1 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
3 * profile-count.h (slow_safe_scale_64bit): New function.
4 (safe_scale_64bit): New inline.
5 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
6 * profile-count.c: Include wide-int.h
7 (slow_safe_scale_64bit): New.
9 2017-10-11 Nathan Sidwell <nathan@acm.org>
11 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
12 HAS_DECL_ASSEMBLER_NAME_P.
13 * gimple-expr.c (gimple_decl_printable_name: Check
14 HAS_DECL_ASSEMBLER_NAME_P too.
15 * ipa-utils.h (type_in_anonymous_namespace_p): Check
16 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
17 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
18 * passes.c (rest_of_decl_compilation): Check
19 HAS_DECL_ASSEMBLER_NAME_P too.
20 * recog.c (verify_changes): Likewise.
21 * tree-pretty-print.c (dump_decl_name): Likewise.
22 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
24 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
25 (SET_DECL_ASSEMBLER_NAME): Use it.
26 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
27 (COPY_DECL_ASSEMBLER_NAME): Likewise.
28 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
30 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
32 * config.gcc (i386, x86_64): Add extra objects.
33 * i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
34 (ix86_min_insn_size): Declare.
35 (ix86_issue_rate): Declare.
36 (ix86_adjust_cost): Declare.
37 (ia32_multipass_dfa_lookahead): Declare.
38 (ix86_macro_fusion_p): Declare.
39 (ix86_macro_fusion_pair_p): Declare.
40 (ix86_bd_has_dispatch): Declare.
41 (ix86_bd_do_dispatch): Declare.
42 (ix86_core2i7_init_hooks): Declare.
43 (ix86_atom_sched_reorder): Declare.
44 * i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
45 (COSTS_N_BYTES): Move to x86-tune-costs.h.
46 (DUMMY_STRINGOP_ALGS):x86-tune-costs.h.
47 (rip_relative_addr_p): Rename to ...
48 (ix86_rip_relative_addr_p): ... this one; export.
49 (memory_address_length): Update.
50 (ix86_issue_rate): Move to x86-tune-sched.c.
51 (ix86_flags_dependent): Move to x86-tune-sched.c.
52 (ix86_agi_dependent): Move to x86-tune-sched.c.
53 (exact_dependency_1): Move to x86-tune-sched.c.
54 (exact_store_load_dependency): Move to x86-tune-sched.c.
55 (ix86_adjust_cost): Move to x86-tune-sched.c.
56 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
57 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
58 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
59 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
60 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
61 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
62 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
63 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
64 (min_insn_size): Rename to ...
65 (ix86_min_insn_size): ... this one; export.
66 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
67 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
68 (core2i7_first_cycle_multipass_backtrack): Move to x86-tune-sched-core.c.
69 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
70 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
71 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
72 (ix86_avoid_jump_mispredicts): Update.
73 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
74 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
75 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
76 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
77 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
78 (MAX_INSN): Move to ix86-tune-sched-bd.c.
79 (MAX_IMM): Move to ix86-tune-sched-bd.c.
80 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
81 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
82 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
83 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
84 (MAX_STORE): Move to ix86-tune-sched-bd.c.
85 (BIG): Move to ix86-tune-sched-bd.c.
86 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
87 (enum insn_path): Move to ix86-tune-sched-bd.c.
88 (get_mem_group): Move to ix86-tune-sched-bd.c.
89 (is_cmp): Move to ix86-tune-sched-bd.c.
90 (dispatch_violation): Move to ix86-tune-sched-bd.c.
91 (is_branch): Move to ix86-tune-sched-bd.c.
92 (is_prefetch): Move to ix86-tune-sched-bd.c.
93 (init_window): Move to ix86-tune-sched-bd.c.
94 (allocate_window): Move to ix86-tune-sched-bd.c.
95 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
96 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
97 (process_end_window): Move to ix86-tune-sched-bd.c.
98 (allocate_next_window): Move to ix86-tune-sched-bd.c.
99 (find_constant): Move to ix86-tune-sched-bd.c.
100 (get_num_immediates): Move to ix86-tune-sched-bd.c.
101 (has_immediate): Move to ix86-tune-sched-bd.c.
102 (get_insn_path): Move to ix86-tune-sched-bd.c.
103 (get_insn_group): Move to ix86-tune-sched-bd.c.
104 (count_num_restricted): Move to ix86-tune-sched-bd.c.
105 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
106 (add_insn_window): Move to ix86-tune-sched-bd.c.
107 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
108 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
109 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
110 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
111 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
112 (do_dispatch): Move to ix86-tune-sched-bd.c.
113 (has_dispatch): Move to ix86-tune-sched-bd.c.
114 * i386/t-i386: Add new object files.
115 * i386/x86-tune-costs.h: New file.
116 * i386/x86-tune-sched-atom.c: New file.
117 * i386/x86-tune-sched-bd.c: New file.
118 * i386/x86-tune-sched-core.c: New file.
119 * i386/x86-tune-sched.c: New file.
121 2017-10-11 Liu Hao <lh_mouse@126.com>
123 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
124 the generic version below instead.
125 (should_colorize): Recognize Windows consoles as terminals
127 * pretty-print.c [__MINGW32__] (write_all): New function.
128 [__MINGW32__] (find_esc_head): Likewise.
129 [__MINGW32__] (find_esc_terminator): Likewise.
130 [__MINGW32__] (eat_esc_sequence): Likewise.
131 [__MINGW32__] (mingw_ansi_fputs): New function that handles
133 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
136 2017-10-11 Richard Biener <rguenther@suse.de>
138 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
139 Properly call analyze_scalar_evolution with the loop of the stmt.
141 2017-10-11 Richard Biener <rguenther@suse.de>
143 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
144 * tree-core.h (tree_base): Add chrec_var union member.
145 * tree.h (CHREC_VAR): Remove.
146 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
147 * tree-chrec.h (build_polynomial_chrec): Adjust.
148 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
149 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
151 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
153 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
154 * match.pd: ... here.
155 ((T) X == (T) Y): Relax condition.
157 2017-10-11 Bin Cheng <bin.cheng@arm.com>
159 PR tree-optimization/82472
160 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
162 (break_alias_scc_partitions): Update postorder number.
164 2017-10-11 Martin Liska <mliska@suse.cz>
167 * opts.c (parse_no_sanitize_attribute): Do not use error_value
169 * opts.h (parse_no_sanitize_attribute): Remove last argument.
171 2017-10-11 Martin Liska <mliska@suse.cz>
173 * print-rtl.c (print_insn): Move declaration of idbuf
174 to same scope as name.
176 2017-10-11 Martin Liska <mliska@suse.cz>
181 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
184 2017-10-11 Martin Liska <mliska@suse.cz>
187 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
190 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
192 * config.gcc (default_gnu_indirect_function): Default to yes for
193 arm*-*-linux* with glibc.
195 2017-10-11 Richard Biener <rguenther@suse.de>
197 * tree-scalar-evolution.c (get_scalar_evolution): Handle
198 default-defs and types we do not want to analyze.
199 (interpret_loop_phi): Replace unreachable code with an assert.
200 (compute_scalar_evolution_in_loop): Remove and inline ...
201 (analyze_scalar_evolution_1): ... here, replacing condition with
202 what makes the intent clearer. Remove handling of cases
203 get_scalar_evolution now handles.
205 2017-10-10 Jim Wilson <wilson@tuliptree.org>
207 PR rtl-optimization/81434
208 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
209 comment for main loop. In sched_group_found if, also add checks for
210 pass and min_cost_group.
212 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
214 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
215 (rs6000_insn_cost): New function.
216 * config/rs6000/rs6000.md (cost): New attribute.
218 2017-10-10 Jakub Jelinek <jakub@redhat.com>
219 H.J. Lu <hongjiu.lu@intel.com>
223 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
224 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
225 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
226 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
227 to OPTION_MASK_ISA_AVX512VL - builtins that have both
228 OPTION_MASK_ISA_MMX and some other bit set require both
229 mmx and the ISAs without the mmx bit.
230 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
231 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
232 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
233 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
234 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
235 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
236 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
237 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
238 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
239 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
240 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
241 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
242 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
243 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
244 Add OPTION_MASK_ISA_MMX.
246 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
248 * config.gcc (armv7*-*-freebsd*): New target.
249 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
251 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
253 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
254 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
257 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
259 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
260 a template parameter.
261 (WIDE_INT_REF_FOR): Update accordingly.
262 * tree.h (wi::int_traits <const_tree>): Delete.
263 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
264 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
265 (wi::tree_to_wide_ref): New typedef.
266 (wi::to_wide): New function.
267 * calls.c (get_size_range): Use wi::to_wide when operating on
269 * cgraph.c (cgraph_node::create_thunk): Likewise.
270 * config/i386/i386.c (ix86_data_alignment): Likewise.
271 (ix86_local_alignment): Likewise.
272 * dbxout.c (stabstr_O): Likewise.
273 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
274 * expr.c (const_vector_from_tree): Likewise.
275 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
276 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
277 (fold_negate_expr_1, int_const_binop_1, const_binop)
278 (fold_convert_const_int_from_real, optimize_bit_field_compare)
279 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
280 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
281 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
282 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
283 (fold_not_const, round_up_loc): Likewise.
284 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
285 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
286 (alloca_call_type): Likewise.
287 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
288 * godump.c (go_output_typedef): Likewise.
289 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
290 * internal-fn.c (get_min_precision): Likewise.
291 * ipa-cp.c (ipcp_store_vr_results): Likewise.
292 * ipa-polymorphic-call.c
293 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
294 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
295 (ipa_modify_call_arguments): Likewise.
296 * match.pd: Likewise.
297 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
298 * print-tree.c (print_node_brief, print_node): Likewise.
299 * stmt.c (expand_case): Likewise.
300 * stor-layout.c (layout_type): Likewise.
301 * tree-affine.c (tree_to_aff_combination): Likewise.
302 * tree-cfg.c (group_case_labels_stmt): Likewise.
303 * tree-data-ref.c (dr_analyze_indices): Likewise.
304 (prune_runtime_alias_test_list): Likewise.
305 * tree-dump.c (dequeue_and_dump): Likewise.
306 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
307 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
308 * tree-pretty-print.c (dump_generic_node): Likewise.
309 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
310 (simple_iv_with_niters): Likewise.
311 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
312 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
313 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
314 * tree-ssa-loop-niter.c (split_to_var_and_offset)
315 (refine_value_range_using_guard, number_of_iterations_ne_max)
316 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
317 (get_cst_init_from_scev, record_nonwrapping_iv)
318 (scev_var_range_cant_overflow): Likewise.
319 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
320 * tree-ssa-pre.c (compute_avail): Likewise.
321 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
322 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
323 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
324 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
325 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
326 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
327 (dump_case_nodes, try_switch_expansion): Likewise.
328 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
329 (vect_do_peeling): Likewise.
330 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
331 * tree-vect-stmts.c (vectorizable_load): Likewise.
332 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
333 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
334 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
335 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
336 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
337 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
338 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
339 (evrp_dom_walker::before_dom_children): Likewise.
340 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
341 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
342 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
343 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
344 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
345 (get_range_pos_neg): Likewise.
346 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
347 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
348 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
349 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
350 * config/avr/avr.c (avr_fold_builtin): Likewise.
351 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
352 * config/msp430/msp430.c (msp430_attr): Likewise.
353 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
354 * config/powerpcspe/powerpcspe-c.c
355 (altivec_resolve_overloaded_builtin): Likewise.
356 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
357 (rs6000_expand_ternop_builtin): Likewise.
358 * config/rs6000/rs6000-c.c
359 (altivec_resolve_overloaded_builtin): Likewise.
360 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
361 (rs6000_expand_ternop_builtin): Likewise.
362 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
364 2017-10-10 Bin Cheng <bin.cheng@arm.com>
366 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
367 when copying loop nest with only one inner loop.
369 2017-10-10 Richard Biener <rguenther@suse.de>
371 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
372 blocks if SCEV is active.
373 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
375 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
376 (scev_initialize): Assert we are not yet initialized.
378 2017-10-10 Bin Cheng <bin.cheng@arm.com>
380 * tree-loop-distribution.c (generate_loops_for_partition): Remove
381 inner loop's exit stmt by making it always exit the loop, otherwise
382 we would generate an infinite empty loop.
384 2017-10-10 Bin Cheng <bin.cheng@arm.com>
386 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
387 renaming variables in new preheader if it's deleted.
389 2017-10-10 Bin Cheng <bin.cheng@arm.com>
391 * tree-loop-distribution.c (struct partition): Remove unused field
392 loops of the structure.
393 (partition_alloc, partition_free): Ditto.
394 (build_rdg_partition_for_vertex): Ditto.
396 2017-10-09 Jeff Law <law@redhat.com>
398 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
399 return type to match prototype and documentation.
401 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
403 * config/rs6000/rs6000.c (processor_costs): Move to ...
404 * config/rs6000/rs6000.h: ... here.
405 (rs6000_cost): Declare.
407 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
409 * except.c (setjmp_fn): New global variable.
410 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
411 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
412 if DONT_USE_BUILTIN_SETJMP is defined.
414 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
416 * target.def (insn_cost): New hook.
417 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
418 * doc/tm.texi: Regenerate.
419 * rtlanal.c (insn_cost): Use the new hook.
421 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
423 * combine.c (combine_validate_cost): Compute the new insn_cost,
424 not just pattern_cost.
425 (try_combine): Adjust comment.
427 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
429 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
431 * combine.c (uid_insn_cost): Adjust comment.
432 (combine_validate_cost): Adjust comment. Use pattern_cost instead
434 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
435 * dse.c (find_shift_sequence): Ditto.
436 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
437 (bb_valid_for_noce_process_p): Use pattern_cost.
438 * rtl.h (insn_rtx_cost): Delete.
439 (pattern_cost): New prototype.
440 (insn_cost): New prototype.
441 * rtlanal.c (insn_rtx_cost): Rename to...
442 (pattern_cost): ... this.
445 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
447 * config/i386/i386.md (*jcc_2): Remove insn pattern.
448 (*jcc<mode>_0_r_i387): Ditto.
449 (*jccxf_r_i387): Ditto.
450 (*jcc<mode>_r_i387): Ditto.
451 (*jccu<mode>_r_i387): Ditto.
452 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
453 (*jcc): Rename from *jcc_1.
455 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
457 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
458 deferred rescans after the lvx/stvx recombination pre-pass.
460 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
462 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
463 memory operation instruction support.
464 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
465 (rs6000-ibm-aix[789]*): Likewise.
466 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
467 Document new functions.
469 2017-10-09 Richard Biener <rguenther@suse.de>
471 PR tree-optimization/82397
472 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
473 equality only for semantically equal trees.
475 2017-10-09 Richard Biener <rguenther@suse.de>
477 PR tree-optimization/82449
478 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
479 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
480 allow constant addresses.
481 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
484 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
486 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
489 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
492 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
495 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
498 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
500 2017-10-09 Jakub Jelinek <jakub@redhat.com>
503 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
504 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
506 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
508 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
509 (WI_BINARY_PREDICATE_RESULT): Likewise.
510 (wi::binary_traits::operator_result): New type.
511 (wi::binary_traits::predicate_result): Likewise.
512 (generic_wide_int::operator~, unary generic_wide_int::operator-)
513 (generic_wide_int::operator==, generic_wide_int::operator!=)
514 (generic_wide_int::operator&, generic_wide_int::and_not)
515 (generic_wide_int::operator|, generic_wide_int::or_not)
516 (generic_wide_int::operator^, generic_wide_int::operator+
517 (binary generic_wide_int::operator-, generic_wide_int::operator*):
519 (operator~, unary operator-, operator==, operator!=, operator&)
520 (operator|, operator^, operator+, binary operator-, operator*): New
522 * expr.c (get_inner_reference): Use wi::bit_and_not.
523 * fold-const.c (fold_binary_loc): Likewise.
524 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
525 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
526 (bit_value_binop): Likewise.
527 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
528 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
529 (extract_range_from_binary_expr_1): Likewise.
530 (masked_increment): Likewise.
531 (simplify_bit_ops_using_ranges): Likewise.
533 2017-10-09 Martin Jambor <mjambor@suse.cz>
536 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
537 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
538 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
540 (hsa_fixup_mov_insn_type): New function.
541 (hsa_op_with_type::get_in_type): Use it.
542 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
543 immediates in an assert.
544 (hsa_op_with_type::extend_int_to_32bit): New method.
545 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
546 types. Convert to dest type if necessary.
547 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
548 (reg_for_gimple_ssa): Pass false as min32int to
549 hsa_type_for_scalar_tree_type.
550 (gen_hsa_addr): Fixup type when creating addresable temporary.
551 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
552 (gen_hsa_unary_operation): Extend operands and convert to dest type if
553 necessary. Call hsa_fixup_mov_insn_type.
554 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
555 extend operands and convert to dest type if necessary.
556 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
557 to dest type if necessary.
558 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
559 if conversion nt necessary and size matches.
560 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
561 to dest type if necessary.
562 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
563 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
565 (gen_hsa_clrsb): Likewise.
566 (gen_hsa_ffs): Likewise.
567 (gen_hsa_divmod): Extend operands and convert to dest type if
569 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
571 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
573 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
574 Remove empty default arguments. Use a brace block as output
576 (conditional return): Ditto.
578 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
579 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
581 (group_ending_nop): Ditto.
583 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
584 (splitters for those): Ditto.
586 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
588 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
589 a conditional jump (and the compare for it) so that pc_rtx is the
591 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
592 for the deleted and renamed ctr<mode>_internal[234] patterns.
593 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
594 Delete second conditional return pattern.
595 (ctr<mode>_internal2): Delete this second bdnz pattern.
596 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
597 (ctr<mode>_internal4): Delete this second bdz pattern.
599 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
601 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
602 (always_initialized_rtx_for_ssa_name_p): New predicate.
603 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
604 (finish_out_of_ssa): Free new field of SA.
605 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
606 * tree-ssa-coalesce.c: Include tree-ssa.h.
607 (get_parm_default_def_partitions): Remove extern keyword.
608 (get_undefined_value_partitions): New function.
609 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
610 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
612 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
614 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
616 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
618 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
619 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
621 * i386.c (ix86_expand_set_or_movmem): Disable 512bit loops for targets
624 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
626 * i386.c (has_dispatch): Disable for Ryzen.
628 2017-10-08 Olivier Hainque <hainque@adacore.com>
630 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
631 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
633 (thumb_set_return_address): Likewise.
635 2017-10-08 Olivier Hainque <hainque@adacore.com>
637 * common/config/arm/arm-common.c (arm_except_unwind_info):
638 Handle DWARF2_UNWIND_INFO.
640 2017-10-07 Michael Collison <michael.collison@arm.com>
642 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
645 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
647 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
648 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
649 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
650 defined, force the creation of a new block for a dispatch label.
652 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
654 * invoke.texi (Wsuggest-attribute=cold): Document.
655 * common.opt (Wsuggest-attribute=cold): New
656 * ipa-pure-const.c (warn_function_cold): New function.
657 * predict.c (compute_function_frequency): Use it.
658 * predict.h (warn_function_cold): Declare.
660 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
662 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
665 2017-10-06 Martin Liska <mliska@suse.cz>
667 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
668 keyword for member functions.
669 (struct sanopt_tree_couple): New struct.
670 (struct sanopt_tree_couple_hash): New function.
671 (struct sanopt_ctx): Add new hash_map.
672 (has_dominating_ubsan_ptr_check): New function.
673 (record_ubsan_ptr_check_stmt): Likewise.
674 (maybe_optimize_ubsan_ptr_ifn): Likewise.
675 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
676 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
678 2017-10-06 Sudakshina Das <sudi.das@arm.com>
681 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
682 aarch64_simd_valid_immediate on CONST_VECTORs.
683 (aarch64_reg_or_bic_imm): Likewise.
685 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
687 PR rtl-optimization/82396
688 * haifa-sched.c (ready_sort_real): Disable qsort checking.
690 2017-10-06 Sebastian Pop <sebpop@gmail.com>
692 * graphite-dependences.c (scop_get_reads): Move code to...
693 (scop_get_must_writes): Move code to...
694 (scop_get_may_writes): Move code to...
695 (scop_get_reads_and_writes): ... here.
696 (scop_get_dependences): Call scop_get_reads_and_writes.
698 2017-10-06 Jakub Jelinek <jakub@redhat.com>
700 PR tree-optimization/82434
701 * fold-const.h (can_native_encode_type_p,
702 can_native_encode_string_p): Remove.
703 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
704 don't encode anything, just return what would be otherwise returned.
705 (native_encode_fixed, native_encode_complex, native_encode_vector):
707 (native_encode_string): Likewise. Inline by hand
708 can_native_encode_string_p.
709 (can_native_encode_type_p): Remove.
710 (can_native_encode_string_p): Remove.
711 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
712 STRING_CSTs using can_native_encode_string_p, test all
713 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
714 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
715 argument from native_encode_expr.
716 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
717 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
720 2017-10-06 Richard Biener <rguenther@suse.de>
722 PR tree-optimization/82397
723 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
724 operand_equal_p but rely on data_ref_compare_tree for detecting
726 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
727 to match up with dr_group_sort_cmp.
729 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
732 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
734 * config/s390/s390-builtin-types.def: Regenerate.
736 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
739 * config/s390/s390-builtin-types.def: Regenerate.
740 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
741 Change flag from B_VXE to B_VX.
742 (s390_vec_min_dbl): Remove B_VXE flag.
744 2017-10-06 Richard Biener <rguenther@suse.de>
746 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
747 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
748 translate_isl_ast_to_gimple::is_valid_rename,
749 translate_isl_ast_to_gimple::get_rename,
750 translate_isl_ast_to_gimple::get_def_bb_for_const,
751 translate_isl_ast_to_gimple::get_new_name,
752 translate_isl_ast_to_gimple::collect_all_ssa_names,
753 translate_isl_ast_to_gimple::copy_loop_phi_args,
754 translate_isl_ast_to_gimple::collect_all_ssa_names,
755 translate_isl_ast_to_gimple::copy_loop_phi_args,
756 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
757 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
758 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
759 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
760 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
761 translate_isl_ast_to_gimple::copy_cond_phi_args,
762 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
763 translate_isl_ast_to_gimple::edge_for_new_close_phis,
764 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
765 translate_isl_ast_to_gimple::rename_uses,
766 translate_isl_ast_to_gimple::rename_all_uses): Remove.
767 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
768 (set_rename_for_each_def): Likewise.
769 (graphite_copy_stmts_from_block): Handle debug stmt resetting
770 here. Handle rewriting SCEV analyzable uses here.
771 (copy_bb_and_scalar_dependences): Generate code for PHI
773 (graphite_regenerate_ast_isl): Adjust.
774 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
775 (add_write, add_read): New functions.
776 (build_cross_bb_scalars_def): Use it and simplify.
777 (build_cross_bb_scalars_use): Likewise.
778 (graphite_find_cross_bb_scalar_vars): Inline into...
779 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
780 simulating out-of-SSA. Compute liveout and add dependencies.
781 (build_scops): Force an empty entry block.
782 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
784 (sese_build_liveouts): Declare.
785 (sese_trivially_empty_bb_p): Likewise.
786 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
787 compute liveout and debug_liveout.
788 (sese_bad_liveouts_use): Remove.
789 (sese_reset_debug_liveouts_bb): Likewise.
790 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
791 (sese_build_liveouts): Build liveout and debug_liveout and store
793 (new_sese_info): Adjust.
794 (free_sese_info): Likewise.
795 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
796 do not build liveout here.
797 (move_sese_in_condition): Adjust region entry.
798 (scev_analyzable_p): Match up with chrec_apply requirements.
799 (sese_trivially_empty_bb_p): New.
800 * tree-into-ssa.c (get_reaching_def): Properly support generating
801 default-defs for incremental rewrite of anonymous names.
803 2017-10-06 Richard Biener <rguenther@suse.de>
805 * graphite-sese-to-poly.c (extract_affine): For casts increasing
806 precision do not perform modulo reduction.
808 2017-10-06 Richard Biener <rguenther@suse.de>
810 PR tree-optimization/82436
811 * tree-vect-slp.c (vect_supported_load_permutation_p): More
812 conservatively choose the vectorization factor when checking
813 whether we can perform the required load permutation.
814 (vect_transform_slp_perm_load): Assert when we may not fail.
816 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
818 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
819 message for incompatible -msdata=* and -mcall-* options.
821 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
823 * i386.c (ia32_multipass_dfa_lookahead): Default to issue rate
824 for post-reload scheduling.
826 2017-10-05 Tamar Christina <tamar.christina@arm.com>
828 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
830 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
832 * i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
833 to improve monte carlo in scimark.
835 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
837 * i386.c (ix86_size_cost, i386_cost, i486_cost, pentium_cost,
838 lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
839 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
840 pentium4_cost, nocona_cost): Set reassociation width to 1.
841 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
842 width to 2 for fp operations and 1 otherwise.
843 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
845 (atom_cost): Set reassociation width to 2.
846 (slm_cost, generic_cost): Set fp reassociation width to 2 and 1 otherwise.
847 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
848 (core_cost): Set fp reassociation width to 4 and vector to 2.
849 (ix86_reassociation_width): Rewrite using cost table; special case
850 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
851 and TARGET_AVX128_OPTIMAL.
852 * i386.h (processor_costs): Add
853 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
854 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
855 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
856 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
857 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
858 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
860 2017-10-05 Nathan Sidwell <nathan@acm.org>
862 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
864 2017-10-05 Tamar Christina <tamar.christina@arm.com>
866 * config/arm/arm.c (arm_test_fpu_data): New.
867 (arm_run_selftests): Call arm_test_fpu_data.
869 2017-10-04 Nathan Sidwell <nathan@acm.org>
871 * toplev.c (toplev::main): Remove excess parens on pretty_printer
873 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
875 2017-10-04 Sudakshina Das <sudi.das@arm.com>
877 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
878 check type for aarch64_simd_valid_immediate.
879 (aarch64_output_simd_mov_immediate): Update prototype.
880 (aarch64_simd_valid_immediate): Update prototype.
881 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
882 support for ORR-immediate.
883 (and<mode>3): modified pattern to add support for BIC-immediate.
884 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
885 now checks for valid immediate for BIC and ORR based on new enum
887 (aarch64_output_simd_mov_immediate): Function now used to output
888 BIC/ORR imm as well based on new enum argument.
889 * config/aarch64/constraints.md (Do): New vector immediate constraint.
891 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
892 (aarch64_reg_or_bic_imm): Likewise.
894 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
896 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
897 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
899 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
903 PR rtl-optimization/82396
904 * haifa-sched.c (autopref_multipass_init): Simplify
906 (autopref_rank_data): Simplify sort order.
907 * sched-int.h (autopref_multipass_data_): Remove
908 multi_mem_insn_p, min_offset and max_offset.
910 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
912 * doc/sourcebuild.texi: Document vect_peeling_profitable.
914 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
916 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
919 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
921 * doc/sourcebuild.texi: Document vect_long_mult.
923 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
925 PR tree-optimization/82413
926 * fold-const.c (build_range_check): Use widest_int when comparing
927 the maximum ETYPE value with HIGH.
929 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
931 PR rtl-optimization/82396
932 * haifa-sched.c (autopref_multipass_init): Simplify
934 (autopref_rank_data): Simplify sort order.
935 * sched-int.h (autopref_multipass_data_): Remove
936 multi_mem_insn_p, min_offset and max_offset.
938 2017-10-04 Jakub Jelinek <jakub@redhat.com>
940 PR tree-optimization/82381
941 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
942 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
945 PR tree-optimization/82374
946 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
947 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
948 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
949 current_function_decl to the new decl.
951 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
953 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
954 helper macro for IEEE float128 hardware built-in functions.
955 (SQRTF128_ODD): Add built-in functions with the round-to-odd
957 (TRUNCF128_ODD): Likewise.
958 (ADDF128_ODD): Likewise.
959 (SUBF128_ODD): Likewise.
960 (MULF128_ODD): Likewise.
961 (DIVF128_ODD): Likewise.
962 (FMAF128_ODD): Likewise.
963 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
964 UNSPEC_TRUNC_ROUND_TO_ODD.
965 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
966 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
967 floating point round to odd instructions.
968 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
969 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
970 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
971 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
972 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
973 (trunc<mode>sf2_hw): Change the truncate with round to odd
974 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
975 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
976 to odd hardware instructions.
977 (sub<mode>3_odd): Likewise.
978 (mul<mode>3_odd): Likewise.
979 (div<mode>3_odd): Likewise.
980 (sqrt<mode>2_odd): Likewise.
981 (fma<mode>4_odd): Likewise.
982 (fms<mode>4_odd): Likewise.
983 (nfma<mode>4_odd): Likewise.
984 (nfms<mode>4_odd): Likewise.
985 (trunc<mode>df2_odd): Change the truncate with round to odd
986 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
988 * doc/extend.texi (PowerPC built-in functions): Update documentation
989 for existing IEEE float128-bit built-in functions. Add built-in
990 functions that generate the IEEE 128-bit floating point round to
993 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
995 PR rtl-optimization/77729
996 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
997 to (X&(C1&~C2))|C2 transformations.
999 2017-10-03 Martin Jambor <mjambor@suse.cz>
1001 PR tree-optimization/82363
1002 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
1003 mismatch, mark lacc written regardless of racc.
1005 2017-10-03 Jakub Jelinek <jakub@redhat.com>
1007 PR tree-optimization/82381
1008 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
1009 stmt_to_insert nor wheather SSA_NAMEs are default defs.
1010 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
1011 fallthrough into reassoc_stmt_dominates_stmt_p.
1014 * combine.c (combine_instructions): Don't combine in unreachable
1017 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
1020 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
1021 function to not use the have_cpu variable. Do not set cpu_index,
1022 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
1024 (rs6000_valid_attribute_p): Remove duplicate initializations of
1025 old_optimize and func_optimize.
1026 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
1027 (rs6000_activate_target_options): Make global.
1028 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
1031 2017-10-02 Jakub Jelinek <jakub@redhat.com>
1033 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
1034 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
1035 Set *poffset to 0 and *psize and *pmax_size to -1 if
1036 *poffset + *psize overflows in HOST_WIDE_INT.
1038 PR tree-optimization/82387
1039 PR tree-optimization/82388
1040 PR tree-optimization/82389
1041 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
1042 instead of live_bytes non-NULL.
1044 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
1047 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
1050 2017-10-02 Richard Biener <rguenther@suse.de>
1052 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
1053 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
1054 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
1056 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
1058 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
1059 requested precision matches the type's.
1060 * calls.c (alloc_max_size): Calculate the new candidate size as
1061 a widest_int and use wi::to_widest when comparing it with the
1062 current candidate size.
1063 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
1064 zero rather than integer_zero_node.
1065 * match.pd: Check for a no-op conversion before using wi::add
1066 rather than after. Use tree_to_uhwi when summing small shift
1067 counts into an unsigned int.
1069 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
1070 Alan Hayward <alan.hayward@arm.com>
1071 David Sherwood <david.sherwood@arm.com>
1074 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
1075 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
1076 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
1077 POINTER_AND_FP_REGS.
1079 2017-10-02 Richard Biener <rguenther@suse.de>
1081 PR tree-optimization/82355
1082 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
1083 a mapping for the enclosing loop but avoid generating one for
1085 (copy_bb_and_scalar_dependences): Remove premature codegen
1086 error on PHIs in blocks duplicated into multiple places.
1087 * graphite-scop-detection.c
1088 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
1089 in the region use it as loop and nest to analyze the DR in.
1090 (try_generate_gimple_bb): Likewise.
1091 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
1092 (add_loop_constraints): For blocks in a loop not in the region
1093 create a dimension with a single iteration.
1094 * sese.h (gbb_loop_at_index): Remove assert.
1096 2017-10-01 Kevin Buettner <kevinb@redhat.com>
1098 * omp-expand.c (adjust_context_scope): New function.
1099 (expand_parallel_call): Call adjust_context_scope.
1101 2017-10-01 Jeff Law <law@redhat.com>
1103 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
1104 dom_opt_dom_walker class with direct access to private members.
1105 Add comments. Call test_for_singularity.
1106 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
1107 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
1108 m_dummy_cond anymore.
1109 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
1111 (pass_dominator:execute): Build the dummy_cond here and pass it
1112 to the dom_opt_dom_walker ctor.
1113 (test_for_singularity): New function.
1115 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
1116 Maya Rashish <coypu@sdf.org>
1118 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
1119 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
1120 (alpha*-*-netbsd*) Use nbsd_tm_file.
1121 (arm*-*-netbsdelf*) Likewise.
1122 (i[34567]86-*-netbsdelf*) Likewise.
1123 (x86_64-*-netbsd*) Likewise.
1124 (mips*-*-netbsd*) Likewise.
1125 (powerpc-*-netbsd*) Likewise.
1126 (sh*-*-netbsd*) Likewise.
1127 (sparc-*-netbsdelf*) Likewise.
1128 (sparc64-*-netbsd*) Likewise.
1129 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
1131 (vax-*-netbsdelf*) Likewise.
1132 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
1133 (UINT_FAST8_TYPE) Likewise.
1134 (INT_FAST16_TYPE) Check CHAR_FAST16.
1135 (UINT_FAST16_TYPE) Likewise.
1137 2017-09-30 Jakub Jelinek <jakub@redhat.com>
1140 * config/i386/i386.md
1141 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
1142 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
1143 *divmodsi4_zext_2): New define_insn_and_split.
1144 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
1145 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
1146 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
1147 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
1148 New define_insn_and_split.
1149 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
1150 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
1151 operands[1] having DImode when mode is SImode.
1153 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
1154 always SImode for DIV and MOD in REG_EQUAL notes.
1156 2017-09-29 Yury Gribov <tetra2005@gmail.com>
1159 * match.pd: Fix handling of NaNs in pattern.
1161 2017-09-29 Jeff Law <law@redhat.com>
1163 * sbitmap.c (bitmap_bit_in_range_p): New function.
1164 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
1165 * tree-ssa-dse.c (live_bytes_read): New function.
1166 (dse_classify_store): Ignore reads of dead bytes.
1168 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
1169 typos and whitespace errors.
1170 * config/i386/predicates.md (address_no_seg_operand): Likewise.
1171 * config/s390/s390.c (s390_emit_prologue): Likewise.
1173 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
1176 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
1177 with a symbol for LRA.
1179 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
1181 PR rtl-optimization/82338
1182 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
1184 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
1186 * genmodes.c (calc_wider_mode): Suppress qsort macro.
1187 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
1188 (qsort_chk): Declare.
1189 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
1190 (qsort_chk): New function.
1192 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1194 PR tree-optimization/82337
1195 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
1196 phi definition if the PHI result appears in an abnormal PHI.
1197 (find_basis_for_base_expr): Don't record a basis if the LHS of the
1198 basis appears in an abnormal PHI.
1200 2017-09-29 Richard Biener <rguenther@suse.de>
1202 * graphite-isl-ast-to-gimple.c
1203 (translate_isl_ast_to_gimple::set_codegen_error): New function.
1204 (binary_op_to_tree): Use it.
1205 (get_rename_from_scev): Likewise.
1206 (copy_loop_phi_nodes): Likewise.
1207 (copy_bb_and_scalar_dependences): Likewise.
1208 (translate_pending_phi_nodes): Likewise.
1210 2017-09-29 Jakub Jelinek <jakub@redhat.com>
1213 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
1214 for movabsq $(i32 << shift), r64.
1216 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
1218 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
1219 index when encoding %esp as %rsp to avoid 0x67 prefix.
1221 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
1223 * config/i386/i386.md (*movsf_internal, *movdf_internal):
1224 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
1226 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
1228 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
1229 Extensions with more than 16 double VFP registers.
1230 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
1231 to_clear_mask and all code related to it. Replace the remaining
1232 entry by a sbitmap and adapt code accordingly.
1234 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
1236 * brig-builtins.def: Change pure attributes to const.
1238 2017-09-28 Joseph Myers <joseph@codesourcery.com>
1240 * config.gcc (default_gnu_indirect_function): Default to yes for
1241 sparc*-*-linux* with glibc.
1243 2017-09-28 Joseph Myers <joseph@codesourcery.com>
1245 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
1246 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
1247 when creating .init_array and .fini_array sections with priority
1250 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
1253 * config/aarch64/aarch64.c
1254 (aarch64_builtin_support_vector_misalignment): Always return false
1255 when misalignment is unknown.
1257 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
1259 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
1260 this function to return false if the definition used by the swap
1261 instruction is artificial, or if the memory address from which the
1262 constant value is loaded is not represented by a base address held
1263 in a register or if the base address register is a frame or stack
1264 pointer. Additionally, return false if the base address of the
1265 loaded constant is a SYMBOL_REF but is not considered to be a
1267 (replace_swapped_load_constant): New function.
1268 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
1269 loaded constant vector with a load of a swapped constant vector.
1271 2017-09-27 Carl Love <cel@us.ibm.com>
1273 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
1274 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
1275 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
1278 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
1280 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
1281 first, always call autopref_rank_data otherwise.
1283 2017-09-27 Richard Biener <rguenther@suse.de>
1285 * graphite-scop-detection.c (find_scop_parameters): Move
1286 loop bound handling ...
1287 (gather_bbs::before_dom_children): ... here, avoiding the need
1288 to build scop_info->loop_nest.
1289 (record_loop_in_sese): Remove.
1290 * sese.h (sese_info_t::loop_nest): Remove.
1291 * sese.c (new_sese_info): Do not allocate loop_nest.
1292 (free_sese_info): Do not free loop_nest.
1294 2017-09-27 Jakub Jelinek <jakub@redhat.com>
1297 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
1298 lhs from calls if the lhs has addressable type.
1300 2017-09-27 Richard Biener <rguenther@suse.de>
1302 * graphite.h (scop::max_alias_set): New member.
1303 * graphite-scop-detection.c: Remove references to non-existing
1304 --param in comments.
1305 (build_alias_sets): Record the maximum alias set used for drs.
1306 (build_scops): Support zero as unlimited for
1307 --param graphite-max-arrays-per-scop.
1308 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
1310 (build_poly_sr_1): ... here. Compute alias set based on the
1311 maximum alias set used for drs rather than
1312 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
1314 2017-09-27 Richard Biener <rguenther@suse.de>
1316 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
1317 --param loop-block-tile-size=0 to disable tiling.
1319 2017-09-27 Richard Biener <rguenther@suse.de>
1321 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
1322 (graphite-max-nb-scop-params): Document special value zero.
1323 * domwalk.h (dom_walker::STOP): New symbolical constant.
1324 (dom_walker::dom_walker): Add optional parameter for bb to
1326 (dom_walker::~dom_walker): Declare.
1327 (dom_walker::before_dom_children): Document STOP return value.
1328 (dom_walker::m_user_bb_to_rpo): New member.
1329 (dom_walker::m_bb_to_rpo): Likewise.
1330 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
1331 mapping here if not provided by the user.
1332 (dom_walker::~dom_walker): Free bb to RPO mapping if not
1333 provided by the user.
1334 (dom_walker::STOP): Define.
1335 (dom_walker::walk): Do not compute bb to RPO mapping here.
1336 Support STOP return value from before_dom_children to stop
1338 * graphite-optimize-isl.c (optimize_isl): If the schedule
1339 is the same still generate code if -fgraphite-identity
1340 or -floop-parallelize-all are given.
1341 * graphite-scop-detection.c: Include cfganal.h.
1342 (gather_bbs::gather_bbs): Get and pass through bb to RPO
1344 (gather_bbs::before_dom_children): Return STOP for BBs
1346 (build_scops): Compute bb to RPO mapping and pass it to
1347 the domwalk. Treat --param graphite-max-nb-scop-params=0
1348 as not limiting the number of params.
1349 * graphite.c (graphite_initialize): Remove limit on the
1350 number of basic-blocks in a function.
1351 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
1352 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
1353 default value of 10.
1355 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
1357 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
1358 Adjust code to eliminate needing to do the shift right 32-bits
1359 operation after XSCVDPSPN.
1361 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1363 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
1364 ((X / Y) != 0 -> X >= Y): Likewise.
1366 2017-09-26 Carl Love <cel@us.ibm.com>
1368 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
1369 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
1370 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
1371 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
1372 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
1373 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
1374 definitions and overloading.
1375 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
1376 statement for P9V_BUILTIN_XST_LEN_R.
1377 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
1378 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
1379 define_expand and define_insn for the instructions and builtins.
1380 * doc/extend.texi: Update the built-in documentation file for the new
1382 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
1383 define_insn for the instructions
1385 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
1388 * gcc/config/netbsd-protos.h: New file.
1389 * gcc/config/netbsd.c: New file.
1390 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
1391 * gcc/config/t-netbsd: New file.
1392 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
1393 (tmake_file) Add t-netbsd.
1394 (extra_objs) Add netbsd.o.
1396 2017-09-26 Janus Weil <janus@gcc.gnu.org>
1400 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
1402 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
1404 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
1405 sign extension from a vector register to a GPR by doing a 32-bit
1406 direct move and then an EXTSW.
1407 (extendsi<mode>2 splitter): Likewise.
1408 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
1409 right or vector extract after doing XSCVDPSPN. Use
1410 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
1412 (movdi_from_sf_zero_ext): Likewise.
1413 (reload_gpr_from_vsxsf): Likewise.
1414 (p8_mfvsrd_4_disf): Delete, no longer used.
1415 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
1416 then needing to move the SFmode to a GPR to use the XSCVDPSP
1417 instruction instead of FRSP and XSCVDPSPN.
1418 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
1419 it is adjacent to the other XSCVSPDP insns.
1420 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
1421 SFmode to be in traditional Altivec registers.
1422 (vsx_xscvdpspn): Eliminate useless alternative constraint.
1423 (vsx_xscvspdpn): Likewise.
1424 (vsx_xscvspdpn_scalar): Likewise.
1426 2017-09-26 Martin Jambor <mjambor@suse.cz>
1428 * tree-sra.c (compare_access_positions): Put integral types first,
1429 stabilize sorting of integral types, remove conditions putting
1430 non-full-precision integers last.
1431 (sort_and_splice_var_accesses): Disable scalarization if a
1432 non-integert would be represented by a non-full-precision integer.
1434 2017-09-26 Joseph Myers <joseph@codesourcery.com>
1436 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
1437 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
1438 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
1439 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
1440 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
1441 conditionals inside the function instead of around it. Call
1442 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
1443 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
1445 2017-09-26 Richard Biener <rguenther@suse.de>
1447 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
1449 (scop_detection::build_scop_breadth): ... this. Removed.
1450 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
1451 (scop_detection::harmful_stmt_in_bb): Likewise.
1452 (scop_detection::graphite_can_represent_stmt): Likewise.
1453 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
1454 (scop_detection::can_represent_loop): Remove recursion, fold in ...
1455 (scop_detection::can_represent_loop_1): ... this. Removed.
1456 (scop_detection::harmful_loop_in_region): Simplify after inlining
1457 the above and remove more quadraticness.
1458 (build_scops): Adjust.
1459 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
1462 2017-09-26 Jakub Jelinek <jakub@redhat.com>
1465 * config/i386/i386.c (ix86_print_operand_address_as): Only test
1466 REGNO (base) == SP_REG if base is a REG.
1469 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
1470 if it is different SSA_NAME.
1471 (optimize_range_tests_cmp_bitwise): New function.
1472 (optimize_range_tests): Call it.
1474 2017-09-26 Richard Biener <rguenther@suse.de>
1476 PR tree-optimization/82321
1477 * graphite.c (canonicalize_loop_closed_ssa): Properly check
1478 for the def being inside the loop.
1480 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1482 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
1484 * config/s390/s390-builtins.def: Fix constraint on op4.
1486 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1488 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
1489 independent expanders.
1490 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
1491 ("vec_ordered", "vec_unordered"): New expanders.
1493 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1495 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
1498 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1500 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
1501 vec_unpacks_lo_v16qi.
1502 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
1504 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1506 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
1507 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
1508 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
1510 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1512 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
1514 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
1516 ("*vec_slb<mode>"): New insn pattern.
1517 ("vec_shr_<mode>"): New expander.
1518 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
1519 and force the shift count operand to V16QImode.
1520 ("vec_srb<mode>"): Set shift count mode to V16QI.
1522 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1524 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
1525 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
1526 ("vec_widen_smult_hi_<mode>"): New expander definitions.
1528 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
1531 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
1533 2017-09-26 Richard Biener <rguenther@suse.de>
1535 PR tree-optimization/82320
1536 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
1539 2017-09-25 Jeff Law <law@redhat.com>
1541 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
1542 prototype for new argument.
1543 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
1544 mostly extracted from rs6000_emit_allocate_stack.
1545 (rs6000_emit_probe_stack_range_stack_clash): New function.
1546 (rs6000_emit_allocate_stack): Call
1547 rs6000_emit_probe_stack_range_stack_clash as needed.
1548 (rs6000_emit_probe_stack_range): Add additional argument
1549 to call to gen_probe_stack_range{si,di}.
1550 (output_probe_stack_range): New.
1551 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
1552 (output_probe_stack_range_stack_clash): New.
1553 (rs6000_emit_prologue): Emit notes into dump file as requested.
1554 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
1555 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
1556 Add additional operand and pass it to output_probe_stack_range.
1558 2017-09-25 Bin Cheng <bin.cheng@arm.com>
1560 PR tree-optimization/82163
1561 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
1562 (checking_verify_loop_closed_ssa): New parameter.
1563 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
1564 (check_loop_closed_ssa_stmt): Delete.
1565 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
1566 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
1567 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
1570 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
1572 * brig-builtins.def: Treat HSAIL barrier builtins as
1573 setjmp/longjump style functions.
1575 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
1577 * target.def (constant_alignment): New hook.
1578 * defaults.h (CONSTANT_ALIGNMENT): Delete.
1579 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
1580 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
1581 * doc/tm.texi: Regenerate.
1582 * targhooks.h (default_constant_alignment): Declare.
1583 (constant_alignment_word_strings): Likewise.
1584 * targhooks.c (default_constant_alignment): New function.
1585 (constant_alignment_word_strings): Likewise.
1586 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
1587 instead of CONSTANT_ALIGNMENT.
1588 * varasm.c (align_variable, get_variable_align, build_constant_desc)
1589 (force_const_mem): Likewise.
1590 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
1591 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
1592 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
1593 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1594 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
1596 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
1597 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1598 constant_alignment_word_strings.
1599 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
1600 (CONSTANT_ALIGNMENT): Likewise.
1601 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1602 (arm_constant_alignment): New function.
1603 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
1604 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1605 constant_alignment_word_strings.
1606 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
1607 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1608 constant_alignment_word_strings.
