead of i386/rtems-64.h.
27834 * config/i386/rtems-64.h: Remove.
27836 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
27840 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
27842 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
27844 2017-01-19 Tamar Christina <tamar.christina@arm.com>
27846 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
27847 Change int to HOST_WIDE_INT.
27848 * config/aarch64/aarch64-protos.h
27849 (aarch64_simd_gen_const_vector_dup): Likewise.
27850 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
27852 2017-01-19 David Malcolm <dmalcolm@redhat.com>
27854 * langhooks-def.h (lhd_type_for_size): New decl.
27855 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
27856 * langhooks.c (lhd_type_for_size): New function, taken from
27859 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
27861 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
27862 define_bypass for CR latency.
27863 (power9-cracked-alu): Update bypass latency and remove power9-branch.
27864 (power9-alu2): Add define_bypass for CR latency.
27866 (power9-mul): Update insn latency.
27867 (power9-mul-compare): Update insn latency, bypass latency and remove
27870 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27872 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
27874 * config/aarch64/aarch64.md
27875 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
27876 aarch64_nopcrelative_literal_loads.
27877 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
27879 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
27881 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
27882 TARGET_LOONGSON_3A.
27883 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
27885 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
27888 * config.gcc (supported_defaults): Add lxc1-sxc1.
27889 (with_lxc1_sxc1): Add validation.
27890 (all_defaults): Add lxc1-sxc1.
27891 * config/mips/mips.opt (mlxc1-sxc1): New option.
27892 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
27894 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
27895 __mips_no_lxc1_sxc1.
27896 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
27897 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
27898 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
27900 2017-01-19 Richard Biener <rguenther@suse.de>
27902 PR tree-optimization/72488
27903 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
27904 sure to restore SSA info.
27905 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
27907 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
27909 PR rtl-optimization/79121
27910 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
27911 of the inner type when shifting an extended value.
27913 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
27916 * symtab.c (symtab_node::equal_address_to): Fix comparing of
27917 interposable aliases.
27919 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
27922 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
27923 Use the evmergelohi instruction.
27924 (mov_si<mode>_e500_subreg4_2_le): Likewise.
27925 (mov_sitf_e500_subreg8_2_be): Likewise.
27926 (mov_sitf_e500_subreg12_2_le): Likewise.
27927 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
27928 (mov_si<mode>_e500_subreg4_2_be): Likewise.
27929 (mov_sitf_e500_subreg8_2_le): Likewise.
27930 (mov_sitf_e500_subreg12_2_be): Likewise.
27932 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27934 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
27935 attribute from vecsimple to vecperm.
27936 (altivec_vbpermq2): Likewise.
27938 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27941 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
27943 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27944 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
27945 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
27946 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
27947 case where N arg is SIZE_MAX.
27948 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
27949 (cmpstrsi): Add pattern.
27951 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27953 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27954 __builtin_vec_revb builtins.
27955 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
27956 built-in functions to support generation of the ISA 3.0 XXBR<x>
27957 vector byte reverse instructions.
27958 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
27959 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
27960 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
27961 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
27962 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
27963 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
27964 (P9V_BUILTIN_VEC_REVB): Likewise.
27965 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
27966 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
27967 (p9_xxbrq_v16qi): Likewise.
27968 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
27969 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
27970 (p9_xxbrh_v8hi): Likewise.
27971 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
27972 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
27973 vec_revb built-in functions.
27975 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
27977 PR rtl-optimization/78952
27978 * config/i386/i386.md (any_extract): New code iterator.
27979 (*insvqi_2): Use any_extract for source operand.
27980 (*insvqi_3): Use any_shiftrt for source operand.
27982 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
27984 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
27986 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
27988 2017-01-18 Matthias Klose <doko@ubuntu.com>
27990 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
27992 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27994 * config/rs6000/altivec.h (vec_bperm): Change #define.
27995 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
27996 (altivec_vbpermq2): New define_insn.
27997 (altivec_vbpermd): Likewise.
27998 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
27999 function interface.
28000 (VBPERMD): Likewise.
28001 (VBPERM): New polymorphic function interface.
28002 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
28003 Add entries for P9V_BUILTIN_VEC_VBPERM.
28004 * doc/extend.texi: Add interfaces for vec_bperm.
28006 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28008 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
28009 first letter of error messages.
28010 (s390_resolve_overloaded_builtin): Likewise.
28011 * config/s390/s390.c (s390_expand_builtin): Likewise.
28012 (s390_invalid_arg_for_unprototyped_fn): Likewise.
28013 (s390_valid_target_attribute_inner_p): Likewise.
28014 * config/s390/s390.md ("tabort"): Likewise.
28016 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
28018 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
28019 (ISA_AVOID_DIV_HILO): New macro.
28020 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
28021 (ISA_HAS_DDIV): Likewise.
28023 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28025 * doc/invoke.texi (fabi-version): Correct number of occurrences.
28027 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28029 * doc/invoke.texi (fabi-version): Spelling fix.
28031 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28034 * doc/invoke.texi (fabi-version): Mention mangling fix for
28037 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28040 * doc/invoke.texi (fabi-version): Document discriminator mangling.
28042 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
28045 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
28046 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
28048 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
28050 (stack_protect_test): Ditto.
28051 * config/rs6000/rs6000.opt (mstack-protector-guard=,
28052 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
28054 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
28055 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
28056 -mstack-protector-guard-offset=.
28057 (RS/6000 and PowerPC Options): Ditto.
28059 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
28061 * config/i386/i386.h (MASK_CLASS_P): New define.
28062 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
28063 there are no registers from different register sets also when
28064 mask registers are used. Update function comment.
28065 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
28066 to (*k/*r) and (*k/*km) alternatives.
28068 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
28070 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
28071 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
28072 (EH_RETURN_HANDLER_RTX): New define.
28073 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
28074 Force frame pointer in EH return functions.
28075 (aarch64_expand_epilogue): Add barrier for eh_return.
28076 (aarch64_final_eh_return_addr): Remove.
28077 (aarch64_eh_return_handler_rtx): New function.
28078 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
28080 (aarch64_eh_return_handler_rtx): New prototype.
28082 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28084 * config/rs6000/altivec.h (vec_rlmi): New #define.
28085 (vec_vrlnm): Likewise.
28086 (vec_rlnm): Likewise.
28087 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
28088 (UNSPEC_VRLNM): Likewise.
28089 (VIlong): New mode iterator.
28090 (altivec_vrl<VI_char>mi): New define_insn.
28091 (altivec_vrl<VI_char>nm): Likewise.
28092 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
28094 (VRLDNM): Likewise.
28095 (RLNM): New polymorphic function entry.
28096 (VRLWMI): New monomorphic function entry.
28097 (VRLDMI): Likewise.
28098 (RLMI): New polymorphic function entry.
28099 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
28100 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
28101 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
28104 2017-01-17 Jakub Jelinek <jakub@redhat.com>
28107 * dwarf2out.c (field_byte_offset): Restore the
28108 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
28109 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
28110 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
28113 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
28116 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
28118 2017-01-17 Jakub Jelinek <jakub@redhat.com>
28121 * dwarf2out.c (add_data_member_location_attribute): For constant
28122 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
28123 instead of DW_AT_data_member_location, DW_AT_bit_offset and
28124 DW_AT_byte_size attributes.
28126 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
28128 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
28129 after forcing to constant memory when the code model is medium.
28131 2017-01-17 Julia Koval <julia.koval@intel.com>
28134 * config/i386/avx512fintrin.h
28135 (_mm512_i32gather_ps): Change __addr type to void const*.
28136 (_mm512_mask_i32gather_ps): Ditto.
28137 (_mm512_i32gather_pd): Ditto.
28138 (_mm512_mask_i32gather_pd): Ditto.