1609 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
1610 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1611 (cris_constant_alignment): New function.
1612 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
1613 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1614 (epiphany_constant_alignment): New function.
1615 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
1616 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1617 constant_alignment_word_strings.
1618 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
1619 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1620 constant_alignment_word_strings.
1621 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
1622 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1623 constant_alignment_word_strings.
1624 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
1625 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
1626 * config/i386/i386.c (ix86_constant_alignment): Make static.
1627 Use the same interface as the target hook.
1628 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1629 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
1630 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1631 constant_alignment_word_strings.
1632 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
1633 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
1634 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1635 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
1636 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1637 constant_alignment_word_strings.
1638 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
1639 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1640 constant_alignment_word_strings.
1641 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
1642 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1643 constant_alignment_word_strings.
1644 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
1645 * config/microblaze/microblaze.c (microblaze_constant_alignment):
1647 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1648 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
1649 * config/mips/mips.c (mips_constant_alignment): New function.
1650 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1651 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
1652 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
1653 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1654 (mmix_constant_alignment): Make static. Use the same interface
1656 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
1657 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1658 constant_alignment_word_strings.
1659 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
1660 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1661 constant_alignment_word_strings.
1662 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
1663 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1664 constant_alignment_word_strings.
1665 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
1666 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1667 (rs6000_constant_alignment): New function.
1668 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
1669 * config/riscv/riscv.c (riscv_constant_alignment): New function.
1670 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1671 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
1672 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1673 (rs6000_constant_alignment): New function.
1674 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
1675 * config/s390/s390.c (s390_constant_alignment): New function.
1676 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1677 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
1678 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1679 constant_alignment_word_strings.
1680 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
1681 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1682 (sparc_constant_alignment): New function.
1683 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
1684 * config/spu/spu.c (spu_constant_alignment): New function.
1685 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1686 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
1687 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1688 constant_alignment_word_strings.
1689 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
1690 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1691 constant_alignment_word_strings.
1692 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
1693 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1694 constant_alignment_word_strings.
1695 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
1696 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1697 (visium_constant_alignment): New function.
1698 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
1699 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1700 (xtensa_constant_alignment): New function.
1701 * system.h (CONSTANT_ALIGNMENT): Poison.
1703 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
1705 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
1706 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
1707 (rs6000_builtin_valid_without_lhs): New helper function.
1708 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1709 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
1711 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
1713 * target.h (vec_perm_indices): Use unsigned short rather than
1715 (auto_vec_perm_indices): Likewise.
1716 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
1717 Use unsigned int rather than unsigned char.
1718 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
1720 2017-09-25 Richard Biener <rguenther@suse.de>
1722 * cfgloop.h (sort_sibling_loops): Declare.
1723 * cfgloop.c (sort_sibling_loops_cmp): New helper.
1724 (sort_sibling_loops): New function sorting the sibling loop list
1726 * graphite.c (graphite_transform_loops): Sort sibling loops.
1728 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
1730 * target.def (vec_perm_const_ok): Change sel parameter to
1732 * optabs-query.c (can_vec_perm_p): Update accordingly.
1733 * doc/tm.texi: Regenerate.
1734 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
1735 auto_vec_perm_indices and remove separate nelt field.
1736 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
1737 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
1738 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
1739 (aarch64_expand_vec_perm_const): Update accordingly.
1740 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
1741 to vec_perm_indices.
1742 * config/arm/arm.c (expand_vec_perm_d): Change perm to
1743 auto_vec_perm_indices and remove separate nelt field.
1744 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
1745 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
1746 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
1748 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
1749 to vec_perm_indices.
1750 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
1751 sel to vec_perm_indices.
1752 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
1753 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
1754 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
1756 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
1759 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
1762 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
1763 on the FUNCTION_DECL function context if it has a DIE that is a
1766 2017-09-25 Richard Biener <rguenther@suse.de>
1768 PR tree-optimization/82285
1769 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
1772 2017-09-25 Tom de Vries <tom@codesourcery.com>
1776 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
1779 2017-09-25 Richard Biener <rguenther@suse.de>
1781 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
1782 ISL errors other than isl_error_quota happen. Dump if the
1783 schedule is the same.
1784 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
1785 errors instead of aborting inside ISL.
1787 2017-09-25 Iain Sandoe <iain@codesourcery.com>
1790 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
1791 of libgcc_eh for m64.
1792 * config/i386/darwin64.h: Likewise.
1794 2017-09-25 Richard Biener <rguenther@suse.de>
1797 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
1798 attribute for incomplete types nor twice for complete ones.
1800 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
1803 * config/i386/i386.c (ix86_print_operand_address_as): Encode
1804 %esp as %rsp to avoid 0x67 prefix if there is no index or base
1807 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
1810 * config/i386/i386.opt (mprefer-avx256): Use
1811 ix86_target_flags variable.
1812 * config/i386/i386.c (ix86_target_string): Move
1813 -mprefer-avx256 to flag2_opts.
1815 2017-09-22 Jakub Jelinek <jakub@redhat.com>
1818 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
1819 and x != -1 | y != -1 into (x & y) != -1.
1821 2017-09-22 Steve Ellcey <sellcey@cavium.com>
1823 * config.gcc: Add new case statement to set
1824 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
1825 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
1826 s390x-*-linux* case statements. Added aarch64 to the list of
1827 supported architectures.
1829 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1831 PR tree-optimization/82289
1832 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
1833 STMT_VINFO_RELEVANT_P.
1835 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1836 Alan Hayward <alan.hayward@arm.com>
1837 David Sherwood <david.sherwood@arm.com>
1839 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
1840 for VR_RANGE only; don't allow VR_ANTI_RANGE.
1841 (extract_range_from_binary_expr_1): Don't call
1842 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
1844 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1845 Alan Hayward <alan.hayward@arm.com>
1846 David Sherwood <david.sherwood@arm.com>
1848 * target.def (preferred_vector_alignment): New hook.
1849 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
1851 * doc/tm.texi: Regenerate.
1852 * targhooks.h (default_preferred_vector_alignment): Declare.
1853 * targhooks.c (default_preferred_vector_alignment): New function.
1854 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
1856 (DR_TARGET_ALIGNMENT): New macro.
1857 (aligned_access_p): Update commentary.
1858 (vect_known_alignment_in_bytes): New function.
1859 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
1861 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
1862 Calculate the misalignment based on the target alignment rather than
1864 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
1865 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
1866 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
1867 the target alignment, rather than masking the element misalignment
1868 with the number of elements in a vector. Also use the target
1869 alignment when calculating the maximum number of peels.
1870 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
1871 instead of TYPE_ALIGN_UNIT.
1872 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
1873 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
1874 (vect_create_addr_base_for_vector_ref): Update call accordingly.
1875 (vect_create_data_ref_ptr): Likewise.
1876 (vect_setup_realignment): Realign by ANDing with
1877 -DR_TARGET_MISALIGNMENT.
1878 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
1879 the number of peels based on DR_TARGET_ALIGNMENT.
1880 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
1881 with the guaranteed alignment boundary when deciding whether
1883 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
1884 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
1885 (ensure_base_align): Remove stmt_info parameter. Get the
1886 target base alignment from DR_TARGET_ALIGNMENT.
1887 (vectorizable_store): Update call accordingly. Interpret
1888 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
1890 (vectorizable_load): Likewise.
1892 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1893 Alan Hayward <alan.hayward@arm.com>
1894 David Sherwood <david.sherwood@arm.com>
1896 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
1897 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
1898 (vect_enhance_data_refs_alignment): Likewise.
1900 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
1902 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
1903 error. Only quit immediately if parsing is complete.
1904 (BEGIN): Initialize fatal_err and parse_done.
1905 (begin fpu, end fpu): Check number of arguments.
1906 (begin arch, end arch): Likewise.
1907 (begin cpu, end cpu): Likewise.
1908 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
1909 (optalias): Likewise.
1911 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
1913 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
1914 * config/arm/arm-isa.h: Delete. Move definitions to ...
1915 * arm-cpus.in: ... here. Use new feature and fgroup values.
1916 * config/arm/arm.c (arm_option_override): Use lower case for feature
1918 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
1919 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
1920 * config/arm/parsecpu.awk (END): Add new command 'isa'.
1922 (print_isa_bits_for): New function.
1923 (gen_isa): New function.
1924 (gen_comm_data): Use print_isa_bits_for.
1925 (define feature): New keyword.
1926 (define fgroup): New keyword.
1927 * config/arm/t-arm (TM_H): Remove.
1928 (GTM_H): Add arm-isa.h.
1929 (arm-isa.h): Add rule to generate file.
1930 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
1931 case for feature bit names.
1933 2017-09-22 Richard Biener <rguenther@suse.de>
1935 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
1937 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
1938 print of no dependency loops ...
1939 * graphite.c (graphite_transform_loops): ... here.
1940 (canonicalize_loop_closed_ssa_form): Work from inner to outer
1942 (same_close_phi_node, remove_duplicate_close_phi,
1943 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
1944 (canonicalize_loop_closed_ssa): ... here and simplify.
1945 * graphite-optimize-isl.c: Include tree-vectorizer.h.
1946 (optimize_isl): Use dump_printf_loc to tell when we stopped
1947 optimizing because of an ISL timeout.
1949 2017-09-22 Richard Biener <rguenther@suse.de>
1951 PR tree-optimization/82291
1952 * tree-if-conv.c (predicate_mem_writes): Make sure to
1953 remove writes in blocks predicated with false.
1955 2017-09-22 Richard Biener <rguenther@suse.de>
1957 * sese.c: Include cfganal.h.
1958 (if_region_set_false_region): Remove.
1959 (create_if_region_on_edge): Likewise.
1960 (move_sese_in_condition): Re-implement without destroying
1963 2017-09-22 Richard Biener <rguenther@suse.de>
1965 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
1966 Verify both BBs contain loop PHI nodes before dispatching to
1968 (graphite_regenerate_ast_isl): Do not recompute dominators,
1969 do not verify three times. Restructure for clarity.
1970 * graphite-scop-detection.c (same_close_phi_node,
1971 remove_duplicate_close_phi, make_close_phi_nodes_unique,
1972 defined_in_loop_p, canonicalize_loop_closed_ssa,
1973 canonicalize_loop_closed_ssa_form): Simplify, remove excess
1974 checking and SSA rewrite, move to ...
1975 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
1976 (graphite_initialize): Do not pass in ctx, do not reset the
1977 SCEV cache, compute only dominators.
1978 (graphite_transform_loops): Allocate ISL ctx after
1979 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
1980 Maintain post-dominators only around build_scops.
1981 * sese.c (if_region_set_false_region): Make static. Free
1982 and recompute dominators.
1983 (move_sese_in_condition): Assert we don't get called with
1984 post-dominators computed.
1985 * sese.h (if_region_set_false_region): Remove.
1987 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
1989 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
1990 mode attribute for TARGET_AVX512VL.
1992 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
1994 * config/i386/i386.opt (mprefer-avx256): New option.
1995 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
1997 (ix86_preferred_simd_mode): Return 256-bit AVX modes
1998 for TARGET_PREFER_AVX256.
1999 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
2001 2017-09-21 Jeff Law <law@redhat.com>
2003 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
2004 Fix dump output if the only stack space is for pushed registers.
2006 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2008 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
2011 2017-09-21 Martin Sebor <msebor@redhat.com>
2014 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
2015 code (in C++) or code that triggers warnings.
2017 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
2019 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
2021 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
2023 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
2024 * sched-rgn.c: Ditto.
2025 * sel-sched-ir.c: Ditto.
2027 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
2029 * toplev.h (set_random_seed): Adjust return type.
2030 * toplev.c (init_local_tick): Move eager initialization of random_seed
2031 to get_random_seed. Adjust comment.
2032 (init_random_seed): Inline to get_random_seed, delete.
2033 (get_random_seed): Initialize random_seed lazily.
2034 (set_random_seed): Do not return previous value.
2035 (print_switch_value): Do not call get_random_seed.
2037 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
2039 * cgraph.c (delete_function_version): New, broken out from...
2040 (cgraph_node::delete_function_version): ...here. Rename to
2041 cgraph_node::delete_function_version_by_decl. Update all uses.
2042 (cgraph_node::remove): Call delete_function_version.
2044 2017-09-21 Jakub Jelinek <jakub@redhat.com>
2047 * tree-inline.c (expand_call_inline): Emit clobber stmts for
2048 VAR_DECLs to which addressable non-volatile parameters are mapped
2049 and for id->retvar after the return value assignment. Clear
2050 id->retval and id->retbnd after inlining.
2052 2017-09-21 Richard Biener <rguenther@suse.de>
2054 PR tree-optimization/82276
2055 PR tree-optimization/82244
2056 * tree-vrp.c (build_assert_expr_for): Set
2057 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
2059 (remove_range_assertions): Revert earlier change.
2061 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
2064 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
2066 2017-09-21 Richard Biener <rguenther@suse.de>
2068 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
2069 Restore valid IL after code generation errors.
2070 * graphite.c (graphite_transform_loops): Diagnose code
2071 generation issues as MSG_MISSED_OPTIMIZATION and continue
2072 with processing SCOPs.
2074 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2075 Alan Hayward <alan.hayward@arm.com>
2076 David Sherwood <david.sherwood@arm.com>
2078 * calls.c (compute_argument_addresses): Use simplify_gen_binary
2079 rather than choosing between plus_constant and gen_rtx_<CODE>.
2080 * expr.c (emit_push_insn): Likewise.
2081 (expand_expr_real_2): Likewise.
2083 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2084 Alan Hayward <alan.hayward@arm.com>
2085 David Sherwood <david.sherwood@arm.com>
2087 * loop-unroll.c (split_iv): Call copy_rtx on the step.
2089 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2090 Alan Hayward <alan.hayward@arm.com>
2091 David Sherwood <david.sherwood@arm.com>
2093 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
2094 calling tree_to_uhwi.
2096 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2097 Alan Hayward <alan.hayward@arm.com>
2098 David Sherwood <david.sherwood@arm.com>
2100 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
2101 INTEGER_CST rather than a negative test for ADDR_EXPR.
2103 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2104 Alan Hayward <alan.hayward@arm.com>
2105 David Sherwood <david.sherwood@arm.com>
2107 * tree-vrp.c (extract_range_from_binary_expr_1): Check
2108 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
2110 2017-09-21 Richard Biener <rguenther@suse.de>
2112 PR tree-optimization/71351
2113 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
2114 graphite_create_new_loop_guard): Remove, fold remaining parts
2116 (translate_isl_ast_node_for): ... here and simplify.
2118 2017-09-21 Jakub Jelinek <jakub@redhat.com>
2121 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
2122 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
2123 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
2124 alternative always use QI mode, for -Os imov (=R,R) alternative
2125 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
2128 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2129 Jeff Law <law@redhat.com>
2131 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
2132 (allocate_stack_space): New function, partially extracted from
2134 (s390_emit_prologue): Track offset to most recent stack probe.
2135 Code to allocate space moved into allocate_stack_space.
2136 Dump actions when no stack is allocated.
2137 (s390_prologue_plus_offset): New function.
2138 (s390_emit_stack_probe): Likewise.
2140 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
2142 * common.opt (Wa, Wl, Wp, g, gz=): Add
2144 (gno-column-info): Remove.
2145 (gcolumn-info): Drop RejectNegative.
2147 (gno-record-gcc-switches): Remove.
2148 (grecord-gcc-switches): Drop RejectNegative.
2149 (gno-split-dwarf): Remove.
2150 (gsplit-dwarf): Drop RejectNegative.
2151 (gno-strict-dwarf): Remove.
2152 (gstrict-dwarf): Drop RejectNegative.
2153 * config/darwin.opt (gfull, gused): Add RejectNegative.
2154 * dwarf2out.c (gen_producer_string): Drop
2155 gno-record-gcc-switches handler.
2156 * optc-gen.awk: Add g to prefixes with negative forms.
2157 * opts-common.c (remapping_prefix_p): New.
2158 (find_opt): Check it.
2159 (generate_canonical_option): Test g prefix.
2160 (option_map): Add -gno- mapping.
2161 (add_misspelling_candidates): Check remapping_prefix_p.
2163 2017-09-20 Jeff Law <law@redhat.com>
2165 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
2166 thinko in stack clash protection support.
2168 * explow.c (compute_stack_clash_protection_loop_data): Use
2169 CONST_INT_P instead of explicit test. Verify object is a
2170 CONST_INT_P before looking at INTVAL.
2171 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
2172 instead of explicit test.
2174 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
2177 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
2178 address instead of to r1 and r11.
2180 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
2182 * config.gcc: Support "knm".
2183 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
2184 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2186 * config/i386/i386.c (m_KNM): Define.
2187 (processor_target_table): Add "knm".
2189 (ix86_option_override_internal): Add "knm".
2190 (ix86_issue_rate): Add PROCESSOR_KNM.
2191 (ix86_adjust_cost): Ditto.
2192 (ia32_multipass_dfa_lookahead): Ditto.
2193 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
2194 (fold_builtin_cpu): Add M_INTEL_KNM.
2195 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
2196 (processor_type): Add PROCESSOR_KNM.
2197 * config/i386/x86-tune.def: Add m_KNM.
2198 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
2200 2017-09-20 Richard Biener <rguenther@suse.de>
2202 PR tree-optimization/80213
2203 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
2204 are allowed in empty BBs as well.
2205 (canonicalize_loop_closed_ssa): Also look for other complex
2207 (scop_detection::get_sese): Include the loop-closed PHI block
2209 (scop_detection::merge_sese): Remove code adding extra blocks.
2210 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
2211 (build_scops): Assert the final returned scop is invalid.
2213 2017-09-20 Richard Biener <rguenther@suse.de>
2215 PR tree-optimization/82264
2216 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
2218 (vn_phi_lookup): Likewise.
2219 (vn_phi_insert): Likewise.
2221 2017-09-20 Jakub Jelinek <jakub@redhat.com>
2223 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
2224 that fits into uhwi or shwi, add DW_AT_const_value regardless
2225 of early_dwarf without going through RTL, using add_AT_unsigned
2228 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
2229 (DEBUG_LTO_ABBREV_SECTION): Likewise.
2230 (DEBUG_LTO_MACINFO_SECTION): Likewise.
2231 (DEBUG_MACRO_SECTION): Likewise.
2232 (DEBUG_LTO_MACRO_SECTION): Likewise.
2233 (DEBUG_STR_DWO_SECTION): Likewise.
2234 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
2235 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
2236 (DEBUG_LTO_DWO_LINE_SECTION): Define.
2237 (DEBUG_LTO_LINE_STR_SECTION): Define.
2238 (init_sections_and_labels): Initialize debug_line_str_section
2239 variable. Initialize debug_loc_section for -gdwarf-5 to
2240 DEBUG_LOCLISTS_SECTION. Formatting fixes.
2242 2017-09-20 Richard Biener <rguenther@suse.de>
2244 * graphite-sese-to-poly.c (extract_affine): Properly handle
2245 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
2247 2017-09-20 Richard Biener <rguenther@suse.de>
2249 PR tree-optimization/81373
2250 * graphite-scop-detection.c (build_cross_bb_scalars_def):
2251 Force SESE live-out defs to be handled even if they are
2254 2017-09-19 Jeff Law <law@redhat.com>
2256 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
2257 nothing for stack adjustments with REG_STACK_CHECK.
2258 * sched-deps.c (parse_add_or_inc): Reject insns with
2259 REG_STACK_CHECK from dependency breaking.
2260 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
2261 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
2262 * reg-notes.def (STACK_CHECK): New note.
2264 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
2265 (ix86_expand_prologue): Dump stack clash info as needed.
2266 Call ix86_adjust_stack_and_probe_stack_clash as needed.
2268 * function.c (dump_stack_clash_frame_info): New function.
2269 * function.h (dump_stack_clash_frame_info): Prototype.
2270 (enum stack_clash_probes): New enum.
2272 * config/alpha/alpha.c (alpha_expand_prologue): Also check
2273 flag_stack_clash_protection.
2274 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
2275 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
2276 (arm_frame_pointer_required): Likewise.
2277 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
2278 (ia64_expand_prologue): Likewise.
2279 * config/mips/mips.c (mips_expand_prologue): Likewise.
2280 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
2281 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
2282 (sparc_flat_expand_prologue): Likewise.
2283 * config/spu/spu.c (spu_expand_prologue): Likewise.
2285 * explow.c: Include "params.h".
2286 (anti_adjust_stack_and_probe_stack_clash): New function.
2287 (get_stack_check_protect): Likewise.
2288 (compute_stack_clash_protection_loop_data): Likewise.
2289 (emit_stack_clash_protection_loop_start): Likewise.
2290 (emit_stack_clash_protection_loop_end): Likewise.
2291 (allocate_dynamic_stack_space): Use get_stack_check_protect.
2292 Use anti_adjust_stack_and_probe_stack_clash.
2293 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
2294 (emit_stack_clash_protection_loop_start): Likewise.
2295 (emit_stack_clash_protection_loop_end): Likewise.
2296 * rtl.h (get_stack_check_protect): Prototype.
2297 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
2298 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
2299 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
2301 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
2303 * doc/tm.texi: Rebuilt.
2304 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
2305 get_stack_check_protect.
2306 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
2307 * config/arm/arm.c (arm_expand_prologue): Likewise.
2308 (arm_frame_pointer_required): Likewise.
2309 * config/i386/i386.c (ix86_expand_prologue): Likewise.
2310 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
2311 * config/mips/mips.c (mips_expand_prologue): Likewise.
2312 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
2313 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
2314 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
2315 (sparc_flat_expand_prologue): Likewise.
2317 * common.opt (-fstack-clash-protection): New option.
2318 * flag-types.h (enum stack_check_type): Note difference between
2319 -fstack-check= and -fstack-clash-protection.
2320 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
2321 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
2322 * toplev.c (process_options): Issue warnings/errors for cases
2323 not handled with -fstack-clash-protection.
2324 * doc/invoke.texi (-fstack-clash-protection): Document new option.
2325 (-fstack-check): Note additional problem with -fstack-check=generic.
2326 Note that -fstack-check is primarily for Ada and refer users
2327 to -fstack-clash-protection for stack-clash-protection.
2328 Document new params for stack clash protection.
2330 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
2332 * config/i386/i386.c (ix86_split_long_move): Do not handle
2333 address used for LEA in a special way.
2335 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
2337 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
2339 2017-09-19 Martin Sebor <msebor@redhat.com>
2342 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
2343 of incompatible types.
2345 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
2347 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
2348 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
2349 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2350 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
2352 2017-09-19 Richard Biener <rguenther@suse.de>
2354 PR tree-optimization/82244
2355 * tree-vrp.c (remove_range_assertions): Do not propagate
2356 a constant to abnormals but replace the assert with a copy.
2358 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
2360 PR rtl-optimization/57878
2361 PR rtl-optimization/68988
2362 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
2363 avoidance test involving non_reload_pseudos. Move frequency test
2364 below the general fragmentation avoidance test.
2366 2017-09-19 Richard Biener <rguenther@suse.de>
2368 PR tree-optimization/69728
2369 * graphite-sese-to-poly.c (schedule_error): New global.
2370 (add_loop_schedule): Handle empty domain by failing the
2372 (build_original_schedule): Handle schedule_error.
2374 2017-09-19 Richard Biener <rguenther@suse.de>
2376 * graphite-scop-detection.c (scop_detection::can_represent_loop):
2377 Do not iterate to sibling loops but only to siblings of inner
2380 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
2383 * config/m68k/m68k.md (moveq feeding equality comparison): Check
2384 that the registers are different.
2386 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
2388 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
2389 to processor_model and "amdfam17h" to arch_names_table.
2390 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
2392 2017-09-18 Jakub Jelinek <jakub@redhat.com>
2395 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
2397 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2398 Alan Hayward <alan.hayward@arm.com>
2399 David Sherwood <david.sherwood@arm.com>
2401 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
2403 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
2405 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
2406 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
2407 vect_schedule_slp_instance.
2408 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
2409 Update call to vect_slp_analyze_node_operations. Simplify return
2411 (vect_slp_analyze_bb_1): Update call accordingly.
2412 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
2413 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
2414 (vect_schedule_slp): Update call accordingly.
2416 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2417 Alan Hayward <alan.hayward@arm.com>
2418 David Sherwood <david.sherwood@arm.com>
2420 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
2421 with types that aren't in fact scalar.
2423 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2425 * tree-vect-slp.c (vect_record_max_nunits): New function,
2427 (vect_build_slp_tree_1): ...here.
2428 (vect_build_slp_tree_2): Call it for phis too.
2430 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2432 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
2433 to vect_get_vec_def_for_operand when getting the mask operand.
2435 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2436 Alan Hayward <alan.hayward@arm.com>
2437 David Sherwood <david.sherwood@arm.com>
2439 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
2442 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2443 Alan Hayward <alan.hayward@arm.com>
2444 David Sherwood <david.sherwood@arm.com>
2446 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
2447 calculation for vector booleans.
2449 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2450 Alan Hayward <alan.hayward@arm.com>
2451 David Sherwood <david.sherwood@arm.com>
2453 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
2455 (vect_transform_stmt): ...here.
2456 (vect_analyze_stmt): Use it instead of calling
2457 vectorizable_live_operation directly.
2459 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
2461 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
2462 non-SIMT targets in acc vector loops.
2464 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
2466 * configure.ac: Add arc and check if assembler supports gdwarf2.
2467 * configure: Regenerate.
2469 2017-09-18 Richard Biener <rguenther@suse.de>
2471 PR tree-optimization/82220
2472 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
2473 epilogue niters from the min_profitable_iters compute.
2475 2017-09-18 Jakub Jelinek <jakub@redhat.com>
2478 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
2479 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
2480 (ix86_init_pic_reg): Revert 2017-09-01 changes.
2482 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
2485 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
2486 switching to a new text section.
2488 2017-09-18 Richard Biener <rguenther@suse.de>
2490 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
2492 (build_alias_set): Reject aliases with no access function.
2494 2017-09-18 Richard Biener <rguenther@suse.de>
2496 PR tree-optimization/79622
2497 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
2499 (build_cross_bb_scalars_use): Likewise.
2501 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
2503 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
2505 2017-09-18 Alan Modra <amodra@gmail.com>
2508 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
2509 stack_pointer_rtx for count 0. Update comments. Break up
2510 large rtl expression.
2512 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
2514 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
2515 Increase to 20 bytes.
2516 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
2517 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
2518 or avx version of the stub.
2520 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
2523 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
2524 compute the minimum stack alignment. Also update preferred stack
2525 boundary for leaf functions.
2527 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
2529 PR tree-optimization/82228
2530 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
2533 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
2535 * common/config/nds32/nds32-common.c
2536 (nds32_option_optimization_table): Refine formatting.
2537 (nds32_option_optimization_table): Use -fsched-pressure and
2538 -fomit-frame-pointer for specific optimization level.
2540 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
2542 * config/nds32/nds32.c: Refine formatting and comments.
2543 * config/nds32/nds32.h: Likewise.
2544 * config/nds32/nds32.md: Likewise.
2545 * config/nds32/nds32-cost.c: Likewise.
2546 * config/nds32/nds32-isr.c: Likewise.
2547 * config/nds32/nds32-md-auxiliary.c: Likewise.
2548 * config/nds32/nds32-multiple.md: Likewise.
2549 * config/nds32/nds32-predicates.c: Likewise.
2551 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
2552 Jakub Jelinek <jakub@redhat.com>
2554 Add support for -std=c++2a.
2555 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
2557 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
2559 2017-09-15 Steve Ellcey <sellcey@cavium.com>
2562 * doc/extend.texi (Common Function Attributes): Add
2563 references to ARM, AArch64, and S/390 specific attributes.
2564 (Function Specific Option Pragmas): Add AArch64 and S/390
2565 to list of back ends that support the target pragma.
2567 2017-09-15 Nathan Sidwell <nathan@acm.org>
2569 * doc/standards.texi: Fix C++17 description. Update URLs for
2572 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
2574 * common.opt (Wcast-align=strict): New warning option.
2575 * doc/invoke.texi: Document -Wcast-align=strict.
2577 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
2579 * cgraph.h (cgraph_thunk_info): Add comments.
2580 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
2581 assert for VIRTUAL_* arguments stricter.
2583 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
2585 PR tree-optimization/71026
2586 * match.pd: Move RDIV patterns from fold-const.c
2587 * fold-const.c (distribute_real_division): Removed.
2588 (fold_binary_loc): Remove calls to distribute_real_divison.
2590 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2592 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
2593 c++1z and gnu++1z as deprecated. Change other references to
2594 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
2595 Change -Wc++1z-compat to -Wc++17-compat.
2596 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
2597 * dwarf2out.c (highest_c_language): Handle C++17.
2598 (gen_compile_unit_die): Likewise.
2600 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2602 PR rtl-optimization/82192
2603 * combine.c (make_extraction): Don't look through non-paradoxical
2604 SUBREGs or TRUNCATE if pos + len is or might be bigger than
2607 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2608 Alan Hayward <alan.hayward@arm.com>
2609 David Sherwood <david.sherwood@arm.com>
2611 * target.def (function_arg_offset): New hook.
2612 * targhooks.h (default_function_arg_offset): Declare.
2613 * targhooks.c (default_function_arg_offset): New function.
2614 * function.c (locate_and_pad_parm): Use
2615 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
2616 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
2617 (TARGET_FUNCTION_ARG_OFFSET): ...this.
2618 * doc/tm.texi: Regenerate.
2619 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
2620 * config/spu/spu.c (spu_function_arg_offset): New function.
2621 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
2622 * system.h (FUNCTION_ARG_OFFSET): Poison.
2624 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2625 Alan Hayard <alan.hayward@arm.com>
2626 David Sherwood <david.sherwood@arm.com>
2628 * target.def (truly_noop_truncation): New hook.
2629 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
2630 than TRULY_NOOP_TRUNCATION.
2631 * hooks.h (hook_bool_uint_uint_true): Declare.
2632 * hooks.c (hook_bool_uint_uint_true): New function.
2633 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
2634 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
2635 * doc/tm.texi: Regenerate.
2636 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
2637 rather than TRULY_NOOP_TRUNCATION in comments.
2638 (simplify_comparison): Likewise.
2639 (record_truncated_value): Likewise.
2640 * expmed.c (extract_bit_field_1): Likewise.
2641 (extract_split_bit_field): Likewise.
2642 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
2643 instead of TRULY_NOOP_TRUNCATION.
2644 * function.c (assign_parm_setup_block): Likewise.
2645 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
2646 * rtlhooks.c: Include target.h.
2647 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
2648 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
2649 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
2650 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
2651 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
2652 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
2653 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
2654 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
2655 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
2656 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
2657 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
2658 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
2659 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
2660 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
2661 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
2662 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
2663 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
2664 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
2665 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
2666 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
2667 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
2668 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
2669 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
2670 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
2671 * config/mips/mips.c (mips_truly_noop_truncation): New function.
2672 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2673 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2674 rather than TRULY_NOOP_TRUNCATION in comments.
2675 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
2676 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
2677 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
2678 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
2679 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
2680 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
2681 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
2682 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
2683 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
2684 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
2685 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
2686 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2687 rather than TRULY_NOOP_TRUNCATION in comments.
2688 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
2689 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
2690 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
2691 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
2692 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
2693 TRULY_NOOP_TRUNCATION condition.
2694 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
2695 (TRULY_NOOP_TRUNCATION): Delete.
2696 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
2697 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
2698 * config/spu/spu.c (spu_truly_noop_truncation): New function.
2699 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2700 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
2701 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
2702 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
2703 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2704 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2705 rather than TRULY_NOOP_TRUNCATION in comments.
2706 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
2707 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
2708 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
2709 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
2710 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
2711 * system.h (TRULY_NOOP_TRUNCATION): Poison.
2713 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
2716 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
2717 (*cmp_ior): Likewise.
2718 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
2719 (*ior_scc_scc_cmp): Likewise.
2720 (*and_scc_scc): Likewise.
2721 (*and_scc_scc_cmp): Likewise.
2723 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2724 Alan Hayard <alan.hayward@arm.com>
2725 David Sherwood <david.sherwood@arm.com>
2727 * target.def (can_change_mode_class): New hook.
2728 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
2729 (hard_regno_nregs): Likewise.
2730 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
2731 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
2732 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
2733 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
2734 (LOAD_EXTEND_OP): Update accordingly.
2735 * doc/tm.texi: Regenerate.
2736 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
2737 CANNOT_CHANGE_MODE_CLASS.
2738 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
2739 (REG_CAN_CHANGE_MODE_P): ...this new macro.
2740 * combine.c (simplify_set): Update accordingly.
2741 * emit-rtl.c (validate_subreg): Likewise.
2742 * recog.c (general_operand): Likewise.
2743 * regcprop.c (mode_change_ok): Likewise.
2744 * reload1.c (choose_reload_regs): Likewise.
2745 (inherit_piecemeal_p): Likewise.
2746 * rtlanal.c (simplify_subreg_regno): Likewise.
2747 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
2748 instead of CANNOT_CHANGE_MODE_CLASS.
2749 (reload_cse_simplify_operands): Likewise.
2750 * reload.c (push_reload): Use targetm.can_change_mode_class
2751 instead of CANNOT_CHANGE_MODE_CLASS.
2752 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
2753 REG_CANNOT_CHANGE_MODE_P.
2754 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2755 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
2756 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2757 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2758 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2759 (arm_can_change_mode_class): New function.
2760 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
2761 than CANNOT_CHANGE_MODE_CLASS in comments.
2762 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2763 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
2764 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
2765 (ix86_can_change_mode_class): ...this new function, inverting the
2766 sense of the return value.
2767 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2768 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2769 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2770 (ia64_can_change_mode_class): New function.
2771 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2772 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
2773 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
2774 (m32c_can_change_mode_class): ...this new function, inverting the
2775 sense of the return value.
2776 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2777 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2778 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
2779 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
2780 (mips_can_change_mode_class): ...this new function, inverting the
2781 sense of the return value.
2782 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2783 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2784 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2785 (msp430_can_change_mode_class): New function.
2786 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2787 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
2788 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2789 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2790 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2791 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
2792 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2793 (pa_cannot_change_mode_class): Replace with...
2794 (pa_can_change_mode_class): ...this new function, inverting the
2795 sense of the return value.
2796 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
2797 than CANNOT_CHANGE_MODE_CLASS in comments.
2798 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2799 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
2800 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2801 (pdp11_cannot_change_mode_class): Replace with...
2802 (pdp11_can_change_mode_class): ...this new function, inverting the
2803 sense of the return value.
2804 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2805 * config/powerpcspe/powerpcspe-protos.h
2806 (rs6000_cannot_change_mode_class_ptr): Delete.
2807 * config/powerpcspe/powerpcspe.c
2808 (rs6000_cannot_change_mode_class_ptr): Delete.
2809 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2810 (rs6000_option_override_internal): Assign to
2811 targetm.can_change_mode_class instead of
2812 rs6000_cannot_change_mode_class_ptr.
2813 (rs6000_cannot_change_mode_class): Replace with...
2814 (rs6000_can_change_mode_class): ...this new function, inverting the
2815 sense of the return value.
2816 (rs6000_debug_cannot_change_mode_class): Replace with...
2817 (rs6000_debug_can_change_mode_class): ...this new function.
2818 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2819 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
2820 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2821 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2822 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
2824 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
2825 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2826 (rs6000_option_override_internal): Assign to
2827 targetm.can_change_mode_class instead of
2828 rs6000_cannot_change_mode_class_ptr.
2829 (rs6000_cannot_change_mode_class): Replace with...
2830 (rs6000_can_change_mode_class): ...this new function, inverting the
2831 sense of the return value.
2832 (rs6000_debug_cannot_change_mode_class): Replace with...
2833 (rs6000_debug_can_change_mode_class): ...this new function.
2834 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2835 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
2836 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
2837 (s390_can_change_mode_class): ...this new function, inverting the
2838 sense of the return value.
2839 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2840 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2841 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
2842 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2843 (sh_cannot_change_mode_class): Replace with...
2844 (sh_can_change_mode_class): ...this new function, inverting the
2845 sense of the return value.
2846 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2847 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2848 (sparc_can_change_mode_class): New function.
2849 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2850 * config/spu/spu.c (spu_can_change_mode_class): New function.
2851 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2852 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2853 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2854 (visium_can_change_mode_class): New function.
2855 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
2857 2017-09-15 Richard Biener <rguenther@suse.de>
2859 PR tree-optimization/82217
2860 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
2861 but not undefined case.
2863 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2866 * postreload.c (reload_cse_simplify_operands): Skip
2867 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
2869 2017-09-15 Richard Biener <rguenther@suse.de>
2871 PR tree-optimization/68823
2872 * graphite-scop-detection.c (build_alias_set): If we have a
2873 possible dependence check whether we can handle them by just
2874 looking at the DRs DR_ACCESS_FNs.
2875 (build_scops): If build_alias_set fails, fail the SCOP.
2877 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
2879 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
2880 to support float128 built-in functions that require the ISA 3.0
2882 (BU_FLOAT128_3_HW): Likewise.
2883 (SQRTF128): Add support for the IEEE 128-bit square root and fma
2885 (FMAF128): Likewise.
2887 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
2888 support for built-in functions that need the ISA 3.0 IEEE 128-bit
2889 floating point instructions.
2890 (rs6000_invalid_builtin): Likewise.
2891 (rs6000_builtin_mask_names): Likewise.
2892 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
2893 (RS6000_BTM_FLOAT128_HW): Likewise.
2894 (RS6000_BTM_COMMON): Likewise.
2895 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
2897 * doc/extend.texi (RS/6000 built-in functions): Document the
2898 IEEE 128-bit floating point square root and fused multiply-add
2901 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
2903 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
2904 reg (r2) isn't in the set of registers defined in the prologue.
2906 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2907 Alan Hayward <alan.hayward@arm.com>
2908 David Sherwood <david.sherwood@arm.com>
2910 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
2911 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
2912 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
2913 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
2914 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
2916 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
2917 max_vectorization_factor.
2918 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
2920 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2921 Alan Hayward <alan.hayward@arm.com>
2922 David Sherwood <david.sherwood@arm.com>
2924 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
2925 (vect_worthwhile_without_simd_p): Declare.
2926 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
2927 (vectorizable_reduction): Use it.
2928 * tree-vect-stmts.c (vectorizable_shift): Likewise.
2929 (vectorizable_operation): Likewise.
2931 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2932 Alan Hayward <alan.hayward@arm.com>
2933 David Sherwood <david.sherwood@arm.com>
2935 * tree-vectorizer.h (vect_get_num_copies): New function.
2936 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
2937 * tree-vect-loop.c (vectorizable_reduction): Likewise.
2938 (vectorizable_induction): Likewise.
2939 (vectorizable_live_operation): Likewise.
2940 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
2941 (vectorizable_bswap): Likewise.
2942 (vectorizable_call): Likewise.
2943 (vectorizable_conversion): Likewise.
2944 (vectorizable_assignment): Likewise.
2945 (vectorizable_shift): Likewise.
2946 (vectorizable_operation): Likewise.
2947 (vectorizable_store): Likewise.
2948 (vectorizable_load): Likewise.
2949 (vectorizable_condition): Likewise.
2950 (vectorizable_comparison): Likewise.
2951 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
2953 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2954 Alan Hayward <alan.hayward@arm.com>
2955 David Sherwood <david.sherwood@arm.com>
2957 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
2958 of vect_init_vector.
2960 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2961 Alan Hayward <alan.hayward@arm.com>
2962 David Sherwood <david.sherwood@arm.com>
2964 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
2965 an inline wrapper that provides a location.
2966 (gimple_build_vector): Likewise.
2967 * gimple-fold.c (gimple_build_vector_from_val): New function.
2968 (gimple_build_vector): Likewise.
2969 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
2970 functions to build the initial value. Always return a gimple value.
2971 (get_initial_defs_for_reduction): Likewise. Only compute
2973 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
2974 vect_init_vector on the results from get_initial_def(s)_for_reduction.
2975 (vectorizable_induction): Use gimple_build_vector rather than
2978 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2979 Alan Hayward <alan.hayward@arm.com>
2980 David Sherwood <david.sherwood@arm.com>
2982 * target.h (vec_perm_indices): New typedef.
2983 (auto_vec_perm_indices): Likewise.
2984 * optabs-query.h: Include target.h
2985 (can_vec_perm_p): Take a vec_perm_indices *.
2986 * optabs-query.c (can_vec_perm_p): Likewise.
2987 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
2988 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
2989 * tree-vect-generic.c (lower_vec_perm): Likewise.
2990 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
2991 (vect_grouped_load_supported): Likewise.
2992 (vect_shift_permute_load_chain): Likewise.
2993 (vect_permute_store_chain): Use auto_vec_perm_indices.
2994 (vect_permute_load_chain): Likewise.
2995 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
2996 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
2997 Update uses of can_vec_perm_p.
2998 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
2999 mode with a number of elements. Take a vec_perm_indices *.
3000 (vect_create_epilog_for_reduction): Update accordingly.
3001 Use auto_vec_perm_indices.
3002 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
3003 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
3004 (vect_transform_slp_perm_load): Likewise.
3005 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
3006 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
3007 (vect_gen_perm_mask_checked): Likewise.
3008 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
3009 (vect_gen_perm_mask_checked): Likewise.
3010 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
3011 (vectorizable_store): Likewise.
3012 (vectorizable_load): Likewise.
3013 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
3014 (vectorizable_bswap): Likewise.
3016 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3017 Alan Hayward <alan.hayward@arm.com>
3018 David Sherwood <david.sherwood@arm.com>
3020 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
3021 * tree.c (build_vector): Likewise.
3022 (build_vector_from_ctor): Update accordingly.
3023 (build_vector_from_val): Likewise.
3024 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
3025 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
3026 * tree-vect-generic.c (add_rshift): Likewise.
3027 (expand_vector_divmod): Likewise.
3028 (optimize_vector_constructor): Likewise.
3029 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
3030 (vect_transform_slp_perm_load): Likewise.
3031 (vect_schedule_slp_instance): Likewise.
3032 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
3033 (vectorizable_call): Likewise.
3034 (vect_gen_perm_mask_any): Likewise. Add elements in order.
3035 * expmed.c (make_tree): Likewise.
3036 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
3037 a vector passed to build_vector.
3038 (fold_convert_const): Likewise.
3039 (exact_inverse): Likewise.
3040 (fold_ternary_loc): Likewise.
3041 (fold_relational_const): Likewise.
3042 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
3043 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
3044 (const_unop): Likewise. Store the reduction accumulator in a
3045 variable rather than an array.
3046 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
3047 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
3048 the new vector, rather than constructing it after the input arrays.
3049 (native_interpret_vector): Use auto_vec<tree> when building
3050 a vector passed to build_vector. Add elements in order.
3051 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
3052 auto_vec<tree> when building a vector passed to build_vector.
3053 (vect_create_epilog_for_reduction): Likewise.
3054 (vectorizable_induction): Likewise.
3055 (get_initial_def_for_reduction): Likewise. Fix indentation of
3057 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
3059 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
3060 passed to build_vector.
3062 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3063 Alan Hayward <alan.hayward@arm.com>
3064 David Sherwood <david.sherwood@arm.com>
3066 * tree-core.h (tree_base::u): Add an "nelts" field.
3067 (tree_vector): Use VECTOR_CST_NELTS as the length.
3068 * tree.c (tree_size): Likewise.
3069 (make_vector): Initialize VECTOR_CST_NELTS.
3070 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
3071 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
3072 TYPE_VECTOR_SUBPARTS.
3073 * expr.c (const_vector_mask_from_tree): Consistently use "units"
3074 as the number of units, setting it from VECTOR_CST_NELTS.
3075 (const_vector_from_tree): Likewise.
3076 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
3077 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
3078 (fold_negate_expr_1): Likewise.
3079 (fold_convert_const): Likewise.
3080 (const_binop): Likewise. Differentiate the number of output and
3082 (const_unop): Likewise.
3083 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
3084 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
3085 in cases that did the opposite.
3087 2017-09-14 Richard Biener <rguenther@suse.de>
3089 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
3092 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
3094 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
3096 2017-09-14 Jakub Jelinek <jakub@redhat.com>
3099 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
3100 if and where to split a bb, except for splitting before debug insn
3101 sequences followed by non-label real insn. Delete debug insns
3102 in between basic blocks.
3104 * combine.c (make_compound_operation_int): Formatting fixes.
3106 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
3107 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3108 * config/netbsd.h (LINK_EH_SPEC): Likewise.
3109 * config/sol2.h (LINK_EH_SPEC): Likewise.
3110 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3111 * config/s390/linux.h (LINK_SPEC): Likewise.
3112 * config/freebsd.h (LINK_EH_SPEC): Likewise.
3113 * config/openbsd.h (LINK_EH_SPEC): Likewise.
3114 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3115 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
3116 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
3117 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3118 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
3119 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
3121 2017-09-13 Jakub Jelinek <jakub@redhat.com>
3123 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
3125 (ENDFILE_LINUX_SPEC): Likewise.
3126 (LINK_EH_SPEC): Likewise.
3127 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
3128 (LINK_OS_LINUX_SPEC32): Likewise.
3129 (LINK_OS_LINUX_SPEC64): Likewise.
3130 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
3131 (LINK_OS_LINUX_SPEC): Likewise.
3133 2017-09-13 Martin Liska <mliska@suse.cz>
3136 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
3137 CASE_HIGH is NULL_TREE.
3139 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
3140 Alan Hayward <alan.hayward@arm.com>
3141 David Sherwood <david.sherwood@arm.com>
3143 * target.def (secondary_memory_needed): New hook.
3144 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
3145 instead of SECONDARY_MEMORY_NEEDED.
3146 (secondary_memory_needed_mode): Likewise.
3147 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
3148 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
3149 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
3150 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
3151 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
3152 * doc/tm.texi: Regenerate.
3153 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
3154 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
3155 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3156 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
3157 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
3158 * config/i386/i386.c (inline_secondary_memory_needed): Put the
3159 mode argument first and change the reg_class arguments to reg_class_t.
3160 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
3161 Make static. Update the call to inline_secondary_memory_needed.
3162 (ix86_register_move_cost): Update the call to
3163 inline_secondary_memory_needed.
3164 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3165 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
3167 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
3168 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
3170 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
3171 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
3172 * config/mips/mips.c (mips_secondary_memory_needed): Make static
3173 and match hook interface. Add comment from mips.h.
3174 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3175 * config/mmix/mmix.md (truncdfsf2): Refer to
3176 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
3178 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
3179 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
3180 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3181 (pa_secondary_memory_needed): New function.