28139 (_mm512_i64gather_ps): Ditto.
28140 (_mm512_mask_i64gather_ps): Ditto.
28141 (_mm512_i64gather_pd): Ditto.
28142 (_mm512_mask_i64gather_pd): Ditto.
28143 (_mm512_i32gather_epi32): Ditto.
28144 (_mm512_mask_i32gather_epi32): Ditto.
28145 (_mm512_i32gather_epi64): Ditto.
28146 (_mm512_mask_i32gather_epi64): Ditto.
28147 (_mm512_i64gather_epi32): Ditto.
28148 (_mm512_mask_i64gather_epi32): Ditto.
28149 (_mm512_i64gather_epi64): Ditto.
28150 (_mm512_mask_i64gather_epi64): Ditto.
28151 (_mm512_i32scatter_ps): Change __addr type to void*.
28152 (_mm512_mask_i32scatter_ps): Ditto.
28153 (_mm512_i32scatter_pd): Ditto.
28154 (_mm512_mask_i32scatter_pd): Ditto.
28155 (_mm512_i64scatter_ps): Ditto.
28156 (_mm512_mask_i64scatter_ps): Ditto.
28157 (_mm512_i64scatter_pd): Ditto.
28158 (_mm512_mask_i64scatter_pd): Ditto.
28159 (_mm512_i32scatter_epi32): Ditto.
28160 (_mm512_mask_i32scatter_epi32): Ditto.
28161 (_mm512_i32scatter_epi64): Ditto.
28162 (_mm512_mask_i32scatter_epi64): Ditto.
28163 (_mm512_i64scatter_epi32): Ditto.
28164 (_mm512_mask_i64scatter_epi32): Ditto.
28165 (_mm512_i64scatter_epi64): Ditto.
28166 (_mm512_mask_i64scatter_epi64): Ditto.
28167 * config/i386/avx512pfintrin.h
28168 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
28169 (_mm512_mask_prefetch_i32gather_ps): Ditto.
28170 (_mm512_mask_prefetch_i64gather_pd): Ditto.
28171 (_mm512_mask_prefetch_i64gather_ps): Ditto.
28172 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
28173 (_mm512_prefetch_i32scatter_ps): Ditto.
28174 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
28175 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
28176 (_mm512_prefetch_i64scatter_pd): Ditto.
28177 (_mm512_prefetch_i64scatter_ps): Ditto.
28178 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
28179 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
28180 * config/i386/avx512vlintrin.h
28181 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
28182 (_mm_mmask_i32gather_ps): Ditto.
28183 (_mm256_mmask_i32gather_pd): Ditto.
28184 (_mm_mmask_i32gather_pd): Ditto.
28185 (_mm256_mmask_i64gather_ps): Ditto.
28186 (_mm_mmask_i64gather_ps): Ditto.
28187 (_mm256_mmask_i64gather_pd): Ditto.
28188 (_mm_mmask_i64gather_pd): Ditto.
28189 (_mm256_mmask_i32gather_epi32): Ditto.
28190 (_mm_mmask_i32gather_epi32): Ditto.
28191 (_mm256_mmask_i32gather_epi64): Ditto.
28192 (_mm_mmask_i32gather_epi64): Ditto.
28193 (_mm256_mmask_i64gather_epi32): Ditto.
28194 (_mm_mmask_i64gather_epi32): Ditto.
28195 (_mm256_mmask_i64gather_epi64): Ditto.
28196 (_mm_mmask_i64gather_epi64): Ditto.
28197 (_mm256_i32scatter_ps): Change __addr type to void*.
28198 (_mm256_mask_i32scatter_ps): Ditto.
28199 (_mm_i32scatter_ps): Ditto.
28200 (_mm_mask_i32scatter_ps): Ditto.
28201 (_mm256_i32scatter_pd): Ditto.
28202 (_mm256_mask_i32scatter_pd): Ditto.
28203 (_mm_i32scatter_pd): Ditto.
28204 (_mm_mask_i32scatter_pd): Ditto.
28205 (_mm256_i64scatter_ps): Ditto.
28206 (_mm256_mask_i64scatter_ps): Ditto.
28207 (_mm_i64scatter_ps): Ditto.
28208 (_mm_mask_i64scatter_ps): Ditto.
28209 (_mm256_i64scatter_pd): Ditto.
28210 (_mm256_mask_i64scatter_pd): Ditto.
28211 (_mm_i64scatter_pd): Ditto.
28212 (_mm_mask_i64scatter_pd): Ditto.
28213 (_mm256_i32scatter_epi32): Ditto.
28214 (_mm256_mask_i32scatter_epi32): Ditto.
28215 (_mm_i32scatter_epi32): Ditto.
28216 (_mm_mask_i32scatter_epi32): Ditto.
28217 (_mm256_i32scatter_epi64): Ditto.
28218 (_mm256_mask_i32scatter_epi64): Ditto.
28219 (_mm_i32scatter_epi64): Ditto.
28220 (_mm_mask_i32scatter_epi64): Ditto.
28221 (_mm256_i64scatter_epi32): Ditto.
28222 (_mm256_mask_i64scatter_epi32): Ditto.
28223 (_mm_i64scatter_epi32): Ditto.
28224 (_mm_mask_i64scatter_epi32): Ditto.
28225 (_mm256_i64scatter_epi64): Ditto.
28226 (_mm256_mask_i64scatter_epi64): Ditto.
28227 (_mm_i64scatter_epi64): Ditto.
28228 (_mm_mask_i64scatter_epi64): Ditto.
28229 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
28230 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
28231 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
28232 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
28233 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
28234 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
28235 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
28236 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
28237 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
28238 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
28239 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
28240 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
28241 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
28242 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
28243 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
28244 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
28245 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
28246 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
28247 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
28248 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
28249 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
28250 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
28251 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
28252 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
28253 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
28254 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
28255 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
28256 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
28257 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
28258 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
28259 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
28260 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
28261 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
28262 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
28263 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
28264 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
28265 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
28266 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
28267 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
28268 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
28269 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
28270 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
28271 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
28272 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
28273 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
28274 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
28275 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
28276 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
28277 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
28278 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
28279 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
28280 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
28281 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
28282 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
28283 definitions accordingly.
28285 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
28286 Kuan-Lin Chen <kuanlinchentw@gmail.com>
28289 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
28292 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
28295 * ira-conflicts.c (ira_build_conflicts): Update total conflict
28296 hard regs for inner regno.
28298 2017-01-17 Martin Liska <mliska@suse.cz>
28301 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
28302 assumption and add comment.
28304 2017-01-17 Nathan Sidwell <nathan@acm.org>
28306 * ipa-visibility.c (localize_node): New function, broken out of ...
28307 (function_and_variable_visibility): ... here. Call it.
28309 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
28311 PR middle-end/77445
28312 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
28313 correctly set frequency of oudgoing edge.
28314 (duplicate_thread_path): Fix profile updating.
28316 2017-01-17 Jakub Jelinek <jakub@redhat.com>
28319 * configure.ac: Add GCC_BASE_VER.
28320 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
28321 version from BASE-VER file.
28322 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
28323 (gcc.o): Depend on $(BASEVER).
28324 * common.opt (dumpfullversion): New option.
28325 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
28326 * doc/invoke.texi: Document -dumpfullversion.
28327 * doc/install.texi: Document --with-gcc-major-version-only.
28328 * configure: Regenerated.
28330 2017-01-17 Richard Biener <rguenther@suse.de>
28332 PR tree-optimization/71433
28333 * tree-vrp.c (register_new_assert_for): Merge same asserts
28334 on all incoming edges.
28335 (process_assert_insertions_for): Handle insertions at the
28338 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
28340 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
28341 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
28343 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
28346 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
28349 2017-01-17 Alan Modra <amodra@gmail.com>
28352 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
28353 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
28354 symbolic stack limit when pic.