3182 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
3183 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
3184 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3185 (pdp11_secondary_memory_needed): Make static and match hook interface.
3186 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
3187 * config/powerpcspe/powerpcspe-protos.h
3188 (rs6000_secondary_memory_needed_ptr): Delete.
3189 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
3191 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3192 (rs6000_option_override_internal): Assign to
3193 targetm.secondary_memory_needed rather than
3194 rs6000_secondary_memory_needed_ptr.
3195 (rs6000_secondary_memory_needed): Match hook interface.
3196 (rs6000_debug_secondary_memory_needed): Likewise.
3197 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
3198 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
3199 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
3200 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3201 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
3202 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
3204 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
3205 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3206 (rs6000_option_override_internal): Assign to
3207 targetm.secondary_memory_needed rather than
3208 rs6000_secondary_memory_needed_ptr.
3209 (rs6000_secondary_memory_needed): Match hook interface.
3210 (rs6000_debug_secondary_memory_needed): Likewise.
3211 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
3212 * config/s390/s390.c (s390_secondary_memory_needed): New function.
3213 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3214 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
3215 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3216 (sparc_secondary_memory_needed): New function.
3217 * lra-constraints.c (check_and_process_move): Refer to
3218 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
3220 (curr_insn_transform): Likewise.
3221 (process_alt_operands): Use targetm.secondary_memory_needed
3222 instead of TARGET_SECONDARY_MEMORY_NEEDED.
3223 (check_secondary_memory_needed_p): Likewise.
3224 (choose_split_class): Likewise.
3225 * reload.c: Unconditionally include code that was previously
3226 conditional on SECONDARY_MEMORY_NEEDED.
3227 (push_secondary_reload): Use targetm.secondary_memory_needed
3228 instead of TARGET_SECONDARY_MEMORY_NEEDED.
3229 (push_reload): Likewise.
3230 * reload1.c: Unconditionally include code that was previously
3231 conditional on SECONDARY_MEMORY_NEEDED.
3232 (choose_reload_regs): Use targetm.secondary_memory_needed
3233 instead of TARGET_SECONDARY_MEMORY_NEEDED.
3234 (gen_reload): Likewise.
3235 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
3237 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
3238 Alan Hayward <alan.hayward@arm.com>
3239 David Sherwood <david.sherwood@arm.com>
3241 * target.def (secondary_memory_needed_mode): New hook:
3242 * targhooks.c (default_secondary_memory_needed_mode): Declare.
3243 * targhooks.h (default_secondary_memory_needed_mode): New function.
3244 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
3245 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
3246 * doc/tm.texi: Regenerate.
3247 * lra-constraints.c (check_and_process_move): Use
3248 targetm.secondary_memory_needed_mode instead of
3249 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
3250 (curr_insn_transform): Likewise.
3251 * reload.c (get_secondary_mem): Likewise.
3252 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3253 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
3255 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3256 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3257 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
3258 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3259 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
3261 * config/powerpcspe/powerpcspe-protos.h
3262 (rs6000_secondary_memory_needed_mode): Delete.
3263 * config/powerpcspe/powerpcspe.c
3264 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3265 (rs6000_secondary_memory_needed_mode): Make static.
3266 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3267 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
3269 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
3271 (rs6000_secondary_memory_needed_mode): Make static.
3272 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3273 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
3274 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3275 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3276 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
3278 (sparc_secondary_memory_needed_mode): New function.
3279 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
3281 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
3283 * config/aarch64/constraints.md (Umq): New constraint.
3284 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
3286 (mov<mode>): Update condition.
3288 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3290 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
3291 when bitposition is the same.
3293 2017-09-13 Richard Biener <rguenther@suse.de>
3295 * dwarf2out.c (output_die_symbol): Remove.
3296 (output_die): Do not output a DIEs symbol.
3298 2017-09-13 Richard Biener <rguenther@suse.de>
3301 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
3302 default-def to avoid breaking iterator update with the weird
3303 interaction with cgraph_update_edges_for_call_stmt_node.
3305 2017-09-13 Richard Biener <rguenther@suse.de>
3307 * tree-cfg.c (verify_gimple_assign_binary): Add verification
3308 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
3309 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
3310 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
3312 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
3314 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
3315 Disable pc relative literal load irrespective of
3316 TARGET_FIX_ERR_A53_84341 for default.
3318 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
3320 * config/sparc/sparc.c (output_return): Output the source location of
3321 the insn in the delay slot, if any.
3322 (output_sibcall): Likewise.
3324 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
3327 * common.opt (-static-pie): New alias.
3328 (shared): Negate static-pie.
3329 (-no-pie): Update help text.
3331 (static-pie): New option.
3332 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
3333 -static-pie support.
3334 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
3335 (LINK_EH_SPEC): Likewise.
3336 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3337 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
3338 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
3339 * gcc.c (LINK_COMMAND_SPEC): Likewise.
3340 (init_gcc_specs): Likewise.
3341 (init_spec): Likewise.
3342 (display_help): Update help message for -pie.
3343 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
3346 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
3348 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
3349 (movdi_aarch64): Likewise.
3350 (movti_aarch64): Likewise.
3352 2017-09-12 Simon Wright <simon@pushface.org>
3355 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
3356 calculation of the minor version, always output as 0.
3358 2017-09-12 Jakub Jelinek <jakub@redhat.com>
3361 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
3362 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
3363 on it early, rather than manual conversion late. For
3364 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
3365 instead of performing manual conversion.
3367 2017-09-12 Carl Love <cel@us.ibm.com>
3369 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
3370 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
3372 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
3373 VMULOSW): Add definitions.
3374 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
3375 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
3376 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
3377 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
3378 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
3380 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
3382 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
3384 (movti_aarch64): Likewise.
3385 (movdf_aarch64): Likewise.
3386 (movtf_aarch64): Likewise.
3387 (load_pairdi): Likewise.
3388 (store_pairdi): Likewise.
3389 (load_pairdf): Likewise.
3390 (store_pairdf): Likewise.
3391 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
3392 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
3393 (ldr_got_small_<mode>): Likewise.
3394 (ldr_got_small_28k_<mode>): Likewise.
3395 (ldr_got_tiny): Likewise.
3396 * config/aarch64/iterators.md (ldst_sz): New.
3397 (ldpstp_sz): Likewise.
3398 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
3400 (thunderx_load): Split load_8 to load_16.
3401 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
3403 (thunderx2t99_storepair_basic): Split store_8 to store_16.
3404 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
3405 (xgene1_store_pair): Split store_8 to store_16.
3406 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
3407 (falkor_st_0_st_sd): Split store_8 to store_16.
3409 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
3411 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
3412 and store1/2/3/4 to store_4/8/12/16.
3413 * config/aarch64/aarch64.md: Update for rename.
3414 * config/arm/arm.md: Likewise.: Likewise.
3415 * config/arm/arm.c: Likewise.
3416 * config/arm/thumb1.md: Likewise.
3417 * config/arm/thumb2.md: Likewise.
3418 * config/arm/vfp.md: Likewise.
3419 * config/arm/arm-generic.md: Likewise.
3420 * config/arm/arm1020e.md: Likewise.
3421 * config/arm/arm1026ejs.md: Likewise.
3422 * config/arm/arm1136jfs.md: Likewise.
3423 * config/arm/arm926ejs.md: Likewise.
3424 * config/arm/cortex-a15.md: Likewise.
3425 * config/arm/cortex-a17.md: Likewise.
3426 * config/arm/cortex-a5.md: Likewise.
3427 * config/arm/cortex-a53.md: Likewise.
3428 * config/arm/cortex-a57.md: Likewise.
3429 * config/arm/cortex-a7.md: Likewise.
3430 * config/arm/cortex-a8.md: Likewise.
3431 * config/arm/cortex-a9.md: Likewise.
3432 * config/arm/cortex-m4.md: Likewise.
3433 * config/arm/cortex-m7.md: Likewise.
3434 * config/arm/cortex-r4.md: Likewise.
3435 * config/arm/exynos-m1.md: Likewise.
3436 * config/arm/fa526.md: Likewise.
3437 * config/arm/fa606te.md: Likewise.
3438 * config/arm/fa626te.md: Likewise.
3439 * config/arm/fa726te.md: Likewise.
3440 * config/arm/fmp626.md: Likewise.
3441 * config/arm/iwmmxt.md: Likewise.
3442 * config/arm/ldmstm.md: Likewise.
3443 * config/arm/marvell-pj4.md: Likewise.
3444 * config/arm/xgene1.md: Likewise.
3445 * config/aarch64/thunderx.md: Likewise.
3446 * config/aarch64/thunderx2t99.md: Likewise.
3447 * config/aarch64/falkor.md: Likewise.
3449 2017-09-12 Martin Liska <mliska@suse.cz>
3451 * attribs.c (private_lookup_attribute): New function.
3452 * attribs.h (private_lookup_attribute): Declared here.
3453 (lookup_attribute): Called from this place.
3455 2017-09-12 Richard Biener <rguenther@suse.de>
3457 PR tree-optimization/82157
3458 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
3459 stmts with side-effects.
3461 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3462 Alan Hayward <alan.hayward@arm.com>
3463 David Sherwood <david.sherwood@arm.com>
3465 * target.def (hard_regno_nregs): New hook.
3466 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
3467 * targhooks.h (default_hard_regno_nregs): Declare.
3468 * targhooks.c (default_hard_regno_nregs): New function.
3469 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
3470 (TARGET_HARD_REGNO_NREGS): ...this hook.
3471 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
3472 (CLASS_MAX_NREGS): Likewise.
3473 * doc/tm.texi: Regenerate.
3474 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
3475 instead of HARD_REGNO_NREGS.
3476 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
3477 HARD_REGNO_NREGS in the comment.
3478 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
3479 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
3480 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
3481 Return an unsigned int.
3482 (TARGET_HARD_REGNO_NREGS): Redefine.
3483 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
3484 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
3485 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
3486 (arc_hard_regno_nregs): New function.
3487 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
3488 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
3489 (arm_hard_regno_nregs): New function.
3490 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
3491 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
3492 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
3493 (TARGET_HARD_REGNO_NREGS): Redefine.
3494 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
3495 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
3496 (HARD_REGNO_NREGS): Delete.
3497 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
3498 (cr16_hard_regno_nregs): New function.
3499 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
3500 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
3501 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
3502 (cris_hard_regno_nregs): New function.
3503 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
3504 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
3505 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
3506 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
3507 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
3508 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
3509 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
3510 (frv_hard_regno_nregs): Make static. Take and return an
3512 (frv_class_max_nregs): Remove outdated copy of documentation.
3513 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
3514 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
3515 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
3516 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
3517 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
3518 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
3519 (TARGET_HARD_REGNO_NREGS): Redefine.
3520 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
3521 (CLASS_MAX_NREGS): Update comment.
3522 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
3523 (ia64_hard_regno_nregs): New function.
3524 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
3525 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
3526 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
3527 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
3528 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
3530 (m32c_hard_regno_nregs): Likewise. Make static.
3531 (TARGET_HARD_REGNO_NREGS): Redefine.
3532 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
3533 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
3534 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
3535 (m68k_hard_regno_nregs): New function.
3536 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
3537 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
3538 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
3539 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
3540 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
3541 Take and return an unsigned int.
3542 (TARGET_HARD_REGNO_NREGS): Redefine.
3543 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
3544 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
3545 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
3546 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
3547 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
3548 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
3549 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
3550 (msp430_hard_regno_nregs): Make static. Take and return an
3552 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
3553 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
3554 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
3555 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
3556 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
3557 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
3558 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
3559 (TARGET_HARD_REGNO_NREGS): Redefine.
3560 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
3561 (PA_HARD_REGNO_NREGS): ...this.
3562 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
3563 (PA_HARD_REGNO_NREGS): ...this.
3564 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
3565 (pa_hard_regno_nregs): New function.
3566 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
3567 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
3568 (pdp11_hard_regno_nregs): New function.
3569 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
3570 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
3571 (rs6000_hard_regno_nregs_hook): New function.
3572 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
3573 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
3574 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
3575 Take and return an unsigned int. Move earlier in file.
3576 (TARGET_HARD_REGNO_NREGS): Redefine.
3577 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
3578 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
3579 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
3580 (rl78_hard_regno_nregs): Make static. Take and return an
3582 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
3583 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
3584 (rs6000_hard_regno_nregs_hook): New function.
3585 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
3586 * config/rx/rx.c (rx_hard_regno_nregs): New function.
3587 (TARGET_HARD_REGNO_NREGS): Redefine.
3588 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
3589 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
3590 instead of HARD_REGNO_NREGS.
3591 (s390_hard_regno_nregs): New function.
3592 (s390_hard_regno_mode_ok): Add comment from s390.h.
3593 (TARGET_HARD_REGNO_NREGS): Redefine.
3594 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
3595 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
3596 (sh_hard_regno_nregs): New function.
3597 (sh_pass_in_reg_p): Use it.
3598 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
3599 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
3600 (sparc_hard_regno_nregs): New function.
3601 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
3602 * config/spu/spu.c (spu_hard_regno_nregs): New function.
3603 (spu_function_arg_advance): Use it, supplying a valid register number.
3604 (TARGET_HARD_REGNO_NREGS): Redefine.
3605 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
3606 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
3607 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
3608 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
3609 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
3610 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
3611 (CLASS_MAX_NREGS): Remove copy of old documentation.
3612 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
3613 (visium_hard_regno_nregs): New function.
3614 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
3615 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
3616 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
3617 xtensa_hard_regno_nregs): New function.
3618 * system.h (HARD_REGNO_NREGS): Poison.
3620 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3622 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
3623 hard_regno_nregs instead of HARD_REGNO_NREGS.
3624 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
3625 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
3626 (c6x_expand_epilogue): Likewise.
3627 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
3628 (frv_read_iacc_argument): Likewise.
3629 * config/sh/sh.c: Include regs.h.
3630 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
3631 (regs_used): Likewise.
3632 (output_stack_adjust): Likewise.
3633 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
3634 * expmed.c: Include regs.h.
3635 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
3636 * ree.c: Include regs.h.
3637 (combine_reaching_defs): Use hard_regno_nregs instead of
3639 (add_removable_extension): Likewise.
3641 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3643 * regs.h (hard_regno_nregs): Turn into a function.
3644 (end_hard_regno): Update accordingly.
3645 * caller-save.c (setup_save_areas): Likewise.
3646 (save_call_clobbered_regs): Likewise.
3647 (replace_reg_with_saved_mem): Likewise.
3648 (insert_restore): Likewise.
3649 (insert_save): Likewise.
3650 * combine.c (can_change_dest_mode): Likewise.
3651 (move_deaths): Likewise.
3652 (distribute_notes): Likewise.
3653 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
3654 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
3655 (rs6000_split_multireg_move): Likewise.
3656 (rs6000_register_move_cost): Likewise.
3657 (rs6000_memory_move_cost): Likewise.
3658 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
3659 (rs6000_split_multireg_move): Likewise.
3660 (rs6000_register_move_cost): Likewise.
3661 (rs6000_memory_move_cost): Likewise.
3662 * cselib.c (cselib_reset_table): Likewise.
3663 (cselib_lookup_1): Likewise.
3664 * emit-rtl.c (set_mode_and_regno): Likewise.
3665 * function.c (aggregate_value_p): Likewise.
3666 * ira-color.c (setup_profitable_hard_regs): Likewise.
3667 (check_hard_reg_p): Likewise.
3668 (calculate_saved_nregs): Likewise.
3669 (assign_hard_reg): Likewise.
3670 (improve_allocation): Likewise.
3671 (calculate_spill_cost): Likewise.
3672 * ira-emit.c (modify_move_list): Likewise.
3673 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
3674 (ira_hard_reg_in_set_p): Likewise.
3675 * ira.c (setup_reg_mode_hard_regset): Likewise.
3676 (clarify_prohibited_class_mode_regs): Likewise.
3677 (check_allocation): Likewise.
3678 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3679 (lra_setup_reg_renumber): Likewise.
3680 (setup_try_hard_regno_pseudos): Likewise.
3681 (spill_for): Likewise.
3682 (assign_hard_regno): Likewise.
3683 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
3684 * lra-constraints.c (in_class_p): Likewise.
3685 (lra_constraint_offset): Likewise.
3686 (simplify_operand_subreg): Likewise.
3687 (lra_constraints): Likewise.
3688 (split_reg): Likewise.
3689 (split_if_necessary): Likewise.
3690 (invariant_p): Likewise.
3691 (inherit_in_ebb): Likewise.
3692 * lra-lives.c (process_bb_lives): Likewise.
3693 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
3694 (get_hard_regs): Likewise.
3695 (do_remat): Likewise.
3696 * lra-spills.c (assign_spill_hard_regs): Likewise.
3697 * mode-switching.c (create_pre_exit): Likewise.
3698 * postreload.c (reload_combine_recognize_pattern): Likewise.
3699 * recog.c (peep2_find_free_register): Likewise.
3700 * regcprop.c (kill_value_regno): Likewise.
3701 (set_value_regno): Likewise.
3702 (copy_value): Likewise.
3703 (maybe_mode_change): Likewise.
3704 (find_oldest_value_reg): Likewise.
3705 (copyprop_hardreg_forward_1): Likewise.
3706 * regrename.c (check_new_reg_p): Likewise.
3707 (regrename_do_replace): Likewise.
3708 * reload.c (push_reload): Likewise.
3709 (combine_reloads): Likewise.
3710 (find_dummy_reload): Likewise.
3711 (operands_match_p): Likewise.
3712 (find_reloads): Likewise.
3713 (find_equiv_reg): Likewise.
3714 (reload_adjust_reg_for_mode): Likewise.
3715 * reload1.c (count_pseudo): Likewise.
3716 (count_spilled_pseudo): Likewise.
3717 (find_reg): Likewise.
3718 (clear_reload_reg_in_use): Likewise.
3719 (free_for_value_p): Likewise.
3720 (allocate_reload_reg): Likewise.
3721 (choose_reload_regs): Likewise.
3722 (reload_adjust_reg_for_temp): Likewise.
3723 (emit_reload_insns): Likewise.
3724 (delete_output_reload): Likewise.
3725 * rtlanal.c (subreg_get_info): Likewise.
3726 * sched-deps.c (sched_analyze_reg): Likewise.
3727 * sel-sched.c (init_regs_for_mode): Likewise.
3728 (mark_unavailable_hard_regs): Likewise.
3729 (choose_best_reg_1): Likewise.
3730 (verify_target_availability): Likewise.
3731 * valtrack.c (dead_debug_insert_temp): Likewise.
3732 * var-tracking.c (track_loc_p): Likewise.
3733 (emit_note_insn_var_location): Likewise.
3734 * varasm.c (make_decl_rtl): Likewise.
3735 * reginfo.c (choose_hard_reg_mode): Likewise.
3736 (init_reg_modes_target): Refer directly to
3737 this_target_regs->x_hard_regno_nregs.
3739 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3741 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
3742 instead of hard_regno_nregs.
3744 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3746 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
3747 end_hard_regno instead of hard_regno_nregs.
3748 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
3749 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
3750 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
3751 * ira-color.c (improve_allocation): Likewise.
3752 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3753 * lra-lives.c (mark_regno_live): Likewise.
3754 (mark_regno_dead): Likewise.
3755 * lra-remat.c (operand_to_remat): Likewise.
3756 * lra.c (collect_non_operand_hard_regs): Likewise.
3757 * postreload.c (reload_combine_note_store): Likewise.
3758 (move2add_valid_value_p): Likewise.
3759 * reload.c (regno_clobbered_p): Likewise.
3761 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3763 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
3765 * config/v850/v850.c (v850_reorg): Likewise.
3766 * reload.c (refers_to_regno_for_reload_p): Likewise.
3767 (find_equiv_reg): Likewise.
3768 * reload1.c (reload_reg_reaches_end_p): Likewise.
3770 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3772 * caller-save.c (add_used_regs): Use REG_NREGS instead of
3774 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
3775 * config/arm/arm.c (output_move_neon): Likewise.
3776 (arm_attr_length_move_neon): Likewise.
3777 (neon_split_vcombine): Likewise.
3778 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
3779 (c6x_mark_reg_written): Likewise.
3780 (c6x_dwarf_register_span): Likewise.
3781 * config/i386/i386.c (ix86_save_reg): Likewise.
3782 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
3783 (rws_access_reg): Likewise.
3784 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
3785 * mode-switching.c (create_pre_exit): Likewise.
3786 * ree.c (combine_reaching_defs): Likewise.
3787 (add_removable_extension): Likewise.
3788 * regcprop.c (find_oldest_value_reg): Likewise.
3789 (copyprop_hardreg_forward_1): Likewise.
3790 * reload.c (reload_inner_reg_of_subreg): Likewise.
3791 (push_reload): Likewise.
3792 (combine_reloads): Likewise.
3793 (find_dummy_reload): Likewise.
3794 (reload_adjust_reg_for_mode): Likewise.
3795 * reload1.c (find_reload_regs): Likewise.
3796 (forget_old_reloads_1): Likewise.
3797 (reload_reg_free_for_value_p): Likewise.
3798 (reload_adjust_reg_for_temp): Likewise.
3799 (emit_reload_insns): Likewise.
3800 (delete_output_reload): Likewise.
3801 * sel-sched.c (choose_best_reg_1): Likewise.
3802 (choose_best_pseudo_reg): Likewise.
3804 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3805 Alan Hayward <alan.hayward@arm.com>
3806 David Sherwood <david.sherwood@arm.com>
3808 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
3809 * target.def (slow_unaligned_access): New hook.
3810 * targhooks.h (default_slow_unaligned_access): Declare.
3811 * targhooks.c (default_slow_unaligned_access): New function.
3812 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
3813 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
3814 * doc/tm.texi: Regenerate.
3815 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
3816 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
3817 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
3819 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
3820 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
3822 (rs6000_slow_unaligned_access): New function.
3823 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
3824 (expand_block_compare): Likewise.
3825 (expand_strn_compare): Likewise.
3826 (rs6000_rtx_costs): Likewise.
3827 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
3828 (riscv_slow_unaligned_access): Likewise.
3829 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
3830 (riscv_slow_unaligned_access_p): ...this and make static.
3831 (riscv_option_override): Update accordingly.
3832 (riscv_slow_unaligned_access): New function.
3833 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
3834 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
3835 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
3836 (rs6000_slow_unaligned_access): New function.
3837 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
3838 (rs6000_rtx_costs): Likewise.
3839 * config/rs6000/rs6000-string.c (expand_block_compare)
3840 (expand_strn_compare): Use targetm.slow_unaligned_access instead
3841 of SLOW_UNALIGNED_ACCESS.
3842 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
3843 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
3844 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
3845 of SLOW_UNALIGNED_ACCESS.
3846 * expmed.c (simple_mem_bitfield_p): Likewise.
3847 * expr.c (alignment_for_piecewise_move): Likewise.
3848 (emit_group_load_1): Likewise.
3849 (emit_group_store): Likewise.
3850 (copy_blkmode_from_reg): Likewise.
3851 (emit_push_insn): Likewise.
3852 (expand_assignment): Likewise.
3853 (store_field): Likewise.
3854 (expand_expr_real_1): Likewise.
3855 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
3856 * lra-constraints.c (simplify_operand_subreg): Likewise.
3857 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
3858 * gimple-ssa-store-merging.c: Likewise in block comment at start
3860 * tree-ssa-strlen.c: Include target.h.
3861 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
3862 of SLOW_UNALIGNED_ACCESS.
3863 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
3865 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3867 PR rtl-optimization/82185
3868 * expmed.c (emit_store_flag_int): Only test tem if it has been
3871 2017-09-12 Richard Biener <rguenther@suse.de>
3874 * match.pd ((FTYPE) N CMP CST): Fix typo.
3876 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
3878 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
3880 (mips_near_type_p): Add 'short_call' attribute as a synonym
3882 * doc/extend.texi (short_call): Document new function attribute.
3884 2017-09-12 Jakub Jelinek <jakub@redhat.com>
3887 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
3888 assertion check that in the condition.
3889 (get_atomic_generic_size): Likewise. Before testing if parameter
3890 has pointer type, if it has array type, call for C++
3891 default_conversion to perform array-to-pointer conversion.
3893 2017-09-12 Richard Biener <rguenther@suse.de>
3895 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
3896 for operations we cannot scalarize.
3898 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
3900 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
3901 vectors heap vectors. Clean up comments.
3902 Make visited_bbs a reference.
3903 (profitable_jump_thread_path): Make GC
3904 vectors heap vectors. Clean up comments.
3906 (convert_and_register_jump_thread_path): Make GC vectors heap
3908 (check_subpath_and_update_thread_path): Same. Clean up comments.
3909 Make visited_bbs a reference.
3910 (handle_phi): Abstract common code to to
3911 register_jump_thread_path_if_profitable.
3912 Rename VAR_BB to DEF_BB.
3914 Make GC vectors heap vectors.
3915 Make visited_bbs a reference.
3916 (handle_assignment): Same.
3917 (register_jump_thread_path_if_profitable): New.
3918 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
3920 Make GC vectors heap vectors. Clean up comments.
3921 Make visited_bbs a reference.
3922 (find_jump_threads_backwards): Make visited_bbs live in the stack.
3923 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
3926 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
3929 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
3930 words of E_DImode object are reachable by xtensa_uimm8x4 access.
3932 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
3934 Revert r251800 and r251799.
3936 2017-09-11 Martin Jambor <mjambor@suse.cz>
3939 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
3940 arguments in advance.
3941 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
3942 use it to find predecessor edges.
3943 (naive_outof_ssa): Collect vector of predecessors.
3945 2017-09-08 Jason Merrill <jason@redhat.com>
3947 PR c++/70029 - ICE with ref-qualifier and -flto
3948 * langhooks.h (struct lang_hooks_for_types): Add
3949 copy_lang_qualifiers.
3950 * attribs.c (build_type_attribute_qual_variant): Use it.
3951 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
3953 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
3954 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
3956 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
3959 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
3960 (*mulsi3_sp64): New instruction.
3961 (mulsi3): New expander.
3963 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
3965 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
3967 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3969 * sancov.c: Include memmodel.h.
3971 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
3974 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
3977 2017-09-07 Carl Love <cel@us.ibm.com>
3979 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
3980 the sldi instruction.
3982 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
3984 * sancov.c: Include tm_p.h.
3986 2017-09-07 Jakub Jelinek <jakub@redhat.com>
3989 * output.h (switch_to_other_text_partition): New declaration.
3990 * varasm.c (switch_to_other_text_partition): New function.
3991 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
3992 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
3993 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
3994 to the other text partition before emitting LCL label and switch back
3995 after emitting the word after it.
3997 2017-09-07 Richard Biener <rguenther@suse.de>
3999 * passes.def (pass_split_crit_edges): Remove instance before PRE.
4000 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
4001 critical edges here, after loop init.
4002 (pass_data_pre): Remove PROP_no_crit_edges flags.
4003 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
4004 for valueization of call args to avoid leaking VN_TOP.
4005 (visit_use): Assert we do not visit default defs.
4006 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
4007 Use error_mark_node to more easily detect leaking VN_TOP.
4008 All default-defs are varying, not VN_TOP. Mark them visited.
4009 (run_scc_vn): Make code match comment.
4011 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
4013 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
4014 OPTION_MASK_FLOAT128_KEYWORD.
4015 (POWERPC_MASKS): Likewise.
4016 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
4017 support for the -mfloat128-type option, and make -mfloat128
4018 default on PowerPC Linux systems. Define or undefine
4019 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
4020 Define __float128 to be __ieee128 if IEEE 128-bit support is
4021 enabled, or undefine it.
4022 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
4023 Delete defining __FLOAT128_TYPE__.
4024 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
4025 -mfloat128-type option and make -mfloat128 default on PowerPC
4027 (TARGET_FLOAT128_TYPE): Likewise.
4028 (-mfloat128-type): Likewise.
4029 * config/rs6000/rs6000.c (rs6000_option_override_internal):
4030 Delete the -mfloat128-type option and make -mfloat128 default on
4031 PowerPC Linux systems. Always use __ieee128 to be the keyword for
4032 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
4033 128-bit floating point is enabled. Change tests from using
4034 -mfloat128-type to -mfloat128.
4035 (rs6000_mangle_type): Use the correct mangling for the __float128
4036 type even if normal long double is restricted to 64-bits.
4037 (floatn_mode): Enable the _Float128 type by default on VSX Linux
4039 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
4040 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
4041 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
4043 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
4044 documentation for -mfloat128.
4046 2017-09-06 Olivier Hainque <hainque@adacore.com>
4048 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
4050 2017-09-06 Wish Wu <wishwu007@gmail.com>
4051 Jakub Jelinek <jakub@redhat.com>
4053 * asan.c (initialize_sanitizer_builtins): Add
4054 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
4055 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
4056 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
4057 BT_FN_VOID_UINT64_PTR variables.
4058 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
4059 (BT_FN_VOID_UINT16_UINT16): Likewise.
4060 (BT_FN_VOID_UINT32_UINT32): Likewise.
4061 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
4062 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
4063 (BT_FN_VOID_UINT64_PTR): Likewise.
4064 * common.opt (flag_sanitize_coverage): New variable.
4065 (fsanitize-coverage=trace-pc): Remove.
4066 (fsanitize-coverage=): Add.
4067 * flag-types.h (enum sanitize_coverage_code): New enum.
4068 * fold-const.c (fold_range_test): Disable non-short-circuit
4069 optimization if flag_sanitize_coverage.
4070 (fold_truth_andor): Likewise.
4071 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
4072 * opts.c (COVERAGE_SANITIZER_OPT): Define.
4073 (coverage_sanitizer_opts): New array.
4074 (get_closest_sanitizer_option): Add OPTS argument, handle also
4075 OPT_fsanitize_coverage_.
4076 (parse_sanitizer_options): Adjusted to also handle
4077 OPT_fsanitize_coverage_.
4078 (common_handle_option): Add OPT_fsanitize_coverage_.
4079 * sancov.c (instrument_comparison, instrument_switch): New function.
4080 (sancov_pass): Add trace-cmp support.
4081 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
4082 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
4083 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
4084 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
4085 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
4086 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
4087 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
4088 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
4089 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
4090 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
4092 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
4094 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
4095 error. Only quit immediately if parsing is complete.
4096 (BEGIN): Initialize fatal_err and parse_done.
4097 (begin fpu, end fpu): Check number of arguments.
4098 (begin arch, end arch): Likewise.
4099 (begin cpu, end cpu): Likewise.
4100 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
4101 (optalias): Likewise.
4103 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
4105 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
4106 * config/arm/arm-isa.h: Delete. Move definitions to ...
4107 * arm-cpus.in: ... here. Use new feature and fgroup values.
4108 * config/arm/arm.c (arm_option_override): Use lower case for feature
4110 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
4111 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
4112 * config/arm/parsecpu.awk (END): Add new command 'isa'.
4114 (print_isa_bits_for): New function.
4115 (gen_isa): New function.
4116 (gen_comm_data): Use print_isa_bits_for.
4117 (define feature): New keyword.
4118 (define fgroup): New keyword.
4119 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
4120 (arm-isa.h): Add rule to generate file.
4121 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
4122 case for feature bit names.
4124 2017-09-06 Richard Biener <rguenther@suse.de>
4126 * tree-ssa-pre.c (NECESSARY): Remove.
4127 (create_expression_by_pieces): Do not touch pass-local flags.
4128 (insert_into_preds_of_block): Likewise.
4129 (do_pre_regular_insertion): Likewise.
4130 (eliminate_insert): Likewise.
4131 (eliminate_dom_walker::before_dom_children): Likewise.
4132 (fini_eliminate): Do not look at inserted_exprs.
4133 (mark_operand_necessary): Remove.
4134 (remove_dead_inserted_code): Replace with simple work-list
4135 algorithm based on inserted_exprs and SSA uses.
4136 (pass_pre::execute): Re-order fini_eliminate and
4137 remove_dead_inserted_code.
4139 2017-09-06 Olivier Hainque <hainque@adacore.com>
4141 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
4142 for VxWorks 7. Adjust surrounding comments.
4144 2017-09-06 Richard Biener <rguenther@suse.de>
4146 * gimple-ssa-strength-reduction.c
4147 (find_candidates_dom_walker::before_dom_children): Also allow
4150 2017-09-06 Richard Biener <rguenther@suse.de>
4152 PR tree-optimization/82108
4153 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
4154 for gap in the non-permutation SLP case.
4156 2017-09-06 Martin Jambor <mjambor@suse.cz>
4158 PR tree-optimization/82078
4159 * tree-sra.c (sort_and_splice_var_accesses): Move call to
4160 add_access_to_work_queue...
4161 (build_accesses_from_assign): ...here.
4162 (propagate_all_subaccesses): Make sure racc is the group
4163 representative, if there is one.
4165 2017-09-06 Jakub Jelinek <jakub@redhat.com>
4168 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
4171 2017-09-06 Richard Biener <rguenther@suse.de>
4173 * gimple-ssa-strength-reduction.c
4174 (find_candidates_dom_walker::before_doom_children): Use a
4175 type and not a mode check.
4177 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
4180 * config/arm/predicates.md (arm_general_adddi_operand): Create new
4182 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
4184 2017-09-05 Jeff Law <law@redhat.com>
4186 PR tree-optimization/64910
4187 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
4188 cases where we have 3 or more operands.
4190 2017-09-05 Jakub Jelinek <jakub@redhat.com>
4193 * omp-low.c (lower_omp_for): Recompute tree invariant if
4194 gimple_omp_for_initial/final is ADDR_EXPR.
4197 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
4198 into gimple val before gimplification fo the COND_EXPR.
4200 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
4202 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
4203 REGION_COPY argument.
4204 (thread_through_all_blocks): Remove unused argument to
4205 duplicate_thread_path.
4207 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4208 Alan Hayward <alan.hayward@arm.com>
4209 David Sherwood <david.sherwood@arm.com>
4211 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
4212 Take a scalar_mode rather than a machine_mode.
4213 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
4214 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
4215 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
4216 (aarch64_gen_adjusted_ldpstp): Likewise.
4217 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
4219 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4220 Alan Hayward <alan.hayward@arm.com>
4221 David Sherwood <david.sherwood@arm.com>
4223 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
4224 Take a scalar_int_mode instead of a machine_mode.
4225 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
4226 (aarch64_output_scalar_simd_mov_immediate): Likewise.
4227 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
4228 (aarch64_simd_attr_length_rglist): Delete.
4229 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
4230 a scalar_int_mode instead of a machine_mode.
4231 (aarch64_add_offset): Likewise.
4232 (aarch64_internal_mov_immediate): Likewise
4233 (aarch64_add_constant_internal): Likewise.
4234 (aarch64_add_constant): Likewise.
4235 (aarch64_movw_imm): Likewise.
4236 (aarch64_rtx_arith_op_extract_p): Likewise.
4237 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
4238 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
4239 Remove assert that the mode isn't a vector.
4240 (aarch64_output_scalar_simd_mov_immediate): Likewise.
4241 (aarch64_expand_mov_immediate): Update calls after above changes.
4242 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
4243 (aarch64_and_bitmask_imm): Check for scalar integer modes.
4244 (aarch64_move_imm): Likewise.
4245 (aarch64_can_const_movi_rtx_p): Likewise.
4246 (aarch64_strip_extend): Likewise.
4247 (aarch64_extr_rtx_p): Likewise.
4248 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
4249 a CONST_INT when the mode parameter is VOIDmode.
4250 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
4252 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4254 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
4255 * stor-layout.c (bitwise_mode_for_mode): Likewise.
4256 (bitwise_type_for_mode): Update accordingly.
4258 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4260 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
4261 * stor-layout.c (mode_for_size_tree): Likewise.
4262 (mode_for_array): Update accordingly.
4263 (layout_decl): Likewise.
4264 (compute_record_mode): Likewise. Only set the mode once.
4266 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4268 * target.def (get_mask_mode): Change return type to opt_mode.
4270 * doc/tm.texi: Regenerate.
4271 * targhooks.h (default_get_mask_mode): Return an opt_mode.
4272 * targhooks.c (default_get_mask_mode): Likewise.
4273 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
4274 * optabs-query.c (can_vec_mask_load_store_p): Update use of
4275 targetm.get_mask_mode.
4276 * tree.c (build_truth_vector_type): Likewise.
4278 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4280 * machmode.h (mode_for_vector): Return an opt_mode.
4281 * stor-layout.c (mode_for_vector): Likewise.
4282 (mode_for_int_vector): Update accordingly.
4283 (layout_type): Likewise.
4284 * config/i386/i386.c (emit_memmov): Likewise.
4285 (ix86_expand_set_or_movmem): Likewise.
4286 (ix86_expand_vector_init): Likewise.
4287 (ix86_get_mask_mode): Likewise.
4288 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
4290 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
4291 * expmed.c (extract_bit_field_1): Likewise.
4292 * expr.c (expand_expr_real_2): Likewise.
4293 * optabs-query.c (can_vec_perm_p): Likewise.
4294 (can_vec_mask_load_store_p): Likewise.
4295 * optabs.c (expand_vec_perm): Likewise.
4296 * targhooks.c (default_get_mask_mode): Likewise.
4297 * tree-vect-stmts.c (vectorizable_store): Likewise.
4298 (vectorizable_load): Likewise.
4299 (get_vectype_for_scalar_type_and_size): Likewise.
4301 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4303 * machmode.h (mode_for_int_vector): New function.
4304 * stor-layout.c (mode_for_int_vector): Likewise.
4305 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
4306 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
4307 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
4308 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
4309 (s390_expand_vcond): Likewise.
4311 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4313 * machmode.h (opt_machine_mode): New type.
4314 (opt_mode<T>): Allow construction from anything that can be
4316 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
4317 (mode_for_size): Return an opt_machine_mode.
4318 * stor-layout.c (mode_for_size): Likewise.
4319 (mode_for_size_tree): Update call accordingly.
4320 (bitwise_mode_for_mode): Likewise.
4321 (make_fract_type): Likewise.
4322 (make_accum_type): Likewise.
4323 * caller-save.c (replace_reg_with_saved_mem): Update call
4325 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4326 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4327 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4328 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4329 * expmed.c (extract_bit_field_1): Likewise.
4330 * reload.c (get_secondary_mem): Likewise.
4331 * varasm.c (assemble_integer): Likewise.
4332 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
4335 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4337 * machmode.h (decimal_float_mode_for_size): New function.
4338 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
4339 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
4340 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
4341 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
4342 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
4343 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
4345 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4347 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
4348 (get_builtin_sync_mode): Likewise.
4349 (expand_ifn_atomic_compare_exchange): Likewise.
4350 (expand_builtin_atomic_clear): Likewise.
4351 (expand_builtin_atomic_test_and_set): Likewise.
4352 (fold_builtin_atomic_always_lock_free): Likewise.
4353 * calls.c (compute_argument_addresses): Likewise.
4354 (emit_library_call_value_1): Likewise.
4355 (store_one_arg): Likewise.
4356 * combine.c (combine_instructions): Likewise.
4357 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
4358 * config/arm/arm.c (arm_function_value): Likewise.
4359 (aapcs_allocate_return_reg): Likewise.
4360 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
4361 * config/i386/i386.c (construct_container): Likewise.
4362 (ix86_gimplify_va_arg): Likewise.
4363 (ix86_expand_sse_cmp): Likewise.
4364 (emit_memmov): Likewise.
4365 (emit_memset): Likewise.
4366 (expand_small_movmem_or_setmem): Likewise.
4367 (ix86_expand_pextr): Likewise.
4368 (ix86_expand_pinsr): Likewise.
4369 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
4370 * config/microblaze/microblaze.c (microblaze_block_move_straight):
4372 * config/mips/mips.c (mips_function_value_1) Likewise.
4373 (mips_block_move_straight): Likewise.
4374 (mips_expand_ins_as_unaligned_store): Likewise.
4375 * config/powerpcspe/powerpcspe.c
4376 (rs6000_darwin64_record_arg_advance_flush): Likewise.
4377 (rs6000_darwin64_record_arg_flush): Likewise.
4378 * config/rs6000/rs6000.c
4379 (rs6000_darwin64_record_arg_advance_flush): Likewise.
4380 (rs6000_darwin64_record_arg_flush): Likewise.
4381 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
4382 (sparc_function_value_1): Likewise.
4383 * config/spu/spu.c (adjust_operand): Likewise.
4384 (spu_emit_branch_or_set): Likewise.
4385 (arith_immediate_p): Likewise.
4386 * emit-rtl.c (gen_lowpart_common): Likewise.
4387 * expr.c (expand_expr_real_1): Likewise.
4388 * function.c (assign_parm_setup_block): Likewise.
4389 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
4390 * reload1.c (alter_reg): Likewise.
4391 * stor-layout.c (mode_for_vector): Likewise.
4392 (layout_type): Likewise.
4394 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4396 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
4397 (spu_convert_move): Likewise.
4398 * lower-subreg.c (resolve_simple_move): Likewise.
4400 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4403 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
4404 define_insn to a define_expand.
4405 (altivec_vsum2sws_direct): New define_insn.
4406 (altivec_vsumsws): Convert from a define_insn to a define_expand.
4408 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
4410 * config/arm/arm.c (arm_option_params_internal): Improve setting of
4413 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
4418 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
4419 realign stack if stack alignment needed is less than incoming
4422 2017-09-05 Marek Polacek <polacek@redhat.com>
4425 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
4428 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
4430 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
4432 2017-09-05 Richard Biener <rguenther@suse.de>
4434 PR tree-optimization/82084
4435 * fold-const.c (can_native_encode_string_p): Handle wide characters.
4437 2017-09-05 Richard Biener <rguenther@suse.de>
4439 PR tree-optimization/82102
4440 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
4442 2017-09-05 Martin Liska <mliska@suse.cz>
4444 PR tree-optimization/82032
4445 * tree-cfg.c (generate_range_test): New function.
4446 * tree-cfg.h (generate_range_test): Declared here.
4447 * tree-cfgcleanup.c (convert_single_case_switch): New function.
4448 (cleanup_control_expr_graph): Use it.
4449 * tree-switch-conversion.c (try_switch_expansion): Remove
4451 (emit_case_nodes): Use generate_range_test.
4453 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
4456 * config/i386/i386.md (*<btsc><mode>_mask): Add
4457 TARGET_USE_BT to insn constraint.
4458 (*btr<mode>_mask): Ditto.
4460 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
4462 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
4463 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
4465 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
4468 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
4469 TARGET_NEON and TARGET_IWMMXT.
4470 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
4471 TARGET_NEON and TARGET_IWMMXT.
4472 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
4474 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
4476 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
4477 (ix86_rewrite_tls_address): Ditto.
4478 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
4479 (ix86_rewrite_tls_address_1): Ditto.
4480 (ix86_rewrite_tls_address): Ditto.
4481 * config/i386/predicates.md (tls_address_pattern): New predicate.
4482 * config/i386/i386.md (TLS address splitter): New splitter.
4484 2017-09-04 Richard Biener <rguenther@suse.de>
4486 PR tree-optimization/82084
4487 * fold-const.h (can_native_encode_string_p): Declare.
4488 * fold-const.c (can_native_encode_string_p): Factor out from ...
4489 (native_encode_string): ... here.
4490 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
4491 vectorizing stores from constants we later cannot handle.
4493 2017-09-04 Marek Polacek <polacek@redhat.com>
4496 * doc/invoke.texi: Update -Wtautological-compare documentation.
4498 2017-09-04 Jeff Law <law@redhat.com>
4500 PR tree-optimization/64910
4501 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
4502 swap the first and last operand if the last is a constant.
4504 2017-09-04 Marek Polacek <polacek@redhat.com>
4507 * convert.c (do_narrow): When sanitizing signed integer overflows,
4508 bail out for signed types.
4509 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
4511 2017-09-04 Richard Biener <rguenther@suse.de>
4513 PR tree-optimization/82060
4514 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
4515 Move devirtualization after stmt folding and before EH/AB/noreturn
4516 cleanup to get the stmt refs canonicalized. Use a bool instead
4517 of gimple_modified_p since that doesn't work for NOPs. Schedule
4518 NOPs generated by folding for removal.
4520 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4521 Alan Hayward <alan.hayward@arm.com>
4522 David Sherwood <david.sherwood@arm.com>
4524 * coretypes.h (pad_direction): New enum.
4525 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
4526 (FUNCTION_ARG_PADDING): Likewise.
4527 * target.def (function_arg_padding): New hook.
4528 * targhooks.h (default_function_arg_padding): Declare.
4529 * targhooks.c (default_function_arg_padding): New function.
4530 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
4531 (TARGET_FUNCTION_ARG_PADDING): ...this.
4532 * doc/tm.texi: Regenerate.
4533 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
4534 instead of direction.
4535 (compute_argument_addresses): Likewise.
4536 (load_register_parameters): Likewise.
4537 (emit_library_call_value_1): Likewise.
4538 (store_one_arg): Use targetm.calls.function_arg_padding instead
4539 of FUNCTION_ARG_PADDING.
4540 (must_pass_in_stack_var_size_or_pad): Likewise.
4541 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
4542 (emit_group_store): Likewise.
4543 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
4544 instead of FUNCTION_ARG_PADDING.
4545 (emit_push_insn): Likewise, and propagate enum change throughout
4547 * function.h (direction): Delete.
4548 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
4550 * function.c (assign_parm_find_stack_rtl): Likewise.
4551 (assign_parm_setup_block_p): Likewise.
4552 (assign_parm_setup_block): Likewise.
4553 (gimplify_parameters): Likewise.
4554 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
4555 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
4557 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
4558 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
4559 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
4560 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
4561 (aarch64_function_arg_padding): ...this new function.
4562 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
4563 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4564 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
4565 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
4566 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
4567 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4568 (arm_pad_arg_upward): Replace with...
4569 (arm_function_arg_padding): ...this new function.
4570 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
4572 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
4573 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
4574 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4575 (ia64_hpux_function_arg_padding): Replace with...
4576 (ia64_function_arg_padding): ...this new function. Use pad_direction
4577 instead of direction. Check for TARGET_HPUX.
4578 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
4579 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4580 (iq2000_function_arg_padding): New function.
4581 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
4582 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
4583 (mips_function_arg_padding): ...this new function.
4584 (mips_pad_reg_upward): Update accordingly.
4585 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4586 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
4587 targetm.calls.function_arg_padding.
4588 (FUNCTION_ARG_PADDING): Delete.
4589 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
4590 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
4591 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
4592 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
4593 (nios2_block_reg_padding): Return pad_direction instead of direction.
4594 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
4595 instead of direction.
4596 (nios2_function_arg_padding): Likewise. Make static.
4597 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4598 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
4599 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
4600 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
4601 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4602 (pa_function_arg_padding): Make static. Return pad_direction instead
4604 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
4605 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
4606 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
4607 instead of direction. Use targetm.calls.function_arg_padding.
4608 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
4609 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
4610 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
4611 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
4612 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
4614 (function_arg_padding): Rename to...
4615 (rs6000_function_arg_padding): ...this. Make static. Return
4616 pad_direction instead of direction.
4617 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
4618 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
4619 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
4620 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
4621 instead of direction. Use targetm.calls.function_arg_padding.
4622 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
4623 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
4624 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
4625 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
4626 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4627 (function_arg_padding): Rename to...
4628 (rs6000_function_arg_padding): ...this. Make static. Return
4629 pad_direction instead of direction.
4630 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
4631 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
4632 * config/s390/s390.c (s390_function_arg_padding): New function.
4633 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4634 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
4635 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
4636 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4637 (function_arg_padding): Rename to...
4638 (sparc_function_arg_padding): ...this. Make static. Return
4639 pad_direction instead of direction.
4640 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
4641 * config/spu/spu.c (spu_function_arg_padding): New function.
4642 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4643 * system.h (FUNCTION_ARG_PADDING): Poison.
4645 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4646 Alan Hayward <alan.hayward@arm.com>
4647 David Sherwood <david.sherwood@arm.com>
4649 * target.def (modes_tieable_p): New hook.
4650 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
4651 (TARGET_MODES_TIEABLE_P): ...this.
4652 * doc/tm.texi.in: Regenerate.
4653 * hooks.h (hook_bool_mode_mode_true): Declare.
4654 * hooks.c (hook_bool_mode_mode_true): New function.
4655 * combine.c (subst): Use targetm.modes_tieable_p instead of
4657 * dse.c (find_shift_sequence): Likewise.
4658 * expmed.c (extract_low_bits): Likewise.
4659 * lower-subreg.c: Include target.h.
4660 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
4662 * rtlanal.c (rtx_cost): Likewise.
4663 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
4664 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
4665 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
4666 (TARGET_MODES_TIEABLE_P): Redefine.
4667 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
4668 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
4669 (TARGET_MODES_TIEABLE_P): Redefine.
4670 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
4671 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
4672 (arc_modes_tieable_p): New function.
4673 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
4674 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
4675 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
4676 (arm_modes_tieable_p): Make static.
4677 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
4678 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
4679 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
4680 (TARGET_MODES_TIEABLE_P): Redefine.
4681 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
4682 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
4683 (TARGET_MODES_TIEABLE_P): Redefine.
4684 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
4685 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
4686 (cr16_modes_tieable_p): New function.
4687 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
4688 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
4689 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
4690 (TRULY_NOOP_TRUNCATION): Update comment.
4691 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
4692 (TRULY_NOOP_TRUNCATION): Update comment.
4693 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
4694 (frv_modes_tieable_p): New function.
4695 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
4696 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
4697 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
4698 (TARGET_MODES_TIEABLE_P): Redefine.
4699 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
4700 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
4701 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
4702 (TARGET_MODES_TIEABLE_P): Redefine.
4703 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
4704 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
4705 (ia64_modes_tieable_p): New function.
4706 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
4707 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
4708 (iq2000_modes_tieable_p): New function.
4709 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
4710 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
4711 (lm32_modes_tieable_p): New function.
4712 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
4713 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
4714 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
4715 (TARGET_MODES_TIEABLE_P): Redefine.
4716 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
4717 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
4718 (m32r_modes_tieable_p): New function.
4719 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
4720 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
4721 (m68k_modes_tieable_p): New function.
4722 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
4723 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
4724 (mcore_modes_tieable_p): New function.
4725 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
4726 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
4728 (TARGET_MODES_TIEABLE_P): Redefine.
4729 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
4730 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
4731 * config/mips/mips.c (mips_modes_tieable_p): Make static.
4732 (TARGET_MODES_TIEABLE_P): Redefine.
4733 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
4734 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
4735 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
4736 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
4737 (mn10300_modes_tieable_p): ...this and make static.
4738 (TARGET_MODES_TIEABLE_P): Redefine.
4739 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
4740 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
4741 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
4742 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
4743 (msp430_modes_tieable_p): Make static.
4744 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
4745 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
4746 (TARGET_MODES_TIEABLE_P): Redefine.
4747 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
4748 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
4749 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
4750 (TARGET_MODES_TIEABLE_P): Redefine.
4751 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
4752 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
4753 * config/pa/pa.c (pa_modes_tieable_p): Make static.
4754 (TARGET_MODES_TIEABLE_P): Redefine.
4755 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
4756 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
4757 (pdp11_modes_tieable_p): New function.
4758 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
4759 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
4760 (rs6000_modes_tieable_p): New function.
4761 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
4762 * config/powerpcspe/powerpcspe.md: Update comment.
4763 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
4764 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
4765 (TARGET_MODES_TIEABLE_P): Redefine.
4766 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
4767 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
4768 (rl78_modes_tieable_p): New function.
4769 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
4770 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
4771 (rs6000_modes_tieable_p): New function.
4772 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
4773 * config/rs6000/rs6000.md: Update comment.
4774 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
4775 * config/rx/rx.c (rx_modes_tieable_p): New function.
4776 (TARGET_MODES_TIEABLE_P): Redefine.
4777 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
4778 * config/s390/s390.c (s390_modes_tieable_p): New function.
4779 (TARGET_MODES_TIEABLE_P): Redefine.
4780 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
4781 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
4782 (sh_modes_tieable_p): New function.
4783 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
4784 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
4785 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
4786 (sparc_modes_tieable_p): Make static.
4787 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
4788 * config/spu/spu.c (spu_modes_tieable_p): New function.
4789 (TARGET_MODES_TIEABLE_P): Redefine.
4790 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
4791 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
4792 (TARGET_MODES_TIEABLE_P): Redefine.
4793 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
4794 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
4795 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
4796 * config/v850/v850.c (v850_modes_tieable_p): New function.
4797 (TARGET_MODES_TIEABLE_P): Redefine.
4798 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
4799 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
4800 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
4801 (visium_modes_tieable_p): New function.
4802 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
4803 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
4804 (xtensa_modes_tieable_p): New function.
4805 * system.h (MODES_TIEABLE_P): Poison.
4807 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4808 Alan Hayward <alan.hayward@arm.com>
4809 David Sherwood <david.sherwood@arm.com>
4811 * target.def (hard_regno_mode_ok): New hook.
4812 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
4813 (TARGET_HARD_REGNO_MODE_OK): ...this.
4814 * doc/tm.texi.in: Regenerate.
4815 * hooks.h (hook_bool_uint_mode_true): Declare.
4816 * hooks.c (hook_bool_uint_mode_true): New function.
4817 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
4819 * genpreds.c (write_insn_preds_c): Add an include of target.h.
4820 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
4821 instead of HARD_REGNO_MODE_OK.
4822 * caller-save.c: Include target.h.
4823 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
4825 * combine.c (can_combine_p): Likewise.
4826 (combinable_i3pat): Likewise.
4827 (can_change_dest_mode): Likewise.
4828 * expr.c (init_expr_target): Likewise.
4829 (convert_move): Likewise.
4830 (convert_modes): Likewise.
4831 * ira.c (setup_prohibited_class_mode_regs): Likewise.
4832 (setup_prohibited_mode_move_regs): Likewise.
4833 * ira.h (target_ira): Likewise.
4834 * lra-assigns.c (find_hard_regno_for_1): Likewise.
4835 * lra-constraints.c (process_alt_operands): Likewise.
4836 (split_reg): Likewise.
4837 * recog.c (peep2_find_free_register): Likewise.
4838 * ree.c (combine_reaching_defs): Likewise.
4839 * regcprop.c (maybe_mode_change): Likewise.
4840 * reginfo.c (init_reg_sets_1): Likewise.
4841 (choose_hard_reg_mode): Likewise.
4842 (simplifiable_subregs): Likewise.
4843 * regrename.c (check_new_reg_p): Likewise.
4844 * reload.c (find_valid_class): Likewise.
4845 (find_valid_class_1): Likewise.
4846 (reload_inner_reg_of_subreg): Likewise.
4847 (push_reload): Likewise.
4848 (combine_reloads): Likewise.
4849 (find_dummy_reload): Likewise.
4850 (find_reloads): Likewise.
4851 * reload1.c (find_reg): Likewise.
4852 (set_reload_reg): Likewise.
4853 (allocate_reload_reg): Likewise.
4854 (choose_reload_regs): Likewise.
4855 (reload_adjust_reg_for_temp): Likewise.
4856 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
4857 (simplify_subreg_regno): Likewise.
4858 * sel-sched.c (init_regs_for_mode): Likewise.
4859 * varasm.c (make_decl_rtl): Likewise.
4860 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
4861 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
4863 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
4864 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
4865 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4866 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
4867 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
4868 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4869 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
4870 (arc_mode_class): Delete.
4871 (HARD_REGNO_MODE_OK): Delete.
4872 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4873 (arc_hard_regno_mode_ok): Rename old array to...
4874 (arc_hard_regno_mode_ok_modes): ...this.
4875 (arc_conditional_register_usage): Update accordingly.
4876 (arc_mode_class): Make static.
4877 (arc_hard_regno_mode_ok): New function.
4878 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
4879 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
4880 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4881 (arm_hard_regno_mode_ok): Make static.
4882 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
4884 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
4885 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
4886 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
4888 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4889 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
4890 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
4891 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
4892 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
4893 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4894 * config/bfin/predicates.md (valid_reg_operand): Use
4895 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
4896 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
4897 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
4898 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4899 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
4900 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
4901 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4902 (cr16_hard_regno_mode_ok): Make static and return a bool.
4903 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
4904 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4905 (cris_hard_regno_mode_ok): New function.
4906 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
4907 (epiphany_mode_class): Delete.
4908 (HARD_REGNO_MODE_OK): Delete.
4909 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
4910 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4911 (hard_regno_mode_ok): Rename to...
4912 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
4913 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
4914 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
4916 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
4917 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
4918 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4919 (frv_hard_regno_mode_ok): Make static and return a bool.
4920 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
4922 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
4923 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
4924 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
4925 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
4927 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4928 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
4929 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
4930 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
4932 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4933 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
4934 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4935 (ia64_hard_regno_mode_ok): New function.
4936 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
4937 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4938 (iq2000_hard_regno_mode_ok): New function.
4939 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
4940 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4941 (lm32_hard_regno_mode_ok): New function.
4942 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
4943 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
4944 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
4945 instead of HARD_REGNO_MODE_OK.
4946 (m32c_hard_regno_ok): Rename to...
4947 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
4948 (m32c_cannot_change_mode_class): Update accordingly.
4949 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4950 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
4951 (m32r_mode_class): Delete.
4952 (HARD_REGNO_MODE_OK): Delete.
4953 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4954 (m32r_hard_regno_mode_ok): Rename to...
4955 (m32r_hard_regno_modes): ...this.
4956 (m32r_mode_class): Make static.
4957 (m32r_hard_regno_mode_ok): New function.
4958 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
4959 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
4960 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4961 (m68k_hard_regno_mode_ok): Make static.
4962 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
4963 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4964 (mcore_hard_regno_mode_ok): New function.
4965 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
4966 (HARD_REGNO_MODE_OK): Delete.
4967 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
4969 (microblaze_hard_regno_mode_ok_p): ...this and make static.
4970 (microblaze_hard_regno_mode_ok): New function.
4971 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4972 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
4973 (mips_hard_regno_mode_ok): Delete.
4974 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
4975 (mips_hard_regno_mode_ok_p): ...this and make static.
4976 (mips_hard_regno_mode_ok_p): Rename to...
4977 (mips_hard_regno_mode_ok_uncached): ...this.
4978 (mips_hard_regno_mode_ok): New function.
4979 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
4980 of HARD_REGNO_MODE_OK.
4981 (mips_option_override): Update after above name changes.
4982 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4983 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
4984 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
4985 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
4986 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
4987 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4988 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
4989 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
4990 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
4991 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4992 (msp430_hard_regno_mode_ok): Make static and return a bool.
4993 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
4994 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
4995 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
4997 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4998 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
4999 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
5000 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
5001 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
5002 (PA_HARD_REGNO_MODE_OK): ...this
5003 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
5004 (PA_HARD_REGNO_MODE_OK): ...this.
5005 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5006 (pa_hard_regno_mode_ok): New function.
5007 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
5008 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5009 (pdp11_hard_regno_mode_ok): New function.
5010 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
5011 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
5013 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
5015 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5016 (rs6000_hard_regno_mode_ok): Rename to...
5017 (rs6000_hard_regno_mode_ok_uncached): ...this.
5018 (rs6000_init_hard_regno_mode_ok): Update accordingly.
5019 (rs6000_hard_regno_mode_ok): New function.
5020 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
5021 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
5022 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
5023 (riscv_hard_regno_mode_ok): ...this and make static.
5024 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5025 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
5026 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
5027 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5028 (rl78_hard_regno_mode_ok): Make static and return bool.
5029 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
5030 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
5032 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
5033 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5034 (rs6000_hard_regno_mode_ok): Rename to...
5035 (rs6000_hard_regno_mode_ok_uncached): ...this.
5036 (rs6000_init_hard_regno_mode_ok): Update accordingly.
5037 (rs6000_hard_regno_mode_ok): New function.
5038 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
5039 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
5040 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5041 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
5042 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
5043 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
5044 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5045 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
5046 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
5047 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5048 (sh_hard_regno_mode_ok): Make static.
5049 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
5050 instead of HARD_REGNO_MODE_OK.
5051 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
5052 (sparc_mode_class): Delete.
5053 (HARD_REGNO_MODE_OK): Delete.
5054 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5055 (hard_regno_mode_classes): Make static.
5056 (sparc_mode_class): Likewise.
5057 (sparc_hard_regno_mode_ok): New function.
5058 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
5059 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
5060 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
5062 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5063 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
5064 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
5065 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
5066 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
5067 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5068 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
5069 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
5070 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5071 (visium_hard_regno_mode_ok): New function.
5072 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
5073 instead of HARD_REGNO_MODE_OK.
5074 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
5075 (HARD_REGNO_MODE_OK): Delete.
5076 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
5077 (xtensa_hard_regno_mode_ok_p): ...this and make static.
5078 (xtensa_option_override): Update accordingly.
5079 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5080 (xtensa_hard_regno_mode_ok): New function.
5081 * system.h (HARD_REGNO_MODE_OK): Poison.
5083 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5084 Alan Hayward <alan.hayward@arm.com>
5085 David Sherwood <david.sherwood@arm.com>
5087 * target.def (hard_regno_call_part_clobbered): New hook.
5088 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
5089 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
5090 * doc/tm.texi: Regenerate.
5091 * hooks.h (hook_bool_uint_mode_false): Declare.
5092 * hooks.c (hook_bool_uint_mode_false): New function.
5093 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5094 * cselib.c (cselib_process_insn): Use
5095 targetm.hard_regno_call_part_clobbered instead of
5096 HARD_REGNO_CALL_PART_CLOBBERED.
5097 * ira-conflicts.c (ira_build_conflicts): Likewise.
5098 * ira-costs.c (ira_tune_allocno_costs): Likewise.
5099 * lra-constraints.c (need_for_call_save_p): Likewise.
5100 * lra-lives.c: Include target.h.
5101 (check_pseudos_live_through_calls): Use
5102 targetm.hard_regno_call_part_clobbered instead of
5103 HARD_REGNO_CALL_PART_CLOBBERED.
5104 * regcprop.c: Include target.h.
5105 (copyprop_hardreg_forward_1): Use
5106 targetm.hard_regno_call_part_clobbered instead of
5107 HARD_REGNO_CALL_PART_CLOBBERED.
5108 * reginfo.c (choose_hard_reg_mode): Likewise.
5109 * regrename.c (check_new_reg_p): Likewise.
5110 * reload.c (find_equiv_reg): Likewise.
5111 * reload1.c (emit_reload_insns): Likewise.
5112 * sched-deps.c (deps_analyze_insn): Likewise.
5113 * sel-sched.c (init_regs_for_mode): Likewise.
5114 (mark_unavailable_hard_regs): Likewise.
5115 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
5116 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5117 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
5119 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5120 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5121 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
5123 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
5125 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5126 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5127 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
5129 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5130 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5131 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
5133 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5134 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
5136 * config/powerpcspe/powerpcspe.c
5137 (rs6000_hard_regno_call_part_clobbered): New function.
5138 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5139 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5140 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
5142 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5143 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5144 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
5146 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5147 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5148 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
5150 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5151 Alan Hayward <alan.hayward@arm.com>
5152 David Sherwood <david.sherwood@arm.com>
5154 * rtl.h (subreg_memory_offset): Declare.
5155 * emit-rtl.c (subreg_memory_offset): New function.
5156 * expmed.c (store_bit_field_1): Use it.
5157 * expr.c (undefined_operand_subword_p): Likewise.
5158 * simplify-rtx.c (simplify_subreg): Likewise.
5160 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
5162 PR rtl-optimization/57448
5165 * optabs.c (expand_atomic_load): Place compiler memory barriers if
5166 using atomic_load pattern.
5167 (expand_atomic_store): Likewise.
5169 2017-09-04 Jakub Jelinek <jakub@redhat.com>
5172 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
5173 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
5174 handling. Use replace_call_with_value with NULL instead of
5175 gsi_replace, unlink_stmt_vdef and release_defs.
5177 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
5180 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
5182 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5185 * rtl.h (emit_library_call_value_1): Declare.
5186 (emit_library_call): Replace declaration with a series of overloads.
5187 Remove the parameter count argument.
5188 (emit_library_call_value): Likewise.
5189 * calls.c (emit_library_call_value_1): Make global. Replace varargs
5190 with an "rtx_mode_t *".
5191 (emit_library_call_value): Delete.
5192 (emit_library_call): Likewise.
5193 * asan.c (asan_emit_stack_protection): Update calls accordingly.
5194 (asan_emit_allocas_unpoison): Likewise.
5195 * builtins.c (expand_builtin_powi): Likewise.
5196 (expand_asan_emit_allocas_unpoison): Likewise.
5197 * cfgexpand.c (expand_main_function): Likewise.
5198 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
5199 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
5200 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
5201 * config/arm/arm.c (arm_trampoline_init): Likewise.
5202 (arm_call_tls_get_addr): Likewise.
5203 (arm_expand_divmod_libfunc): Likewise.
5204 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
5205 (smulsi3_highpart): Likewise.
5206 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
5207 (c6x_expand_compare): Likewise.
5208 (c6x_expand_movmem): Likewise.
5209 * config/frv/frv.c (frv_trampoline_init): Likewise.
5210 * config/i386/i386.c (ix86_trampoline_init): Likewise.
5211 (ix86_expand_divmod_libfunc): Likewise.
5212 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
5213 (ia64_expand_compare): Likewise.
5214 (ia64_profile_hook): Likewise.
5215 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
5216 (nonlocal_goto): Likewise.
5217 (restore_stack_nonlocal): Likewise.
5218 * config/m32r/m32r.c (block_move_call): Likewise.
5219 (m32r_trampoline_init): Likewise.
5220 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
5221 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
5222 (m68k_call_m68k_read_tp): Likewise.
5223 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
5224 (microblaze_expand_divide): Likewise.
5225 * config/mips/mips.h (mips_args): Likewise.
5226 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
5227 (MIPS_ICACHE_SYNC): Likewise.
5228 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
5229 (nios2_trampoline_init): Likewise.
5230 * config/pa/pa.c (hppa_tls_call): Likewise.
5231 (pa_trampoline_init): Likewise.
5232 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
5233 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
5234 (expand_strn_compare): Likewise.
5235 (rs6000_generate_compare): Likewise.
5236 (rs6000_expand_float128_convert): Likewise.
5237 (output_profile_hook): Likewise.
5238 (rs6000_trampoline_init): Likewise.
5239 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
5240 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
5241 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
5242 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
5243 (rs6000_generate_compare): Likewise.
5244 (rs6000_expand_float128_convert): Likewise.
5245 (output_profile_hook): Likewise.
5246 (rs6000_trampoline_init): Likewise.
5247 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
5248 * config/sh/sh.c (sh_trampoline_init): Likewise.
5249 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
5250 (sparc_emit_float_lib_cmp): Likewise.
5251 (sparc32_initialize_trampoline): Likewise.
5252 (sparc64_initialize_trampoline): Likewise.
5253 (sparc_profile_hook): Likewise.
5254 * config/spu/spu.c (ea_load_store): Likewise.
5255 * config/spu/spu.md (floatunssidf2): Likewise.
5256 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
5257 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
5258 * config/visium/visium.c (expand_block_move_4): Likewise.
5259 (expand_block_move_2): Likewise.
5260 (expand_block_move_1): Likewise.
5261 (expand_block_set_4): Likewise.
5262 (expand_block_set_2): Likewise.
5263 (expand_block_set_1): Likewise.
5264 (visium_trampoline_init): Likewise.
5265 (visium_profile_hook): Likewise.
5266 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
5267 (xtensa_setup_frame_addresses): Likewise.
5268 (xtensa_trampoline_init): Likewise.
5269 * except.c (sjlj_emit_function_enter): Likewise.
5270 (sjlj_emit_function_exit): Likewise.
5271 * explow.c (allocate_dynamic_stack_space): Likewise.
5272 (probe_stack_range): Likewise.
5273 * expr.c (convert_mode_scalar): Likewise.
5274 * optabs.c (expand_binop): Likewise.
5275 (expand_twoval_binop_libfunc): Likewise.
5276 (expand_unop): Likewise.
5277 (prepare_cmp_insn): Likewise.
5278 (prepare_float_lib_cmp): Likewise.
5279 (expand_float): Likewise.
5280 (expand_fix): Likewise.
5281 (expand_fixed_convert): Likewise.
5282 (maybe_emit_sync_lock_test_and_set): Likewise.
5283 (expand_atomic_compare_and_swap): Likewise.
5284 (expand_mem_thread_fence): Likewise.
5285 (expand_atomic_fetch_op): Likewise.
5287 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
5289 * doc/generic.texi (OpenACC): Adjust URL.
5290 * doc/invoke.texi (C Dialect Options): Ditto.
5292 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
5294 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
5295 predicate for operand 1. Add (m,<S>) constraint.
5296 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
5297 Prevent memory operand 1 with register operand 2.
5299 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
5301 PR rtl-optimization/82024
5302 * combine.c (try_combine): If the combination result is a PARALLEL,
5303 and we only need to retain the SET in there that would be placed
5304 at I2, check that we can place that at I3 instead, before doing so.
5306 2017-09-01 Jakub Jelinek <jakub@redhat.com>
5309 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
5311 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
5312 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
5315 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
5316 Jeff Law <law@redhat.com>
5318 * varasm.c (bss_initializer_p): Do not put constants into .bss
5319 (categorize_decl_for_section): Handle bss_initializer_p returning
5320 false when DECL_INITIAL is NULL.
5322 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5325 * config/s390/s390.c (s390_can_inline_p): New function.
5327 2017-09-01 Jeff Law <law@redhat.com>
5329 PR tree-optimization/82052
5330 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
5331 Always initialize the returned slot after a hash table miss
5332 when INSERT is true.
5334 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
5336 * config/s390/s390.md (mem_signal_fence): Remove.
5337 * doc/md.texi (mem_signal_fence): Remove.
5338 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
5340 * target-insns.def (mem_signal_fence): Remove.
5342 2017-09-01 Jakub Jelinek <jakub@redhat.com>
5345 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
5348 * asan.c (create_odr_indicator): Strip name encoding from assembler
5349 name before appending it after __odr_asan_.
5351 2017-09-01 Martin Liska <mliska@suse.cz>
5353 PR tree-optimization/82059
5354 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
5355 frequency only when an edge is redirected.
5357 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5359 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
5360 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
5361 (arc_conditional_register_usage): Remove ARC600 lp_count
5363 (arc_file_start): Emit Tag_ARC_CPU_variation.
5364 (arc_can_use_doloop_p): New conditions to use ZOLs.
5365 (hwloop_fail): New function.
5366 (hwloop_optimize): Likewise.
5367 (hwloop_pattern_reg): Likewise.
5368 (arc_doloop_hooks): New struct, to be used with reorg_loops.
5369 (arc_reorg_loops): New function, calls reorg_loops.
5370 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
5371 (arc600_corereg_hazard): Remove ZOL checking, case handled by
5373 (arc_loop_hazard): Remove function, functionality moved into
5375 (arc_hazard): Remove arc_loop_hazard call.
5376 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
5377 into hwloop_optimize.
5378 (arc_label_align): Remove ZOL handling.
5379 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
5380 * config/arc/arc.md (doloop_begin): Remove pattern.
5381 (doloop_begin_i): Likewise.
5382 (doloop_end_i): Likewise.
5383 (doloop_fallback): Likewise.
5384 (doloop_fallback_m): Likewise.
5385 (doloop_end): Reimplement expand.
5386 (arc_lp): New pattern for LP instruction.
5387 (loop_end): New pattern.
5388 (loop_fail): Likewise.
5389 (decrement_and_branch_until_zero): Likewise.
5390 * config/arc/arc.opt (mlpc-width): New option.
5391 * doc/invoke.texi (mlpc-width): Document option.
5393 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5395 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
5396 (arc_ccfsm_advance): Fix checking for delay slots.
5397 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
5399 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5401 * config/arc/arc.md (movqi_insn): Add stores to save constant long
5403 (movhi_insn): Update store instruction constraint which are saving
5404 6-bit short immediates.
5405 (movsi_insn): Consider also short scaled load operations.
5406 (zero_extendhisi2_i): Use Usd constraint instead of T.
5407 (extendhisi2_i): Add q constraint.
5408 (arc_clzsi2): Add type and length attributes.
5409 (arc_ctzsi2): Likewise.
5410 * config/arc/constraints.md (Usc): Update constraint, the
5411 assembler can parse two relocations for a single instruction.
5413 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5415 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
5416 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
5418 2017-08-31 Olivier Hainque <hainque@adacore.com>
5420 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
5421 match as powerpc-wrs-vxworks*.
5423 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
5425 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
5426 register constraint for by-element operand.
5427 (aarch64_mls_elt_merge<mode>): Likewise.
5429 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
5431 * config/arc/arc.c (arc_can_follow_jump): Check for short
5434 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
5436 * config.gcc: Use g.opt for arc.
5437 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
5438 functionality moved to ...
5439 (legitimate_scaled_address_p): New function, ...here.
5440 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
5441 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
5442 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
5444 (arc_override_options): Handle G option.
5445 (arc_output_pic_addr_const): Correct function definition.
5446 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
5447 (arc_decl_anon_ns_mem_p): Delete.
5448 (arc_in_small_data_p): Overhaul this function to take into
5449 consideration the value given via G option.
5450 (arc_rewrite_small_data_1): Renamed and corrected old
5451 arc_rewrite_small_data function.
5452 (arc_rewrite_small_data): New function.
5453 (small_data_pattern): Don't use pic_offset_table_rtx.
5454 * config/arc/arc.h (CC1_SPEC): Recognize G option.
5455 * config/arc/simdext.md (movmisalignv2hi): Use
5456 prepare_move_operands function.
5458 (movmisalign*): Likewise.
5459 * doc/invoke.texi (ARC options): Document -G option.
5461 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
5463 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
5465 * config/arc/arc.c (arc_print_operand): Output scalled address for
5466 sdata whenever is possible.
5467 (arc_in_small_data_p): Allow sdata for 64bit datum when double
5468 load/stores are available.
5469 (compact_sda_memory_operand): Check for the alignment required by
5470 code density instructions.
5471 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
5473 * config/arc/constraints.md (Usd): Update constraint.
5474 (Us0): New constraint.
5475 (Usc): Update constraint.
5477 2017-08-31 Richard Biener <rguenther@suse.de>
5480 * dwarf2out.c (dwarf2out_early_global_decl): Process each
5483 2017-08-31 Tamar Christina <tamar.christina@arm.com>
5485 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
5486 Resize type_signature.
5488 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
5489 Alan Hayward <alan.hayward@arm.com>
5490 David Sherwood <david.sherwood@arm.com>
5492 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
5493 subregs whose inner modes can be stored in GPRs.
5494 (aarch64_classify_index): Likewise.
5496 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
5497 Alan Hayward <alan.hayward@arm.com>
5498 David Sherwood <david.sherwood@arm.com>
5500 * config/aarch64/iterators.md (V_cmp_result): Rename to...
5501 (V_INT_EQUIV): ...this.
5502 (v_cmp_result): Rename to...
5503 (v_int_equiv): ...this.
5504 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
5505 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
5506 (copysign<mode>3): Likewise.
5507 (aarch64_simd_bsl<mode>_internal): Likewise.
5508 (aarch64_simd_bsl<mode>): Likewise.
5509 (vec_cmp<mode><mode>): Likewise.
5510 (vcond<mode><mode>): Likewise.
5511 (vcond<v_cmp_mixed><mode>): Likewise.
5512 (vcondu<mode><v_cmp_mixed>): Likewise.
5513 (aarch64_cm<optab><mode>): Likewise.
5514 (aarch64_cmtst<mode>): Likewise.
5515 (aarch64_fac<optab><mode>): Likewise.
5516 (vec_perm_const<mode>): Likewise.
5517 (vcond_mask_<mode><v_cmp_result>): Rename to...
5518 (vcond_mask_<mode><v_int_equiv>): ...this.
5519 (vec_cmp<mode><v_cmp_result>): Rename to...
5520 (vec_cmp<mode><v_int_equiv>): ...this.
5522 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
5523 Alan Hayward <alan.hayward@arm.com>
5524 David Sherwood <david.sherwood@arm.com>
5526 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
5528 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
5529 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
5530 (UNSPEC_LD4_DREG): New unspecs.
5531 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
5532 (aarch64_ld2<mode>_dreg_be): Replace with...
5533 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
5535 (aarch64_ld3<mode>_dreg_le)
5536 (aarch64_ld3<mode>_dreg_be): Replace with...
5537 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
5539 (aarch64_ld4<mode>_dreg_le)
5540 (aarch64_ld4<mode>_dreg_be): Replace with...
5541 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
5544 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5546 PR tree-optimization/81987
5547 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
5548 insert an initializer in a location not dominated by the stride
5551 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
5553 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
5554 on the entire header of the finally block in the fallthru case.
5556 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
5558 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
5560 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
5562 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
5563 rs6000_emit_move_from_cr and call renamed function.
5564 (rs6000_emit_prologue): Call renamed functions.
5565 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
5566 movesi_from_cr, remove volatile CRs.
5568 2017-08-30 Jon Beniston <jon@beniston.com>
5569 Richard Biener <rguenther@suse.de>
5571 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
5572 of VECTOR_MODE_P check.
5573 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
5574 element vector types.
5576 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5578 * df.h (df_read_modify_subreg_p): Remove in favor of...
5579 * rtl.h (read_modify_subreg_p): ...this new function. Take a
5580 const_rtx instead of an rtx.
5581 * cprop.c (local_cprop_find_used_regs): Update accordingly.
5582 * df-problems.c (df_word_lr_mark_ref): Likewise.
5583 * ira-lives.c (mark_pseudo_reg_live): Likewise.
5584 (mark_pseudo_reg_dead): Likewise.
5585 (mark_ref_dead): Likewise.
5586 * reginfo.c (init_subregs_of_mode): Likewise.
5587 * sched-deps.c (sched_analyze_1): Likewise.
5588 * df-scan.c (df_def_record_1): Likewise.
5589 (df_uses_record): Likewise.
5590 (df_read_modify_subreg_p): Remove in favor of...
5591 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
5592 const_rtx instead of an rtx.
5594 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5595 Alan Hayward <alan.hayward@arm.com>
5596 David Sherwood <david.sherwood@arm.com>
5598 * rtl.h (partial_subreg_p): New function.
5599 * caller-save.c (save_call_clobbered_regs): Use it.
5600 * calls.c (expand_call): Likewise.
5601 * combine.c (combinable_i3pat): Likewise.
5602 (simplify_set): Likewise.
5603 (make_extraction): Likewise.
5604 (make_compound_operation_int): Likewise.
5605 (gen_lowpart_or_truncate): Likewise.
5606 (force_to_mode): Likewise.
5607 (make_field_assignment): Likewise.
5608 (reg_truncated_to_mode): Likewise.
5609 (record_truncated_value): Likewise.
5610 (move_deaths): Likewise.
5611 * cse.c (record_jump_cond): Likewise.
5612 (cse_insn): Likewise.
5613 * cselib.c (cselib_lookup_1): Likewise.
5614 * expmed.c (extract_bit_field_using_extv): Likewise.
5615 * function.c (assign_parm_setup_reg): Likewise.
5616 * ifcvt.c (noce_convert_multiple_sets): Likewise.
5617 * ira-build.c (create_insn_allocnos): Likewise.
5618 * lra-coalesce.c (merge_pseudos): Likewise.
5619 * lra-constraints.c (match_reload): Likewise.
5620 (simplify_operand_subreg): Likewise.
5621 (curr_insn_transform): Likewise.
5622 * lra-lives.c (process_bb_lives): Likewise.
5623 * lra.c (new_insn_reg): Likewise.
5624 (lra_substitute_pseudo): Likewise.
5625 * regcprop.c (mode_change_ok): Likewise.
5626 (maybe_mode_change): Likewise.
5627 (copyprop_hardreg_forward_1): Likewise.
5628 * reload.c (push_reload): Likewise.
5629 (find_reloads): Likewise.
5630 (find_reloads_subreg_address): Likewise.
5631 * reload1.c (alter_reg): Likewise.
5632 (eliminate_regs_1): Likewise.
5633 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5635 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
5637 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
5638 back to if statements, including unpack.
5640 2017-08-30 Martin Liska <mliska@suse.cz>
5643 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
5645 (func_checker::compare_asm_inputs_outputs): ... this function.
5646 (func_checker::compare_gimple_asm): Use the function to compare
5647 also ASM constrains.
5648 * ipa-icf-gimple.h: Rename the function.
5650 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5651 Alan Hayward <alan.hayward@arm.com>
5652 David Sherwood <david.sherwood@arm.com>
5654 * coretypes.h (complex_mode): New type.
5655 * gdbhooks.py (build_pretty_printer): Handle it.
5656 * machmode.h (complex_mode): New class.
5657 (complex_mode::includes_p): New function.
5658 (is_complex_int_mode): Likewise.
5659 (is_complex_float_mode): Likewise.
5660 * genmodes.c (get_mode_class): Handle complex mode classes.
5661 * function.c (expand_function_end): Use is_complex_int_mode.
5663 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5664 Alan Hayward <alan.hayward@arm.com>
5665 David Sherwood <david.sherwood@arm.com>
5667 * coretypes.h (scalar_mode_pod): New typedef.
5668 * gdbhooks.py (build_pretty_printer): Handle it.
5669 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
5670 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
5671 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
5672 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
5675 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5676 Alan Hayward <alan.hayward@arm.com>
5677 David Sherwood <david.sherwood@arm.com>
5679 * machmode.h (mode_for_vector): Take a scalar_mode instead
5681 * stor-layout.c (mode_for_vector): Likewise.
5682 * explow.c (promote_mode): Use as_a <scalar_mode>.
5683 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
5685 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5686 Alan Hayward <alan.hayward@arm.com>
5687 David Sherwood <david.sherwood@arm.com>
5689 * target.def (preferred_simd_mode): Take a scalar_mode
5690 instead of a machine_mode.
5691 * targhooks.h (default_preferred_simd_mode): Likewise.
5692 * targhooks.c (default_preferred_simd_mode): Likewise.
5693 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
5694 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
5695 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
5696 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
5697 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
5698 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
5699 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
5700 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
5702 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
5703 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
5704 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
5705 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
5706 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
5707 * doc/tm.texi: Regenerate.
5708 * optabs-query.c (can_vec_mask_load_store_p): Return false for
5711 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5712 Alan Hayward <alan.hayward@arm.com>
5713 David Sherwood <david.sherwood@arm.com>
5715 * target.def (scalar_mode_supported_p): Take a scalar_mode
5716 instead of a machine_mode.
5717 * targhooks.h (default_scalar_mode_supported_p): Likewise.
5718 * targhooks.c (default_scalar_mode_supported_p): Likewise.
5719 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
5720 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
5721 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
5722 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
5723 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
5724 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
5725 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
5726 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
5727 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
5728 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
5729 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
5730 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
5732 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
5733 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
5734 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
5735 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
5736 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
5738 * doc/tm.texi: Regenerate.
5740 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5741 Alan Hayward <alan.hayward@arm.com>
5742 David Sherwood <david.sherwood@arm.com>
5744 * coretypes.h (opt_scalar_mode): New typedef.
5745 * gdbhooks.py (build_pretty_printers): Handle it.
5746 * machmode.h (mode_iterator::get_2xwider): Add overload for
5748 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
5750 * expr.c (convert_mode_scalar): Likewise.
5751 * omp-low.c (omp_clause_aligned_alignment): Likewise.
5752 * optabs.c (expand_float): Likewise.
5753 (expand_fix): Likewise.
5754 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5756 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5757 Alan Hayward <alan.hayward@arm.com>
5758 David Sherwood <david.sherwood@arm.com>
5760 * optabs.c (expand_float): Explicitly check for scalars before
5761 using a branching expansion.
5762 (expand_fix): Likewise.
5764 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5765 Alan Hayward <alan.hayward@arm.com>
5766 David Sherwood <david.sherwood@arm.com>
5768 * expr.c (convert_mode): Split scalar handling out into...
5769 (convert_mode_scalar): ...this new function. Treat the modes
5772 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5773 Alan Hayward <alan.hayward@arm.com>
5774 David Sherwood <david.sherwood@arm.com>
5776 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
5778 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
5780 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5781 Alan Hayward <alan.hayward@arm.com>
5782 David Sherwood <david.sherwood@arm.com>
5784 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
5785 rather than a machine_mode.
5786 (fixed_from_string): Likewise.
5787 (fixed_convert): Likewise.
5788 (fixed_convert_from_int): Likewise.
5789 (fixed_convert_from_real): Likewise.
5790 (real_convert_from_fixed): Likewise.
5791 * fixed-value.c (fixed_from_double_int): Likewise.
5792 (fixed_from_string): Likewise.
5793 (fixed_convert): Likewise.
5794 (fixed_convert_from_int): Likewise.
5795 (fixed_convert_from_real): Likewise.
5796 (real_convert_from_fixed): Likewise.
5797 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
5799 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5800 Alan Hayward <alan.hayward@arm.com>
5801 David Sherwood <david.sherwood@arm.com>
5803 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
5804 of separate mode class checks. Do not allow vector modes here.
5805 (immed_wide_int_const): Use as_a <scalar_mode>.
5806 * explow.c (trunc_int_for_mode): Likewise.
5807 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
5808 (wi::shwi): Likewise.
5809 (wi::min_value): Likewise.
5810 (wi::max_value): Likewise.
5811 * dwarf2out.c (loc_descriptor): Likewise.
5812 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
5815 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5816 Alan Hayward <alan.hayward@arm.com>
5817 David Sherwood <david.sherwood@arm.com>
5819 * tree.h (SCALAR_TYPE_MODE): New macro.
5820 * expr.c (expand_expr_addr_expr_1): Use it.
5821 (expand_expr_real_2): Likewise.
5822 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
5823 (fold_convert_const_fixed_from_int): Likewise.
5824 (fold_convert_const_fixed_from_real): Likewise.
5825 (native_encode_fixed): Likewise
5826 (native_encode_complex): Likewise
5827 (native_encode_vector): Likewise.
5828 (native_interpret_fixed): Likewise.
5829 (native_interpret_real): Likewise.
5830 (native_interpret_complex): Likewise.
5831 (native_interpret_vector): Likewise.
5832 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
5833 (simd_clone_adjust_argument_types): Likewise.
5834 (simd_clone_init_simd_arrays): Likewise.
5835 (simd_clone_adjust): Likewise.
5836 * stor-layout.c (layout_type): Likewise.
5837 * tree.c (build_minus_one_cst): Likewise.
5838 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
5839 * tree-inline.c (estimate_move_cost): Likewise.
5840 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
5841 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
5842 (vectorizable_reduction): Likewise.
5843 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
5844 (vect_recog_mixed_size_cond_pattern): Likewise.
5845 (check_bool_pattern): Likewise.
5846 (adjust_bool_pattern): Likewise.
5847 (search_type_for_mask_1): Likewise.
5848 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
5849 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5850 (vectorizable_load): Likewise.
5851 (vectorizable_store): Likewise.
5852 * ubsan.c (ubsan_encode_value): Likewise.
5853 * varasm.c (output_constant): Likewise.
5855 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5856 Alan Hayward <alan.hayward@arm.com>
5857 David Sherwood <david.sherwood@arm.com>
5859 * coretypes.h (scalar_mode): New class.
5860 * machmode.h (scalar_mode): Likewise.
5861 (scalar_mode::includes_p): New function.
5862 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
5863 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
5864 * genmodes.c (get_mode_class): Handle remaining scalar modes.
5865 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
5866 * expmed.c (store_bit_field_1): Likewise.
5867 (extract_bit_field_1): Likewise.
5868 * expr.c (write_complex_part): Likewise.
5869 (read_complex_part): Likewise.
5870 (emit_move_complex_push): Likewise.
5871 (expand_expr_real_2): Likewise.
5872 * function.c (assign_parm_setup_reg): Likewise.
5873 (assign_parms_unsplit_complex): Likewise.
5874 * optabs.c (expand_binop): Likewise.
5875 * rtlanal.c (subreg_get_info): Likewise.
5876 * simplify-rtx.c (simplify_immed_subreg): Likewise.
5877 * varasm.c (output_constant_pool_2): Likewise.
5879 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5880 Alan Hayward <alan.hayward@arm.com>
5881 David Sherwood <david.sherwood@arm.com>
5883 * expmed.c (extract_high_half): Use scalar_int_mode and remove
5885 (expmed_mult_highpart_optab): Likewise.
5886 (expmed_mult_highpart): Likewise.
5888 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5889 Alan Hayward <alan.hayward@arm.com>
5890 David Sherwood <david.sherwood@arm.com>
5892 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
5893 instead of a machine_mode.
5894 (builtin_memset_read_str): Likewise.
5895 * builtins.c (c_readstr): Likewise.
5896 (builtin_memcpy_read_str): Likewise.
5897 (builtin_strncpy_read_str): Likewise.
5898 (builtin_memset_read_str): Likewise.
5899 (builtin_memset_gen_str): Likewise.
5900 (expand_builtin_signbit): Use scalar_int_mode for local variables.
5901 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
5902 instead of a machine_mode.
5903 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
5905 (make_extraction): Likewise.
5906 (try_widen_shift_mode): Take and return scalar_int_modes instead
5908 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
5909 a scalar_int_mode instead of a machine_mode.
5910 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
5911 (avr_addr_space_pointer_mode): Likewise.
5912 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
5913 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
5914 (msp430_unwind_word_mode): Likewise.
5915 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
5916 (spu_addr_space_pointer_mode): Likewise.
5917 (spu_addr_space_address_mode): Likewise.
5918 (spu_libgcc_cmp_return_mode): Likewise.
5919 (spu_libgcc_shift_count_mode): Likewise.
5920 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
5921 (rl78_addr_space_pointer_mode): Likewise.
5922 (fl78_unwind_word_mode): Likewise.
5923 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
5925 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
5926 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
5927 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
5928 (mips_valid_pointer_mode): Likewise.
5929 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
5930 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
5931 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
5933 (ft32_addr_space_address_mode): Likewise.
5934 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
5935 scalar_int_mode instead of a machine_mode.
5936 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
5938 (m32c_addr_space_address_mode): Likewise.
5939 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
5940 (rs6000_eh_return_filter_mode): Likewise.
5941 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
5942 (rs6000_eh_return_filter_mode): Likewise.
5943 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
5944 (s390_libgcc_shift_count_mode): Likewise.
5945 (s390_unwind_word_mode): Likewise.
5946 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
5948 * target.def (mode_rep_extended): Likewise.
5949 (valid_pointer_mode): Likewise.
5950 (addr_space.valid_pointer_mode): Likewise.
5951 (eh_return_filter_mode): Return a scalar_int_mode rather than
5953 (libgcc_cmp_return_mode): Likewise.
5954 (libgcc_shift_count_mode): Likewise.
5955 (unwind_word_mode): Likewise.
5956 (addr_space.pointer_mode): Likewise.
5957 (addr_space.address_mode): Likewise.
5958 * doc/tm.texi: Regenerate.
5959 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
5961 (do_jump): Use scalar_int_mode for local variables.
5962 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
5963 rather than a machine_mode.
5964 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
5965 (scompare_loc_descriptor_wide): Likewise.
5966 (scompare_loc_descriptor_narrow): Likewise.
5967 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
5969 * except.c (sjlj_emit_dispatch_table): Likewise.
5970 (expand_builtin_eh_copy_values): Likewise.
5971 * explow.c (convert_memory_address_addr_space_1): Likewise.
5972 Take a scalar_int_mode rather than a machine_mode.
5973 (convert_memory_address_addr_space): Take a scalar_int_mode rather
5974 than a machine_mode.
5975 (memory_address_addr_space): Use scalar_int_mode for local variables.
5976 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
5977 rather than a machine_mode.
5978 * expmed.c (mask_rtx): Likewise.
5979 (init_expmed_one_conv): Likewise.
5980 (expand_mult_highpart_adjust): Likewise.
5981 (extract_high_half): Likewise.
5982 (expmed_mult_highpart_optab): Likewise.
5983 (expmed_mult_highpart): Likewise.
5984 (expand_smod_pow2): Likewise.
5985 (expand_sdiv_pow2): Likewise.
5986 (emit_store_flag_int): Likewise.
5987 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
5989 (extract_low_bits): Likewise.
5990 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
5992 * expr.c (pieces_addr::adjust): Likewise.
5993 (can_store_by_pieces): Likewise.
5994 (store_by_pieces): Likewise.
5995 (clear_by_pieces_1): Likewise.
5996 (expand_expr_addr_expr_1): Likewise.
5997 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
5998 (expand_expr_real_1): Likewise.
5999 (try_casesi): Likewise.
6000 * final.c (shorten_branches): Likewise.
6001 * fold-const.c (fold_convert_const_int_from_fixed): Change the
6002 type of "mode" to machine_mode.
6003 * internal-fn.c (expand_arith_overflow_result_store): Take a
6004 scalar_int_mode rather than a machine_mode.
6005 (expand_mul_overflow): Use scalar_int_mode for local variables.
6006 * loop-doloop.c (doloop_modify): Likewise.