28356 2017-01-16 Martin Sebor <msebor@redhat.com>
28358 PR tree-optimization/78608
28359 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
28361 2017-01-16 Jeff Law <law@redhat.com>
28364 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
28365 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
28366 for several include directories that may be relative to sysroot.
28367 * config/i386/x-mingw32 (gplus_includedir): Define.
28368 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
28369 (native_system_includedir): Likewise.
28370 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
28371 override if TARGET_SYSTEM_ROOT is defined.
28372 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
28374 PR tree-optimization/79090
28375 PR tree-optimization/33562
28376 PR tree-optimization/61912
28377 PR tree-optimization/77485
28378 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
28379 and computed trims into the dump file.
28381 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
28383 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
28385 2017-01-16 Jakub Jelinek <jakub@redhat.com>
28388 * gimplify.c (gimplify_init_constructor): If want_value and
28389 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
28393 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
28394 sequence. Formatting fixes.
28395 (doloop_optimize): Formatting fixes.
28398 * gcc.c (debug_level_greater_than_spec_func): New function.
28399 (static_spec_functions): Add debug-level-gt spec function.
28400 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
28402 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
28403 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
28404 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
28405 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
28406 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
28407 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
28409 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
28411 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
28412 QImode fixups to general and mask registers only.
28414 2017-01-16 Carl Love <cel@us.ibm.com>
28416 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
28417 for built-in functions
28418 vector signed char vec_nabs (vector signed char)
28419 vector signed short vec_nabs (vector signed short)
28420 vector signed int vec_nabs (vector signed int)
28421 vector signed long long vec_nabs (vector signed long long)
28422 vector float vec_nabs (vector float)
28423 vector double vec_nabs (vector double)
28424 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
28426 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
28427 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
28428 * doc/extend.texi: Update the documentation file for the new built-in
28431 2017-01-16 Martin Sebor <msebor@redhat.com>
28433 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
28436 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28438 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
28439 UNSPEC_VSX__XXSPLTD to require special splat handling.
28441 2017-01-16 David Malcolm <dmalcolm@redhat.com>
28444 * system.h: Poison strndup.
28446 2017-01-16 Alan Modra <amodra@gmail.com>
28449 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
28452 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
28454 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
28456 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
28458 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
28459 call recog here. Assert that INSN_CODE (insn) is non-negative.
28461 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
28464 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
28466 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
28467 in the currently scheduled RTL fragment.
28469 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
28471 PR rtl-optimization/78751
28472 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
28475 2017-01-14 Jeff Law <law@redhat.com>
28477 PR tree-optimization/79090
28478 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
28479 variable length stores.
28480 (compute_trims): Delete dead assignment to *trim_tail.
28481 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
28484 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
28486 PR rtl-optimization/78626
28487 PR rtl-optimization/78727
28488 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
28489 of a block, and split such blocks after everything else is finished.
28491 2017-01-14 Alan Modra <amodra@gmail.com>
28494 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
28495 target legitimate_combined_insn.
28496 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
28497 (rs6000_legitimate_combined_insn): New function.
28498 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
28500 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
28501 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
28502 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
28504 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
28506 * doc/frontends.texi (G++ and GCC): Remove references to Java.
28508 2017-01-13 Jeff Law <law@redhat.com>
28510 PR tree-optimization/33562
28511 PR tree-optimization/61912
28512 PR tree-optimization/77485
28513 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
28515 (delete_dead_assignment): Likewise.
28516 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
28517 statement to delete_dead_call and delete_dead_assignment.
28519 2017-01-13 David Malcolm <dmalcolm@redhat.com>
28522 * substring-locations.c (format_warning_va): Strengthen case 1 so
28523 that both endpoints of the substring must be within the format
28524 range for just the substring to be printed.
28526 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
28528 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
28529 * config/i386/i386.c (ix86_target_string): Add missing options
28530 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
28531 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
28532 flags_other and ix86_target_other to flags2_other. Display unknown
28534 (ix86_valid_target_attribute_inner_p): Add missing options and
28535 reorder options by implied ISAs, as in ix86_target_string.
28537 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
28539 * hash-table.h (hash_table::too_empty_p): New function.
28540 (hash_table::expand): Use it.
28541 (hash_table::traverse): Likewise.
28542 (hash_table::empty_slot): Use sizeof (value_type) instead of
28543 sizeof (PTR) to convert bytes to elements. Shrink the table
28544 if the current size is excessive for the current number of
28547 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
28549 * ira-costs.c (record_reg_classes): Break from the inner loop
28550 early once alt_fail is known to be true. Update outer loop
28551 handling accordingly.
28553 2017-01-13 Jeff Law <law@redhat.com>
28555 * tree-ssa-dse.c (decrement_count): New function.
28556 (increment_start_addr, maybe_trim_memstar_call): Likewise.
28557 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
28558 when we know the partially dead statement is a mem* function.
28560 PR tree-optimization/61912
28561 PR tree-optimization/77485
28562 * tree-ssa-dse.c: Include expr.h.
28563 (maybe_trim_constructor_store): New function.
28564 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
28566 PR tree-optimization/33562
28567 PR tree-optimization/61912
28568 PR tree-optimization/77485
28569 * doc/invoke.texi: Document new dse-max-object-size param.
28570 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
28571 * tree-ssa-dse.c: Include params.h.
28572 (dse_store_status): New enum.
28573 (initialize_ao_ref_for_dse): New, partially extracted from
28575 (valid_ao_ref_for_dse, normalize_ref): New.
28576 (setup_live_bytes_from_ref, compute_trims): Likewise.
28577 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
28578 (maybe_trim_partially_dead_store): Likewise.
28579 (maybe_trim_complex_store): Likewise.
28580 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
28581 Track what bytes live from the original store. Return tri-state
28582 for dead, partially dead or live.
28583 (dse_dom_walker): Add constructor, destructor and new private members.
28584 (delete_dead_call, delete_dead_assignment): New extracted from
28586 (dse_optimize_stmt): Make a member of dse_dom_walker.
28587 Use initialize_ao_ref_for_dse.
28589 PR tree-optimization/33562
28590 PR tree-optimization/61912
28591 PR tree-optimization/77485
28592 * sbitmap.h (bitmap_count_bits): Prototype.
28593 (bitmap_clear_range, bitmap_set_range): Likewise.
28594 * sbitmap.c (bitmap_clear_range): New function.
28595 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
28597 2017-01-13 Martin Liska <mliska@suse.cz>
28600 * function.c (set_cfun): Add new argument force.
28601 * function.h (set_cfun): Likewise.
28602 * ipa-inline-transform.c (inline_call): Use the function when
28603 strict alising from is dropped for function we inline to.
28605 2017-01-13 Richard Biener <rguenther@suse.de>
28607 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
28608 for dumping GIMPLE INTEGER_CSTs.
28610 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28612 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
28613 to 201112L since C++17.
28615 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
28618 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
28619 if -fsanitize=kernel-address is present.
28621 2017-01-13 Richard Biener <rguenther@suse.de>
28623 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
28624 as _Literal ( type ) number in case usual suffixes do not
28625 preserve all information.
28627 2017-01-13 Richard Biener <rguenther@suse.de>
28629 PR tree-optimization/77283
28630 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
28631 and ssa-iterators.h.
28632 (is_feasible_trace): Implement a cost model based on joiner
28635 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
28638 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
28639 char or short to __float128/_Float128 directly.
28641 2017-01-12 Martin Sebor <msebor@redhat.com>
28643 to -Wformat-overflow.
28644 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
28645 (min_bytes_remaining): Same.
28646 (get_string_length): Same.
28647 (format_string): Same.
28648 (format_directive): Same.
28650 (pass_sprintf_length::handle_gimple_call): Same.