6007 (doloop_optimize): Likewise.
6008 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
6009 than a machine_mode.
6010 (expand_doubleword_shift_condmove): Likewise.
6011 (expand_doubleword_shift): Likewise.
6012 (expand_doubleword_clz): Likewise.
6013 (expand_doubleword_popcount): Likewise.
6014 (expand_doubleword_parity): Likewise.
6015 (expand_absneg_bit): Use scalar_int_mode for local variables.
6016 (prepare_float_lib_cmp): Likewise.
6017 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
6018 rather than a machine_mode.
6019 (convert_memory_address_addr_space): Likewise.
6020 (get_mode_bounds): Likewise.
6021 (get_address_mode): Return a scalar_int_mode rather than a
6023 * rtlanal.c (get_address_mode): Likewise.
6024 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
6025 than a machine_mode.
6026 * targhooks.c (default_mode_rep_extended): Likewise.
6027 (default_valid_pointer_mode): Likewise.
6028 (default_addr_space_valid_pointer_mode): Likewise.
6029 (default_eh_return_filter_mode): Return a scalar_int_mode rather
6030 than a machine_mode.
6031 (default_libgcc_cmp_return_mode): Likewise.
6032 (default_libgcc_shift_count_mode): Likewise.
6033 (default_unwind_word_mode): Likewise.
6034 (default_addr_space_pointer_mode): Likewise.
6035 (default_addr_space_address_mode): Likewise.
6036 * targhooks.h (default_eh_return_filter_mode): Likewise.
6037 (default_libgcc_cmp_return_mode): Likewise.
6038 (default_libgcc_shift_count_mode): Likewise.
6039 (default_unwind_word_mode): Likewise.
6040 (default_addr_space_pointer_mode): Likewise.
6041 (default_addr_space_address_mode): Likewise.
6042 (default_mode_rep_extended): Take a scalar_int_mode rather than
6044 (default_valid_pointer_mode): Likewise.
6045 (default_addr_space_valid_pointer_mode): Likewise.
6046 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
6048 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
6049 rather than a machine_mode.
6050 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
6051 for local variables.
6052 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6053 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
6054 than a machine_mode.
6056 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6057 Alan Hayward <alan.hayward@arm.com>
6058 David Sherwood <david.sherwood@arm.com>
6060 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
6061 the mode argument to scalar_int_mode.
6062 (do_jump_by_parts_zero_rtx): Likewise.
6063 (do_jump_by_parts_equality_rtx): Likewise.
6064 (do_jump_by_parts_greater): Take a mode argument.
6065 (do_jump_by_parts_equality): Likewise.
6066 (do_jump_1): Update calls accordingly.
6068 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6069 Alan Hayward <alan.hayward@arm.com>
6070 David Sherwood <david.sherwood@arm.com>
6072 * is-a.h (safe_dyn_cast): New function.
6073 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
6074 (jump_table_for_label): Likewise.
6075 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
6076 instead of an rtx_insn *.
6077 (shorten_branches): Use dyn_cast instead of LABEL_P and
6078 JUMP_TABLE_DATA_P. Use jump_table_for_label and
6079 rtx_jump_table_data::get_data_mode.
6080 (final_scan_insn): Likewise.
6082 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6083 Alan Hayward <alan.hayward@arm.com>
6084 David Sherwood <david.sherwood@arm.com>
6086 * combine.c (try_combine): Use is_a <scalar_int_mode> when
6087 trying to combine a full-register integer set with a subreg
6090 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6091 Alan Hayward <alan.hayward@arm.com>
6092 David Sherwood <david.sherwood@arm.com>
6094 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
6095 that is always either address_mode or pointer_mode.
6097 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6098 Alan Hayward <alan.hayward@arm.com>
6099 David Sherwood <david.sherwood@arm.com>
6101 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
6102 when the two are known to be equal.
6104 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6105 Alan Hayward <alan.hayward@arm.com>
6106 David Sherwood <david.sherwood@arm.com>
6108 * simplify-rtx.c (simplify_const_unary_operation): Use
6109 is_a <scalar_int_mode> instead of checking for a nonzero
6110 precision. Forcibly convert op_mode to a scalar_int_mode
6111 in that case. More clearly differentiate the operand and
6112 result modes and use the former when deciding what the value
6113 of a count-bits operation should be. Use is_int_mode instead
6114 of checking for a MODE_INT. Remove redundant check for whether
6115 this mode has a zero precision.
6117 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6118 Alan Hayward <alan.hayward@arm.com>
6119 David Sherwood <david.sherwood@arm.com>
6121 * optabs.c (widen_leading): Change the type of the mode argument
6122 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
6123 (widen_bswap): Likewise.
6124 (expand_parity): Likewise.
6125 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
6126 (expand_ffs): Likewise.
6127 (epand_unop): Check for scalar integer modes before calling the
6130 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6131 Alan Hayward <alan.hayward@arm.com>
6132 David Sherwood <david.sherwood@arm.com>
6134 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
6136 (expand_expr_real_1): Update call accordingly.
6138 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6139 Alan Hayward <alan.hayward@arm.com>
6140 David Sherwood <david.sherwood@arm.com>
6142 * expmed.c (store_bit_field_using_insv): Add op0_mode and
6143 value_mode arguments. Use scalar_int_mode internally.
6144 (store_bit_field_1): Rename the new integer mode from imode
6145 to op0_mode and use it instead of GET_MODE (op0). Update calls
6146 to store_split_bit_field, store_bit_field_using_insv and
6147 store_fixed_bit_field.
6148 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
6149 Use scalar_int_mode internally. Use a bit count rather than a mode
6150 when calculating the largest bit size for get_best_mode.
6151 Update calls to store_split_bit_field and store_fixed_bit_field_1.
6152 (store_fixed_bit_field_1): Add mode and value_mode arguments.
6153 Remove assertion that OP0 has a scalar integer mode.
6154 (store_split_bit_field): Add op0_mode and value_mode arguments.
6155 Update calls to extract_fixed_bit_field.
6156 (extract_bit_field_using_extv): Add an op0_mode argument.
6157 Use scalar_int_mode internally.
6158 (extract_bit_field_1): Rename the new integer mode from imode to
6159 op0_mode and use it instead of GET_MODE (op0). Update calls to
6160 extract_split_bit_field, extract_bit_field_using_extv and
6161 extract_fixed_bit_field.
6162 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
6163 to extract_split_bit_field and extract_fixed_bit_field_1.
6164 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
6165 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
6167 (extract_split_bit_field): Add an op0_mode argument. Update call
6168 to extract_fixed_bit_field.
6170 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6171 Alan Hayward <alan.hayward@arm.com>
6172 David Sherwood <david.sherwood@arm.com>
6174 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
6175 * explow.c (hard_function_value): Likewise.
6176 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
6177 convert_to_mode call outside the loop.
6178 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
6179 for the mode iterator. Require the mode specified by max_pieces
6181 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
6183 (copy_blkmode_to_reg): Likewise.
6184 (set_storage_via_setmem): Likewise.
6185 * optabs.c (prepare_cmp_insn): Likewise.
6186 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6187 * stor-layout.c (finish_bitfield_representative): Likewise.
6189 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6190 Alan Hayward <alan.hayward@arm.com>
6191 David Sherwood <david.sherwood@arm.com>
6193 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
6194 * expr.c (convert_move): Use them.
6195 (convert_modes): Likewise.
6196 (store_expr_with_bounds): Likewise.
6198 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6199 Alan Hayward <alan.hayward@arm.com>
6200 David Sherwood <david.sherwood@arm.com>
6202 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
6203 parameter for the mode of "x". Remove the "known_x", "known_mode"
6204 and "known_ret" arguments. Change the type of the mode argument
6206 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
6207 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
6208 (reg_num_sign_bit_copies_for_combine): Likewise.
6209 * rtlanal.c (nonzero_bits1): Likewise.
6210 (num_sign_bit_copies1): Likewise.
6211 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
6212 (reg_num_sign_bit_copies_general): Likewise.
6213 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
6214 (reg_nonzero_bits_general): Likewise.
6216 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6217 Alan Hayward <alan.hayward@arm.com>
6218 David Sherwood <david.sherwood@arm.com>
6220 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
6221 than in subroutines. Return 1 for non-integer modes.
6222 (cached_num_sign_bit_copies): Change the type of the mode parameter
6224 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
6225 classes. Handle CONST_INT_P first and then check whether X also
6226 has a scalar integer mode. Check the same thing for inner registers
6227 of a SUBREG and for values that are being extended or truncated.
6229 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6230 Alan Hayward <alan.hayward@arm.com>
6231 David Sherwood <david.sherwood@arm.com>
6233 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
6234 in subroutines. Return the mode mask for non-integer modes.
6235 (cached_nonzero_bits): Change the type of the mode parameter
6237 (nonzero_bits1): Likewise. Remove early exit for other mode
6238 classes. Handle CONST_INT_P first and then check whether X
6239 also has a scalar integer mode.
6241 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6242 Alan Hayward <alan.hayward@arm.com>
6243 David Sherwood <david.sherwood@arm.com>
6245 * expr.c (widest_int_mode_for_size): Make the comment match the code.
6246 Return a scalar_int_mode and assert that the size is greater than
6248 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
6249 (op_by_pieces_d::op_by_pieces_d): Likewise.
6250 (op_by_pieces_d::run): Likewise.
6251 (can_store_by_pieces): 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 (extract_left_shift): Add a mode argument and update
6259 (make_compound_operation_int): Change the type of the mode parameter
6260 to scalar_int_mode and update the call to extract_left_shift.
6262 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6263 Alan Hayward <alan.hayward@arm.com>
6264 David Sherwood <david.sherwood@arm.com>
6266 * combine.c (simplify_and_const_int): Change the type of the mode
6267 parameter to scalar_int_mode.
6268 (simplify_and_const_int_1): Likewise. Update recursive call.
6270 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6271 Alan Hayward <alan.hayward@arm.com>
6272 David Sherwood <david.sherwood@arm.com>
6274 * combine.c (simplify_compare_const): Check that the mode is a
6275 scalar_int_mode (rather than VOIDmode) before testing its
6277 (simplify_comparison): Move COMPARISON_P handling out of the
6278 loop and restrict the latter part of the loop to scalar_int_modes.
6279 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
6280 and when considering SUBREG_REGs. Use is_int_mode instead of
6281 checking GET_MODE_CLASS against MODE_INT.
6283 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6284 Alan Hayward <alan.hayward@arm.com>
6285 David Sherwood <david.sherwood@arm.com>
6287 * combine.c (try_widen_shift_mode): Move check for equal modes to...
6288 (simplify_shift_const_1): ...here. Use scalar_int_mode for
6289 shift_unit_mode and for modes involved in scalar shifts.
6291 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6292 Alan Hayward <alan.hayward@arm.com>
6293 David Sherwood <david.sherwood@arm.com>
6295 * combine.c (force_int_to_mode): New function, split out from...
6296 (force_to_mode): ...here. Keep xmode up-to-date and use it
6297 instead of GET_MODE (x).
6299 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6300 Alan Hayward <alan.hayward@arm.com>
6301 David Sherwood <david.sherwood@arm.com>
6303 * optabs-query.h (extraction_insn::struct_mode): Change type to
6304 opt_scalar_int_mode and update comment.
6305 (extraction_insn::field_mode): Change type to scalar_int_mode.
6306 (extraction_insn::pos_mode): Likewise.
6307 * combine.c (make_extraction): Update accordingly.
6308 * optabs-query.c (get_traditional_extraction_insn): Likewise.
6309 (get_optab_extraction_insn): Likewise.
6310 * recog.c (simplify_while_replacing): Likewise.
6311 * expmed.c (narrow_bit_field_mem): Change the type of the mode
6312 parameter to opt_scalar_int_mode.
6314 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6315 Alan Hayward <alan.hayward@arm.com>
6316 David Sherwood <david.sherwood@arm.com>
6318 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
6319 to a scalar_int_mode instead of a machine_mode.
6320 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
6321 (get_best_mode): Return a boolean and use a pointer argument to store
6322 the selected mode. Replace the limit mode parameter with a bit limit.
6323 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
6324 for the values returned by bit_field_mode_iterator::next_mode.
6325 (store_bit_field): Update call to get_best_mode.
6326 (store_fixed_bit_field): Likewise.
6327 (extract_fixed_bit_field): Likewise.
6328 * expr.c (optimize_bitfield_assignment_op): Likewise.
6329 * fold-const.c (optimize_bit_field_compare): Likewise.
6330 (fold_truth_andor_1): Likewise.
6331 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
6332 Update for new type of m_mode.
6333 (get_best_mode): As above.
6335 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6336 Alan Hayward <alan.hayward@arm.com>
6337 David Sherwood <david.sherwood@arm.com>
6339 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
6340 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
6341 (store_bit_field): Check is_a <scalar_int_mode> before calling
6342 strict_volatile_bitfield_p.
6343 (extract_bit_field): Likewise.
6345 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6346 Alan Hayward <alan.hayward@arm.com>
6347 David Sherwood <david.sherwood@arm.com>
6349 * target.def (cstore_mode): Return a scalar_int_mode.
6350 * doc/tm.texi: Regenerate.
6351 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
6352 * targhooks.h (default_cstore_mode): Likewise.
6353 * targhooks.c (default_cstore_mode): Likewise, using a forced
6355 * expmed.c (emit_cstore): Expect the target of the cstore to be
6358 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6359 Alan Hayward <alan.hayward@arm.com>
6360 David Sherwood <david.sherwood@arm.com>
6362 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
6364 (niter_desc): Likewise mode.
6365 (iv_analyze): Add a mode parameter.
6367 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
6368 and change its type to scalar_int_mode.
6369 * loop-iv.c: Update commentary at head of file.
6370 (iv_constant): Pass the mode paraeter before the rtx it describes
6371 and change its type to scalar_int_mode. Remove VOIDmode handling.
6372 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
6373 (iv_extend): Likewise.
6374 (shorten_into_mode): Likewise.
6375 (iv_add): Use scalar_int_mode.
6376 (iv_mult): Likewise.
6377 (iv_shift): Likewise.
6378 (canonicalize_iv_subregs): Likewise.
6379 (get_biv_step_1): Pass the outer_mode parameter before the rtx
6380 it describes and change its mode to scalar_int_mode. Also change
6381 the type of the returned inner_mode to scalar_int_mode.
6382 (get_biv_step): Likewise, turning outer_mode from a pointer
6383 into a direct parameter. Update call to get_biv_step_1.
6384 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
6385 iv_constant and get_biv_step.
6386 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
6387 and change its type to scalar_int_mode. Don't initialise iv->mode
6388 to VOIDmode and remove later checks for its still being VOIDmode.
6389 Update calls to iv_analyze_op and iv_analyze_expr. Check
6390 is_a <scalar_int_mode> when changing the mode under consideration.
6391 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
6392 Update call to iv_analyze_expr.
6393 (iv_analyze_op): Add a mode parameter. Reject subregs whose
6394 inner register is not also a scalar_int_mode. Update call to
6396 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
6397 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
6398 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
6399 separate mode class checks. Update calls to iv_analyze. Remove
6400 fix-up of VOIDmodes after iv_analyze_biv.
6401 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
6402 don't have a scalar_int_mode. Update call to biv_p.
6404 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6405 Alan Hayward <alan.hayward@arm.com>
6406 David Sherwood <david.sherwood@arm.com>
6408 * cfgexpand.c (convert_debug_memory_address): Use
6409 as_a <scalar_int_mode>.
6410 * combine.c (expand_compound_operation): Likewise.
6411 (make_extraction): Likewise.
6412 (change_zero_ext): Likewise.
6413 (simplify_comparison): Likewise.
6414 * cse.c (cse_insn): Likewise.
6415 * dwarf2out.c (minmax_loc_descriptor): Likewise.
6416 (mem_loc_descriptor): Likewise.
6417 (loc_descriptor): Likewise.
6418 * expmed.c (init_expmed_one_mode): Likewise.
6419 (synth_mult): Likewise.
6420 (emit_store_flag_1): Likewise.
6421 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
6422 of a comparison with size.
6423 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
6424 (reduce_to_bit_field_precision): Likewise.
6425 * function.c (expand_function_end): Likewise.
6426 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
6427 * loop-doloop.c (doloop_modify): Likewise.
6428 * optabs.c (expand_binop): Likewise.
6429 (expand_unop): Likewise.
6430 (expand_copysign_absneg): Likewise.
6431 (prepare_cmp_insn): Likewise.
6432 (maybe_legitimize_operand): Likewise.
6433 * recog.c (const_scalar_int_operand): Likewise.
6434 * rtlanal.c (get_address_mode): Likewise.
6435 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6436 (simplify_cond_clz_ctz): Likewise.
6437 * tree-nested.c (get_nl_goto_field): Likewise.
6438 * tree.c (build_vector_type_for_mode): Likewise.
6439 * var-tracking.c (use_narrower_mode): Likewise.
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 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
6446 * builtins.c (expand_builtin_signbit): Use it.
6447 * cfgexpand.c (expand_debug_expr): Likewise.
6448 * dojump.c (do_jump): Likewise.
6449 (do_compare_and_jump): Likewise.
6450 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
6451 * expmed.c (make_tree): Likewise.
6452 * expr.c (expand_expr_real_2): Likewise.
6453 (expand_expr_real_1): Likewise.
6454 (try_casesi): Likewise.
6455 * fold-const-call.c (fold_const_call_ss): Likewise.
6456 * fold-const.c (unextend): Likewise.
6457 (extract_muldiv_1): Likewise.
6458 (fold_single_bit_test): Likewise.
6459 (native_encode_int): Likewise.
6460 (native_encode_string): Likewise.
6461 (native_interpret_int): Likewise.
6462 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
6463 * internal-fn.c (expand_addsub_overflow): Likewise.
6464 (expand_neg_overflow): Likewise.
6465 (expand_mul_overflow): Likewise.
6466 (expand_arith_overflow): Likewise.
6467 * match.pd: Likewise.
6468 * stor-layout.c (layout_type): Likewise.
6469 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
6470 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
6471 * tree-ssanames.c (get_range_info): Likewise.
6472 * tree-switch-conversion.c (array_value_type) Likewise.
6473 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
6474 (vect_recog_divmod_pattern): Likewise.
6475 (vect_recog_mixed_size_cond_pattern): Likewise.
6476 * tree-vrp.c (extract_range_basic): Likewise.
6477 (simplify_float_conversion_using_ranges): Likewise.
6478 * tree.c (int_fits_type_p): Likewise.
6479 * ubsan.c (instrument_bool_enum_load): Likewise.
6480 * varasm.c (mergeable_string_section): Likewise.
6481 (narrowing_initializer_constant_valid_p): Likewise.
6482 (output_constant): Likewise.
6484 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6485 Alan Hayward <alan.hayward@arm.com>
6486 David Sherwood <david.sherwood@arm.com>
6488 * machmode.h (NARROWEST_INT_MODE): New macro.
6489 * expr.c (alignment_for_piecewise_move): Use it instead of
6490 GET_CLASS_NARROWEST_MODE (MODE_INT).
6491 (push_block): Likewise.
6492 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
6494 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6496 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6497 Alan Hayward <alan.hayward@arm.com>
6498 David Sherwood <david.sherwood@arm.com>
6500 * postreload.c (move2add_valid_value_p): Change the type of the
6501 mode parameter to scalar_int_mode.
6502 (move2add_use_add2_insn): Add a mode parameter and use it instead
6504 (move2add_use_add3_insn): Likewise.
6505 (reload_cse_move2add): Update accordingly.
6507 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6508 Alan Hayward <alan.hayward@arm.com>
6509 David Sherwood <david.sherwood@arm.com>
6511 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
6513 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
6514 * optabs.c (expand_unop): 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 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
6521 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
6522 (popcount_loc_descriptor): Likewise.
6523 (bswap_loc_descriptor): Likewise.
6524 (rotate_loc_descriptor): Likewise.
6525 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
6526 calling the functions above.
6528 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6529 Alan Hayward <alan.hayward@arm.com>
6530 David Sherwood <david.sherwood@arm.com>
6532 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
6534 (try_combine): Likewise.
6535 (simplify_if_then_else): Likewise.
6536 * cse.c (cse_insn): Likewise.
6537 * dwarf2out.c (mem_loc_descriptor): Likewise.
6538 * emit-rtl.c (gen_lowpart_common): Likewise.
6539 * simplify-rtx.c (simplify_truncation): Likewise.
6540 (simplify_binary_operation_1): Likewise.
6541 (simplify_const_relational_operation): Likewise.
6542 (simplify_ternary_operation): Likewise.
6543 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
6545 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6546 Alan Hayward <alan.hayward@arm.com>
6547 David Sherwood <david.sherwood@arm.com>
6549 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
6550 * reload.c (push_reload): Likewise.
6551 (find_reloads): Likewise.
6553 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6554 Alan Hayward <alan.hayward@arm.com>
6555 David Sherwood <david.sherwood@arm.com>
6557 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
6558 (make_compound_operation_int): Likewise.
6559 (change_zero_ext): Likewise.
6560 * expr.c (convert_move): Likewise.
6561 (convert_modes): Likewise.
6562 * fwprop.c (forward_propagate_subreg): Likewise.
6563 * loop-iv.c (get_biv_step_1): Likewise.
6564 * optabs.c (widen_operand): Likewise.
6565 * postreload.c (move2add_valid_value_p): Likewise.
6566 * recog.c (simplify_while_replacing): Likewise.
6567 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6568 (simplify_binary_operation_1): Likewise. Remove redundant
6569 mode equality check.
6571 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6572 Alan Hayward <alan.hayward@arm.com>
6573 David Sherwood <david.sherwood@arm.com>
6575 * combine.c (combine_simplify_rtx): Add checks for
6576 is_a <scalar_int_mode>.
6577 (simplify_if_then_else): Likewise.
6578 (make_field_assignment): Likewise.
6579 (simplify_comparison): Likewise.
6580 * ifcvt.c (noce_try_bitop): Likewise.
6581 * loop-invariant.c (canonicalize_address_mult): Likewise.
6582 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6584 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6585 Alan Hayward <alan.hayward@arm.com>
6586 David Sherwood <david.sherwood@arm.com>
6588 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
6589 is_a <scalar_int_mode> instead of != BLKmode.
6591 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6592 Alan Hayward <alan.hayward@arm.com>
6593 David Sherwood <david.sherwood@arm.com>
6595 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
6596 instead of != VOIDmode.
6597 * combine.c (if_then_else_cond): Likewise.
6598 (change_zero_ext): Likewise.
6599 * dwarf2out.c (mem_loc_descriptor): Likewise.
6600 (loc_descriptor): Likewise.
6601 * rtlanal.c (canonicalize_condition): Likewise.
6602 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
6604 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6605 Alan Hayward <alan.hayward@arm.com>
6606 David Sherwood <david.sherwood@arm.com>
6608 * simplify-rtx.c (simplify_binary_operation_1): Use
6609 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
6611 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6612 Alan Hayward <alan.hayward@arm.com>
6613 David Sherwood <david.sherwood@arm.com>
6615 * wide-int.h (int_traits<unsigned char>) New class.
6616 (int_traits<unsigned short>) Likewise.
6617 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
6618 Use GET_MODE_UNIT_PRECISION and remove redundant test for
6620 * combine.c (set_nonzero_bits_and_sign_copies): Use
6621 is_a <scalar_int_mode>.
6622 (find_split_point): Likewise.
6623 (combine_simplify_rtx): Likewise.
6624 (simplify_logical): Likewise.
6625 (expand_compound_operation): Likewise.
6626 (expand_field_assignment): Likewise.
6627 (make_compound_operation): Likewise.
6628 (extended_count): Likewise.
6629 (change_zero_ext): Likewise.
6630 (simplify_comparison): Likewise.
6631 * dwarf2out.c (scompare_loc_descriptor): Likewise.
6632 (ucompare_loc_descriptor): Likewise.
6633 (minmax_loc_descriptor): Likewise.
6634 (mem_loc_descriptor): Likewise.
6635 (loc_descriptor): Likewise.
6636 * expmed.c (init_expmed_one_mode): Likewise.
6637 * lra-constraints.c (lra_constraint_offset): Likewise.
6638 * optabs.c (prepare_libcall_arg): Likewise.
6639 * postreload.c (move2add_note_store): Likewise.
6640 * reload.c (operands_match_p): Likewise.
6641 * rtl.h (load_extend_op): Likewise.
6642 * rtlhooks.c (gen_lowpart_general): Likewise.
6643 * simplify-rtx.c (simplify_truncation): Likewise.
6644 (simplify_unary_operation_1): Likewise.
6645 (simplify_binary_operation_1): Likewise.
6646 (simplify_const_binary_operation): Likewise.
6647 (simplify_const_relational_operation): Likewise.
6648 (simplify_subreg): Likewise.
6649 * stor-layout.c (bitwise_mode_for_mode): Likewise.
6650 * var-tracking.c (adjust_mems): Likewise.
6651 (prepare_call_arguments): Likewise.
6653 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6654 Alan Hayward <alan.hayward@arm.com>
6655 David Sherwood <david.sherwood@arm.com>
6657 * machmode.h (is_int_mode): New fuction.
6658 * combine.c (find_split_point): Use it.
6659 (combine_simplify_rtx): Likewise.
6660 (simplify_if_then_else): Likewise.
6661 (simplify_set): Likewise.
6662 (simplify_shift_const_1): Likewise.
6663 (simplify_comparison): Likewise.
6664 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
6665 * cse.c (notreg_cost): Likewise.
6666 (cse_insn): Likewise.
6667 * cselib.c (cselib_lookup_1): Likewise.
6668 * dojump.c (do_jump_1): Likewise.
6669 (do_compare_rtx_and_jump): Likewise.
6670 * dse.c (get_call_args): Likewise.
6671 * dwarf2out.c (rtl_for_decl_init): Likewise.
6672 (native_encode_initializer): Likewise.
6673 * expmed.c (emit_store_flag_1): Likewise.
6674 (emit_store_flag): Likewise.
6675 * expr.c (convert_modes): Likewise.
6676 (store_field): Likewise.
6677 (expand_expr_real_1): Likewise.
6678 * fold-const.c (fold_read_from_constant_string): Likewise.
6679 * gimple-ssa-sprintf.c (get_format_string): Likewise.
6680 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
6681 * optabs.c (expand_binop): Likewise.
6682 (expand_unop): Likewise.
6683 (expand_abs_nojump): Likewise.
6684 (expand_one_cmpl_abs_nojump): Likewise.
6685 * simplify-rtx.c (mode_signbit_p): Likewise.
6686 (val_signbit_p): Likewise.
6687 (val_signbit_known_set_p): Likewise.
6688 (val_signbit_known_clear_p): Likewise.
6689 (simplify_relational_operation_1): Likewise.
6690 * tree.c (vector_type_mode): Likewise.
6692 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6693 Alan Hayward <alan.hayward@arm.com>
6694 David Sherwood <david.sherwood@arm.com>
6696 * machmode.h (smallest_mode_for_size): Fix formatting.
6697 (smallest_int_mode_for_size): New function.
6698 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
6699 instead of smallest_mode_for_size.
6700 * combine.c (make_extraction): Likewise.
6701 * config/arc/arc.c (arc_expand_movmem): Likewise.
6702 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
6703 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
6704 * config/s390/s390.c (s390_expand_insv): Likewise.
6705 * config/sparc/sparc.c (assign_int_registers): Likewise.
6706 * config/spu/spu.c (spu_function_value): Likewise.
6707 (spu_function_arg): Likewise.
6708 * coverage.c (get_gcov_type): Likewise.
6709 (get_gcov_unsigned_t): Likewise.
6710 * dse.c (find_shift_sequence): Likewise.
6711 * expmed.c (store_bit_field_1): Likewise.
6712 * expr.c (convert_move): Likewise.
6713 (store_field): Likewise.
6714 * internal-fn.c (expand_arith_overflow): Likewise.
6715 * optabs-query.c (get_best_extraction_insn): Likewise.
6716 * optabs.c (expand_twoval_binop_libfunc): Likewise.
6717 * stor-layout.c (layout_type): Likewise.
6718 (initialize_sizetypes): Likewise.
6719 * targhooks.c (default_get_mask_mode): Likewise.
6720 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
6722 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6723 Alan Hayward <alan.hayward@arm.com>
6724 David Sherwood <david.sherwood@arm.com>
6726 * machmode.h (opt_mode::else_blk): New function.
6727 (int_mode_for_mode): Declare.
6728 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
6729 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
6731 * cfgexpand.c (expand_debug_expr): Likewise.
6732 * combine.c (gen_lowpart_or_truncate): Likewise.
6733 (gen_lowpart_for_combine): Likewise.
6734 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
6735 * config/avr/avr.c (avr_to_int_mode): Likewise.
6736 (avr_out_plus_1): Likewise.
6737 (avr_out_plus): Likewise.
6738 (avr_out_round): Likewise.
6739 * config/i386/i386.c (ix86_split_to_parts): Likewise.
6740 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
6741 (s390_expand_vcond): Likewise.
6742 * config/spu/spu.c (spu_split_immediate): Likewise.
6743 (spu_expand_mov): Likewise.
6744 * dse.c (get_stored_val): Likewise.
6745 * expmed.c (store_bit_field_1): Likewise.
6746 (convert_extracted_bit_field): Use int_mode_for_mode instead of
6748 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
6749 (extract_low_bits): Likewise.
6750 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
6751 handling rather than repeating the check.
6752 (emit_group_store): Likewise.
6753 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
6754 * optabs.c (expand_absneg_bit): Likewise.
6755 (expand_copysign_absneg): Likewise.
6756 (expand_copysign_bit): Likewise.
6757 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
6758 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
6759 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
6760 * var-tracking.c (prepare_call_arguments): Likewise.
6761 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
6762 int_mode_for_mode instead of mode_for_size.
6763 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
6765 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6766 Alan Hayward <alan.hayward@arm.com>
6767 David Sherwood <david.sherwood@arm.com>
6769 * machmode.h (int_mode_for_size): New function.
6770 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
6771 instead of mode_for_size.
6772 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
6774 * combine.c (expand_field_assignment): Use int_mode_for_size
6775 instead of mode_for_size.
6776 (make_extraction): Likewise.
6777 (simplify_shift_const_1): Likewise.
6778 (simplify_comparison): Likewise.
6779 * dojump.c (do_jump): Likewise.
6780 * dwarf2out.c (mem_loc_descriptor): Likewise.
6781 * emit-rtl.c (init_derived_machine_modes): Likewise.
6782 * expmed.c (flip_storage_order): Likewise.
6783 (convert_extracted_bit_field): Likewise.
6784 * expr.c (copy_blkmode_from_reg): Likewise.
6785 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
6786 * internal-fn.c (expand_mul_overflow): Likewise.
6787 * lower-subreg.c (simple_move): Likewise.
6788 * optabs-libfuncs.c (init_optabs): Likewise.
6789 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6790 * tree.c (vector_type_mode): Likewise.
6791 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
6792 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
6793 * tree-vect-generic.c (expand_vector_parallel): Likewise.
6794 * tree-vect-stmts.c (vectorizable_load): Likewise.
6795 (vectorizable_store): Likewise.
6797 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6798 Alan Hayward <alan.hayward@arm.com>
6799 David Sherwood <david.sherwood@arm.com>
6801 * coretypes.h (pod_mode): New type.
6802 (scalar_int_mode_pod): New typedef.
6803 * machmode.h (pod_mode): New class.
6804 (int_n_data_t::m): Change type to scalar_int_mode_pod.
6805 * genmodes.c (emit_mode_int_n): Update accordingly.
6806 * lower-subreg.h (target_lower_subreg): Change type to
6807 scalar_int_mode_pod.
6808 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
6809 scalar_int_mode_pod.
6811 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6812 Alan Hayward <alan.hayward@arm.com>
6813 David Sherwood <david.sherwood@arm.com>
6815 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
6816 machine_mode to scalar_int_mode.
6817 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
6818 (rs6000_option_override_internal): Remove cast to int.
6819 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
6820 machine_mode to scalar_int_mode.
6821 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
6822 (rs6000_option_override_internal): Remove cast to int.
6823 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
6824 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
6826 * config/s390/s390.c (s390_expand_builtin): Likewise.
6827 * coretypes.h (scalar_int_mode): New type.
6828 (opt_scalar_int_mode): New typedef.
6829 * machmode.h (scalar_int_mode): New class.
6830 (scalar_int_mode::includes_p): New function.
6831 (byte_mode): Change type to scalar_int_mode.
6832 (word_mode): Likewise.
6833 (ptr_mode): Likewise.
6834 * emit-rtl.c (byte_mode): Likewise.
6835 (word_mode): Likewise.
6836 (ptr_mode): Likewise.
6837 (init_derived_machine_modes): Update accordingly.
6838 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
6839 and MODE_PARTIAL_INT.
6840 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
6841 opt_scalar_int_mode.
6843 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6844 Alan Hayward <alan.hayward@arm.com>
6845 David Sherwood <david.sherwood@arm.com>
6847 * target.def (libgcc_floating_mode_supported_p): Take a
6849 * doc/tm.texi: Regenerate.
6850 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
6852 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
6853 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
6856 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6857 Alan Hayward <alan.hayward@arm.com>
6858 David Sherwood <david.sherwood@arm.com>
6860 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
6861 * doc/tm.texi: Regenerate.
6862 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
6863 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
6864 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
6865 * targhooks.h (default_floatn_mode): Likewise.
6866 * targhooks.c (default_floatn_mode): Likewise.
6867 * tree.c (build_common_tree_nodes): Update accordingly.
6869 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6870 Alan Hayward <alan.hayward@arm.com>
6871 David Sherwood <david.sherwood@arm.com>
6873 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
6874 (mode_iterator::iterate_p): Likewise.
6875 (mode_iterator::get_wider): Likewise.
6876 * expr.c (init_expr_target): Use opt_scalar_float_mode.
6878 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6879 Alan Hayward <alan.hayward@arm.com>
6880 David Sherwood <david.sherwood@arm.com>
6882 * coretypes.h (opt_scalar_float_mode): New typedef.
6883 * machmode.h (float_mode_for_size): New function.
6884 * emit-rtl.c (double_mode): Delete.
6885 (init_emit_once): Use float_mode_for_size.
6886 * stor-layout.c (layout_type): Likewise.
6887 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
6889 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6890 Alan Hayward <alan.hayward@arm.com>
6891 David Sherwood <david.sherwood@arm.com>
6893 * output.h (assemble_real): Take a scalar_float_mode.
6894 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
6895 * config/arm/arm.md (consttable_4): Likewise.
6896 (consttable_8): Likewise.
6897 (consttable_16): Likewise.
6898 * config/mips/mips.md (consttable_float): Likewise.
6899 * config/s390/s390.c (s390_output_pool_entry): Likewise.
6900 * varasm.c (assemble_real): Take a scalar_float_mode.
6901 (output_constant_pool_2): Update accordingly.
6902 (output_constant): Likewise.
6904 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6905 Alan Hayward <alan.hayward@arm.com>
6906 David Sherwood <david.sherwood@arm.com>
6908 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
6909 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
6910 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
6911 (native_encode_real): Likewise.
6912 (native_interpret_real): Likewise.
6913 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
6914 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6916 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6917 Alan Hayward <alan.hayward@arm.com>
6918 David Sherwood <david.sherwood@arm.com>
6920 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
6921 <scalar_float_mode>. Simplify.
6922 (gen_extend_conv_libfunc): Likewise.
6924 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6925 Alan Hayward <alan.hayward@arm.com>
6926 David Sherwood <david.sherwood@arm.com>
6928 * coretypes.h (scalar_float_mode): New type.
6929 * machmode.h (mode_traits::from_int): Use machine_mode if
6930 USE_ENUM_MODES is defined.
6931 (is_a): New function.
6933 (dyn_cast): Likewise.
6934 (scalar_float_mode): New class.
6935 (scalar_float_mode::includes_p): New function.
6936 (is_float_mode): Likewise.
6937 * gdbhooks.py (MachineModePrinter): New class.
6938 (build_pretty_printer): Use it for scalar_float_mode.
6939 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
6940 (format_helper::format_helper): Turn into a template.
6941 * genmodes.c (get_mode_class): New function.
6942 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
6943 or machine_mode if none.
6944 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
6945 as_a <scalar_float_mode>.
6946 * dwarf2out.c (mem_loc_descriptor): Likewise.
6947 (insert_float): Likewise.
6948 (add_const_value_attribute): Likewise.
6949 * simplify-rtx.c (simplify_immed_subreg): Likewise.
6950 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
6951 (expand_unop): Update accordingly.
6952 (expand_abs_nojump): Likewise.
6953 (expand_copysign_absneg): Take a scalar_float_mode.
6954 (expand_copysign_bit): Likewise.
6955 (expand_copysign): Update accordingly.
6957 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6958 Alan Hayward <alan.hayward@arm.com>
6959 David Sherwood <david.sherwood@arm.com>
6961 * coretypes.h (opt_mode): New class.
6962 * machmode.h (opt_mode): Likewise.
6963 (opt_mode::else_void): New function.
6964 (opt_mode::require): Likewise.
6965 (opt_mode::exists): Likewise.
6966 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
6967 (GET_MODE_2XWIDER_MODE): Likewise.
6968 (mode_iterator::get_wider): Update accordingly.
6969 (mode_iterator::get_2xwider): Likewise.
6970 (mode_iterator::get_known_wider): Likewise, turning into a template.
6971 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
6972 forcing a wider mode to exist.
6973 * config/cr16/cr16.h (LONG_REG_P): Likewise.
6974 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6975 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
6976 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
6977 * lower-subreg.c (init_lower_subreg): Likewise.
6978 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
6979 on the final iteration.
6980 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
6981 a wider mode exists before asking for a move pattern.
6982 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
6983 forcing a wider mode to exist.
6984 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
6985 returning false if no such mode exists.
6986 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
6987 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
6988 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
6989 Avoid checking for a MODE_INT if we already know the mode is not a
6991 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
6992 forcing a wider mode to exist.
6993 (expmed_mult_highpart_optab): Likewise.
6994 (expmed_mult_highpart): Likewise.
6995 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
6997 * lto-streamer-in.c (lto_input_mode_table): Likewise.
6998 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
6999 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
7000 * internal-fn.c (expand_mul_overflow): Update use of
7001 GET_MODE_2XWIDER_MODE.
7002 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7003 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
7004 GET_MODE_WIDER_MODE.
7005 (convert_plusminus_to_widen): Likewise.
7006 * tree-switch-conversion.c (array_value_type): Likewise.
7007 * var-tracking.c (emit_note_insn_var_location): Likewise.
7008 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7009 Return false inside rather than outside the loop if no wider mode
7011 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
7012 and GET_MODE_2XWIDER_MODE
7013 (can_compare_p): Use else_void.
7014 * gdbhooks.py (OptMachineModePrinter): New class.
7015 (build_pretty_printer): Use it for opt_mode.
7017 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7018 Alan Hayward <alan.hayward@arm.com>
7019 David Sherwood <david.sherwood@arm.com>
7021 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
7022 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
7024 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7025 Alan Hayward <alan.hayward@arm.com>
7026 David Sherwood <david.sherwood@arm.com>
7028 * machmode.h (mode_traits): New structure.
7029 (get_narrowest_mode): New function.
7030 (mode_iterator::start): Likewise.
7031 (mode_iterator::iterate_p): Likewise.
7032 (mode_iterator::get_wider): Likewise.
7033 (mode_iterator::get_known_wider): Likewise.
7034 (mode_iterator::get_2xwider): Likewise.
7035 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
7036 (FOR_EACH_MODE): Likewise.
7037 (FOR_EACH_MODE_FROM): Likewise.
7038 (FOR_EACH_MODE_UNTIL): Likewise.
7039 (FOR_EACH_WIDER_MODE): Likewise.
7040 (FOR_EACH_2XWIDER_MODE): Likewise.
7041 * builtins.c (expand_builtin_strlen): Use new mode iterators.
7042 * combine.c (simplify_comparison): Likewise
7043 * config/i386/i386.c (type_natural_mode): Likewise.
7044 * cse.c (cse_insn): Likewise.
7045 * dse.c (find_shift_sequence): Likewise.
7046 * emit-rtl.c (init_derived_machine_modes): Likewise.
7047 (init_emit_once): Likewise.
7048 * explow.c (hard_function_value): Likewise.
7049 * expmed.c (extract_fixed_bit_field_1): Likewise.
7050 (extract_bit_field_1): Likewise.
7051 (expand_divmod): Likewise.
7052 (emit_store_flag_1): Likewise.
7053 * expr.c (init_expr_target): Likewise.
7054 (convert_move): Likewise.
7055 (alignment_for_piecewise_move): Likewise.
7056 (widest_int_mode_for_size): Likewise.
7057 (emit_block_move_via_movmem): Likewise.
7058 (copy_blkmode_to_reg): Likewise.
7059 (set_storage_via_setmem): Likewise.
7060 (compress_float_constant): Likewise.
7061 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7062 * optabs-query.c (get_best_extraction_insn): Likewise.
7063 * optabs.c (expand_binop): Likewise.
7064 (expand_twoval_unop): Likewise.
7065 (expand_twoval_binop): Likewise.
7066 (widen_leading): Likewise.
7067 (widen_bswap): Likewise.
7068 (expand_parity): Likewise.
7069 (expand_unop): Likewise.
7070 (prepare_cmp_insn): Likewise.
7071 (prepare_float_lib_cmp): Likewise.
7072 (expand_float): Likewise.
7073 (expand_fix): Likewise.
7074 (expand_sfix_optab): Likewise.
7075 * postreload.c (move2add_use_add2_insn): Likewise.
7076 * reg-stack.c (reg_to_stack): Likewise.
7077 * reginfo.c (choose_hard_reg_mode): Likewise.
7078 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
7079 * stor-layout.c (mode_for_size): Likewise.
7080 (smallest_mode_for_size): Likewise.
7081 (mode_for_vector): Likewise.
7082 (finish_bitfield_representative): Likewise.
7083 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
7084 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
7085 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7086 * var-tracking.c (prepare_call_arguments): Likewise.
7088 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7089 Alan Hayward <alan.hayward@arm.com>
7090 David Sherwood <david.sherwood@arm.com>
7092 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
7093 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
7094 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
7095 * machmode.h (mode_size): Move earlier in file.
7096 (mode_precision): Likewise.
7097 (mode_inner): Likewise.
7098 (mode_nunits): Likewise.
7099 (mode_unit_size): Likewise.
7100 (unit_unit_precision): Likewise.
7101 (mode_wider): Likewise.
7102 (mode_2xwider): Likewise.
7103 (machine_mode): New class.
7104 (mode_to_bytes): New function.
7105 (mode_to_bits): Likewise.
7106 (mode_to_precision): Likewise.
7107 (mode_to_inner): Likewise.
7108 (mode_to_unit_size): Likewise.
7109 (mode_to_unit_precision): Likewise.
7110 (mode_to_nunits): Likewise.
7111 (GET_MODE_SIZE): Use mode_to_bytes.
7112 (GET_MODE_BITSIZE): Use mode_to_bits.
7113 (GET_MODE_PRECISION): Use mode_to_precision.
7114 (GET_MODE_INNER): Use mode_to_inner.
7115 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
7116 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
7117 (GET_MODE_NUNITS): Use mode_to_nunits.
7118 * system.h (ALWAYS_INLINE): New macro.
7119 * config/powerpcspe/powerpcspe-c.c
7120 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
7121 int for arg1_mode and arg2_mode.
7123 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7124 Alan Hayward <alan.hayward@arm.com>
7125 David Sherwood <david.sherwood@arm.com>
7127 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
7128 Prefix mode names with E_ in case statements.
7129 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7130 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
7131 (aarch64_split_simd_move): Likewise.
7132 (aarch64_gen_storewb_pair): Likewise.
7133 (aarch64_gen_loadwb_pair): Likewise.
7134 (aarch64_gen_store_pair): Likewise.
7135 (aarch64_gen_load_pair): Likewise.
7136 (aarch64_get_condition_code_1): Likewise.
7137 (aarch64_constant_pool_reload_icode): Likewise.
7138 (get_rsqrte_type): Likewise.
7139 (get_rsqrts_type): Likewise.
7140 (get_recpe_type): Likewise.
7141 (get_recps_type): Likewise.
7142 (aarch64_gimplify_va_arg_expr): Likewise.
7143 (aarch64_simd_container_mode): Likewise.
7144 (aarch64_emit_load_exclusive): Likewise.
7145 (aarch64_emit_store_exclusive): Likewise.
7146 (aarch64_expand_compare_and_swap): Likewise.
7147 (aarch64_gen_atomic_cas): Likewise.
7148 (aarch64_emit_bic): Likewise.
7149 (aarch64_emit_atomic_swap): Likewise.
7150 (aarch64_emit_atomic_load_op): Likewise.
7151 (aarch64_evpc_trn): Likewise.
7152 (aarch64_evpc_uzp): Likewise.
7153 (aarch64_evpc_zip): Likewise.
7154 (aarch64_evpc_ext): Likewise.
7155 (aarch64_evpc_rev): Likewise.
7156 (aarch64_evpc_dup): Likewise.
7157 (aarch64_gen_ccmp_first): Likewise.
7158 (aarch64_gen_ccmp_next): Likewise.
7159 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
7160 (alpha_emit_xfloating_libcall): Likewise.
7161 (emit_insxl): Likewise.
7162 (alpha_arg_type): Likewise.
7163 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
7164 (arc_preferred_simd_mode): Likewise.
7165 (arc_secondary_reload): Likewise.
7166 (get_arc_condition_code): Likewise.
7167 (arc_print_operand): Likewise.
7168 (arc_legitimate_constant_p): Likewise.
7169 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7170 * config/arc/arc.md (casesi_load): Likewise.
7171 (casesi_compact_jump): Likewise.
7172 * config/arc/predicates.md (proper_comparison_operator): Likewise.
7173 (cc_use_register): Likewise.
7174 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7175 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
7176 (arm_init_iwmmxt_builtins): Likewise.
7177 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
7178 (neon_expand_vector_init): Likewise.
7179 (arm_attr_length_move_neon): Likewise.
7180 (maybe_get_arm_condition_code): Likewise.
7181 (arm_emit_vector_const): Likewise.
7182 (arm_preferred_simd_mode): Likewise.
7183 (arm_output_iwmmxt_tinsr): Likewise.
7184 (thumb1_output_casesi): Likewise.
7185 (thumb2_output_casesi): Likewise.
7186 (arm_emit_load_exclusive): Likewise.
7187 (arm_emit_store_exclusive): Likewise.
7188 (arm_expand_compare_and_swap): Likewise.
7189 (arm_evpc_neon_vuzp): Likewise.
7190 (arm_evpc_neon_vzip): Likewise.
7191 (arm_evpc_neon_vrev): Likewise.
7192 (arm_evpc_neon_vtrn): Likewise.
7193 (arm_evpc_neon_vext): Likewise.
7194 (arm_validize_comparison): Likewise.
7195 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
7196 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
7197 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
7198 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
7199 (c6x_preferred_simd_mode): Likewise.
7200 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
7201 (epiphany_rtx_costs): Likewise.
7202 * config/epiphany/predicates.md (proper_comparison_operator):
7204 * config/frv/frv.c (condexec_memory_operand): Likewise.
7205 (frv_emit_move): Likewise.
7206 (output_move_single): Likewise.
7207 (output_condmove_single): Likewise.
7208 (frv_hard_regno_mode_ok): Likewise.
7209 (frv_matching_accg_mode): Likewise.
7210 * config/h8300/h8300.c (split_adds_subs): Likewise.
7211 (h8300_rtx_costs): Likewise.
7212 (h8300_print_operand): Likewise.
7213 (compute_mov_length): Likewise.
7214 (output_logical_op): Likewise.
7215 (compute_logical_op_length): Likewise.
7216 (compute_logical_op_cc): Likewise.
7217 (h8300_shift_needs_scratch_p): Likewise.
7218 (output_a_shift): Likewise.
7219 (compute_a_shift_length): Likewise.
7220 (compute_a_shift_cc): Likewise.
7221 (expand_a_rotate): Likewise.
7222 (output_a_rotate): Likewise.
7223 * config/i386/i386.c (classify_argument): Likewise.
7224 (function_arg_advance_32): Likewise.
7225 (function_arg_32): Likewise.
7226 (function_arg_64): Likewise.
7227 (function_value_64): Likewise.
7228 (ix86_gimplify_va_arg): Likewise.
7229 (ix86_legitimate_constant_p): Likewise.
7230 (put_condition_code): Likewise.
7231 (split_double_mode): Likewise.
7232 (ix86_avx256_split_vector_move_misalign): Likewise.
7233 (ix86_expand_vector_logical_operator): Likewise.
7234 (ix86_split_idivmod): Likewise.
7235 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
7236 (ix86_build_const_vector): Likewise.
7237 (ix86_build_signbit_mask): Likewise.
7238 (ix86_match_ccmode): Likewise.
7239 (ix86_cc_modes_compatible): Likewise.
7240 (ix86_expand_branch): Likewise.
7241 (ix86_expand_sse_cmp): Likewise.
7242 (ix86_expand_sse_movcc): Likewise.
7243 (ix86_expand_int_sse_cmp): Likewise.
7244 (ix86_expand_vec_perm_vpermi2): Likewise.
7245 (ix86_expand_vec_perm): Likewise.
7246 (ix86_expand_sse_unpack): Likewise.
7247 (ix86_expand_int_addcc): Likewise.
7248 (ix86_split_to_parts): Likewise.
7249 (ix86_vectorize_builtin_gather): Likewise.
7250 (ix86_vectorize_builtin_scatter): Likewise.
7251 (avx_vpermilp_parallel): Likewise.
7252 (inline_memory_move_cost): Likewise.
7253 (ix86_tieable_integer_mode_p): Likewise.
7254 (x86_maybe_negate_const_int): Likewise.
7255 (ix86_expand_vector_init_duplicate): Likewise.
7256 (ix86_expand_vector_init_one_nonzero): Likewise.
7257 (ix86_expand_vector_init_one_var): Likewise.
7258 (ix86_expand_vector_init_concat): Likewise.
7259 (ix86_expand_vector_init_interleave): Likewise.
7260 (ix86_expand_vector_init_general): Likewise.
7261 (ix86_expand_vector_set): Likewise.
7262 (ix86_expand_vector_extract): Likewise.
7263 (emit_reduc_half): Likewise.
7264 (ix86_emit_i387_round): Likewise.
7265 (ix86_mangle_type): Likewise.
7266 (ix86_expand_round_sse4): Likewise.
7267 (expand_vec_perm_blend): Likewise.
7268 (canonicalize_vector_int_perm): Likewise.
7269 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
7270 (expand_vec_perm_1): Likewise.
7271 (expand_vec_perm_interleave3): Likewise.
7272 (expand_vec_perm_even_odd_pack): Likewise.
7273 (expand_vec_perm_even_odd_1): Likewise.
7274 (expand_vec_perm_broadcast_1): Likewise.
7275 (ix86_vectorize_vec_perm_const_ok): Likewise.
7276 (ix86_expand_vecop_qihi): Likewise.
7277 (ix86_expand_mul_widen_hilo): Likewise.
7278 (ix86_expand_sse2_abs): Likewise.
7279 (ix86_expand_pextr): Likewise.
7280 (ix86_expand_pinsr): Likewise.
7281 (ix86_preferred_simd_mode): Likewise.
7282 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
7283 * config/i386/sse.md (*andnot<mode>3): Likewise.
7284 (<mask_codefor><code><mode>3<mask_name>): Likewise.
7285 (*<code><mode>3): Likewise.
7286 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
7287 (ia64_expand_atomic_op): Likewise.
7288 (ia64_arg_type): Likewise.
7289 (ia64_mode_to_int): Likewise.
7290 (ia64_scalar_mode_supported_p): Likewise.
7291 (ia64_vector_mode_supported_p): Likewise.
7292 (expand_vec_perm_broadcast): Likewise.
7293 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
7294 (iq2000_function_arg_advance): Likewise.
7295 (iq2000_function_arg): Likewise.
7296 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
7297 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
7298 (m68k_libcall_value): Likewise.
7299 (m68k_function_value): Likewise.
7300 (sched_attr_op_type): Likewise.
7301 * config/mcore/mcore.c (mcore_output_move): Likewise.
7302 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
7304 (microblaze_function_arg): Likewise.
7305 * config/mips/mips.c (mips16_build_call_stub): Likewise.
7306 (mips_print_operand): Likewise.
7307 (mips_mode_ok_for_mov_fmt_p): Likewise.
7308 (mips_vector_mode_supported_p): Likewise.
7309 (mips_preferred_simd_mode): Likewise.
7310 (mips_expand_vpc_loongson_even_odd): Likewise.
7311 (mips_expand_vec_unpack): Likewise.
7312 (mips_expand_vi_broadcast): Likewise.
7313 (mips_expand_vector_init): Likewise.
7314 (mips_expand_vec_reduc): Likewise.
7315 (mips_expand_msa_cmp): Likewise.
7316 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
7317 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
7318 (cc_flags_for_mode): Likewise.
7319 * config/msp430/msp430.c (msp430_print_operand): Likewise.
7320 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
7321 (nds32_output_casesi_pc_relative): Likewise.
7322 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7323 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
7324 (nvptx_gen_unpack): Likewise.
7325 (nvptx_gen_pack): Likewise.
7326 (nvptx_gen_shuffle): Likewise.
7327 (nvptx_gen_wcast): Likewise.
7328 (nvptx_preferred_simd_mode): Likewise.
7329 * config/pa/pa.c (pa_secondary_reload): Likewise.
7330 * config/pa/predicates.md (base14_operand): Likewise.
7331 * config/powerpcspe/powerpcspe-c.c
7332 (altivec_resolve_overloaded_builtin): Likewise.
7333 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
7335 (rs6000_preferred_simd_mode): Likewise.
7336 (output_vec_const_move): Likewise.
7337 (rs6000_expand_vector_extract): Likewise.
7338 (rs6000_split_vec_extract_var): Likewise.
7339 (reg_offset_addressing_ok_p): Likewise.
7340 (rs6000_legitimate_offset_address_p): Likewise.
7341 (rs6000_legitimize_address): Likewise.
7342 (rs6000_emit_set_const): Likewise.
7343 (rs6000_const_vec): Likewise.
7344 (rs6000_emit_move): Likewise.
7345 (spe_build_register_parallel): Likewise.
7346 (rs6000_darwin64_record_arg_recurse): Likewise.
7347 (swap_selector_for_mode): Likewise.
7348 (spe_init_builtins): Likewise.
7349 (paired_init_builtins): Likewise.
7350 (altivec_init_builtins): Likewise.
7351 (do_load_for_compare): Likewise.
7352 (rs6000_generate_compare): Likewise.
7353 (rs6000_expand_float128_convert): Likewise.
7354 (emit_load_locked): Likewise.
7355 (emit_store_conditional): Likewise.
7356 (rs6000_output_function_epilogue): Likewise.
7357 (rs6000_handle_altivec_attribute): Likewise.
7358 (rs6000_function_value): Likewise.
7359 (emit_fusion_gpr_load): Likewise.
7360 (emit_fusion_p9_load): Likewise.
7361 (emit_fusion_p9_store): Likewise.
7362 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
7363 (fusion_gpr_mem_load): Likewise.
7364 (fusion_addis_mem_combo_load): Likewise.
7365 (fusion_addis_mem_combo_store): Likewise.
7366 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
7367 (fusion_gpr_mem_load): Likewise.
7368 (fusion_addis_mem_combo_load): Likewise.
7369 (fusion_addis_mem_combo_store): Likewise.
7370 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7372 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
7373 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
7374 (rs6000_preferred_simd_mode): Likewise.
7375 (output_vec_const_move): Likewise.
7376 (rs6000_expand_vector_extract): Likewise.
7377 (rs6000_split_vec_extract_var): Likewise.
7378 (reg_offset_addressing_ok_p): Likewise.
7379 (rs6000_legitimate_offset_address_p): Likewise.
7380 (rs6000_legitimize_address): Likewise.
7381 (rs6000_emit_set_const): Likewise.
7382 (rs6000_const_vec): Likewise.
7383 (rs6000_emit_move): Likewise.
7384 (rs6000_darwin64_record_arg_recurse): Likewise.
7385 (swap_selector_for_mode): Likewise.
7386 (paired_init_builtins): Likewise.
7387 (altivec_init_builtins): Likewise.
7388 (rs6000_expand_float128_convert): Likewise.
7389 (emit_load_locked): Likewise.
7390 (emit_store_conditional): Likewise.
7391 (rs6000_output_function_epilogue): Likewise.
7392 (rs6000_handle_altivec_attribute): Likewise.
7393 (rs6000_function_value): Likewise.
7394 (emit_fusion_gpr_load): Likewise.
7395 (emit_fusion_p9_load): Likewise.
7396 (emit_fusion_p9_store): Likewise.
7397 * config/rx/rx.c (rx_gen_move_template): Likewise.
7398 (flags_from_mode): Likewise.
7399 * config/s390/predicates.md (s390_alc_comparison): Likewise.
7400 (s390_slb_comparison): Likewise.
7401 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
7402 (s390_vector_mode_supported_p): Likewise.
7403 (s390_cc_modes_compatible): Likewise.
7404 (s390_match_ccmode_set): Likewise.
7405 (s390_canonicalize_comparison): Likewise.
7406 (s390_emit_compare_and_swap): Likewise.
7407 (s390_branch_condition_mask): Likewise.
7408 (s390_rtx_costs): Likewise.
7409 (s390_secondary_reload): Likewise.
7410 (__SECONDARY_RELOAD_CASE): Likewise.
7411 (s390_expand_cs): Likewise.
7412 (s390_preferred_simd_mode): Likewise.
7413 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
7414 * config/sh/sh.c (sh_print_operand): Likewise.
7415 (dump_table): Likewise.
7416 (sh_secondary_reload): Likewise.
7417 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7418 * config/sh/sh.md (casesi_worker_1): Likewise.
7419 (casesi_worker_2): Likewise.
7420 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
7421 (fcc_comparison_operator): Likewise.
7422 * config/sparc/sparc.c (sparc_expand_move): Likewise.
7423 (emit_soft_tfmode_cvt): Likewise.
7424 (sparc_preferred_simd_mode): Likewise.
7425 (output_cbranch): Likewise.
7426 (sparc_print_operand): Likewise.
7427 (sparc_expand_vec_perm_bmask): Likewise.
7428 (vector_init_bshuffle): Likewise.
7429 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
7430 (spu_vector_mode_supported_p): Likewise.
7431 (spu_expand_insv): Likewise.
7432 (spu_emit_branch_or_set): Likewise.
7433 (spu_handle_vector_attribute): Likewise.
7434 (spu_builtin_splats): Likewise.
7435 (spu_builtin_extract): Likewise.
7436 (spu_builtin_promote): Likewise.
7437 (spu_expand_sign_extend): Likewise.
7438 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
7439 (tilegx_simd_int): Likewise.
7440 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
7441 (tilepro_simd_int): Likewise.
7442 * config/v850/v850.c (const_double_split): Likewise.
7443 (v850_print_operand): Likewise.
7444 (ep_memory_offset): Likewise.
7445 * config/vax/vax.c (vax_rtx_costs): Likewise.
7446 (vax_output_int_move): Likewise.
7447 (vax_output_int_add): Likewise.
7448 (vax_output_int_subtract): Likewise.
7449 * config/visium/predicates.md (visium_branch_operator): Likewise.
7450 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
7451 (visium_print_operand_address): Likewise.
7452 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7453 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
7454 (xtensa_expand_conditional_branch): Likewise.
7455 (xtensa_copy_incoming_a7): Likewise.
7456 (xtensa_output_literal): Likewise.
7457 * dfp.c (decimal_real_maxval): Likewise.
7458 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
7460 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7461 Alan Hayward <alan.hayward@arm.com>
7462 David Sherwood <david.sherwood@arm.com>
7464 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
7465 (mode_nunits_inline): Likewise.
7466 (mode_inner_inline): Likewise.
7467 (mode_unit_size_inline): Likewise.
7468 (mode_unit_precision_inline): Likewise.
7469 (emit_insn_modes_h): Likewise. Also emit a #define of the
7471 (emit_mode_wider): Add an E_ prefix to mode names.
7472 (emit_mode_complex): Likewise.
7473 (emit_mode_inner): Likewise.
7474 (emit_mode_adjustments): Likewise.
7475 (emit_mode_int_n): Likewise.
7476 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
7477 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
7478 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
7479 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
7480 (CRC32_BUILTIN, ENTRY): Likewise.
7481 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
7482 (aarch64_pop_regs): Likewise.
7483 (aarch64_process_components): Likewise.
7484 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
7485 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
7486 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
7487 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
7488 * config/arm/arm.c (arm_init_libfuncs): Likewise.
7489 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
7491 * config/i386/i386-builtin.def (pcmpestr): Likewise.
7492 (pcmpistr): Likewise.
7493 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
7494 * config/mmix/mmix.c (mmix_output_condition): Likewise.
7495 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
7497 * config/rl78/rl78.c (mduc_regs): Likewise.
7498 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
7499 (htm_expand_builtin): Likewise.
7500 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
7501 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
7502 * config/xtensa/xtensa.c (print_operand): Likewise.
7503 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
7504 (NUM_MODE_VECTOR_INT): Likewise.
7505 * genoutput.c (null_operand): Likewise.
7506 (output_operand_data): Likewise.
7507 * genrecog.c (print_parameter_value): Likewise.
7508 * lra.c (debug_operand_data): Likewise.
7510 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7512 * dwarf2out.c (scompare_loc_descriptor_wide)
7513 (scompare_loc_descriptor_narrow): New functions, split out from...
7514 (scompare_loc_descriptor): ...here.
7515 * expmed.c (emit_store_flag_int): New function, split out from...
7516 (emit_store_flag): ...here.
7518 2017-08-30 Richard Biener <rguenther@suse.de>
7520 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
7521 (dwarf2out_early_finish): Move setting of AT_pubnames from
7522 early debug output to early finish.
7524 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
7526 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
7527 and -mdata-region to the assembler.
7529 2017-08-30 Richard Biener <rguenther@suse.de>
7531 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
7533 (gen_subprogram_die): Add DW_AT_object_pointer only early.
7534 (dwarf2out_early_global_decl): Only generate a DIE for the
7535 abstract origin if it doesn't already exist or is a declaration DIE.
7536 (resolve_addr): Do not add the linkage name twice when
7537 generating a stub DIE for the DW_TAG_GNU_call_site target.
7539 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7541 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7542 Use machine_mode rather than int for arg1_mode.
7544 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
7547 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
7548 that the second argument of the built-in functions to unpack
7549 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
7550 switch statement instead a lot of if statements.
7551 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
7552 Allow 64-bit values to be in Altivec registers as well as
7553 traditional floating point registers.
7554 (pack<mode>, FMOVE128_VSX iterator): Likewise.
7556 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
7558 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
7559 MAX_REGS_PER_ADDRESS == 1.
7561 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
7563 * config/i386/i386.opt (flag_fentry): Do not init to -1.
7564 * config/i386/i386.c (ix86_option_override_internal): Simplify
7565 setting of opts->x_flag_entry.
7567 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7568 Jakub Jelinek <jakub@redhat.com>
7569 Richard Biener <rguenther@suse.de>
7571 PR tree-optimization/81503
7572 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
7573 folded constant fits in the target type; reorder tests for clarity.
7575 2017-08-29 Martin Liska <mliska@suse.cz>
7577 * passes.def: Include pass_lower_switch.
7578 * stmt.c (dump_case_nodes): Remove and move to
7579 tree-switch-conversion.
7580 (case_values_threshold): Likewise.
7581 (expand_switch_as_decision_tree_p): Likewise.
7582 (emit_case_decision_tree): Likewise.
7583 (expand_case): Likewise.
7584 (balance_case_nodes): Likewise.
7585 (node_has_low_bound): Likewise.
7586 (node_has_high_bound): Likewise.
7587 (node_is_bounded): Likewise.
7588 (emit_case_nodes): Likewise.
7589 (struct simple_case_node): New struct.
7590 (add_case_node): Remove.
7591 (emit_case_dispatch_table): Use vector instead of case_list.
7592 (reset_out_edges_aux): Remove.
7593 (compute_cases_per_edge): Likewise.
7594 (expand_case): Build list of simple_case_node.
7595 (expand_sjlj_dispatch_table): Use it.
7596 * tree-switch-conversion.c (struct case_node): Moved from
7597 stmt.c and adjusted.
7598 (emit_case_nodes): Likewise.
7599 (node_has_low_bound): Likewise.
7600 (node_has_high_bound): Likewise.
7601 (node_is_bounded): Likewise.
7602 (case_values_threshold): Likewise.
7603 (reset_out_edges_aux): Likewise.
7604 (compute_cases_per_edge): Likewise.
7605 (add_case_node): Likewise.
7606 (dump_case_nodes): Likewise.
7607 (balance_case_nodes): Likewise.
7608 (expand_switch_as_decision_tree_p): Likewise.
7609 (emit_jump): Likewise.
7610 (emit_case_decision_tree): Likewise.
7611 (try_switch_expansion): Likewise.
7612 (do_jump_if_equal): Likewise.
7613 (emit_cmp_and_jump_insns): Likewise.
7614 (fix_phi_operands_for_edge): New function.
7615 (record_phi_operand_mapping): Likewise.
7616 (class pass_lower_switch): New pass.
7617 (pass_lower_switch::execute): New function.
7618 (make_pass_lower_switch): Likewise.
7619 (conditional_probability):
7620 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
7621 * tree-pass.h: Add make_pass_lower_switch.
7623 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
7626 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
7629 2017-08-29 Richard Biener <rguenther@suse.de>
7631 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
7632 we do not add a DW_AT_inline attribute twice.
7633 (gen_subprogram_die): Remove code setting DW_AT_inline on
7634 DECL_ABSTRACT_P nodes.
7636 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
7638 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
7639 calls to internal functions.
7640 (gimplify_modify_expr): Likewise.
7641 * tree-call-cdce.c (use_internal_fn): Likewise.
7642 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
7643 (convert_to_divmod): Set the nothrow flag.
7644 * tree-if-conv.c (predicate_mem_writes): Likewise.
7645 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7646 (vectorizable_call): Likewise.
7647 (vectorizable_store): Likewise.
7648 (vectorizable_load): Likewise.
7649 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
7650 (vect_recog_mask_conversion_pattern): Likewise.
7652 2017-08-29 Martin Liska <mliska@suse.cz>
7655 * gcc.c (driver_handle_option): Add new argument.
7656 * opts-common.c (handle_option): Pass
7657 target_option_override_hook.
7658 * opts-global.c (lang_handle_option): Add new option.
7659 (set_default_handlers): Add new argument.
7660 (decode_options): Likewise.
7661 * opts.c (target_handle_option): Likewise.
7662 (common_handle_option): Call target_option_override_hook.
7663 * opts.h (struct cl_option_handler_func): Add hook for
7664 target option override.
7665 (struct cl_option_handlers): Likewise.
7666 (set_default_handlers): Add new argument.
7667 (decode_options): Likewise.
7668 (common_handle_option): Likewise.
7669 (target_handle_option): Likewise.
7670 * toplev.c (toplev::main): Pass targetm.target_option.override
7673 2017-08-29 Richard Biener <rguenther@suse.de>
7674 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7676 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
7677 life to the active subtree.
7679 2017-08-28 Jeff Law <law@redhat.com>
7681 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
7682 derive_equivalences.
7683 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
7685 (edge_info::derive_equivalences): New private member function
7687 * tree-ssa-dom.c (class edge_info): Changed from a struct
7688 to a class. Add ctor/dtor, methods and data members.
7689 (edge_info::edge_info): Renamed from allocate_edge_info.
7690 Initialize additional members.
7691 (edge_info::~edge_info): New.
7692 (free_dom_edge_info): Delete the edge info.
7693 (record_edge_info): Use new class & associated member functions.
7694 Tighten forms for testing for edge equivalences.
7695 (record_temporary_equivalences): Iterate over the simple
7696 equivalences rather than assuming there's only one per edge.
7697 (cprop_into_successor_phis): Iterate over the simple
7698 equivalences rather than assuming there's only one per edge.
7699 (optimize_stmt): Use operand_equal_p rather than pointer
7700 equality for mini-DSE code.
7702 2017-08-28 Nathan Sidwell <nathan@acm.org>
7704 * gcc.c (execute): Fold SIGPIPE handling into switch
7705 statement. Adjust internal error message.
7707 2017-08-28 Richard Biener <rguenther@suse.de>
7710 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
7711 Do nothing for removed DIEs.
7713 2017-08-28 Richard Biener <rguenther@suse.de>
7715 PR tree-optimization/81977
7716 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
7719 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
7722 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
7723 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
7724 using targetm.gen_mem_thread_fence.
7726 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
7729 * config/i386/i386.md (*<btsc><mode>): Change operand 2
7730 predicate to register_operand. Reorder operands.
7731 (*btr<mode>): Ditto.
7732 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
7733 (*btr<mode>_mask): Ditto.
7735 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
7737 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
7738 * config/rs6000/xmmintrin.h: New file.
7739 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
7741 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7744 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
7745 parameter and_insn and return it.
7746 (recombine_lvx_pattern): Insert a copy to ensure availability of
7747 the base register of the copied masking operation at the point of
7748 the instruction replacement.
7749 (recombine_stvx_pattern): Likewise.
7751 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
7753 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
7754 undocumented switches.
7755 (-mpower9-dform-vector): Likewise.
7756 (-mpower9-dform): Likewise.
7757 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
7758 comments to delete references to -mpower9-dform* switches.
7759 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
7760 Delete reference to -mpower9-dform* switches, test for
7761 -mpower9-vector instead.
7762 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
7763 (OTHER_P9_VECTOR_MASKS): Likewise.
7764 (POWERPC_MASKS): Likewise.
7765 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
7766 tests against -mpower9-dform* to -mpower9-vector. Delete code
7767 that checked for -mpower9-dform* consistancy with other options.
7768 Add test for -mpower9-misc to enable other power9 switches.
7769 (rs6000_init_hard_regno_mode_ok): Likewise.
7770 (rs6000_option_override_internal): Likewise.
7771 (rs6000_emit_prologue): Likewise.
7772 (rs6000_emit_epilogue): Likewise.
7773 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
7774 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
7775 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
7777 (emit_fusion_p9_store): Likewise.
7778 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
7779 resetting these macros if the assembler does not support ISA 3.0
7781 (TARGET_P9_DFORM_VECTOR): Likewise.
7782 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
7783 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
7785 2017-08-25 Alan Modra <amodra@gmail.com>
7787 PR rtl-optimization/81747
7788 * cse.c (cse_extended_basic_block): Don't attempt to record
7789 equivalences for degenerate conditional branches that branch
7790 to their fall-through.
7792 2017-08-24 Martin Sebor <msebor@redhat.com>
7795 * gimple-fold.c (size_must_be_zero_p): New function.
7796 (gimple_fold_builtin_memory_op): Call it.
7798 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
7800 * config/rs6000/mm_malloc.h: New file.
7802 2017-08-24 Bin Cheng <bin.cheng@arm.com>
7804 PR tree-optimization/81913
7805 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
7806 analysis when either IVs in condition can wrap.
7808 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
7810 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
7811 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
7813 2017-08-24 Richard Biener <rguenther@suse.de>
7816 * targhooks.c (default_target_can_inline_p): Properly
7817 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
7818 is present and always compare.
7819 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
7820 infer -mfpmath=sse from TARGET_SSE_P.
7821 (ix86_can_inline_p): Properly use target_option_default_node when
7822 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
7824 2017-08-24 Richard Biener <rguenther@suse.de>
7827 * dwarf2out.c (output_die): Handle flag_generate_offload like
7829 (output_comp_unit): Likewise.
7830 (gen_array_type_die): Likewise.
7831 (dwarf2out_early_finish): Likewise.
7832 (note_variable_value_in_expr): Likewise.
7833 (dwarf2out_finish): Likewise. Adjust assert.
7834 * cgraphunit.c (symbol_table::compile): Move setting of
7835 flag_generate_offload earlier ...
7836 (symbol_table::finalize_compilation_unit): ... here, before
7837 early debug finalization.
7839 2017-08-24 Richard Biener <rguenther@suse.de>
7841 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
7842 and ipa-fnsummary.h.
7843 (ix86_can_inline_p): When ix86_fpmath flags do not match
7844 check whether the callee uses FP math at all.
7846 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
7849 * tree-ssanames.c (get_nonzero_bits): Use element_precision
7850 instead of TYPE_PRECISION.
7852 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
7853 Alan Hayward <alan.hayward@arm.com>
7854 David Sherwood <david.sherwood@arm.com>
7856 * combine.c (make_extraction): Use subreg_offset_from_lsb.
7858 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
7860 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
7862 (ix86_frame::stack_realign_allocate): New field.
7863 (struct machine_frame_state): Modify comments.
7864 (machine_frame_state::sp_realigned_fp_end): New field.
7865 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
7867 (sp_valid_at): Add assertion to assure no attempt to access invalid
7868 offset of a realigned stack.
7869 (fp_valid_at): Likewise.
7870 (choose_baseaddr): Modify comments.
7871 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
7872 ix86_expand_prologue.
7873 (ix86_expand_prologue): Modify stack realignment and allocation.
7874 (ix86_expand_epilogue): Modify comments.
7875 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
7876 avx2_runtime, avx512f, and avx512f_runtime.
7878 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
7880 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
7881 (mstackrealign): Do not init to -1.
7882 * config/i386/i386.c (ix86_option_override_internal):
7883 Check opts_set, not opts when setting default value of
7884 opts->x_ix86_force_align_arg_pointer.
7886 2017-08-23 Richard Biener <rguenther@suse.de>
7888 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
7889 lang_hooks.decl_printable_name.
7890 * print-rtl-function.c (print_rtx_function): Likewise.
7891 * tree-pretty-print.c (dump_function_header): Likewise.
7893 2017-08-23 Richard Biener <rguenther@suse.de>
7896 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
7897 -g0 at compile-time.
7899 2017-08-23 Tamar Christina <tamar.christina@arm.com>
7902 * doc/sourcebuild.texi (Other hardware attributes):
7905 2017-08-23 Tamar Christina <tamar.christina@arm.com>
7908 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
7909 Fix single-use check.
7911 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
7913 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
7915 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
7917 * doc/install.texi: Modify to add more details on running selected
7920 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
7922 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
7923 is combined with -mabi=ms.
7924 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
7927 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7929 PR tree-optimization/81488
7930 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
7931 and cached_basis fields.
7932 (MAX_SPREAD): New constant.
7933 (alloc_cand_and_find_basis): Initialize new fields.
7934 (clear_visited): New function.
7935 (create_phi_basis_1): Rename from create_phi_basis, set visited
7936 and cached_basis fields.
7937 (create_phi_basis): New wrapper function.
7938 (phi_add_costs_1): Rename from phi_add_costs, add spread
7939 parameter, set visited field, short-circuit when limits reached.
7940 (phi_add_costs): New wrapper function.
7941 (record_phi_increments_1): Rename from record_phi_increments, set
7943 (record_phi_increments): New wrapper function.
7944 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
7945 (phi_incr_cost): New wrapper function.
7946 (all_phi_incrs_profitable_1): Rename from
7947 all_phi_incrs_profitable, set visited field.
7948 (all_phi_incrs_profitable): New wrapper function.
7950 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
7951 Alan Hayward <alan.hayward@arm.com>
7952 David Sherwood <david.sherwood@arm.com>
7954 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
7955 that takes the outer and inner modes.
7956 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
7957 comparison as the canonical test for a paradoxical subreg.
7958 * combine.c (simplify_set): Use paradoxical_subreg_p.
7959 (make_extraction): Likewise.
7960 (force_to_mode): Likewise.
7961 (rtx_equal_for_field_assignment_p): Likewise.
7962 (gen_lowpart_for_combine): Likewise.
7963 (simplify_comparison): Likewise.
7964 * cse.c (equiv_constant): Likewise.
7965 * expmed.c (store_bit_field_1): Likewise.
7966 * final.c (alter_subreg): Likewise.
7967 * fwprop.c (propagate_rtx): Likewise.
7968 (forward_propagate_subreg): Likewise.
7969 * ira-conflicts.c (ira_build_conflicts): Likewise.
7970 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
7971 * lra-constraints.c (curr_insn_transform): Likewise.
7972 (split_reg): Likewise.
7973 * lra-eliminations.c (move_plus_up): Likewise.
7974 (lra_eliminate_regs_1): Likewise.
7975 * recog.c (general_operand): Likewise.
7976 * ree.c (combine_reaching_defs): Likewise.
7977 * reload.c (push_reload): Likewise.
7978 (find_reloads): Likewise.
7979 * reload1.c (elimination_effects): Likewise.
7980 (compute_reload_subreg_offset): Likewise.
7981 (choose_reload_regs): Likewise.
7982 * rtlanal.c (subreg_lsb_1): Likewise.
7983 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7984 (simplify_subreg): Likewise.
7985 * var-tracking.c (track_loc_p): Likewise.
7986 * emit-rtl.c (byte_lowpart_offset): Likewise.
7987 (paradoxical_subreg_p): Delete out-of-line definition.
7989 2017-08-22 Jeff Law <law@redhat.com>
7991 PR tree-optimization/81741
7992 PR tree-optimization/71947
7993 * tree-ssa-dom.c: Include tree-inline.h.
7994 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
7995 equivalences if one is more expensive to compute than the other.
7996 * tree-ssa-scopedtables.h (class const_or_copies): Make
7997 record_const_or_copy_raw method private.
7998 (class avail_exprs_stack): New method simplify_binary_operation.
7999 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
8000 avail_exprs_stack::simplify_binary_operation as needed.
8001 (avail_exprs_stack::simplify_binary_operation): New function.
8003 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
8005 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
8006 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
8007 (DOT_SYMBOLS): Likewise.
8008 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
8009 (RELOCATABLE_NEEDS_FIXUP): Likewise.
8010 (RS6000_ABI_NAME): Likewise.
8011 (TARGET_CMODEL): Likewise.
8012 (TOC_SECTION_ASM_OP): Likewise.
8013 (SET_CMODEL): New macro.
8014 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
8016 2017-08-22 Richard Biener <rguenther@suse.de>
8018 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
8019 to allow for free-lang-data replacements similar to verify_type_variant.
8021 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
8022 Alan Hayward <alan.hayward@arm.com>
8023 David Sherwood <david.sherwood@arm.com>
8025 * config/aarch64/aarch64.md (casesi): Use DImode rather than
8026 VOIDmode for the LABEL_REF.
8028 2017-08-22 Richard Biener <rguenther@suse.de>
8030 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
8032 2017-08-22 Richard Biener <rguenther@suse.de>
8034 * common.opt (feliminate-dwarf2-dups): Ignore.
8035 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
8036 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
8037 same_die_p_wrap, compute_section_prefix,
8038 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
8039 (comdat_symbol_id, comdat_symbol_number): Likewise.
8040 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
8042 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
8043 (output_die): Mark unreachable path unreachable.
8044 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
8045 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
8046 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
8047 (dwarf2out_early_finish): Likewise.
8049 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
8051 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
8053 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
8056 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
8057 not VAR_P. Filter attribute warnings with OPT_Wattributes.
8058 (avr_attribute_table) <io, io_low, address>: Initialize
8059 .decl_required with true.
8061 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
8063 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
8064 undocumented debugging options.
8065 (-mvsx-scalar-double): Likewise.
8066 (-mallow-df-permute): Likewise.
8067 (-mvectorize-builtins): Likewise.
8068 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8069 (rs6000_builtin_vectorized_function): Likewise.
8070 (rs6000_builtin_md_vectorized_function): Likewise.
8071 (rs6000_opt_vars): Likewise.
8073 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
8076 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
8077 (*btrq_imm): Rename from *btrq.
8078 (*btcq_imm): Rename from *btcq.
8079 (btsc): New code attribute.
8080 (*<btsc><mode>): New insn pattern.
8081 (*btr<mode>): Ditto.
8082 (*<btsc><mode>_mask): New insn_and_split pattern.
8083 (*btr<mode>_mask): Ditto.
8085 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8086 Alan Hayward <alan.hayward@arm.com>
8087 David Sherwood <david.sherwood@arm.com>
8089 * function.c (pad_below): Simplify padding calculation.
8091 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8092 Alan Hayward <alan.hayward@arm.com>
8093 David Sherwood <david.sherwood@arm.com>
8095 * target.def (function_prologue): Remove frame size argument.
8096 (function_epilogue): Likewise.
8097 * doc/tm.texi: Regenerate.
8098 * final.c (final_start_function): Update call to function_prologue.
8099 (final_end_function): Update call to function_epilogue.
8100 (default_function_pro_epilogue): Remove frame size argument.
8101 * output.h (default_function_pro_epilogue): Likewise.
8102 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
8103 (arm_output_function_prologue): Likewise.
8104 * config/frv/frv.c (frv_function_prologue): Likewise.
8105 (frv_function_epilogue): Likewise.
8106 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
8107 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
8108 (ia64_output_function_epilogue): Likewise.
8109 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
8110 (m32r_output_function_epilogue): Likewise.
8111 * config/microblaze/microblaze.c (microblaze_function_prologue)
8112 (microblaze_function_epilogue): Likewise.
8113 * config/mips/mips.c (mips_output_function_prologue): Likewise.
8114 (mips_output_function_epilogue): Likewise.
8115 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
8116 (mmix_target_asm_function_epilogue): Likewise.
8117 * config/msp430/msp430.c (msp430_start_function): Likewise.
8118 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
8119 (nds32_asm_function_epilogue): Likewise.
8120 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
8121 * config/pa/pa.c (pa_output_function_prologue): Likewise.
8122 (pa_output_function_epilogue): Likewise.
8123 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
8124 (rs6000_output_function_epilogue): Likewise.
8125 * config/rl78/rl78.c (rl78_start_function): Likewise.
8126 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
8127 (rs6000_output_function_epilogue): Likewise.
8128 * config/rx/rx.c (rx_output_function_prologue): Likewise.
8129 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
8130 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
8131 (sparc_asm_function_epilogue): Likewise.
8133 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8135 * tree.h (type_has_mode_precision_p): New function.
8136 * convert.c (convert_to_integer_1): Use it.
8137 * expr.c (expand_expr_real_2): Likewise.
8138 (expand_expr_real_1): Likewise.
8139 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
8140 * match.pd: Likewise.
8141 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
8142 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
8143 * tree-tailcall.c (process_assignment): Likewise.
8144 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8145 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
8146 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
8147 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8148 (vectorizable_assignment): Likewise.
8149 (vectorizable_shift): Likewise.
8150 (vectorizable_operation): Likewise.
8151 * tree-vrp.c (register_edge_assert_for_2): Likewise.
8153 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
8155 * match.pd: Add pow (C, x) simplification.
8157 2017-08-21 Richard Biener <rguenther@suse.de>
8159 PR tree-optimization/81900
8160 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
8161 for blocks with abnormal predecessors.
8162 (compute_antic): Do not set visited flag prematurely.
8164 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
8167 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
8169 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8171 * stor-layout.h (vector_type_mode): Move to...
8172 * tree.h (vector_type_mode): ...here.
8173 * stor-layout.c (vector_type_mode): Move to...
8174 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
8176 2017-08-21 Richard Biener <rguenther@suse.de>
8178 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
8179 register_external_die hooks.
8180 (debug_false_tree_charstarstar_uhwistar): Declare.
8181 (debug_nothing_tree_charstar_uhwi): Likewise.
8182 * debug.c (do_nothing_debug_hooks): Adjust.
8183 (debug_false_tree_charstarstar_uhwistar): New do nothing.
8184 (debug_nothing_tree_charstar_uhwi): Likewise.
8185 * dbxout.c (dbx_debug_hooks): Adjust.
8186 (xcoff_debug_hooks): Likewise.
8187 * sdbout.c (sdb_debug_hooks): Likewise.
8188 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
8189 * dwarf2out.c (macinfo_label_base): New global.
8190 (dwarf2out_register_external_die): New function for the
8191 register_external_die hook.
8192 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
8193 (dwarf2_debug_hooks): Use them.
8194 (dwarf2_lineno_debug_hooks): Adjust.
8195 (struct die_struct): Add with_offset flag.
8196 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
8197 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
8198 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
8199 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
8200 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
8201 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
8202 defining section names for the early LTO debug variants.
8203 (reset_indirect_string): New helper.
8204 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
8205 (print_dw_val): Add support for offsetted symbol references.
8206 (get_ultimate_context): Split out from is_cxx.
8207 (is_cxx): Use get_ultimate_context.
8208 (is_fortran): Add decl overload.
8209 (compute_comp_unit_symbol): Split out worker from
8210 compute_section_prefix.
8211 (compute_section_prefix): Call compute_comp_unit_symbol and
8212 set comdat_type_p here.
8213 (output_die): Skip DIE symbol output for the LTO added one.
8214 Handle DIE symbol references with offset.
8215 (output_comp_unit): Guard section name mangling properly.
8216 For LTO debug sections emit a symbol at the section beginning
8217 which we use to refer to its DIEs.
8218 (add_abstract_origin_attribute): For DIEs registered via
8219 dwarf2out_register_external_die directly refer to the early
8220 DIE rather than indirectly through the shadow one we created.
8221 Remove obsolete call to dwarf2out_abstract_function for
8222 non-function/block origins.
8223 (gen_array_type_die): When generating early LTO debug do
8224 not emit DW_AT_string_length.
8225 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
8226 late when in LTO. As suggested place a gcc_unreachable for
8227 the DECL_ABSTRACT_P case.
8228 (gen_subprogram_die): Avoid another specification DIE
8229 for early built declarations/definitions for the late LTO case.
8230 (gen_variable_die): Add type references for late duplicated VLA dies
8232 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
8233 we have the abstract instance already.
8234 (process_scope_var): Adjust decl DIE contexts in LTO which
8235 first puts them in limbo.
8236 (gen_decl_die): Do not generate type DIEs late apart from
8237 types for VLAs or for decls we do not yet have a DIE. Do not
8238 call dwarf2out_abstract_function late.
8239 (dwarf2out_early_global_decl): Make sure to create DIEs
8240 for abstract instances of a decl first.
8241 (dwarf2out_late_global_decl): Adjust comment.
8242 (output_macinfo_op): With multiple macro sections use
8243 macinfo_label_base to distinguish labels.
8244 (output_macinfo): Likewise. Update macinfo_label_base.
8245 Pass in the line info label.
8246 (note_variable_value_in_expr): When generating LTO resolve
8247 all variable values here by generating DIEs as needed.
8248 (init_sections_and_labels): Add early LTO debug flag parameter
8249 and generate different sections and names if set. Add generation
8250 counter for the labels so we can have multiple of them.
8251 (reset_dies): Helper to allow DIEs to be output multiple times.
8252 (dwarf2out_finish): When outputting DIEs to the fat part of an
8253 LTO object first reset DIEs.
8254 (dwarf2out_early_finish): Output early DIEs when generating LTO.
8255 (modified_type_die): Check for decl_ultimate_origin being self
8257 (gen_type_die_with_usage): Likewise.
8258 (gen_typedef_die): Allow decl_ultimate_origin being self.
8259 (set_decl_abstract_flags): Remove.
8260 (set_block_abstract_flags): Likewise.
8261 (dwarf2out_abstract_function): Treat the early generated DIEs
8262 as the abstract copy and only add DW_AT_inline and
8263 DW_AT_artificial here and call set_decl_origin_self.
8264 If the DIE has an abstract origin don't do anything.
8265 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
8266 if we have none yet (Go fails to build one, PR78628).
8267 (variably_modified_type_p): Prevent endless recursion for Ada
8268 cyclic pointer types.
8269 * lto-streamer-in.c: Include debug.h.
8270 (dref_queue): New global.
8271 (lto_read_tree_1): Stream in DIE references.
8272 (lto_input_tree): Register DIE references.
8273 (input_function): Stream DECL_DEBUG_ARGS.
8274 * lto-streamer-out.c: Include debug.h.
8275 (lto_write_tree_1): Output DIE references.
8276 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
8277 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
8278 (output_function): Stream DECL_DEBUG_ARGS.
8279 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
8280 Stream DECL_ABSTRACT_ORIGIN.
8281 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
8282 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
8283 DECL_CONTEXT for file-scope decls.
8284 * lto-streamer.h (struct dref_entry): Declare.
8285 (dref_queue): Likewise.
8286 * cfgexpand.c (pass_expand::execute): Do not call the
8287 outlining_inline_function hook here.
8288 * lto-wrapper.c (debug_obj): New global.
8289 (tool_cleanup): Unlink it if required.
8290 (debug_objcopy): New function.
8291 (run_gcc): Handle early debug sections in the IL files by
8292 extracting them to separate files, partially linkin them and
8293 feeding the result back as result to the linker.
8294 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
8295 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
8296 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
8297 sections into a separate segment.
8298 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
8300 (darwin_asm_dwarf_section): Likewise.
8301 (darwin_asm_output_dwarf_offset): Likewise.
8302 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
8304 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8305 Alan Hayward <alan.hayward@arm.com>
8306 David Sherwood <david.sherwood@arm.com>
8308 * read-md.h (md_reader::record_potential_iterator_use): Replace
8309 pointer argument with an rtx and an index.
8310 * read-rtl.c (iterator_group::apply_iterator): Likewise.
8311 (apply_mode_iterator): Likewise.
8312 (apply_code_iterator): Likewise.
8313 (apply_int_iterator): Likewise.
8314 (apply_subst_iterator): Likewise.
8315 (record_iterator_use): Likewise.
8316 (record_attribute_use): Likewise.
8317 (md_reader::record_potential_iterator_use): Likewise. Update calls
8318 to record_iterator_use and apply_iterator.
8319 (iterator_use): Replace ptr with x and index.
8320 (attribute_use): Likewise.
8321 (apply_attribute_uses): Update calls to apply_iterator.
8322 (apply_iterators): Likewise. Update initialization of iterator_use.
8323 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
8324 and record_potential_iterator_use.
8325 (rtx_reader::read_rtx_operand): Likewise.
8327 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8328 Alan Hayward <alan.hayward@arm.com>
8329 David Sherwood <david.sherwood@arm.com>
8331 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
8334 2017-08-21 Richard Biener <rguenther@suse.de>
8337 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
8338 at struct end conservatively when comparing common bases.
8340 2017-08-21 Richard Biener <rguenther@suse.de>
8342 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
8343 (mem_ref_in_stmt): Remove.
8344 (determine_max_movement): Use ref index to get at the reference.
8345 (invariantness_dom_walker::before_dom_children): Deal with
8346 lim data already initialized.
8347 (gather_mem_refs_stmt): Initialize lim data and record ref index.
8349 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
8351 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
8352 (TARGET_ISA_ROUND): Ditto.
8353 (TARGET_ROUND): Ditto.
8354 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
8355 * config/i386/i386.md: Ditto.
8356 * config/i386/sse.md: Ditto.
8357 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
8358 with OPTION_MASK_ISA_SSE4_1.
8360 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
8363 * doc/extend.texi (x86 Built-in Functions): Correct the name of
8364 __builtin_ia32_lzcnt_u16.
8366 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
8369 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
8370 (rs6000_set_current_function): Rewrite function to use it.
8372 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
8375 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
8376 and TYPE_WARN_IF_NOT_ALIGN.
8377 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
8378 (handle_warn_if_not_align): New.
8379 (place_union_field): Call handle_warn_if_not_align.
8380 (place_field): Call handle_warn_if_not_align.
8381 Copy TYPE_WARN_IF_NOT_ALIGN.
8382 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
8383 (layout_type): Likewise.
8384 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
8386 (tree_decl_common): Add warn_if_not_align.
8387 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
8388 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
8389 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
8390 (DECL_WARN_IF_NOT_ALIGN): Likewise.
8391 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
8392 * doc/extend.texi: Document warn_if_not_aligned attribute.
8393 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
8395 2017-08-17 Martin Liska <mliska@suse.cz>
8398 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
8399 (get_data_dependence): Use it as pointer type.
8400 (distribute_loop): Likewise.
8402 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8404 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
8405 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
8406 (p8_vmrgow_<mode>_direct): New define_insn.
8407 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
8408 handle endianness for vmrgew and vmrgow permute patterns.
8410 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
8412 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
8413 * config/rs6000/rs6000-cpus.def: Remove comment.
8414 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
8415 (POWERPC_MASKS): Likewise.
8416 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
8417 use of TARGET_VSX_TIMODE.
8418 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
8419 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
8420 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
8421 (rs6000_option_override_internal): Remove dead code.
8422 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
8423 (rs6000_legitimize_reload_address): Likewise.
8424 (rs6000_legitimate_address_p): Likewise.
8425 (rs6000_opt_masks): Delete "vsx-timode".
8426 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
8427 from function comment.
8428 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
8429 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
8430 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
8432 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
8433 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
8434 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
8437 (define_peephole2): Likewise.