28652 2017-01-12 Jakub Jelinek <jakub@redhat.com>
28654 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
28655 info.nowrite calls with no lhs that can't throw. Return bool
28656 whether gsi_remove has been called or not.
28657 (pass_sprintf_length::handle_gimple_call): Return bool whether
28658 try_substitute_return_value called gsi_remove. Formatting fix.
28659 (pass_sprintf_length::execute): Don't use gsi_remove if
28660 handle_gimple_call returned true.
28663 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
28664 be removed due to side-effects, don't remove following barrier nor
28665 turn the successor edge into fallthru edge.
28667 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28670 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
28671 element-reversing loads and stores as not swappable.
28673 2017-01-12 Nathan Sidwell <nathan@acm.org>
28674 Nicolai Stange <nicstange@gmail.com>
28676 * combine.c (try_combine): Don't ignore result of overlap checking
28677 loop. Combine overlap & asm check into single loop.
28679 2017-01-12 Richard Biener <rguenther@suse.de>
28681 * tree-pretty-print.c (dump_generic_node): Provide -gimple
28682 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
28684 2017-01-12 Richard Biener <rguenther@suse.de>
28686 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
28687 and TS_TARGET_OPTION directly derive from TS_BASE.
28688 * tree-core.h (tree_optimization_option): Derive from tree_base.
28689 (tree_target_option): Likewise.
28691 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
28693 * config/i386/i386.c (memory_address_length): Increase len
28694 only when rip_relative_addr_p returns false.
28696 2017-01-11 Julia Koval <julia.koval@intel.com>
28698 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
28699 (OPTION_MASK_ISA_SGX_SET): New.
28700 (ix86_handle_option): Handle OPT_msgx.
28701 * config.gcc: Added sgxintrin.h.
28702 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
28703 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
28704 * config/i386/i386.c (ix86_target_string): Add -msgx.
28706 (ix86_option_override_internal): Handle new options.
28707 (ix86_valid_target_attribute_inner_p): Add sgx.
28708 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
28709 * config/i386/i386.opt: Add msgx.
28710 * config/i386/sgxintrin.h: New file.
28711 * config/i386/x86intrin.h: Add sgxintrin.h.
28713 2017-01-11 Jakub Jelinek <jakub@redhat.com>
28716 * fold-const.c (maybe_nonzero_address): Return 1 for function
28718 (tree_single_nonzero_warnv_p): Don't handle function local objects
28722 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
28725 2017-01-11 David Malcolm <dmalcolm@redhat.com>
28728 * opts.c: Include "spellcheck.h"
28729 (struct string_fragment): New struct.
28730 (struct edit_distance_traits<const string_fragment &>): New
28732 (get_closest_sanitizer_option): New function.
28733 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
28735 2017-01-11 Jakub Jelinek <jakub@redhat.com>
28737 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
28739 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
28740 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
28741 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
28742 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
28743 for initial die_offset if dwarf_split_debug_info.
28744 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
28745 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
28747 (output_skeleton_debug_sections): Formatting fix. Use
28748 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
28749 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
28751 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
28753 * config/arm/cortex-a53.md: Add bypasses for
28754 cortex_a53_r2f_cvt.
28755 (cortex_a53_r2f): Only use for transfers.
28756 (cortex_a53_f2r): Likewise.
28757 (cortex_a53_r2f_cvt): Add reservation for conversions.
28758 (cortex_a53_f2r_cvt): Likewise.
28760 2017-01-11 Tamar Christina <tamar.christina@arm.com>
28762 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
28763 to all inlined functions, change static to extern.
28765 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
28768 * config/arm/arm.c (legitimize_pic_address): Handle reference to
28770 (arm_assemble_integer): Likewise.
28772 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
28774 * config.gcc: Use new awk script to check CPU, FPU and architecture
28775 parameters for --with-... options.
28776 * config/arm/parsecpu.awk: New file
28777 * config/arm/arm-cpus.in: New file.
28778 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
28780 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
28782 * config/arm/t-arm: Update dependency rules.
28783 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
28784 of processing .def files.
28785 * config/arm/genopt.sh: Deleted.
28786 * config/arm/gentune.sh: Deleted.
28787 * config/arm/arm-cores.def: Deleted.
28788 * config/arm/arm-arches.def: Deleted.
28789 * config/arm/arm-fpus.def: Deleted.
28790 * config/arm/arm-tune.md: Regenerated.
28791 * config/arm/arm-tables.opt: Regenerated.
28792 * config/arm/arm-cpu.h: New generated file.
28793 * config/arm/arm-cpu-data.h: New generated file.
28794 * config/arm/arm-cpu-cdata.h: New generated file.
28796 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
28799 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
28801 (input_varpool_node): Unpack dynamically_initialized bit.
28803 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
28805 PR rtl-optimization/79032
28806 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
28807 the alignment of the adjusted memory reference against that of MODE,
28808 instead of the alignment of the original memory reference.
28810 2017-01-11 Martin Jambor <mjambor@suse.cz>
28812 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
28814 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
28815 decorated functions.
28817 2017-01-11 Richard Biener <rguenther@suse.de>
28819 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
28820 set range/nonnull info for PHI results. Do not set it on
28821 stmts marked for removal.
28823 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
28825 * expr.c (store_field): In the bitfield case, fetch the return value
28826 from the registers before applying a single big-endian adjustment.
28827 Always do a final load for a BLKmode value not larger than a word.
28829 2017-01-10 David Malcolm <dmalcolm@redhat.com>
28832 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
28833 that we correctly handle column numbers greater than
28834 LINE_MAP_MAX_COLUMN_NUMBER.
28836 2017-01-10 Martin Sebor <msebor@redhat.com>
28838 PR middle-end/78245
28839 * gimple-ssa-sprintf.c (get_destination_size): Call
28840 {init,fini}object_sizes.
28841 * tree-object-size.c (addr_object_size): Adjust.
28842 (pass_through_call): Adjust.
28843 (pass_object_sizes::execute): Adjust.
28844 * tree-object-size.h (fini_object_sizes): Declare.
28846 2017-01-10 Martin Sebor <msebor@redhat.com>
28848 PR tree-optimization/78775
28849 * builtins.c (get_size_range): Move...
28850 * calls.c: ...to here.
28851 (alloc_max_size): Accept zero argument.
28852 (operand_signed_p): Remove.
28853 (maybe_warn_alloc_args_overflow): Call get_size_range.
28854 * calls.h (get_size_range): Declare.
28856 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
28858 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
28859 from TI's devices.csv file as of September 2016.
28860 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
28862 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
28864 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
28865 * doc/invoke.texi: Likewise.
28866 * doc/md.texi: Likewise.
28867 * doc/objc.texi: Likewise.
28869 2017-01-10 Joshua Conner <joshconner@google.com>
28871 * config/arm/fuchsia-elf.h: New file.
28872 * config/fuchsia.h: New file.
28873 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
28874 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
28876 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
28878 2016-01-10 Richard Biener <rguenther@suse.de>
28880 PR tree-optimization/79034
28881 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28882 Propagate out degenerate PHIs in the joiner.
28884 2017-01-10 Martin Liska <mliska@suse.cz>
28886 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
28887 (sort_congruence_classes_by_decl_uid): Likewise.
28888 (sort_congruence_class_groups_by_decl_uid): Likewise.
28889 (sem_item_optimizer::merge_classes): Sort class, groups in these
28890 classes and members in the groups by DECL_UID of declarations.
28891 This would make merge operations stable.
28893 2017-01-10 Martin Liska <mliska@suse.cz>
28895 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
28896 usage of m_classes_vec.
28897 (sem_item_optimizer::~sem_item_optimizer): Likewise.
28898 (sem_item_optimizer::get_group_by_hash): Likewise.
28899 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
28900 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
28901 (sem_item_optimizer::verify_classes): Likewise.
28902 (sem_item_optimizer::process_cong_reduction): Likewise.
28903 (sem_item_optimizer::dump_cong_classes): Likewise.
28904 (sem_item_optimizer::merge_classes): Likewise.
28905 * ipa-icf.h (congruence_class_hash): Rename from
28906 congruence_class_group_hash. Remove declaration of m_classes_vec.
28908 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
28910 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
28911 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
28912 * config.gcc: Add avx512vpopcntdqintrin.h.
28913 * config/i386/avx512vpopcntdqintrin.h: New.
28914 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
28915 * config/i386/i386-builtin-types.def: Add new types.
28916 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
28917 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
28918 __builtin_ia32_vpopcountq_v8di_mask): New.
28919 * config/i386/i386-c.c (ix86_target_macros_internal): Define
28920 __AVX512VPOPCNTDQ__.
28921 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
28922 (PTA_AVX512VPOPCNTDQ): Define.
28923 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
28924 TARGET_AVX512VPOPCNTDQ_P): Define.
28925 * config/i386/i386.opt: Add mavx512vpopcntdq.
28926 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
28927 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
28929 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28931 PR middle-end/77484
28932 * predict.def (PRED_CALL): Set to 67.
28934 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
28936 * expr.c (store_field): In the bitfield case, if the value comes from
28937 a function call and is of an aggregate type returned in registers, do
28938 not modify the field mode; extract the value in all cases if the mode
28939 is BLKmode and the size is not larger than a word.
28941 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
28944 * config/i386/cpuid.h: Fix undefined behavior.
28946 2017-01-04 Jeff Law <law@redhat.com>
28948 PR tree-optimization/79007
28949 PR tree-optimization/67955
28950 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
28951 conservative for pt.null when flag_non_call_exceptions is on.
28953 2017-01-09 Jakub Jelinek <jakub@redhat.com>
28955 PR translation/79019
28956 PR translation/79020
28957 * params.def (PARAM_INLINE_MIN_SPEEDUP,
28958 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
28959 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
28961 * config/avr/avr.opt (maccumulate-args): Likewise.
28962 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
28963 * common.opt (freport-bug): Likewise.
28964 * cif-code.def (CIF_FINAL_ERROR): Likewise.
28965 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
28966 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
28967 translatable string.
28968 * config/i386/i386.c (function_value_32): Likewise.
28969 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
28970 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
28972 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
28973 * common/config/msp430/msp430-common.c (msp430_handle_option):
28975 * symtab.c (symtab_node::verify_base): Likewise.
28976 * opts.c (set_debug_level): Likewise.
28977 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
28978 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
28979 missing whitespace to translatable strings.
28980 * config/avr/avr.md (bswapsi2): Fix typo in comment.
28981 * config/sh/superh.h: Likewise.
28982 * config/i386/xopintrin.h: Likewise.
28983 * config/i386/znver1.md: Likewise.
28984 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
28985 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
28986 * double-int.h (struct double_int): Likewise.
28987 * double-int.c (div_and_round_double): Likewise.
28988 * wide-int.cc: Likewise.
28989 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
28990 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
28991 * cfgcleanup.c (crossjumps_occured): Renamed to ...
28992 (crossjumps_occurred): ... this.
28993 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
28996 PR tree-optimization/78899
28997 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
28998 returning bool return struct loop *, NULL for failure and the new
29000 (versionable_outer_loop_p): Don't version outer loop if it has
29001 dont_vectorized bit set.
29002 (tree_if_conversion): When versioning outer loop, ensure
29003 tree_if_conversion is performed also on the inner loop of the
29004 non-vectorizable outer loop copy.
29005 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
29006 LOOP_VECTORIZED in inner loop of the scalar outer loop and
29007 prevent vectorization of it.
29008 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
29009 the outer loop vectorization of the non-scalar version is attempted
29010 before vectorization of the inner loop in scalar version. If
29011 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
29012 vectorization of its inner loop.
29013 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
29014 has 2 inner loops, rename also on edges from bb whose single pred
29015 is outer_loop->header. Fix typo in function comment.
29017 2017-01-09 Martin Sebor <msebor@redhat.com>
29020 * asan.c (asan_emit_stack_protection): Increase local buffer size
29021 to avoid snprintf truncation warning.
29023 2017-01-09 Andrew Pinski <apinski@cavium.com>
29025 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
29026 to reference thunderx2t99 for the tuning structure
29027 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
29029 (thunderx2t99_extra_costs): This.
29030 * config/aarch64/aarch64-tune.md: Regenerate.
29031 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
29032 (vulcan_addrcost_table): This.
29033 (vulcan_regmove_cost): Rename to ...
29034 (thunderx2t99_regmove_cost): This.
29035 (vulcan_vector_cost): Rename to ...
29036 (thunderx2t99_vector_cost): this.
29037 (vulcan_branch_cost): Rename to ...
29038 (thunderx2t99_branch_cost): This.
29039 (vulcan_tunings): Rename to ...
29040 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
29041 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
29043 2017-01-09 Martin Jambor <mjambor@suse.cz>
29047 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
29048 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
29049 (propagate_vr_accross_jump_function): Use the above function for all
29050 value range computations for pass-through jump functions and type
29051 converasion from explicit value range values.
29052 (ipcp_propagate_stage): Do not attempt to deduce types of formal
29053 parameters from TYPE_ARG_TYPES.
29054 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
29055 (ipa_write_node_info): Stream type of the actual argument.
29056 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
29058 2017-01-09 Martin Liska <mliska@suse.cz>
29061 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
29062 (lookup_compiler): Do not show error message with have_E.
29064 2017-01-09 Jakub Jelinek <jakub@redhat.com>
29066 PR tree-optimization/78938
29067 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
29068 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
29069 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
29070 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
29073 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29075 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
29078 2017-01-09 Richard Biener <rguenther@suse.de>
29080 PR tree-optimization/78997
29081 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
29082 name condition properly.
29084 2017-01-09 Richard Biener <rguenther@suse.de>
29087 * dwarf2out.c (is_cxx): New overload with context.
29088 (is_naming_typedef_decl): Use it.
29090 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
29092 * invoke.texi (Option Summary): Correct spacing in option lists
29093 and add line breaks to fix over-long lines.
29095 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
29097 PR middle-end/17660
29099 * extend.texi (Common Variable Attributes): Add xref to GCC
29100 Internals manual to explain mode attribute keywords.
29102 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
29105 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
29106 and Preprocessor Options.
29107 (Options for Linking): Document -pthread here....
29108 (RS/6000 and PowerPC Options): ...not here.
29109 (Solaris 2 Options): ...or here.
29110 * doc/cppopts.texi: Document -pthread.
29112 2017-01-08 Martin Sebor <msebor@redhat.com>
29114 PR middle-end/77708
29115 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
29116 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
29117 New member functions.
29118 (format_directive): Used them.
29120 (pass_sprintf_length::handle_gimple_call): Same.
29121 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
29122 to avoid truncation for any argument.
29123 (extract_affine_mul): Same.
29124 * tree.c (get_file_function_name): Same.
29126 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29128 PR middle-end/77484
29129 * predict.def (PRED_INDIR_CALL): Set to 86.
29131 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
29133 PR preprocessor/54124
29134 * doc/cppopts.texi: Reformat -d subtable to list the full name
29135 of the options. Add cross-reference to the docs for the general
29136 compiler -d options.
29137 * doc/invoke.texi (Developer Options): Add cross-reference to the
29138 preprocessor-specific -d option documentation.
29140 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
29142 PR preprocessor/13498
29143 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
29144 redudant material, and reflect new command-line options.
29145 (System Headers): Likewise.
29147 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
29149 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
29150 -isystem, and -idirafter. Copy-edit.