8439 2017-08-17 Martin Sebor <msebor@redhat.com>
8442 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
8443 past the end of an array.
8444 (test_pp_format): Add test cases.
8446 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
8448 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
8449 missing ECF_NOTHROW flags.
8451 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
8454 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
8455 operands residing in integer registers.
8456 (*vsx_le_perm_load_<mode>): Likewise.
8457 (*vsx_le_perm_store_<mode>): Likewise.
8458 (define_peephole2): Add peepholes to optimize the above.
8460 2017-08-17 Marek Polacek <polacek@redhat.com>
8463 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
8464 to mimic what shorten_compare did. Change the return type to bool.
8465 (fold_cond_expr_with_comparison): Update call to
8466 operand_equal_for_comparison_p.
8467 (fold_ternary_loc): Likewise.
8469 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
8471 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
8473 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
8475 2017-08-17 Richard Biener <rguenther@suse.de>
8477 * tree-ssa-structalias.c (solve_graph): When propagating
8478 to successors update the graphs succ edges and avoid duplicate work.
8480 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
8483 * config/i386/i386.c (ix86_option_override_internal): Save target
8484 specific options after ix86_stack_protector_guard_reg was changed.
8486 2017-08-17 Richard Biener <rguenther@suse.de>
8488 PR tree-optimization/81827
8489 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
8490 (new_var_info): Initialize it conservatively.
8491 (get_call_vi): Mark register vars.
8492 (new_scalar_tmp_constraint_exp): Likewise.
8493 (handle_rhs_call): Likewise.
8494 (handle_const_call): Likewise.
8495 (create_function_info_for): Likewise.
8496 (solve_constraints): Sort varinfos to separate register from
8497 non-register vars to pack points-to solution bitmaps during
8500 2017-08-17 Marek Polacek <polacek@redhat.com>
8502 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
8504 2017-08-17 Richard Biener <rguenther@suse.de>
8506 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
8507 to true when overflow is undefined and we saturated the result.
8509 2017-08-17 Alan Modra <amodra@gmail.com>
8512 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
8513 Don't use store multiple if only one reg needs saving.
8514 (interesting_frame_related_regno): New function.
8515 (rs6000_frame_related): Don't emit frame info for regs that
8517 (rs6000_emit_epilogue): Likewise.
8519 2017-08-16 Nathan Sidwell <nathan@acm.org>
8521 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
8522 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
8523 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
8524 * tree.c (free_lang_data_in_type): Use else-if chain. Always
8525 clear TYPE_LANG_1. Remove obsolete member-function stripping.
8526 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
8527 (verify_type): Adjust for TYPE_BINFO move.
8528 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
8529 process TYPE_BINFO directly.
8530 (hash_tree): Likewise.
8531 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
8533 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
8536 2017-08-16 David Malcolm <dmalcolm@redhat.com>
8538 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
8540 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
8543 * config/i386/i386.md (*anddi_1_btr): Change predicates of
8544 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
8545 Add ix86_binary_operator_ok to insn constraint.
8546 (*iordi_1_bts): Ditto.
8547 (*xordi_1_btc): Ditto.
8548 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
8549 Update corresponding peephole2 pattern.
8553 2017-08-16 Bin Cheng <bin.cheng@arm.com>
8555 PR tree-optimization/81832
8556 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
8557 copy loop header which has IFN_LOOP_DIST_ALIAS call.
8559 2017-08-16 Marek Polacek <polacek@redhat.com>
8562 * fold-const.c (fold_indirect_ref_1): Restore original behavior
8563 regarding size_zero_node.
8565 2017-08-16 Martin Liska <mliska@suse.cz>
8568 * config.gcc: Respect previously set extra_objs in case
8571 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
8573 PR tree-optimization/81835
8574 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
8575 the phi SSA_NAME. Check that the condition in a COND_EXPR does
8576 not depend on the phi.
8578 2017-08-16 Alan Modra <amodra@gmail.com>
8580 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
8583 2017-08-16 Alan Modra <amodra@gmail.com>
8585 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
8586 (save_reg_p): ..into this. Update all callers.
8587 (first_reg_to_save): Simplify.
8589 2017-08-16 Alan Modra <amodra@gmail.com>
8591 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
8594 2017-08-15 Joseph Myers <joseph@codesourcery.com>
8598 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
8599 constant count if that count is less than 32.
8601 2017-08-15 Nathan Sidwell <nathan@acm.org>
8603 * gcc.c (execute): Emit friendlier message if inferior is killed
8604 by an external cause.
8606 2017-08-15 Richard Biener <rguenther@suse.de>
8608 PR tree-optimization/81790
8609 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
8610 CONSTRUCTORs from simplifying and VN.
8612 2017-08-14 Martin Sebor <msebor@redhat.com>
8614 * builtin-attrs.def: Add comments.
8616 2017-08-14 Martin Sebor <msebor@redhat.com>
8619 * doc/extend.texi (attribute nonstring): Document new attribute.
8621 2017-08-14 Martin Sebor <msebor@redhat.com>
8624 * tree-diagnostic.c (default_tree_printer): Handle %G.
8625 * gimple-pretty-print.h (percent_G_format): Declare new function.
8626 * gimple-pretty-print.c (percent_G_format): Define.
8627 * tree-pretty-print.c (percent_K_format): Add argument.
8629 2017-08-14 Martin Sebor <msebor@redhat.com>
8631 PR translation/79998
8632 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
8633 Remove a stray space.
8635 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
8638 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
8639 (*iordi_1_bts): Ditto.
8640 (*xordi_1_btc): Ditto.
8642 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8645 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
8646 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8648 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
8649 quoted strings, and make more translator-friendly.
8650 (darwin_rs6000_override_options): Likewise.
8651 (rs6000_option_override_internal): Likewise.
8652 (rs6000_return_in_memory): Fix overlong line.
8653 (init_cmulative_args): Use quoted strings, and make more
8654 translator-friendly.
8655 (rs6000_pass_by_reference): Fix overlong line.
8656 (def_builtin): Use quoted strings.
8657 (altivec_expand_predicate_builtin): Use quoted strings, and make
8658 more translator-friendly.
8659 (htm_expand_builtin): Use quoted strings.
8660 (cpu_expand_builtin): Use quoted strings, and make more
8661 translator-friendly.
8662 (altivec_expand_builtin): Likewise.
8663 (paired_expand_predicate_builtin): Likewise.
8664 (rs6000_invalid_builtin): Likewise.
8665 (builtin_function_type): Use quoted strings.
8666 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
8667 more translator-friendly.
8668 (rs6000_trampoline_init): Likewise.
8669 (rs6000_handle_altivec_attribute): Likewise.
8670 (rs6000_inner_target_options): Use quoted strings.
8671 (rs6000_disable_incompatible_switches): Likewise.
8672 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
8673 strings, and make more translator-friendly.
8674 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
8676 2017-08-14 Bin Cheng <bin.cheng@arm.com>
8678 PR tree-optimization/81799
8679 * tree-loop-distribution.c (version_loop_by_alias_check): Force
8680 cond_expr to simple gimple operand.
8682 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
8685 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
8687 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
8691 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
8692 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
8693 TARGET_GASISR_PROLOGUES.
8694 * config/avr/avr.c (avr_option_override): Same.
8695 (avr_pass_pre_proep::execute): Same.
8697 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
8700 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
8701 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
8703 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
8705 * config/i386/i386.md (*load_tp_<mode>): Redefine as
8706 define_insn_and_split. Split to a memory load from 0 in
8707 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
8708 using PTR mode iterator.
8709 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
8710 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
8711 (*add_tp_<mode>): Redefine as define_insn_and_split.
8712 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
8713 address space. Merge with *add_tp_x32 using PTR mode iterator.
8714 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
8715 Split to an add with a memory load from 0 in
8716 DEFAULT_TLS_SEG_REG address space.
8718 2017-08-12 Andrew Pinski <apinski@cavium.com>
8720 * config/aarch64/aarch64-option-extensions.def (rdma):
8721 Fix feature string to what Linux prints out in /proc/cpuinfo.
8723 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
8726 * dwarf2out.c (modified_type_die): For C typedef types that have
8727 an ultimate origin, process the ultimate origin instead of the
8729 (gen_typedef_die): Assert that input DECLs have no ultimate
8731 (gen_type_die_with_usage): For typedef variants that have an
8732 ultimate origin, just call gen_decl_die on the original DECL.
8733 (process_scope_var): Avoid creating DIEs for local typedefs and
8734 concrete static variables.
8736 2017-08-12 Alan Modra <amodra@gmail.com>
8740 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
8741 match gnu-user.h startfile.
8742 (ENDFILE_LINUX_SPEC): Similarly.
8744 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
8747 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
8749 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
8751 2017-08-11 Tamar Christina <tamar.christina@arm.com>
8752 * config/aarch64/aarch64.md (mov<mode>): Change.
8753 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
8754 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
8755 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
8757 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
8759 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
8760 for storage order barriers.
8762 2017-08-11 Martin Liska <mliska@suse.cz>
8765 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
8766 variables of void type.
8768 2017-08-11 Martin Liska <mliska@suse.cz>
8770 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
8771 TARGET_SUPPORTS_ALIASES.
8772 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
8773 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
8774 (optimize_weakref): Likewise.
8775 * symtab.c (symtab_node::noninterposable_alias): Likewise.
8776 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
8777 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
8779 2017-08-11 Martin Liska <mliska@suse.cz>
8782 * config/i386/i386.c (make_resolver_func): Do complete
8783 refactoring of the function.
8785 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
8788 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
8789 * config/i386/i386.c (ix86_stack_protect_guard): Use
8790 ix86_stack_protect_guard_symbol_str to generate varible declaration.
8791 * doc/invoke.texi (x86 Options): Document
8792 -mstack-protector-guard-symbol= option.
8794 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
8796 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
8797 * config/i386/i386.c (ix86_split_stack_guard): New function.
8798 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
8799 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
8800 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
8801 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
8802 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
8803 (split_stack_space_check): Call ix86_split_stack_guard.
8805 2017-08-10 Martin Sebor <msebor@redhat.com>
8807 * print-tree.c (print_node): Print location using the established
8809 Replace spaces with colons.
8810 (debug_raw, debug): Ditto.
8812 2017-08-10 Martin Sebor <msebor@redhat.com>
8815 * pretty-print.c (pp_format): Correct the handling of %s precision.
8817 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
8820 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
8822 (ix86_finalize_stack_frame_flags): This. Also clear
8823 frame_pointer_needed if -fno-omit-frame-pointer is used without
8825 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
8826 with ix86_finalize_stack_frame_flags.
8827 (ix86_expand_epilogue): Likewise.
8828 (ix86_expand_split_stack_prologue): Likewise.
8829 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
8831 2017-08-10 Martin Liska <mliska@suse.cz>
8834 * c-attribs.c (handle_target_attribute):
8835 Report warning for an empty string argument of target attribute.
8837 2017-08-09 Jakub Jelinek <jakub@redhat.com>
8840 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
8842 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
8843 or DECL_NONLOCAL labels.
8844 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
8845 or DECL_NONLOCAL labels here.
8847 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
8849 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
8850 to indicate when early gimple folding has been disabled.
8851 (rs6000_gimple_fold_builtin): Add debug content.
8852 (rs6000_invalid_builtin): Fix whitespace.
8853 (rs6000_expand_builtin): Fix whitespace.
8854 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
8856 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
8859 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
8860 SAVE_MULTIPLE if not all the registers that saves, should be saved.
8862 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
8864 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
8865 (qdf24xx): Likewise.
8866 * config/aarch64/aarch64.md: Include falkor.md.
8867 * config/aarch64/falkor.md: New.
8869 2017-08-09 Marek Polacek <polacek@redhat.com>
8872 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
8873 * diagnostic.c (emit_diagnostic): Add a comment.
8874 (emit_diagnostic_valist): New function.
8876 2017-08-09 Marek Polacek <polacek@redhat.com>
8879 * input.c (make_location): New overload.
8880 * input.h (make_location): Declare.
8882 2017-08-08 Alan Modra <amodra@gmail.com>
8883 H.J. Lu <hongjiu.lu@intel.com>
8886 * gcc.c (NO_PIE_SPEC): Delete.
8887 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
8889 (LINK_PIE_SPEC): ..to here.
8890 (LINK_COMMAND_SPEC): Support -no-pie.
8891 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
8892 chain of crtbegin*.o selection, update for PIE_SPEC changes and
8894 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
8895 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
8896 (ENDFILE_CRTEND_SPEC): Similarly.
8898 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
8901 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
8902 (mstack-protector-guard-offset=): Ditto.
8903 * config/i386/i386.c (ix86_option_override): Handle
8904 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
8906 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
8907 ix86_stack_protect_guard_offset variables.
8908 (TARGET_STACK_PROTECT_GUARD): Always define.
8909 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
8910 and -mstack-protector-guard-offset= options.
8912 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8914 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
8915 boundary case for the last candidate.
8917 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8919 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
8920 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
8922 2017-08-08 Tamar Christina <tamar.christina@arm.com>
8925 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
8926 * config/aarch64/aarch64-builtins.c
8927 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
8928 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
8929 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
8931 2017-08-08 Tamar Christina <tamar.christina@arm.com>
8932 Andrew Pinski <pinskia@gmail.com>
8935 * internal-fn.def (XORSIGN): New.
8936 * optabs.def (xorsign_optab): New.
8937 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
8938 (convert_expand_mult_copysign): New.
8939 (pass_optimize_widening_mul::execute): Call
8940 convert_expand_mult_copysign.
8942 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8944 PR tree-optimization/81354
8945 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
8946 Insert on edges rather than explicitly creating landing pads.
8947 (analyze_candidates_and_replace): Commit edge inserts.
8949 2017-08-08 Richard Biener <rguenther@suse.de>
8952 * tree-ssa-loop-niter.c: Include tree-dfa.h.
8953 (expand_simple_operations): Also look through ADDR_EXPRs with
8954 MEM_REF bases treating them as POINTER_PLUS_EXPR.
8956 2017-08-08 Richard Biener <rguenther@suse.de>
8958 PR tree-optimization/81723
8959 * tree-vect-slp.c (struct bst_traits): New hash traits.
8960 (bst_fail): New global.
8961 (vect_build_slp_tree_2): New worker, split out from ...
8962 (vect_build_slp_tree): ... this now wrapping it with using
8963 bst_fail set to cache SLP tree build fails. Properly handle
8965 (vect_analyze_slp_instance): Allocate and free bst_fail.
8967 2017-08-08 Martin Liska <mliska@suse.cz>
8970 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
8971 LABEL_DECLs that can be from a different function.
8973 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8975 PR tree-optimization/81744
8976 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
8977 loop's number of iterations.
8979 2017-08-08 Martin Liska <mliska@suse.cz>
8981 * asan.c: Include header files.
8982 * attribs.c (build_decl_attribute_variant): New function moved
8984 (build_type_attribute_qual_variant): Likewise.
8985 (cmp_attrib_identifiers): Likewise.
8986 (simple_cst_list_equal): Likewise.
8987 (omp_declare_simd_clauses_equal): Likewise.
8988 (attribute_value_equal): Likewise.
8989 (comp_type_attributes): Likewise.
8990 (build_type_attribute_variant): Likewise.
8991 (lookup_ident_attribute): Likewise.
8992 (remove_attribute): Likewise.
8993 (merge_attributes): Likewise.
8994 (merge_type_attributes): Likewise.
8995 (merge_decl_attributes): Likewise.
8996 (merge_dllimport_decl_attributes): Likewise.
8997 (handle_dll_attribute): Likewise.
8998 (attribute_list_equal): Likewise.
8999 (attribute_list_contained): Likewise.
9000 * attribs.h (lookup_attribute): New function moved from tree.[ch].
9001 (lookup_attribute_by_prefix): Likewise.
9002 * bb-reorder.c: Include header files.
9003 * builtins.c: Likewise.
9004 * calls.c: Likewise.
9005 * cfgexpand.c: Likewise.
9006 * cgraph.c: Likewise.
9007 * cgraphunit.c: Likewise.
9008 * convert.c: Likewise.
9009 * dwarf2out.c: Likewise.
9010 * final.c: Likewise.
9011 * fold-const.c: Likewise.
9012 * function.c: Likewise.
9013 * gimple-expr.c: Likewise.
9014 * gimple-fold.c: Likewise.
9015 * gimple-pretty-print.c: Likewise.
9016 * gimple.c: Likewise.
9017 * gimplify.c: Likewise.
9018 * hsa-common.c: Likewise.
9019 * hsa-gen.c: Likewise.
9020 * internal-fn.c: Likewise.
9021 * ipa-chkp.c: Likewise.
9022 * ipa-cp.c: Likewise.
9023 * ipa-devirt.c: Likewise.
9024 * ipa-fnsummary.c: Likewise.
9025 * ipa-inline.c: Likewise.
9026 * ipa-visibility.c: Likewise.
9028 * lto-cgraph.c: Likewise.
9029 * omp-expand.c: Likewise.
9030 * omp-general.c: Likewise.
9031 * omp-low.c: Likewise.
9032 * omp-offload.c: Likewise.
9033 * omp-simd-clone.c: Likewise.
9034 * opts-global.c: Likewise.
9035 * passes.c: Likewise.
9036 * predict.c: Likewise.
9037 * sancov.c: Likewise.
9038 * sanopt.c: Likewise.
9039 * symtab.c: Likewise.
9040 * toplev.c: Likewise.
9041 * trans-mem.c: Likewise.
9042 * tree-chkp.c: Likewise.
9043 * tree-eh.c: Likewise.
9044 * tree-into-ssa.c: Likewise.
9045 * tree-object-size.c: Likewise.
9046 * tree-parloops.c: Likewise.
9047 * tree-profile.c: Likewise.
9048 * tree-ssa-ccp.c: Likewise.
9049 * tree-ssa-live.c: Likewise.
9050 * tree-ssa-loop.c: Likewise.
9051 * tree-ssa-sccvn.c: Likewise.
9052 * tree-ssa-structalias.c: Likewise.
9053 * tree-ssa.c: Likewise.
9054 * tree-streamer-in.c: Likewise.
9055 * tree-vectorizer.c: Likewise.
9056 * tree-vrp.c: Likewise.
9058 * ubsan.c: Likewise.
9059 * varasm.c: Likewise.
9060 * varpool.c: Likewise.
9061 * tree.c: Remove functions moved to attribs.[ch].
9063 * config/aarch64/aarch64.c: Add attrs.h header file.
9064 * config/alpha/alpha.c: Likewise.
9065 * config/arc/arc.c: Likewise.
9066 * config/arm/arm.c: Likewise.
9067 * config/avr/avr.c: Likewise.
9068 * config/bfin/bfin.c: Likewise.
9069 * config/c6x/c6x.c: Likewise.
9070 * config/cr16/cr16.c: Likewise.
9071 * config/cris/cris.c: Likewise.
9072 * config/darwin.c: Likewise.
9073 * config/epiphany/epiphany.c: Likewise.
9074 * config/fr30/fr30.c: Likewise.
9075 * config/frv/frv.c: Likewise.
9076 * config/ft32/ft32.c: Likewise.
9077 * config/h8300/h8300.c: Likewise.
9078 * config/i386/winnt.c: Likewise.
9079 * config/ia64/ia64.c: Likewise.
9080 * config/iq2000/iq2000.c: Likewise.
9081 * config/lm32/lm32.c: Likewise.
9082 * config/m32c/m32c.c: Likewise.
9083 * config/m32r/m32r.c: Likewise.
9084 * config/m68k/m68k.c: Likewise.
9085 * config/mcore/mcore.c: Likewise.
9086 * config/microblaze/microblaze.c: Likewise.
9087 * config/mips/mips.c: Likewise.
9088 * config/mmix/mmix.c: Likewise.
9089 * config/mn10300/mn10300.c: Likewise.
9090 * config/moxie/moxie.c: Likewise.
9091 * config/msp430/msp430.c: Likewise.
9092 * config/nds32/nds32-isr.c: Likewise.
9093 * config/nds32/nds32.c: Likewise.
9094 * config/nios2/nios2.c: Likewise.
9095 * config/nvptx/nvptx.c: Likewise.
9096 * config/pa/pa.c: Likewise.
9097 * config/pdp11/pdp11.c: Likewise.
9098 * config/powerpcspe/powerpcspe.c: Likewise.
9099 * config/riscv/riscv.c: Likewise.
9100 * config/rl78/rl78.c: Likewise.
9101 * config/rx/rx.c: Likewise.
9102 * config/s390/s390.c: Likewise.
9103 * config/sh/sh.c: Likewise.
9104 * config/sol2.c: Likewise.
9105 * config/sparc/sparc.c: Likewise.
9106 * config/spu/spu.c: Likewise.
9107 * config/stormy16/stormy16.c: Likewise.
9108 * config/tilegx/tilegx.c: Likewise.
9109 * config/tilepro/tilepro.c: Likewise.
9110 * config/v850/v850.c: Likewise.
9111 * config/vax/vax.c: Likewise.
9112 * config/visium/visium.c: Likewise.
9113 * config/xtensa/xtensa.c: Likewise.
9115 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9118 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
9119 constraints since the -mupper-regs-* switches have been
9121 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
9122 into a vector from a double word element that was extracted from
9123 another vector, and eliminate extra XXPERMDI instructions.
9124 (vsx_concat_<mode>_2): Likewise.
9125 (vsx_concat_<mode>_3): Likewise.
9126 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
9127 concat to allow optimizing inserts from previous extracts.
9129 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
9131 * config/i386/i386.c (ix86_stack_protect_guard): Generate
9132 memory reference to a SSP offset in TLS address space.
9133 (ix86_print_operand) <case '@'>: Remove.
9134 (ix86_print_operand_punct_valid_p): Remove '@' code.
9135 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
9137 (stack_tls_protect_set_<mode>): Remove.
9138 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
9139 (stack_tls_protect_test_<mode>): Remove.
9140 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
9142 2017-08-07 Olivier Hainque <hainque@adacore.com>
9145 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
9147 2017-08-07 Douglas Rupp <rupp@adacore.com>
9149 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
9150 variable was referenced as multidir in command.
9152 2017-08-07 Jakub Jelinek <jakub@redhat.com>
9155 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
9158 2017-08-07 Martin Liska <mliska@suse.cz>
9160 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
9161 * config/rl78/rl78.c: Add include of attribs.h.
9162 * config/sh/sh.c: Likewise.
9163 * config/v850/v850.c: Likewise.
9165 2017-08-07 Tom de Vries <tom@codesourcery.com>
9168 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
9170 2017-08-07 Martin Liska <mliska@suse.cz>
9172 * config/mips/mips.c: Include attribs.h.
9174 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
9177 * doc/invoke.texi: Document change in behvaior for -Ofast for
9180 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
9182 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
9184 (aarch64_pop_regs): Likewise.
9185 (aarch64_gen_load_pair): Likewise.
9186 (aarch64_save_callee_saves): Likewise.
9187 (aarch64_restore_callee_saves): Likewise.
9189 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
9191 * config/i386/i386.c: Revert the last change.
9193 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
9196 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
9198 (ix86_finalize_stack_frame_flags): This. Also clear
9199 frame_pointer_needed if -fno-omit-frame-pointer is used without
9201 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
9202 with ix86_finalize_stack_frame_flags.
9203 (ix86_expand_epilogue): Likewise.
9204 (ix86_expand_split_stack_prologue): Likewise.
9206 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
9209 * config/i386/i386.c (get_builtin_code_for_version): Set priority
9210 to P_AES for Westmere.
9212 2017-08-07 Jonathan Yong <10walls@gmail.com>
9214 * config/i386/mingw.opt (fset-stack-executable): Removed.
9215 * config/i386/cygming.opt (fset-stack-executable): Moved
9217 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
9219 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
9221 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
9223 2017-08-07 Marek Polacek <polacek@redhat.com>
9226 * fold-const.c (fold_indirect_ref_1): Check type_domain.
9228 2017-08-07 Martin Liska <mliska@suse.cz>
9230 * attribs.h (canonicalize_attr_name): New function.
9231 (cmp_attribs): Move from c-format.c and adjusted.
9232 (is_attribute_p): Moved from tree.h.
9233 * tree-inline.c: Add new includes.
9234 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
9235 (private_is_attribute_p): Remove.
9236 (private_lookup_attribute): Likewise.
9237 (private_lookup_attribute_by_prefix): Simplify.
9238 (remove_attribute): Use is_attribute_p.
9239 * tree.h: Remove removed declarations.
9241 2017-08-07 Jakub Jelinek <jakub@redhat.com>
9244 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
9245 instead of computing it in the function. Formatting fix.
9246 (expand_case): Don't rely on default_edge being the first edge,
9247 clear it if removing it, pass default_edge to
9248 emit_case_dispatch_table.
9249 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
9252 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
9254 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
9255 insn in the function, emit NOP after the insn.
9257 2017-08-06 Tom de Vries <tom@codesourcery.com>
9259 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
9262 2017-08-06 Tom de Vries <tom@codesourcery.com>
9264 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
9267 2017-08-04 Yury Gribov <tetra2005@gmail.com>
9269 PR tree-optimization/57371
9270 * match.pd: New pattern.
9272 2017-08-04 Marek Polacek <polacek@redhat.com>
9275 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
9276 perform the computation in offset_int.
9278 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
9280 PR tree-optimization/81136
9281 * tree-vectorizer.h: Include tree-hash-traits.h.
9282 (vec_base_alignments): New typedef.
9283 (vec_info): Add a base_alignments field.
9284 (vect_record_base_alignments): Declare.
9285 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
9287 (DR_IS_CONDITIONAL_IN_STMT): New macro.
9288 (create_data_ref): Add an is_conditional_in_stmt argument.
9289 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
9290 the is_conditional_in_stmt field.
9291 (data_ref_loc): Add an is_conditional_in_stmt field.
9292 (get_references_in_stmt): Set the is_conditional_in_stmt field.
9293 (find_data_references_in_stmt): Update call to create_data_ref.
9294 (graphite_find_data_references_in_stmt): Likewise.
9295 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
9296 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
9297 (vect_record_base_alignment): New function.
9298 (vect_record_base_alignments): Likewise.
9299 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
9300 for nested statements even if we fail to compute a misalignment.
9301 Use pooled base alignments for unconditional references.
9302 (vect_find_same_alignment_drs): Compare base addresses instead
9304 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
9305 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
9307 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
9309 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
9310 Add an explicit name for the enum. Use auto_vec for slp_instances
9312 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
9314 (_bb_vec_info): Add a constructor and destructor.
9315 (vinfo_for_stmt): Return NULL for uids of -1 as well.
9316 (destroy_loop_vec_info): Delete.
9317 (vect_destroy_datarefs): Likewise.
9318 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
9319 (vec_info::vec_info): New function.
9320 (vec_info::~vec_info): Likewise.
9321 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
9322 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
9323 destroy_loop_vec_info.
9324 * tree-vect-loop.c (new_loop_vec_info): Replace with...
9325 (_loop_vec_info::_loop_vec_info): ...this.
9326 (destroy_loop_vec_info): Replace with...
9327 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
9328 the stmt_vec_infos. Leave handling of vec_info information to its
9329 destructor. Remove explicit vector releases.
9330 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
9331 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
9332 * tree-vect-slp.c (new_bb_vec_info): Replace with...
9333 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
9334 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
9335 (destroy_bb_vec_info): Replace with...
9336 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
9337 information to its destructor.
9338 (vect_slp_analyze_bb_1): Use new and delete instead of
9339 new_bb_vec_info and destroy_bb_vec_info.
9340 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
9343 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
9345 * tree-data-ref.h (subscript): Add access_fn field.
9346 (data_dependence_relation): Add could_be_independent_p.
9347 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
9348 (same_access_functions): Move to tree-data-ref.c.
9349 * tree-data-ref.c (ref_contains_union_access_p): New function.
9350 (access_fn_component_p): Likewise.
9351 (access_fn_components_comparable_p): Likewise.
9352 (dr_analyze_indices): Add a reference to access_fn_component_p.
9353 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
9355 (constant_access_functions): Likewise.
9356 (add_other_self_distances): Likewise.
9357 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
9358 (initialize_data_dependence_relation): Use XCNEW and remove
9359 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
9360 of access functions that have the same type. Allow the
9361 subsequence to end with different bases in some circumstances.
9362 Record the chosen access functions in SUB_ACCESS_FN.
9363 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
9364 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
9365 (subscript_dependence_tester_1): Likewise dra and drb.
9366 (build_classic_dist_vector): Update calls accordingly.
9367 (subscript_dependence_tester): Likewise.
9368 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
9369 DDR_COULD_BE_INDEPENDENT_P.
9370 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
9371 comp_alias_ddrs instead of may_alias_ddrs.
9372 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
9374 (vect_analyze_data_ref_dependence): Use it if
9375 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
9376 distance vectors if that fails.
9377 (dependence_distance_ge_vf): New function.
9378 (vect_prune_runtime_alias_test_list): Use it. Don't clear
9379 LOOP_VINFO_MAY_ALIAS_DDRS.
9381 2017-08-04 Richard Biener <rguenther@suse.de>
9384 * fold-const.c (fold_binary_loc): Properly restrict
9385 minus_var0 && minus_var1 case when associating undefined overflow
9388 2017-08-04 Tom de Vries <tom@codesourcery.com>
9390 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
9392 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9394 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9395 Don't start diagnostic messages with a capital letter.
9396 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9398 (rs6000_invalid_builtin): Likewise.
9399 (rs6000_trampoline_init): Likewise.
9401 2017-08-03 Jakub Jelinek <jakub@redhat.com>
9404 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
9405 after setting changeable df flags.
9407 2017-08-03 Richard Biener <rguenther@suse.de>
9409 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
9410 up if the use is in USE - X.
9412 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
9414 * toplev.c (dumpfile.h): New include.
9415 (internal_error_reentered): New static function. Use it...
9416 (internal_error_function): ...here to handle reentered internal_error.
9418 2017-08-03 Richard Biener <rguenther@suse.de>
9421 * fold-const.c (split_tree): Add minus_var and minus_con
9422 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
9423 here but always use minus_*.
9424 (associate_trees): Assert we never associate with MINUS_EXPR
9425 and NULL first operand. Do not recurse for PLUS_EXPR operands
9426 when associating as MINUS_EXPR either.
9427 (fold_binary_loc): Track minus_var and minus_con.
9429 2017-08-03 Tom de Vries <tom@codesourcery.com>
9432 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
9433 ACCEL_COMPILER, apply finish_options on
9434 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
9436 2017-08-03 Tom de Vries <tom@codesourcery.com>
9439 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
9440 function_entry_patch_area_size > 0.
9442 2017-08-03 Jakub Jelinek <jakub@redhat.com>
9445 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
9446 instead of 10??LU, perform unit multiplication in wide_int,
9447 don't change alloc_object_size_limit if the limit is larger
9450 PR tree-optimization/81655
9451 PR tree-optimization/81588
9452 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
9453 the case when ranges[i].low and high are 1 for unsigned type with
9457 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
9458 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
9460 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9462 * tree-vrp.h: Add include guard.
9464 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
9467 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
9468 (ud2): New insn pattern.
9469 * config/i386/i386.c (ix86_expand_epilogue):
9470 For naked functions, generate ud2 instead of trap insn.
9472 2017-08-02 Marek Polacek <polacek@redhat.com>
9475 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
9477 2017-08-02 Tom de Vries <tom@codesourcery.com>
9478 Cesar Philippidis <cesar@codesourcery.com>
9480 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
9481 Add missing edge probabilities.
9483 2017-08-02 Tamar Christina <tamar.christina@arm.com>
9485 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
9488 2017-08-02 Jakub Jelinek <jakub@redhat.com>
9491 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
9492 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
9493 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
9494 prologue_seq sequences - if any.
9496 2017-08-02 Richard Biener <rguenther@suse.de>
9498 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
9499 via vectors if supported, integer extracts via punning if supported
9500 or otherwise vector extracts.
9502 2017-08-02 Richard Biener <rguenther@suse.de>
9504 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
9506 (bitmap_insert_into_set): ... this.
9508 2017-08-02 Richard Biener <rguenther@suse.de>
9510 PR tree-optimization/81633
9512 2015-08-17 Alan Hayward <alan.hayward@arm.com>
9514 PR tree-optimization/71752
9515 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
9517 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
9519 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
9520 (machine_function::call_ms2sysv_pad_out): Remove field.
9521 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
9522 (ix86_compute_frame_layout): Likewise.
9524 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
9527 * config/i386/i386.c (ix86_set_func_type): Disallow naked
9528 attribute with interrupt attribute.
9530 2017-08-01 Andrew Pinski <apinski@cavium.com>
9532 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
9533 BIT_INSERT_EXPR's operand 1
9534 to see if the types precision matches.
9536 2017-08-01 Martin Liska <mliska@suse.cz>
9539 * builtins.c (expand_builtin_memcpy_args): Remove.
9540 (expand_builtin_memcpy): Call newly added function
9541 expand_builtin_memory_copy_args.
9542 (expand_builtin_memcpy_with_bounds): Likewise.
9543 (expand_builtin_mempcpy): Remove last argument.
9544 (expand_builtin_mempcpy_with_bounds): Likewise.
9545 (expand_builtin_memory_copy_args): New function created from
9546 expand_builtin_mempcpy_args with small modifications.
9547 (expand_builtin_mempcpy_args): Remove.
9548 (expand_builtin_stpcpy): Remove unused argument.
9549 (expand_builtin): Likewise.
9550 (expand_builtin_with_bounds): Likewise.
9552 2017-08-01 Martin Liska <mliska@suse.cz>
9555 Make mempcpy more optimal (PR middle-end/70140).
9557 2017-08-01 Jakub Jelinek <jakub@redhat.com>
9560 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9561 __builtin_vec_cmpne verify both arguments are compatible vectors
9562 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
9563 verify arg1_type is a pointer or array type. For __builtin_vec_st,
9564 move computation of aligned to after checking the argument types.
9568 * config/rs6000/vsx.md (vextract_fp_from_shorth,
9569 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
9572 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
9573 Doug Rupp <rupp@adacore.com>
9574 Olivier Hainque <hainque@adacore.com>
9576 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
9577 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
9579 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
9580 for TARGET_OS_CPP_BUILTIN.
9581 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
9582 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
9584 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
9585 passing required abi options to the assembler for EABI configurations.
9586 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
9587 of .text.hot and .text.unlikely sections for kernel modules when
9588 using ARM style exceptions.
9589 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
9590 options. Add EXTRA_CC1_SPEC.
9591 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
9593 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
9595 (ARM_TARGET2_DWARF_FORMAT): Define.
9596 * config/arm/t-vxworks: Adjust multilib control to removal of the
9597 Diab command line options.
9599 2017-08-01 Martin Liska <mliska@suse.cz>
9601 PR gcov-profile/81561
9602 * gcov.c (unblock): Make unblocking safe as we need to preserve
9603 index correspondence of blocks and block_lists.
9605 2017-08-01 Richard Biener <rguenther@suse.de>
9607 PR tree-optimization/81181
9608 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9609 (compute_antic): ... end of iteration here.
9611 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
9613 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
9614 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
9615 (ftree-slp-vectorize): Likewise.
9616 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
9617 can no longer be set independent of flag_tree_loop_vectorize.
9618 * omp-general.c (emp_max_vf): Likewise.
9619 * opts.c (enable_fdo_optimizations): Remove references to
9620 flag_tree_vectorize, these are now implicit.
9621 (common_handle_option): Remove handling for OPT_ftree_vectorize,
9622 and leave it for the options machinery.
9624 2017-08-01 Martin Liska <mliska@suse.cz>
9627 * builtins.c (expand_builtin_memcpy_args): Remove.
9628 (expand_builtin_memcpy): Call newly added function
9629 expand_builtin_memory_copy_args.
9630 (expand_builtin_memcpy_with_bounds): Likewise.
9631 (expand_builtin_mempcpy): Remove last argument.
9632 (expand_builtin_mempcpy_with_bounds): Likewise.
9633 (expand_builtin_memory_copy_args): New function created from
9634 expand_builtin_mempcpy_args with small modifications.
9635 (expand_builtin_mempcpy_args): Remove.
9636 (expand_builtin_stpcpy): Remove unused argument.
9637 (expand_builtin): Likewise.
9638 (expand_builtin_with_bounds): Likewise.
9640 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
9643 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
9644 print "ds:" only for immediates in generic address space.
9646 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
9649 * config/i386/i386.c (ix86_funciton_naked): New prototype.
9650 (ix86_function_ok_for_sibcall): Return false for naked functions.
9652 2017-08-01 Richard Biener <rguenther@suse.de>
9654 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
9655 (compute_antic): Seed worklist with exit block predecessors.
9656 * cfganal.c (dfs_find_deadend): For a cycle return the source
9657 of the edge closing it.
9659 2017-08-01 Tamar Christina <tamar.christina@arm.com>
9661 * config/aarch64/aarch64.c
9662 (aarch64_can_const_movi_rtx_p): Move 0 check.
9664 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9666 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
9667 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
9669 * match.pd: Ditto in address comparison pattern.
9671 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9673 PR tree-optimization/81627
9674 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
9675 closed ssa form for store-store chain.
9677 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9679 PR tree-optimization/81620
9680 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
9681 for store-store chain.
9683 2017-08-01 Jakub Jelinek <jakub@redhat.com>
9685 PR tree-optimization/81588
9686 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
9687 ranges[i].in_p, invert comparison code ccode. For >/>=,
9688 swap rhs1 and rhs2 and comparison code unconditionally,
9689 for </<= don't do that. Don't swap rhs1/rhs2 again if
9690 ranges[i].in_p, instead invert comparison code ccode if
9691 opcode or oe->rank is BIT_IOR_EXPR.
9694 * optabs.def (vec_extract_optab, vec_init_optab): Change from
9695 a direct optab to conversion optab.
9696 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
9697 with GET_MODE_INNER as last argument instead of optab_handler.
9698 * expmed.c (extract_bit_field_1): Likewise. Use vector from
9699 vector extraction if possible and optab is available.
9700 * expr.c (store_constructor): Use convert_optab_handler instead
9701 of optab_handler. Use vector initialization from smaller
9702 vectors if possible and optab is available.
9703 * tree-vect-stmts.c (vectorizable_load): Likewise.
9704 * doc/md.texi (vec_extract, vec_init): Document that the optabs
9706 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
9707 of vec_init from half-sized vectors with the same element mode.
9708 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
9709 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
9710 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
9711 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
9712 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
9713 after mode in gen_vec_extract* calls.
9714 (vec_extract<mode>): Renamed to ...
9715 (vec_extract<mode><ssescalarmodelower>): ... this.
9716 (vec_extract<mode><ssehalfvecmodelower>): New expander.
9717 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
9718 element mode after mode in gen_vec_init* calls.
9719 (VEC_INIT_HALF_MODE): New mode iterator.
9720 (vec_init<mode>): Renamed to ...
9721 (vec_init<mode><ssescalarmodelower>): ... this.
9722 (vec_init<mode><ssehalfvecmodelower>): New expander.
9723 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
9724 (vec_extractv2sfsf): ... this.
9725 (vec_initv2sf): Renamed to ...
9726 (vec_initv2sfsf): ... this.
9727 (vec_extractv2si): Renamed to ...
9728 (vec_extractv2sisi): ... this.
9729 (vec_initv2si): Renamed to ...
9730 (vec_initv2sisi): ... this.
9731 (vec_extractv4hi): Renamed to ...
9732 (vec_extractv4hihi): ... this.
9733 (vec_initv4hi): Renamed to ...
9734 (vec_initv4hihi): ... this.
9735 (vec_extractv8qi): Renamed to ...
9736 (vec_extractv8qiqi): ... this.
9737 (vec_initv8qi): Renamed to ...
9738 (vec_initv8qiqi): ... this.
9739 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
9740 (vec_init<mode>): Renamed to ...
9741 (vec_init<mode><VEC_base_l>): ... this.
9742 (vec_extract<mode>): Renamed to ...
9743 (vec_extract<mode><VEC_base_l>): ... this.
9744 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
9745 (vec_initv2sfsf): ... this.
9746 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
9747 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
9748 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
9749 element mode after mode in gen_vec_init* calls.
9750 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
9751 (vec_init<mode><Vel>): ... this.
9752 (vec_extract<mode>): Renamed to ...
9753 (vec_extract<mode><Vel>): ... this.
9754 * config/aarch64/iterators.md (Vel): New mode attribute.
9755 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
9756 Add element mode after mode in gen_vec_extract* calls.
9757 * config/s390/vector.md (non_vec_l): New mode attribute.
9758 (vec_extract<mode>): Renamed to ...
9759 (vec_extract<mode><non_vec_l>): ... this.
9760 (vec_init<mode>): Renamed to ...
9761 (vec_init<mode><non_vec_l>): ... this.
9762 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
9763 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
9765 * config/arm/iterators.md (V_elem_l): New mode attribute.
9766 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
9767 (vec_extract<mode><V_elem_l>): ... this.
9768 (vec_extractv2di): Renamed to ...
9769 (vec_extractv2didi): ... this.
9770 (vec_init<mode>): Renamed to ...
9771 (vec_init<mode><V_elem_l>): ... this.
9772 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
9773 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
9774 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
9775 Add element mode after gen_vec_extract* calls.
9776 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
9777 (vec_init<mode><unitmode>): ... this.
9778 (vec_extract<mode>): Renamed to ...
9779 (vec_extract<mode><unitmode>): ... this.
9780 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
9781 (vec_init<mode><unitmode>): ... this.
9782 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
9783 (vec_initv2sfsf): ... this.
9784 (vec_extractv2sf): Renamed to ...
9785 (vec_extractv2sfsf): ... this.
9786 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
9787 Add element mode after gen_vec_extract* calls.
9788 * config/mips/mips.md (unitmode): New mode iterator.
9789 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
9790 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
9791 * config/spu/spu.md (inner_l): New mode attribute.
9792 (vec_init<mode>): Renamed to ...
9793 (vec_init<mode><inner_l>): ... this.
9794 (vec_extract<mode>): Renamed to ...
9795 (vec_extract<mode><inner_l>): ... this.
9796 * config/sparc/sparc.md (veltmode): New mode iterator.
9797 (vec_init<VMALL:mode>): Renamed to ...
9798 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
9799 * config/ia64/vect.md (vec_initv2si): Renamed to ...
9800 (vec_initv2sisi): ... this.
9801 (vec_initv2sf): Renamed to ...
9802 (vec_initv2sfsf): ... this.
9803 (vec_extractv2sf): Renamed to ...
9804 (vec_extractv2sfsf): ... this.
9805 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
9806 (vec_init<mode>): Renamed to ...
9807 (vec_init<mode><VEC_base_l>): ... this.
9808 (vec_extract<mode>): Renamed to ...
9809 (vec_extract<mode><VEC_base_l>): ... this.
9810 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
9811 (vec_initv2sfsf): ... this.
9812 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
9813 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
9814 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
9815 gen_vec_init* calls.
9817 2017-08-01 Richard Biener <rguenther@suse.de>
9819 PR tree-optimization/81297
9820 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
9821 TREE_OVERFLOW from INTEGER_CSTs.
9823 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
9825 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
9827 2017-07-31 Carl Love <cel@us.ibm.com>
9829 * config/rs6000/rs6000-c: Add support for built-in functions
9830 vector signed char vec_xl_be (signed long long, signed char *);
9831 vector unsigned char vec_xl_be (signed long long, unsigned char *);
9832 vector signed int vec_xl_be (signed long long, signed int *);
9833 vector unsigned int vec_xl_be (signed long long, unsigned int *);
9834 vector signed long long vec_xl_be (signed long long, signed long long *);
9835 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
9836 vector signed short vec_xl_be (signed long long, signed short *);
9837 vector unsigned short vec_xl_be (signed long long, unsigned short *);
9838 vector double vec_xl_be (signed long long, double *);
9839 vector float vec_xl_be (signed long long, float *);
9840 * config/rs6000/altivec.h (vec_xl_be): Add #define.
9841 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
9842 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
9844 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
9845 (altivec_expand_builtin): Add switch statement to call
9846 altivec_expand_xl_be for each builtin.
9847 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
9848 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
9849 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
9850 __builtin_vsx_le_be_v16qi.
9851 * doc/extend.texi: Update the built-in documentation file for the
9852 new built-in functions.
9854 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
9857 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
9859 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
9861 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9863 * config.gcc: Add z14.
9864 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
9865 CPU model numbers for z13s and z14.
9866 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
9868 * config/s390/s390-opts.h (enum processor_type): Rename
9869 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
9870 * config/s390/s390.c (processor_table): Add field for CPU name to
9871 be passed to Binutils.
9872 (s390_asm_output_machine_for_arch): Use the new field in
9873 processor_table for Binutils.
9874 (s390_expand_builtin): Replace arch12 with z14.
9875 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
9876 (s390_get_sched_attrmask): Likewise.
9877 (s390_get_unit_mask): Likewise.
9878 * config/s390/s390.opt: Add z14 to processor_type enum.
9880 2017-07-31 Martin Jambor <mjambor@suse.cz>
9883 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
9884 regardless of optimization level.
9886 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
9887 Martin Liska <mliska@suse.cz>
9889 * predict.def: Remove old comment and adjust probability.
9890 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
9893 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
9896 * config/i386/i386.c (ix86_function_naked): New function.
9897 (ix86_can_use_return_insn_p): Return false for naked functions.
9898 (ix86_expand_prologue): Skip prologue for naked functions.
9899 (ix86_expand_epilogue): Skip epilogue for naked functions
9900 and emit trap instruction.
9901 (ix86_warn_func_return): New function.
9902 (ix86_attribute_table): Add "naked" attribute specification.
9903 (TARGET_WARN_FUNC_RETURN): Define.
9904 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
9906 2017-07-31 Martin Liska <mliska@suse.cz>
9908 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
9909 (dump_gimple_bb_header): Always dump BB info.
9910 (pp_cfg_jump): Do not append info about BB when dumping a jump.
9912 2017-07-31 Martin Liska <mliska@suse.cz>
9915 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
9916 also with current_function_decl non-null equality.
9918 2017-07-31 Jakub Jelinek <jakub@redhat.com>
9921 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
9922 change type to the element type, instead add eltype variable and
9923 use it where we are interested in the element type.
9925 PR tree-optimization/81603
9926 * ipa-polymorphic-call.c
9927 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
9928 offset arithmetic in offset_int, bail out if the resulting bit offset
9929 doesn't fit into shwi.
9931 2017-07-31 Martin Liska <mliska@suse.cz>
9933 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
9934 (gimplify_save_expr): Fix comment.