29151 * doc/cppopts.texi: Copy-edit. Remove contradiction about
29152 default for -ftrack-macro-expansion. Delete obsolete and
29153 badly-formatted implementation details about -fdebug-cpp output.
29154 * doc/cppwarnopts.texi: Copy-edit.
29156 2017-01-07 David Malcolm <dmalcolm@redhat.com>
29159 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
29160 that the transition from a max line width >= 1<<10 to narrower
29161 lines works correctly.
29163 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
29165 * doc/options.texi (PerFunction): New.
29166 * opt-functions.awk (switch_flags): Map both Optimization and
29167 PerFunction to CL_OPTIMIZATION.
29168 * opth-gen.awk: Test for PerFunction flag along with
29170 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
29171 it only when the latter is present. Skip those that don't in
29172 the hash function generator.
29173 * common.opt (fvar-tracking): Mark as PerFunction instead of
29175 (fvar-tracking-assignments): Likewise.
29176 (fvar-tracking-assignments-toggle): Likewise.
29177 (fvar-tracking-uninit): Likewise.
29179 2017-01-07 Jakub Jelinek <jakub@redhat.com>
29181 PR translation/79018
29182 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
29185 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
29188 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
29189 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
29190 TARGET_LONG_JUMP_TABLE_OFFSETS.
29191 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
29192 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
29193 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
29194 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
29195 * config/m68k/m68k.md (tablejump expander): Likewise.
29196 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
29197 TARGET_LONG_JUMP_TABLE_OFFSETS.
29198 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
29199 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
29201 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
29202 David Holsgrove <david.holsgrove@xilinx.com>
29204 * common/config/microblaze/microblaze-common.c
29205 (TARGET_EXCEPT_UNWIND_INFO): Remove.
29206 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
29208 * config/microblaze/microblaze.c (microblaze_must_save_register)
29209 (microblaze_expand_epilogue, microblaze_return_addr): Handle
29211 (microblaze_eh_return): New function.
29212 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
29213 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
29214 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
29215 * config/microblaze/microblaze.md (eh_return): New pattern.
29217 2017-01-06 Jakub Jelinek <jakub@redhat.com>
29219 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
29220 GCC_DIAGNOSTIC_STRINGIFY): Define.
29222 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
29224 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29226 * config/arm/arm.md (<mcrr>): New.
29228 * config/arm/arm.c (arm_arch5te): New.
29229 (arm_option_override): Set arm_arch5te.
29230 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
29232 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
29233 (arm_mcrr_qualifiers): ... this. New.
29234 (MRRC_QUALIFIERS): Define to...
29235 (arm_mrrc_qualifiers): ... this. New.
29236 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
29238 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
29239 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
29240 (MRRCI, mrrc, MRRC): New.
29241 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
29242 VUNSPEC_MRRC2): New.
29244 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29246 * config/arm/arm.md (<mcr>): New.
29248 * config/arm/arm.c (arm_coproc_builtin_available): Add
29249 support for mcr, mrc, mcr2 and mrc2.
29250 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
29251 (arm_mcr_qualifiers): ... this. New.
29252 (MRC_QUALIFIERS): Define to ...
29253 (arm_mrc_qualifiers): ... this. New.
29254 (MCR_QUALIFIERS): Define to ...
29255 (arm_mcr_qualifiers): ... this. New.
29256 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
29258 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
29259 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
29260 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
29261 VUNSPEC_MRC2): New.
29263 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29265 * config/arm/arm.md (*ldc): New.
29269 * config/arm/arm.c (arm_coproc_builtin_available): Add
29270 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
29271 (arm_coproc_ldc_stc_legitimate_address): New.
29272 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
29273 'qualifier_const_pointer'.
29274 (LDC_QUALIFIERS): Define to...
29275 (arm_ldc_qualifiers): ... this. New.
29276 (STC_QUALIFIERS): Define to...
29277 (arm_stc_qualifiers): ... this. New.
29278 * config/arm/arm-protos.h
29279 (arm_coproc_ldc_stc_legitimate_address): New.
29280 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
29281 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
29282 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
29283 stc2, stcl, stc2l): New.
29284 * config/arm/constraints.md (Uz): New.
29285 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
29286 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
29287 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
29288 VUNSPEC_STC2L): New.
29290 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29292 * config/arm/arm.md (<cdp>): New.
29293 * config/arm/arm.c (neon_const_bounds): Rename this ...
29294 (arm_const_bounds): ... this.
29295 (arm_coproc_builtin_available): New.
29296 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
29297 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
29298 (CDP_QUALIFIERS): Define to...
29299 (arm_cdp_qualifiers): ... this. New.
29301 (arm_expand_builtin_args): Add case for 6 arguments.
29302 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
29303 (arm_const_bounds): ... this.
29304 (arm_coproc_builtin_available): New.
29305 * config/arm/arm_acle.h (__arm_cdp): New.
29307 * config/arm/arm_acle_builtins.def (cdp): New.
29309 * config/arm/iterators.md (CDPI,CDP,cdp): New.
29310 * config/arm/neon.md: Rename all 'neon_const_bounds' to
29311 'arm_const_bounds'.
29312 * config/arm/types.md (coproc): New.
29313 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
29314 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
29315 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
29316 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
29318 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29320 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
29321 (UBINOP_QUALIFIERS): New.
29323 (acle_builtin_data): New. Change comment.
29324 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
29325 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
29326 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
29327 arm_acle_builtins.def.
29328 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
29329 (arm_init_acle_builtins): New.
29330 (CRC32_BUILTIN): Remove.
29331 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
29332 crc32cb, crc32ch and crc32cw.
29333 (arm_init_crc32_builtins): Remove.
29334 (arm_init_builtins): Use arm_init_acle_builtins rather
29335 than arm_init_crc32_builtins.
29336 (arm_expand_acle_builtin): New.
29337 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
29338 * config/arm/arm_acle_builtins.def: New.
29340 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29342 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
29343 (arm_builtin_datum): ... this.
29344 (arm_init_neon_builtin): Rename to ...
29345 (arm_init_builtin): ... this. Add a new parameters PREFIX
29346 and USE_SIG_IN_NAME.
29347 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
29348 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
29349 'arm_builtin_datum'.
29350 (arm_init_vfp_builtins): Likewise.
29351 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
29352 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
29353 (arm_expand_neon_args): Rename to ...
29354 (arm_expand_builtin_args): ... this. Rename builtin_arg
29355 enum values and differentiate between ARG_BUILTIN_MEMORY
29356 and ARG_BUILTIN_NEON_MEMORY.
29357 (arm_expand_neon_builtin_1): Rename to ...
29358 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
29359 values, arm_expand_builtin_args and add bool parameter NEON.
29360 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
29361 (arm_expand_vfp_builtin): Likewise.
29362 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
29364 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29366 PR middle-end/77484
29367 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
29368 * predict.c (tree_estimate_probability_bb): Reverse direction of
29369 polymorphic call predictor.
29371 2017-01-06 David Malcolm <dmalcolm@redhat.com>
29373 * passes.c (execute_one_pass): Split out pass-skipping logic into...
29374 (determine_pass_name_match): ...this new function and...
29375 (should_skip_pass_p): ...this new function.
29377 2017-01-06 Nathan Sidwell <nathan@acm.org>
29379 * ipa-visibility.c (function_and_variable_visibility): Reformat
29380 comments and long lines. Remove extrneous if.
29381 * symtab.c (symtab_node::make_decl_local): Fix code format.
29382 (symtab_node::set_section_for_node): Fix comment typo.
29384 2017-01-06 Martin Liska <mliska@suse.cz>
29387 * lra-constraints.c: Rename invariant to lra_invariant.
29388 * predict.c (set_even_probabilities): Initialize e to NULL.
29390 2017-01-05 Martin Sebor <msebor@redhat.com>
29392 PR tree-optimization/78910
29393 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
29394 (format_integer): Correct off-by-one error in the handling
29395 of precision with negative numbers in signed conversions..
29397 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
29399 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
29401 2017-01-05 Jakub Jelinek <jakub@redhat.com>
29403 PR tree-optimization/71016
29404 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
29405 factor_out_conditional_conversion. Formatting fix.
29406 (factor_out_conditional_conversion): Add cond_stmt argument.
29407 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
29408 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
29411 2017-01-05 David Malcolm <dmalcolm@redhat.com>
29413 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
29414 read-rtl-function.o, and selftest-rtl.o.
29415 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
29416 (selftest::aarch64_test_loading_full_dump): New function.
29417 (selftest::aarch64_run_selftests): New function.
29418 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
29419 selftest::aarch64_run_selftests.
29420 * config/i386/i386.c
29421 (selftest::ix86_test_loading_dump_fragment_1): New function.
29422 (selftest::ix86_test_loading_call_insn): New function.
29423 (selftest::ix86_test_loading_full_dump): New function.
29424 (selftest::ix86_test_loading_unspec): New function.
29425 (selftest::ix86_run_selftests): Call the new functions.
29426 * emit-rtl.c (maybe_set_max_label_num): New function.
29427 * emit-rtl.h (maybe_set_max_label_num): New decl.
29428 * function.c (instantiate_decls): Guard call to
29429 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
29430 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
29432 * gensupport.c (gen_reader::gen_reader): Pass "false"
29433 for new "compact" param of rtx_reader.
29434 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
29435 rather than an empty string for NULL strings.
29436 * read-md.c: Potentially include config.h rather than bconfig.h.
29437 Wrap include of errors.h with #ifdef GENERATOR_FILE.
29438 (have_error): New global, copied from errors.c.
29439 (md_reader::read_name): Rename to...
29440 (md_reader::read_name_1): ...this, adding "out_loc" param,
29441 and converting "missing name or number" to returning false, rather
29443 (md_reader::read_name): Reimplement in terms of read_name_1.
29444 (md_reader::read_name_or_nil): New function.
29445 (md_reader::read_string): Handle "(nil)" by returning NULL.
29446 (md_reader::md_reader): Add new param "compact".
29447 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
29448 (md_reader::read_file): New method.
29449 * read-md.h (md_reader::md_reader): Add new param "compact".
29450 (md_reader::read_file): New method.
29451 (md_reader::is_compact): New accessor.
29452 (md_reader::read_name): Convert return type from void to file_location.
29453 (md_reader::read_name_or_nil): New decl.
29454 (md_reader::read_name_1): New decl.
29455 (md_reader::m_compact): New field.
29456 (noop_reader::noop_reader): Pass "false" for new "compact" param
29458 (rtx_reader::rtx_reader): Add new "compact" param.
29459 (rtx_reader::read_rtx_operand): Make virtual and convert return
29460 type from void to rtx.
29461 (rtx_reader::read_until): New decl.
29462 (rtx_reader::handle_any_trailing_information): New virtual function.
29463 (rtx_reader::postprocess): New virtual function.
29464 (rtx_reader::finalize_string): New virtual function.
29465 (rtx_reader::m_in_call_function_usage): New field.
29466 (rtx_reader::m_reuse_rtx_by_id): New field.
29467 * read-rtl-function.c: New file.
29468 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
29469 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
29470 (selftest::verify_three_block_rtl_cfg): New decl.
29471 * read-rtl-function.h: New file.
29472 * read-rtl.c: Potentially include config.h rather than bconfig.h.
29473 For host, include function.h, memmodel.h, and emit-rtl.h.
29474 (one_time_initialization): New function.
29475 (struct compact_insn_name): New struct.
29476 (compact_insn_names): New array.
29477 (find_code): Handle insn codes in compact dumps.
29478 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
29479 (bind_subst_iter_and_attr): Likewise.
29480 (add_condition_to_string): Likewise.
29481 (add_condition_to_rtx): Likewise.
29482 (apply_attribute_uses): Likewise.
29483 (add_current_iterators): Likewise.
29484 (apply_iterators): Likewise.
29485 (initialize_iterators): Guard usage of apply_subst_iterator with
29486 #ifdef GENERATOR_FILE.
29487 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
29488 (md_reader::read_mapping): Likewise.
29489 (add_define_attr_for_define_subst): Likewise.
29490 (add_define_subst_attr): Likewise.
29491 (read_subst_mapping): Likewise.
29492 (check_code_iterator): Likewise.
29493 (rtx_reader::read_rtx): Likewise. Move one-time initialization
29495 (one_time_initialization): New function.
29496 (rtx_reader::read_until): New method.
29497 (read_flags): New function.
29498 (parse_reg_note_name): New function.
29499 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
29500 Handle reuse_rtx ids.
29501 Wrap iterator lookup within #ifdef GENERATOR_FILE.
29502 Add parsing support for RTL dumps, mirroring the special-cases in
29503 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
29504 values, and calling handle_any_trailing_information.
29505 (rtx_reader::read_rtx_operand): Convert return type from void
29506 to rtx, returning return_rtx. Handle case 'e'. Call
29507 finalize_string on XSTR and XTMPL fields.
29508 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
29509 "(nil)" values were omitted. Call the postprocess vfunc on the
29511 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
29512 class ctor. Initialize m_in_call_function_usage. Call
29513 one_time_initialization.
29514 * rtl-tests.c (selftest::test_uncond_jump): Call
29515 set_new_first_and_last_insn.
29516 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
29517 * selftest-rtl.c: New file.
29518 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
29519 (selftest::get_insn_by_uid): New decl.
29520 * selftest-run-tests.c (selftest::run_tests): Call
29521 read_rtl_function_c_tests.
29522 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
29523 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
29526 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
29528 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
29529 operands in a special way. Assert that pos+len <= mode precision.
29531 2017-01-05 Jakub Jelinek <jakub@redhat.com>
29533 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
29534 3 argument Alias with unlimited for the negative form.
29535 (fno-vect-cost-model): Removed.
29537 2017-01-05 Martin Liska <mliska@suse.cz>
29539 * hsa-gen.c (gen_hsa_divmod): New function.
29540 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
29542 2017-01-05 Martin Liska <mliska@suse.cz>
29545 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
29548 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29550 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
29551 small constant length operands.
29553 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29555 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
29556 between loop iterations.
29558 2017-01-05 Martin Liska <mliska@suse.cz>
29561 * gimplify.c (gimplify_decl_expr): Compare to
29562 asan_poisoned_variables instread of checking flags.
29563 (gimplify_target_expr): Likewise.
29564 (gimplify_expr): Likewise.
29565 (gimplify_function_tree): Conditionally initialize
29566 asan_poisoned_variables.
29568 2017-01-04 Jeff Law <law@redhat.com>
29570 PR tree-optimizatin/78812
29571 * rtl.h (contains_mem_rtx_p): Prototype.
29572 * ifcvt.c (containts_mem_rtx_p): Move from here to...
29573 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
29574 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
29575 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
29576 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
29578 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29580 * input.c (assert_char_at_range): Default-initialize actual_range.
29582 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29584 * df-scan.c (df_ref_create_structure): Make regno unsigned,
29585 to match the caller.
29587 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29589 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
29590 insns after final jump in test to emit dummy move.
29592 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29594 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
29595 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
29597 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29599 * multiple_target.c (create_dispatcher_calls): Init e_next.
29600 * tree-ssa-loop-split.c (split_loop): Init border.
29601 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
29604 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
29609 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
29610 (altivec_register_operand): Do not return true if the operand
29611 contains a SUBREG mixing SImode and SFmode.
29612 (vsx_register_operand): Likewise.
29613 (vsx_reg_sfsubreg_ok): New predicate.
29614 (vfloat_operand): Do not return true if the operand contains a
29615 SUBREG mixing SImode and SFmode.
29616 (vint_operand): Likewise.
29617 (vlogical_operand): Likewise.
29618 (gpc_reg_operand): Likewise.
29619 (int_reg_operand): Likewise.
29620 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
29621 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
29622 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
29624 (rs6000_emit_move_si_sf_subreg): New helper function.
29625 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
29626 fixup SUBREGs involving SImode and SFmode.
29627 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
29628 numbers for the new peephole2 optimization.
29629 (peephole2 for SFmode unions): New peephole2 to optimize cases in
29630 the GLIBC math library that do AND/IOR/XOR operations on single
29631 precision floating point.
29632 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
29633 target macros to say whether we need to avoid SUBREGs mixing
29635 (TARGET_ALLOW_SF_SUBREG): Likewise.
29636 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
29637 (UNSPEC_SI_FROM_SF): Likewise.
29638 (iorxor): Change spacing.
29639 (and_ior_xor): New iterator for AND, IOR, and XOR.
29640 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
29641 (movdi_from_sf_zero_ext): Likewise.
29642 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
29643 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
29644 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
29645 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
29646 (fms<mode>4): Likewise.
29647 (fnma<mode>4): Likewise.
29648 (fnms<mode>4): Likewise.
29649 (nfma<mode>4): Likewise.
29650 (nfms<mode>4): Likewise.
29652 2017-01-04 Marek Polacek <polacek@redhat.com>
29655 * doc/invoke.texi: Document -Wpointer-compare.
29657 2017-01-04 Jakub Jelinek <jakub@redhat.com>
29659 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
29662 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
29663 descriptions for -gdwarf-5 and emit them as uleb128 instead of
29666 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
29669 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
29670 documentation of the powerpc_popcntb_ok attribute.
29671 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29672 code to issue warning messages if a requested CPU configuration is
29673 not supported by the binary (assembler and loader) toolchain.
29674 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
29675 made to define a built-in function that has been disabled.
29676 (paired_init_builtins): Add assertion to prevent ICE if attempt is
29677 made to define a built-in function that has been disabled.
29678 (altivec_init_builtins): Add comment explaining why definition
29679 of the DST built-in functions is not preceded by an assertion
29680 check. Add assertions to prevent ICE if attempts are made to
29681 define an altivec predicate or an abs* built-in function that has
29683 (htm_init_builtins): Add comment explaining why definition of the
29684 htm built-in functions is not preceded by an assertion check.
29686 2017-01-04 Jeff Law <law@redhat.com>
29688 PR tree-optimizatin/67955
29689 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
29690 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
29691 the points-to solution does not include pt_null. Use DECL_PT_UID
29694 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
29696 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
29697 Use gen_int_mode instead of gen_lopwart for const_int operands.
29699 2017-01-04 Jakub Jelinek <jakub@redhat.com>
29701 PR tree-optimization/71563
29702 * match.pd: Simplify X << Y into X if Y is known to be 0 or
29703 out of range value - has low bits known to be zero.
29705 2017-01-04 Alan Modra <amodra@gmail.com>
29707 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
29708 * configure: Regenerate.
29709 * config.in: Regenerate.
29711 2017-01-04 Jakub Jelinek <jakub@redhat.com>
29714 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
29715 a substring of the message, but strcmp with the whole message. Ifdef
29716 ENABLE_NLS, translate the message first using dgettext.
29718 2017-01-03 Jeff Law <law@redhat.com>
29720 PR tree-optimizatin/78856
29721 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
29722 (mark_threaded_blocks): Remove code to truncate thread paths that
29723 cross multiple loop headers. Instead invalidate the cached loop
29724 iteration information and handle case of a thread path walking
29725 into an irreducible region.
29727 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
29730 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
29731 assertions. Add support for doing the signbit if the IEEE 128-bit
29732 floating point value is in a GPR.
29733 * config/rs6000/rs6000.md (Fsignbit): Delete.
29734 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
29735 Update the length attribute if the value is in a GPR.
29736 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
29737 the sign or zero extension instruction, since the value is always 0/1.
29738 (signbit<mode>2_dm2): Delete using <Fsignbit>.
29741 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
29742 extracting SImode to a GPR register so that we can generate a
29743 store, limit the vector to be in a traditional Altivec register
29744 for the vextuwrx instruction.
29746 2017-01-03 Ian Lance Taylor <iant@google.com>
29748 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
29750 2017-01-03 Martin Sebor <msebor@redhat.com>
29752 PR tree-optimization/78696
29753 * gimple-ssa-sprintf.c (format_floating): Correct handling of
29754 precision. Use MPFR for %f for greater fidelity. Correct handling
29756 (pass_sprintf_length::compute_format_length): Set width and precision
29757 specified by asrerisk to void_node for vararg functions.
29758 (try_substitute_return_value): Adjust dump output.
29760 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
29762 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
29764 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
29766 * doc/invoke.texi (SPARC options): Document -mlra as the default.
29767 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
29768 -mlra/-mno-lra was passed to the compiler.
29770 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
29772 PR rtl-optimization/65618
29773 * emit-rtl.c (try_split): Move initialization of "before" and
29774 "after" to just before the call to emit_insn_after_setloc.
29776 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
29778 * doc/md.texi (Standard Names): Remove reference to Java frontend.
29780 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
29782 * dwarf2out.c (gen_enumeration_type_die): When
29783 -gno-strict-dwarf, add a DW_AT_encoding attribute.
29785 2017-01-03 Jakub Jelinek <jakub@redhat.com>
29787 PR tree-optimization/78965
29788 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
29789 Change first argument from const call_info & to call_info &. For %n
29790 set info.nowrite to false.
29792 PR middle-end/78901
29793 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
29794 possibly throwing calls.
29796 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
29797 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
29798 and fns handling, rather than in a separate case SSA_NAME.
29800 2017-01-02 Jeff Law <law@redhat.com>
29802 * config/darwin-driver.c (darwin_driver_init): Const-correctness
29803 fixes for first_period and second_period variables.
29805 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
29808 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
29809 (*insvqi_1): New insn pattern.
29810 (*insvqi_1_mem_rex64): Ditto.
29811 (*insvqi_2): Ditto.
29812 (*insvqi_3): Rename from *insvqi.
29814 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
29816 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
29818 * doc/cfg.texi (Edges): Remove reference to Java.
29819 (Maintaining the CFG): Ditto.
29821 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29823 PR middle-end/77674
29824 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
29825 transparent aliases.
29827 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29829 PR middle-end/77484
29830 * predict.def (PRED_CALL): Update hitrate.
29831 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
29832 * predict.c (tree_estimate_probability_bb): Split CALL predictor
29833 into direct/indirect/polymorphic variants.
29835 2017-01-01 Jakub Jelinek <jakub@redhat.com>
29837 Update copyright years.
29839 * gcc.c (process_command): Update copyright notice dates.
29840 * gcov-dump.c (print_version): Ditto.
29841 * gcov.c (print_version): Ditto.
29842 * gcov-tool.c (print_version): Ditto.
29843 * gengtype.c (create_file): Ditto.
29844 * doc/cpp.texi: Bump @copying's copyright year.
29845 * doc/cppinternals.texi: Ditto.
29846 * doc/gcc.texi: Ditto.
29847 * doc/gccint.texi: Ditto.
29848 * doc/gcov.texi: Ditto.
29849 * doc/install.texi: Ditto.
29850 * doc/invoke.texi: Ditto.
29852 Copyright (C) 2017 Free Software Foundation, Inc.
29854 Copying and distribution of this file, with or without modification,
29855 are permitted in any medium without royalty provided the copyright
29856 notice and this notice are preserved.