9936 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
9939 * config/i386/i386.c (ix86_function_arg): Update arguments for
9941 (ix86_compute_frame_layout): Set the initial stack offset to
9942 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
9943 INCOMING_FRAME_SP_OFFSET.
9944 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
9945 stack before exception handler returns.
9946 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
9947 the 'ERROR_CODE' for exception handler.
9949 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
9951 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
9952 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
9953 (ASM_OUTPUT_REG_POP): Ditto.
9954 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
9955 instead of asm_fprintf to output pure string.
9957 2017-07-29 Jakub Jelinek <jakub@redhat.com>
9959 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
9960 to imported_module_or_decl hook.
9961 (debug_nothing_tree_tree_tree_bool): Remove.
9962 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
9963 * debug.c (do_nothing_debug_hooks): Use
9964 debug_nothing_tree_tree_tree_bool_bool instead of
9965 debug_nothing_tree_tree_tree_bool.
9966 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
9967 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
9968 * sdbout.c (sdb_debug_hooks): Likewise.
9969 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
9970 (gen_namespace_die): Add DW_AT_export_symbols attribute if
9972 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
9973 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
9974 attribute, don't add anything.
9976 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9978 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
9979 (fold_build2_stat_loc): Likewise.
9980 (fold_build3_stat_loc): Likewise.
9981 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
9982 (fold_build1_loc): Remove macro.
9983 (fold_build2_loc): Likewise.
9984 (fold_build3_loc): Likewise.
9986 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9988 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
9989 (gimple_build_debug_bind_source_stat): Likewise.
9990 * gimple.h (gimple_build_debug_bind): Remove macro.
9991 (gimple_build_debug_bind_source): Likewise.
9993 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9995 * bitmap.c (bitmap_alloc): Adjust.
9996 (bitmap_gc_alloc): Likewise.
9997 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
9999 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10001 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
10002 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
10003 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
10004 (bitmap_gc_alloc_stat): Likewise.
10005 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
10007 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10009 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
10010 * rtl.h (shallow_copy_rtx): Remove macro.
10012 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10014 * emit-rtl.c (gen_raw_REG): Adjust.
10015 * gengenrtl.c (gendef): Likewise.
10016 * rtl.c (rtx_alloc_stat): Remove _stat from name.
10017 * rtl.h (rtx_alloc): Remove macro.
10019 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10021 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
10022 (build_tree_list_stat): Likewise.
10023 * tree.h (build_tree_list): Remove macro.
10024 (build_tree_list_vec): Likewise.
10026 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10028 * tree.c (make_vector_stat): Remove _stat from name.
10029 (build_vector_stat): Likewise.
10030 * tree.h (make_vector_stat): Remove macro.
10031 (build_vector_stat): Likewise.
10033 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10035 * tree.h (build_var_debug_value): Remove prototype.
10037 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10039 * tree.c (tree_cons_stat): Remove _stat from name.
10040 * tree.h (tree_cons): Remove macro.
10042 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10044 * tree.c (build_vl_exp_stat): Remove _stat from name.
10045 * tree.h (build_vl_exp): Remove macro.
10047 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10049 * tree.c (build_decl_stat): Remove _stat from name.
10050 * tree.h (build_decl): Remove macro.
10052 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10054 * gimple.c (gimple_build_with_ops_stat): Adjust.
10055 (gimple_alloc_stat): Remove _stat from name.
10056 * gimple.h (gimple_alloc): Remove macro.
10058 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10060 * tree.c (make_tree_vec_stat): Remove _stat from name.
10061 (grow_tree_vec_stat): Likewise.
10062 * tree.h (make_tree_vec_stat): Adjust prototype.
10063 (grow_tree_vec_stat): Likewise.
10064 (make_tree_vec): Remove macro.
10065 (grow_tree_vec): Likewise.
10067 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10069 * fold-const.c (fold_build1_stat_loc): Adjust.
10070 (fold_build2_stat_loc): Likewise.
10071 (fold_build3_stat_loc): Likewise.
10072 * tree.c (build0_stat): Remove _stat from name.
10073 (build1_stat): Likewise.
10074 (build2_stat): Likewise.
10075 (build3_stat): Likewise.
10076 (build4_stat): Likewise.
10077 (build5_stat): Likewise.
10078 * tree.h (build1_loc): Remove macro, and rename _stat function
10080 (build2_loc): Likewise.
10081 (build3_loc): Likewise.
10082 (build4_loc): Likewise.
10083 (build5_loc): Likewise.
10085 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10087 * tree.c (make_int_cst_stat): Remove _stat from name.
10088 * tree.h (make_int_cst_stat): Adjust prototype.
10089 (make_int_cst): Remove macro.
10091 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10093 * tree.c (make_tre_binfo_stat): Remove _stat from name.
10094 * tree.h (make_tree_binfo_stat): Adjust prototype.
10095 (make_tree_binfo): Remove.
10097 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10099 * tree.c (copy_node_stat): Rename to copy_node.
10100 (build_distinct_type_copy): Adjust.
10101 * tree.h (copy_node_stat): Adjust prototype.
10102 (copy_node): Remove macro.
10104 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10106 * tree.c (make_node_stat): rename to make_node.
10107 (build_tree_list_stat): Adjust.
10108 (build0_stat): Likewise.
10109 (build2_stat): Likewise.
10110 (build3_stat): Likewise.
10111 (build4_stat): Likewise.
10112 (build5_stat): Likewise.
10113 (build_decl_stat): Likewise.
10114 * tree.h (make_node_stat): Adjust prototype.
10115 (make_node): remove macro.
10117 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
10119 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
10120 (PPC_FEATURE2_SCV): Likewise.
10121 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
10123 2017-07-28 Tamar Christina <tamar.christina@arm.com>
10125 * config/aarch64/aarch64.c
10126 (aarch64_internal_mov_immediate): Add new special pattern.
10127 * config/aarch64/aarch64.md (*movdi_aarch64):
10128 Add reg/32bit const mov case.
10130 2017-07-28 Tamar Christina <tamar.christina@arm.com>
10131 Richard Sandiford <richard.sandiford@linaro.org>
10133 * config/aarch64/aarch64.md (mov<mode>): Generalize.
10134 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
10135 Add integer and movi cases.
10136 (movi-split-hf-df-sf split, fp16): New.
10137 (enabled): Added TARGET_FP_F16INST.
10138 * config/aarch64/iterators.md (GPF_HF): New.
10139 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
10141 2017-07-28 Tamar Christina <tamar.christina@arm.com>
10143 * config/aarch64/aarch64.c
10144 (aarch64_simd_container_mode): Add prototype.
10145 (aarch64_expand_mov_immediate): Add HI support.
10146 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
10147 (aarch64_can_const_movi_rtx_p): New.
10148 (aarch64_preferred_reload_class):
10149 Remove restrictions of using FP registers for certain SIMD operations.
10150 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
10151 (aarch64_valid_floating_const): Add integer move validation.
10152 (aarch64_simd_imm_scalar_p): Remove.
10153 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
10154 (aarch64_legitimate_constant_p): Expand list of supported cases.
10155 * config/aarch64/aarch64-protos.h
10156 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
10157 (aarch64_reinterpret_float_as_int): New.
10158 (aarch64_simd_imm_scalar_p): Remove.
10159 * config/aarch64/constraints.md (Uvi): New.
10160 (Dd): Split into Ds and new Dd.
10161 * config/aarch64/aarch64.md (*movsi_aarch64):
10163 (*movdi_aarch64): Add SIMD mov case.
10165 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10167 * tree-predcom.c: (struct chain): Handle store-store chain in which
10168 stores for elimination only store loop invariant values.
10169 (execute_pred_commoning_chain): Ditto.
10170 (prepare_initializers_chain_store_elim): Ditto.
10171 (prepare_finalizers): Ditto.
10172 (is_inv_store_elimination_chain): New function.
10173 (initialize_root_vars_store_elim_1): New function.
10175 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10177 * tree-predcom.c: Revise general description of the pass.
10178 (enum chain_type): New enum type for store elimination.
10179 (struct chain): New field supporting store elimination.
10180 (struct component): Ditto.
10181 (dump_chain): Dump store-stores chain.
10182 (release_chain): Release resources.
10183 (split_data_refs_to_components): Compute and create component
10184 contains only stores for elimination.
10185 (get_chain_last_ref_at): New function.
10186 (make_invariant_chain): Initialization.
10187 (make_rooted_chain): Specify chain type in parameter and record it.
10188 (add_looparound_copies): Skip for store-stores chain.
10189 (determine_roots_comp): Compute type of chain and pass it to
10191 (initialize_root_vars_store_elim_2): New function.
10192 (finalize_eliminated_stores): New function.
10193 (remove_stmt): Handle store for elimination.
10194 (execute_pred_commoning_chain): Execute predictive commoning on
10195 store-store chains.
10196 (determine_unroll_factor): Skip unroll for store-stores chain.
10197 (prepare_initializers_chain_store_elim): New function.
10198 (prepare_initializers_chain): Hanlde store-store chain.
10199 (prepare_finalizers_chain, prepare_finalizers): New function.
10200 (tree_predictive_commoning_loop): Return integer value indicating
10201 if loop is unrolled or lcssa form is corrupted.
10202 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
10204 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10206 * tree-predcom.c (initialize_root): Delete.
10207 (execute_pred_commoning_chain): Initialize root vars and replace
10208 reference of non-combined chain directly, rather than call above
10211 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10213 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
10214 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
10216 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10218 * tree-predcom.c (struct chain): New field init_seq.
10219 (release_chain): Release init_seq.
10220 (prepare_initializers_chain): Record intialization stmts in above
10222 (insert_init_seqs): New function.
10223 (tree_predictive_commoning_loop): Call insert_init_seqs.
10225 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10227 * tree-predcom.c (determine_roots_comp): Skip trivial components.
10229 2017-07-28 Richard Biener <rguenther@suse.de>
10231 * match.pd: Remove superfluous :c.
10232 * genmatch.c (simplify::id): Add member.
10233 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
10235 (current_id): New global.
10236 (dt_node::parent): Move from ...
10237 (dt_operand::parent): ... here. Add for_id member.
10238 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
10239 (decision_tree::find_node): Relax order requirement when
10240 merging DT_TRUE nodes to ones inbetween the current simplify
10241 and the one we try to merge with. Add diagnostic whenever
10242 we need to enforce pattern order by not merging.
10243 (decision_tree::insert): Set current_id.
10244 (decision_tree::print_node): Dump parent node and for_id.
10245 (parser::last_id): Add member.
10246 (parser::push_simplify): Assign unique id.
10247 (parser::parser): Initialize last_id.
10249 2017-07-28 Martin Liska <mliska@suse.cz>
10252 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
10253 gimple_build_debug_bind.
10255 2017-07-28 Richard Biener <rguenther@suse.de>
10257 PR tree-optimization/81502
10258 * match.pd: Add pattern combining BIT_INSERT_EXPR with
10260 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
10262 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
10263 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
10264 for BIT_FIELD_REF args.
10265 * fold-const.c (make_bit_field_ref): Likewise.
10266 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
10268 2017-07-28 Jakub Jelinek <jakub@redhat.com>
10271 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
10272 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
10273 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
10274 Or it into SANITIZER_UNDEFINED.
10275 * ubsan.c: Include gimple-fold.h and varasm.h.
10276 (ubsan_expand_ptr_ifn): New function.
10277 (instrument_pointer_overflow): New function.
10278 (maybe_instrument_pointer_overflow): New function.
10279 (instrument_object_size): Formatting fix.
10280 (pass_ubsan::execute): Call instrument_pointer_overflow
10281 and maybe_instrument_pointer_overflow.
10282 * internal-fn.c (expand_UBSAN_PTR): New function.
10283 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
10284 * sanitizer.def (__ubsan_handle_pointer_overflow,
10285 __ubsan_handle_pointer_overflow_abort): New builtins.
10286 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
10287 * internal-fn.def (UBSAN_PTR): New internal function.
10288 * opts.c (sanitizer_opts): Add pointer-overflow.
10289 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
10290 * fold-const.c (build_range_check): Compute pointer range check in
10291 integral type if pointer arithmetics would be needed. Formatting
10294 2017-07-28 Martin Liska <mliska@suse.cz>
10297 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
10298 parameters that are of a variable-length.
10300 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
10302 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
10304 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
10305 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
10306 (CRT_CALL_STATIC_FUNCTION): Likewise.
10307 (ASM_DEFAULT_SPEC): New define.
10308 (ASM_SPEC32): Likewise.
10309 (ASM_SPEC64): Likewise.
10310 (ASM_SPEC_COMMON): Likewise.
10311 (ASM_SPEC): Likewise.
10312 (INVALID_64BIT): Likewise.
10313 (LINK_OS_DEFAULT_SPEC): Likewise.
10314 (LINK_OS_SPEC32): Likewise.
10315 (LINK_OS_SPEC64): Likewise.
10316 (POWERPC_LINUX): Likewise.
10317 (PTRDIFF_TYPE): Likewise.
10318 (RESTORE_FP_PREFIX): Likewise.
10319 (RESTORE_FP_SUFFIX): Likewise.
10320 (SAVE_FP_PREFIX): Likewise.
10321 (SAVE_FP_SUFFIX): Likewise.
10322 (SIZE_TYPE): Likewise.
10323 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
10324 (TARGET_64BIT): Likewise.
10325 (TARGET_64BIT): Likewise.
10326 (TARGET_AIX): Likewise.
10327 (WCHAR_TYPE_SIZE): Likewise.
10328 (WCHAR_TYPE): Undefine.
10329 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
10330 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
10331 (CPP_OS_RTEMS_SPEC): Delete.
10332 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
10333 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
10335 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
10337 2017-07-28 Jakub Jelinek <jakub@redhat.com>
10339 PR tree-optimization/81578
10340 * tree-parloops.c (build_new_reduction): Bail out if
10341 reduction_code isn't one of the standard OpenMP reductions.
10342 Move the details printing after that decision.
10344 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
10346 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
10347 related to reload_in_progress.
10348 (splat_input_operand): Likewise.
10349 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
10351 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
10353 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
10354 (TARGET_INSTANTIATE_DECLS): Likewise.
10355 (legitimate_indexed_address_p): Delete reload_in_progress code.
10356 (rs6000_debug_legitimate_address_p): Likewise.
10357 (rs6000_eliminate_indexed_memrefs): Likewise.
10358 (rs6000_emit_le_vsx_store): Likewise.
10359 (rs6000_emit_move_si_sf_subreg): Likewise.
10360 (rs6000_emit_move): Likewise.
10361 (register_to_reg_type): Likewise.
10362 (rs6000_pre_atomic_barrier): Likewise.
10363 (rs6000_machopic_legitimize_pic_address): Likewise.
10364 (rs6000_allocate_stack_temp): Likewise.
10365 (rs6000_address_for_fpconvert): Likewise.
10366 (rs6000_address_for_altivec): Likewise.
10367 (rs6000_secondary_memory_needed_rtx): Delete function.
10368 (rs6000_check_sdmode): Likewise.
10369 (rs6000_alloc_sdmode_stack_slot): Likewise.
10370 (rs6000_instantiate_decls): Likewise.
10371 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
10372 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
10373 Delete reload_in_progress.
10374 (*vec_reload_and_plus_<mptrsize>): Likewise.
10375 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
10376 (vsx_div_v2di): Likewise.
10377 (vsx_udiv_v2di): Likewise.
10379 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
10381 * config/rs6000/rs6000.opt (mlra): Replace with stub.
10382 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
10383 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
10384 (rs6000_debug_reg_global): Delete print of LRA status.
10385 (rs6000_option_override_internal): Delete dead LRA related code.
10386 (rs6000_lra_p): Delete function.
10387 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
10389 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
10391 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
10392 * config/riscv/rtems.h: New file.
10394 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10395 Sudakshina Das <sudi.das@arm.com>
10397 * config/aarch64/aarch64.md
10398 (define_split for and<mode>3nr_compare): Move
10399 non aarch64_logical_operand to a register.
10400 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
10401 register immediate operand to a register.
10402 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
10404 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
10406 PR middle-end/81564
10407 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
10409 2017-07-27 Richard Biener <rguenther@suse.de>
10411 PR tree-optimization/81573
10412 PR tree-optimization/81494
10413 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
10414 multi defuse cycle case.
10416 2017-07-27 Richard Biener <rguenther@suse.de>
10418 PR tree-optimization/81571
10419 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
10422 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
10424 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
10425 earlier and only if MASK_FPU is set. Adjust formatting.
10427 2017-07-27 Martin Liska <mliska@suse.cz>
10429 * opt-functions.awk: Add validation of value of Init.
10430 * optc-gen.awk: Pass new argument.
10432 2017-07-27 Martin Liska <mliska@suse.cz>
10434 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
10435 Fix wrong condition.
10437 2017-07-27 Martin Liska <mliska@suse.cz>
10439 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
10440 BBs and edges seen by autoFDO.
10442 2017-07-27 Richard Biener <rguenther@suse.de>
10444 PR tree-optimization/81502
10445 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
10446 with incompatible but same sized type.
10447 (execute_update_addresses_taken): Likewise.
10449 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
10451 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
10452 flag_tree_loop_vectorize rather than flag_tree_vectorize.
10454 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10457 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
10458 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
10459 Change s_operand to memory_operand.
10461 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
10463 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
10464 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
10465 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
10466 Emit instructions rather than returning an expression. Handle TFmode
10467 and KFmode by casting to TImode.
10468 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
10469 (rs6000_emit_le_vsx_store): Likewise.
10470 * config/rs6000/vsx.md (VSX_TI): New iterator.
10471 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
10472 (*vsx_le_undo_permute_<mode>): Likewise.
10473 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
10474 emit the split sequence.
10475 (*vsx_le_perm_store_<mode>): Likewise.
10477 2017-07-27 Jakub Jelinek <jakub@redhat.com>
10479 PR tree-optimization/81555
10480 PR tree-optimization/81556
10481 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
10482 if true, force CHANGED for the recursive invocation.
10483 (reassociate_bb): Remember original length of ops array, pass
10484 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
10486 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
10487 attributes for noipa attribute. For naked attribute use
10488 lookup_attribute first before lookup_attribute_spec.
10489 * final.c (rest_of_handle_final): Disable IPA RA for functions with
10491 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
10492 for functions with noipa attribute.
10493 (cgraph_externally_visible_p): Return true for functions with noipa
10495 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
10496 for functions with noipa attribute.
10497 * doc/extend.texi: Document noipa function attribute.
10498 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
10499 also for functions with noipa attribute.
10500 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
10502 2017-07-26 Andrew Pinski <apinski@cavium.com>
10504 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
10505 vec_unalign_load_cost and vec_unalign_store_cost.
10507 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10509 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
10510 -mvsx-small-integer option.
10511 (ISA_3_0_MASKS_IEEE): Likewise.
10512 (OTHER_VSX_VECTOR_MASKS): Likewise.
10513 (POWERPC_MASKS): Likewise.
10514 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
10515 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
10516 code, only testing for DImode being allowed in non-VSX floating
10518 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
10519 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
10521 (rs6000_option_override_internal): Delete -mvsx-small-integer.
10522 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
10523 TARGET_P8_VECTOR test.
10524 (rs6000_secondary_reload_simple_move): Likewise.
10525 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
10526 since TARGET_P9_VECTOR was already tested.
10527 (rs6000_opt_masks): Remove -mvsx-small-integer.
10528 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
10529 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
10531 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
10532 test for TARGET_VEXTRACTUB was used, and that uses
10534 (p9 extract splitter): Likewise.
10535 (vsx_extract_<mode>_di_p9): Likewise.
10536 (vsx_extract_<mode>_store_p9): Likewise.
10537 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
10538 for TARGET_P9_VECTOR was used. Delete code that is now dead with
10539 the elimination of TARGET_VSX_SMALL_INTEGER.
10540 (vsx_extract_<mode>_p8): Likewise.
10541 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
10542 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
10543 (vsx_set_<mode>_p9): Likewise.
10544 (vsx_set_v4sf_p9): Likewise.
10545 (vsx_set_v4sf_p9_zero): Likewise.
10546 (vsx_insert_extract_v4sf_p9): Likewise.
10547 (vsx_insert_extract_v4sf_p9_2): Likewise.
10548 * config/rs6000/rs6000.md (sign extend splitter): Change
10549 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
10550 (floatsi<mode>2_lfiwax_mem): Likewise.
10551 (floatunssi<mode>2_lfiwzx_mem): Likewise.
10552 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
10553 since a test for TARGET_P9_VECTOR was used.
10554 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10555 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
10556 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10557 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
10558 TARGET_P8_VECTOR test.
10559 (fix_trunc<mode>si2_stfiwx): Likewise.
10560 (fix_trunc<mode>si2_internal): Likewise.
10561 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
10562 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
10564 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
10565 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
10566 TARGET_P8_VECTOR test.
10567 (fixuns_trunc<mode>si2_stfiwx): Likewise.
10568 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
10569 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
10571 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
10572 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
10573 since a test for TARGET_P9_VECTOR was used.
10574 (splitter for loading small constants): Likewise.
10576 2017-07-26 Andrew Pinski <apinski@cavium.com>
10578 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
10581 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
10584 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
10585 (fp_valid_at): Likewise.
10587 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
10589 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
10590 (qdf24xx_addrcost_table): Likewise.
10591 (cortexa57_tunings): Update to use generic_branch_cost.
10592 (cortexa72_tunings): Likewise.
10593 (cortexa73_tunings): Likewise.
10594 (qdf24xx_tunings): Likewise.
10596 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
10598 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
10599 (thunderx2t99_branch_cost): Likewise.
10600 (cortexa35_tunings): Update to use generic_branch_cost.
10601 (cortexa53_tunings): Likewise.
10602 (cortexa57_tunings): Likewise.
10603 (cortexa72_tunings): Likewise.
10604 (cortexa73_tunings): Likewise.
10605 (thunderx2t99_tunings): Likewise.
10607 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
10609 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
10610 (sparc_option_override): Honour MASK_FSMULD.
10611 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
10612 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
10613 * config/sparc/sparc.opt (mfsmuld): New option.
10614 * doc/invoke.texi (mfsmuld): Document option.
10616 2017-07-26 Marek Polacek <polacek@redhat.com>
10618 PR middle-end/70992
10619 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
10621 2017-07-26 Richard Biener <rguenther@suse.de>
10623 * gimple-match-head.c (do_valueize): Return OP if valueize
10625 (get_def): New helper to get at the def stmt of a SSA name
10626 if valueize allows.
10627 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
10628 do_valueize to get at the def stmt.
10629 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
10631 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
10633 PR middle-end/46932
10634 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
10636 2017-07-26 Martin Liska <mliska@suse.cz>
10639 * function.c (expand_function_start): Make expansion of
10640 nonlocal_goto_save_area after parm_birth_insn.
10642 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
10644 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
10645 from all CPU target flags enable members.
10647 2017-07-26 Richard Biener <rguenther@suse.de>
10649 * genmatch.c (dt_simplify::gen): Make iterator vars const.
10650 (decision_tree::gen): Make 'type' const.
10651 (write_predicate): Likewise.
10653 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
10655 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
10656 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
10657 (rs6000_option_override_internal): Likewise.
10658 (rs6000_expand_vector_set): Likewise.
10659 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
10660 (TARGET_UPPER_REGS_SF): Likewise.
10661 (TARGET_UPPER_REGS_DI): Likewise.
10662 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
10663 (TARGET_DIRECT_MOVE_64BIT): Likewise.
10664 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
10665 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10666 (Splitters for DI constants in Altivec registers): Likewise.
10667 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
10668 (vsx_set_v4sf_p9): Likewise.
10669 (vsx_set_v4sf_p9_zero): Likewise.
10670 (vsx_insert_extract_v4sf_p9): Likewise.
10671 (vsx_insert_extract_v4sf_p9_2): Likewise.
10673 2017-07-25 Carl Love <cel@us.ibm.com>
10675 * doc/extend.texi: Update the built-in documentation file for the
10676 existing built-in functions
10677 vector signed char vec_cnttz (vector signed char);
10678 vector unsigned char vec_cnttz (vector unsigned char);
10679 vector signed short vec_cnttz (vector signed short);
10680 vector unsigned short vec_cnttz (vector unsigned short);
10681 vector signed int vec_cnttz (vector signed int);
10682 vector unsigned int vec_cnttz (vector unsigned int);
10683 vector signed long long vec_cnttz (vector signed long long);
10684 vector unsigned long long vec_cnttz (vector unsigned long long);
10686 2017-07-25 Andrew Pinski <apinski@cavium.com>
10688 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
10689 accesses where the use is for the first operand of a BIT_INSERT.
10691 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
10694 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
10695 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
10697 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
10699 * config/i386/gstabs.h: Delete.
10700 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
10702 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
10704 * config/i386/i386.c (ix86_decompose_address): Do not check for
10705 register RTX when looking at index_reg or base_reg.
10706 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
10708 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
10710 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
10711 to update EH info here.
10713 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10715 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
10717 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10719 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
10721 2017-07-25 Torsten Duwe <duwe@suse.de>
10723 * common.opt: Introduce -fpatchable-function-entry
10724 command line option, and its variables function_entry_patch_area_size
10725 and function_entry_patch_area_start.
10726 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
10727 including a two-value parser.
10728 * target.def (print_patchable_function_entry): New target hook.
10729 * targhooks.h (default_print_patchable_function_entry): New function.
10730 * targhooks.c (default_print_patchable_function_entry): Likewise.
10731 * toplev.c (process_options): Switch off IPA-RA if
10732 patchable function entries are being generated.
10733 * varasm.c (assemble_start_function): Look at the
10734 patchable-function-entry command line switch and current
10735 function attributes and maybe generate NOP instructions by
10736 calling the print_patchable_function_entry hook.
10737 * doc/extend.texi: Document patchable_function_entry attribute.
10738 * doc/invoke.texi: Document -fpatchable_function_entry
10739 command line option.
10740 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
10742 * doc/tm.texi: Re-generate.
10744 2017-07-25 Jakub Jelinek <jakub@redhat.com>
10747 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
10748 TARGET_AVX512DQ rather than TARGET_AVX512BW.
10750 2017-07-25 Tamar Christina <tamar.christina@arm.com>
10752 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
10754 2017-07-25 Richard Biener <rguenther@suse.de>
10756 PR tree-optimization/81455
10757 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
10758 not walk in cycles when looking for guards.
10760 2017-07-25 Richard Biener <rguenther@suse.de>
10762 PR tree-optimization/81529
10763 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
10764 when optimizing backedge uses.
10766 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
10768 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
10770 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
10771 to dl_section_ref. On AIX, append an expression to subtract
10772 the size of the section length to dl_section_ref.
10774 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
10776 * configure.ac: If any of the config.* scripts fail, exit 1.
10777 * configure: Regenerate.
10779 2017-07-25 Richard Biener <rguenther@suse.de>
10781 PR middle-end/81546
10782 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
10783 of immediate uses, be more verbose on errors.
10785 2017-07-25 Richard Biener <rguenther@suse.de>
10787 PR tree-optimization/81510
10788 * tree-vect-loop.c (vect_is_simple_reduction): When the
10789 reduction stmt is not inside the loop bail out.
10791 2017-07-25 Richard Biener <rguenther@suse.de>
10793 PR tree-optimization/81303
10794 * tree-vect-loop-manip.c (vect_loop_versioning): Build
10795 profitability check against LOOP_VINFO_NITERSM1.
10797 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10799 * domwalk.c (cmp_bb_postorder): Simplify.
10800 (sort_bbs_postorder): New function. Use it...
10801 (dom_walker::walk): ...here to optimize common cases.
10803 2017-07-25 Martin Liska <mliska@suse.cz>
10806 * ipa-visibility.c (function_and_variable_visibility): Make the
10807 redirection just on target that supports aliasing.
10808 Fix GNU coding style.
10810 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
10813 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
10815 * config/arm/rtems.h: Likewise.
10816 * config/bfin/rtems.h: Likewise.
10817 * config/i386/rtemself.h: Likewise.
10818 * config/lm32/rtems.h: Likewise.
10819 * config/m32c/rtems.h: Likewise.
10820 * config/m68k/rtemself.h: Likewise.
10821 * config/microblaze/rtems.h: Likewise.
10822 * config/mips/rtems.h: Likewise.
10823 * config/moxie/rtems.h: Likewise.
10824 * config/nios2/rtems.h: Likewise.
10825 * config/powerpcspe/rtems.h: Likewise.
10826 * config/rs6000/rtems.h: Likewise.
10827 * config/rtems.h: Likewise.
10828 * config/sh/rtems.h: Likewise.
10829 * config/sh/rtemself.h: Likewise.
10830 * config/sparc/rtemself.h: Likewise.
10832 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
10835 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
10836 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
10837 * tree-ssa-structalias.c (alias_get_name): Same.
10839 2017-07-25 Bin Cheng <bin.cheng@arm.com>
10842 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
10843 instructions if no du chain is found.
10845 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
10847 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
10849 2017-07-25 Richard Biener <rguenther@suse.de>
10851 PR middle-end/81505
10852 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
10855 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
10857 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
10858 upper-regs options.
10859 (ISA_2_7_MASKS_SERVER): Likewise.
10860 (ISA_3_0_MASKS_IEEE): Likewise.
10861 (OTHER_P8_VECTOR_MASKS): Likewise.
10862 (OTHER_VSX_VECTOR_MASKS): Likewise.
10863 (POWERPC_MASKS): Likewise.
10864 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
10865 duplicate list of options.
10866 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
10867 explicit -mupper-regs options.
10868 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
10869 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
10870 alias for -mupper-regs-df.
10871 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
10872 (rs6000_init_hard_regno_mode_ok): Likewise.
10873 (rs6000_option_override_internal): Likewise.
10874 (rs6000_opt_masks): Likewise.
10875 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
10876 options in terms of whether -mvsx or -mpower8-vector was used.
10877 (TARGET_UPPER_REGS_DI): Likewise.
10878 (TARGET_UPPER_REGS_SF): Likewise.
10879 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
10880 -mupper-regs-* options.
10882 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
10884 * passes.c (emergency_dump_function): Print some empty lines and a
10885 header before the RTL dump.
10887 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
10889 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
10891 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
10894 * config/aarch64/aarch64.c (aarch64_classify_symbol):
10895 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
10897 2017-07-24 Carl Love <cel@us.ibm.com>
10899 * config/rs6000/rs6000-c.c: Add support for built-in functions
10900 vector float vec_extract_fp32_from_shorth (vector unsigned short);
10901 vector float vec_extract_fp32_from_shortl (vector unsigned short);
10902 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
10903 vec_extract_fp_from_shortl): Add defines for the two builtins.
10904 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
10905 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
10907 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
10908 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
10909 * doc/extend.texi: Update the built-in documentation file for the
10910 new built-in function.
10912 2017-07-24 Jakub Jelinek <jakub@redhat.com>
10915 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
10917 * doc/generic.texi: Likewise.
10918 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
10919 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
10921 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
10923 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
10924 (aarch64_mls_elt_merge<mode>): Likewise.
10926 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
10928 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
10929 having __cxa_atexit.
10931 2017-07-23 Michael Collison <michael.collison@arm.com>
10933 * config/arm/arm.c (arm_option_override): Deprecate
10934 use of -mstructure-size-boundary.
10935 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
10936 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
10938 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10941 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
10942 Reduce cost estimate for direct moves.
10944 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
10947 * config/i386/i386.c (ix86_option_override_internal): Disable
10948 BMI, BMI2 and TBM instructions for -m16.
10950 2017-07-21 Carl Love <cel@us.ibm.com>
10952 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10953 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10954 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10955 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10956 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10957 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10958 VMULOSW): New enum "unspec" values.
10959 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10960 altivec_vmulosw): New patterns.
10961 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10962 VMULOSW): Add definitions.
10964 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
10966 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
10967 (qdf24xx): Likewise.
10968 * config/aarch64/aarch64-options-extensions.def (rdma); New.
10969 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
10970 (AARCH64_FL_V8_1): Renumber.
10971 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
10972 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
10973 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
10974 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
10975 rdma to feature modifiers list.
10977 2017-07-21 Yury Gribov <tetra2005@gmail.com>
10979 PR middle-end/56727
10980 * ipa-visibility (function_and_variable_visibility): Convert
10981 recursive PLT call to direct call if appropriate.
10983 2017-07-21 Andrew Pinski <apinski@cavium.com>
10985 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
10986 operand 1 to see if the types precision matches.
10987 * fold-const.c (operand_equal_p): Likewise.
10989 2017-07-21 Richard Biener <rguenther@suse.de>
10991 PR tree-optimization/81303
10992 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
10993 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
10994 (vect_peeling_hash_get_lowest_cost): Adjust.
10995 (vect_enhance_data_refs_alignment): Likewise. Use
10996 vect_get_peeling_costs_all_drs to compute the penalty for no
10997 peeling to match up costs.
10999 2017-07-21 Richard Biener <rguenther@suse.de>
11001 PR tree-optimization/81500
11002 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
11003 we didn't identify a reduction path.
11005 2017-07-21 Tom de Vries <tom@codesourcery.com>
11006 Cesar Philippidis <cesar@codesourcery.com>
11008 PR gcov-profile/81442
11009 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
11012 2017-07-21 Tom de Vries <tom@codesourcery.com>
11015 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
11017 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
11018 nvptx_override_options_after_change.
11020 2017-07-21 Ulrich Drepper <drepper@redhat.com>
11022 * dwarf2out.c (output_file_names): Avoid double testing for
11023 dwarf_version >= 5.
11025 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
11027 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
11029 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
11031 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
11033 (try_crossjump_to_edge): Do not punt on partitioned functions.
11035 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
11037 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
11038 Put all BBs reachable only via paths crossing cold region to cold
11040 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
11042 2016-07-21 Richard Biener <rguenther@suse.de>
11044 PR tree-optimization/81303
11045 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
11046 into account prologue and epilogue iterations when raising
11047 min_profitable_iters to sth at least covering one vector iteration.
11049 2017-07-21 Tamar Christina <tamar.christina@arm.com>
11051 * config/arm/arm.c (arm_test_cpu_arch_dat):
11054 2017-07-20 Nathan Sidwell <nathan@acm.org>
11056 Remove TYPE_METHODS.
11057 * tree.h (TYPE_METHODS): Delete.
11058 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
11059 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
11060 (dbxout_type_methods): Scan TYPE_FIELDS.
11061 (dbxout_type): Don't check TYPE_METHODS here.
11062 * function.c (use_register_for_decl): Always ignore register for
11063 class types when not optimizing.
11064 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
11065 * tree.c (free_lang_data_in_type): Stitch out member functions and
11066 templates from TYPE_FIELDS.
11067 (build_distinct_type_copy, verify_type_variant,
11068 verify_type): Member fns are on TYPE_FIELDS.
11069 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
11070 * tree-pretty-print.c (dump_generic_node): Likewise.
11072 2017-07-20 Jakub Jelinek <jakub@redhat.com>
11075 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
11076 V2TImode and V4TImode.
11077 (ix86_expand_vector_extract): Likewise.
11078 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
11079 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
11080 (ssescalarmode): Handle V4TImode and V2TImode.
11081 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
11082 (*vec_extractv2ti, *vec_extractv4ti): New insns.
11083 (VEXTRACTI128_MODE): New mode iterator.
11084 (splitter for *vec_extractv?ti first element): New.
11085 (VEC_INIT_MODE): New mode iterator.
11086 (vec_init<mode>): Consolidate 3 expanders into one using
11087 VEC_INIT_MODE mode iterator.
11089 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
11091 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
11092 non_spilled_static_chain_regno_p.
11094 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
11096 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
11098 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
11100 * bb-reorder.c (connect_traces): Allow copying of blocks within
11103 2017-07-20 Richard Biener <rguenther@suse.de>
11105 * gimple.h (gimple_phi_result): Add gphi * overload.
11106 (gimple_phi_result_ptr): Likewise.
11107 (gimple_phi_arg): Likewise. Adjust index assert to only
11108 allow actual argument accesses rather than all slots available
11110 (gimple_phi_arg_def): Add gphi * overload.
11111 * tree-phinodes.c (make_phi_node): Initialize only actual
11113 (resize_phi_node): Clear memory not covered by old node,
11114 do not initialize excess argument slots.
11115 (reserve_phi_args_for_new_edge): Initialize new argument slot
11118 2017-07-20 Bin Cheng <bin.cheng@arm.com>
11120 PR tree-optimization/81388
11122 2016-07-21 Bin Cheng <bin.cheng@arm.com>
11124 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
11125 by removing computation of may_be_zero.
11127 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
11128 Tom de Vries <tom@codesourcery.com>
11130 PR middle-end/81030
11131 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
11132 when gimple level profile disagrees with what RTL expander did.
11134 2017-07-20 Richard Biener <rguenther@suse.de>
11136 PR tree-optimization/61171
11137 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
11138 (vect_analyze_stmt): Add slp instance parameter.
11139 (vectorizable_reduction): Likewise.
11140 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
11141 (vect_is_simple_reduction): Deal with chains not detected
11142 as SLP reduction chain, specifically not properly associated
11143 chains containing a mix of plus/minus.
11144 (get_reduction_op): Remove.
11145 (get_initial_defs_for_reduction): Simplify, pass in whether
11146 this is a reduction chain, pass in the SLP node for the PHIs.
11147 (vect_create_epilog_for_reduction): Get the SLP instance as
11149 (vectorizable_reduction): Get the SLP instance as arg.
11150 During analysis remember the SLP node with the PHIs in the
11151 instance. Simplify getting at the vectorized reduction PHIs.
11152 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
11153 through SLP instance.
11154 (vect_slp_analyze_operations): Likewise.
11155 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
11156 (vect_transform_stmt): Likewise.
11158 2017-07-20 Tom de Vries <tom@codesourcery.com>
11160 PR tree-optimization/81489
11161 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
11162 read of phi arg location to before loop that modifies phi.
11164 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11166 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
11169 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
11171 PR middle-end/81331
11172 * except.c (execute): Fix ordering issue.
11174 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
11176 PR rtl-optimization/81423
11177 * combine.c (make_compound_operation_int): Don't try to optimize
11178 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
11180 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
11182 PR rtl-optimization/81423
11183 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
11184 with a constant that is -1 in the truncated to mode.
11186 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
11188 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
11189 (determine_unlikely_bbs): ... here.
11190 * predict.h (propagate_unlikely_bbs_forward): Declare.
11191 * cfgexpand.c (pass_expand::execute): Use it.
11192 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
11194 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
11195 propagate_unlikely_bbs_forward.
11197 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
11199 PR middle-end/81331
11200 * except.c (maybe_add_nop_after_section_switch): New function.
11203 2017-07-19 Tom de Vries <tom@codesourcery.com>
11205 * gimple.h (gimple_phi_set_arg): Make assert more strict.
11207 2017-07-19 Tom de Vries <tom@codesourcery.com>
11209 * gimple.h (gimple_phi_arg): Make assert more strict.
11211 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
11213 * config.gcc (powerpc*-*-*): Add mmintrin.h.
11214 * config/rs6000/mmintrin.h: New file.
11215 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
11217 2017-07-19 Jakub Jelinek <jakub@redhat.com>
11219 PR tree-optimization/81346
11220 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
11222 2017-07-19 Tom de Vries <tom@codesourcery.com>
11224 * config/nvptx/nvptx.md (VECIM): Add V2DI.
11226 2017-07-19 Tom de Vries <tom@codesourcery.com>
11228 * config/nvptx/nvptx-modes.def: Add V2DImode.
11229 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
11230 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
11231 (nvptx_output_mov_insn): Handle lack of mov.b128.
11232 (nvptx_print_operand): Handle 'H' and 'L' codes.
11233 (nvptx_vector_mode_supported): Allow V2DImode.
11234 (nvptx_preferred_simd_mode): New function.
11235 (nvptx_data_alignment): New function.
11236 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
11237 nvptx_preferred_simd_mode.
11238 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
11240 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
11242 2017-07-19 Tom de Vries <tom@codesourcery.com>
11244 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
11245 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
11246 (nvptx_vector_mode_supported): New function. Allow V2SImode.
11247 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
11248 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
11249 (mov<VECIM>_insn): New define_insn.
11250 (define_expand "mov<VECIM>): New define_expand.
11252 2017-07-19 Tom de Vries <tom@codesourcery.com>
11254 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
11256 2017-07-19 Jakub Jelinek <jakub@redhat.com>
11258 PR tree-optimization/81346
11259 * fold-const.h (fold_div_compare, range_check_type): Declare.
11260 * fold-const.c (range_check_type): New function.
11261 (build_range_check): Use range_check_type.
11262 (fold_div_compare): No longer static, rewritten into
11263 a match.pd helper function.
11264 (fold_comparison): Don't call fold_div_compare here.
11265 * match.pd (X / C1 op C2): New optimization using fold_div_compare
11266 as helper function.
11268 2017-07-19 Nathan Sidwell <nathan@acm.org>
11270 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
11271 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
11272 * tree.c (find_decls_types_r, verify_type): Use
11273 TYPE_{MIN,MAX}_VALUE_RAW.
11274 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
11275 (hash_tree): Likewise.
11276 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
11278 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
11281 2017-07-18 Tom de Vries <tom@codesourcery.com>
11283 PR middle-end/81464
11284 * omp-expand.c (expand_omp_for_static_chunk): Handle
11285 equal-argument loop exit phi.
11287 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
11290 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
11291 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
11292 operand 2 predicate.
11293 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
11294 operand 2 predicate.
11295 (ror,rol -> rorx splitters): Use const_int_operand as
11296 operand 2 predicate.
11298 2017-06-18 Richard Biener <rguenther@suse.de>
11300 PR tree-optimization/81410
11301 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
11302 the gap in the ! slp_perm SLP case after each group.
11304 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
11306 PR middle-end/81463
11307 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
11310 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
11312 PR middle-end/81462
11313 * predict.c (set_even_probabilities): Cleanup; do not affect
11314 probabilities that are already known.
11315 (combine_predictions_for_bb): Call even when count is set.
11317 2017-07-18 Nathan Sidwell <nathan@acm.org>
11319 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
11322 2017-07-18 Bin Cheng <bin.cheng@arm.com>
11325 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
11326 optimization for loop niter analysis.
11328 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
11331 * config/avr/avr.c (avr_optimize_casesi): Don't use
11332 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
11334 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11336 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
11337 body_cost_vec from _vect_peel_extended_info.
11338 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
11339 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
11342 2017-07-18 Bin Cheng <bin.cheng@arm.com>
11344 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
11346 2017-07-18 Richard Biener <rguenther@suse.de>
11348 PR tree-optimization/80620
11349 PR tree-optimization/81403
11350 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
11351 info when re-using a VN table entry.
11353 2017-07-18 Richard Biener <rguenther@suse.de>
11355 PR tree-optimization/81418
11356 * tree-vect-loop.c (vectorizable_reduction): Properly compute
11357 vectype_in. Verify that with lane-reducing reduction operations
11358 we have a single def-use cycle.
11360 2017-07-17 Carl Love <cel@us.ibm.com>
11362 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
11364 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11365 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11366 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11367 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11368 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11369 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11370 VMULOSW): New enum "unspec" values.
11371 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
11372 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
11373 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11374 altivec_vmulosw): New patterns.
11375 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11376 VMULOSW): Add definitions.
11377 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
11379 * config/alpha/alpha.c: Include predict.h.
11381 2017-07-17 Yury Gribov <tetra2005@gmail.com>
11383 * tree-vrp.c (compare_assert_loc): Fix comparison function
11384 to return predictable results.
11386 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11388 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
11390 (subdi3): Likewise.
11391 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
11392 * doc/invoke.texi (mexpand-adddi): Update text.
11394 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11396 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
11397 that also clobbers the CC register. The old expand code is moved
11399 (*arc_clzsi2): ... here.
11400 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
11401 the CC register. The old expand code is moved to ...
11402 (arc_ctzsi2): ... here.
11404 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11406 * config/arc/arc.opt (mindexed-loads): Use initial value
11407 TARGET_INDEXED_LOADS_DEFAULT.
11408 (mauto-modify-reg): Use initial value
11409 TARGET_AUTO_MODIFY_REG_DEFAULT.
11410 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
11411 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
11412 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
11413 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
11415 2017-07-17 Martin Liska <mliska@suse.cz>
11418 * opts.c (finish_options): Do not allow -fgnu-tm
11419 w/ -fsanitize={kernel-,}address. Say sorry.
11421 2017-07-17 Bin Cheng <bin.cheng@arm.com>
11424 * tree-loop-distribution.c (classify_partition): Only assert on
11425 numer of iterations.
11426 (merge_dep_scc_partitions): Delete prameter. Update function call.
11427 (distribute_loop): Remove code handling loop with unknown niters.
11428 (pass_loop_distribution::execute): Skip loop with unknown niters.
11430 2017-07-17 Bin Cheng <bin.cheng@arm.com>
11433 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
11434 function sort_partitions_by_post_order.
11436 2017-07-17 Bin Cheng <bin.cheng@arm.com>
11438 PR tree-optimization/81374
11439 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
11440 the max index of basic blocks, rather than number of basic blocks.
11442 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11444 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
11446 (arc_legitimate_pic_operand_p): Likewise.
11447 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
11449 (arc_needs_pcl_p): Likewise.
11450 (arc_legitimate_pc_offset_p): Likewise.
11451 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
11452 function is also used in constrains.md.
11453 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
11454 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
11455 PLUS. Only return true/false in known cases, otherwise assert.
11456 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
11457 is already called in arc_legitimate_constant_p.
11458 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
11460 (LEGITIMATE_PIC_OPERAND_P): Use
11461 arc_raw_symbolic_reference_mentioned_p function.
11462 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
11467 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11468 Andrew Burgess <andrew.burgess@embecosm.com>
11470 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
11471 (arc_return_address_register): New function.
11472 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
11473 (arc_handle_fndecl_attribute): Add naked attribute.
11474 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
11475 (TARGET_WARN_FUNC_RETURN): Likewise.
11476 (arc_allocate_stack_slots_for_args): New function.
11477 (arc_warn_func_return): Likewise.
11478 (machine_function): Change type fn_type.
11479 (arc_compute_function_type): Consider new naked function type,
11480 change function return type.
11481 (arc_must_save_register): Adapt to handle new
11482 arc_compute_function_type's return type.
11483 (arc_expand_prologue): Likewise.
11484 (arc_expand_epilogue): Likewise.
11485 (arc_return_address_regs): Delete.
11486 (arc_return_address_register): New function.
11487 (arc_epilogue_uses): Use above function.
11488 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
11489 (arc_function_type): Change encoding, add naked type.
11490 (ARC_INTERRUPT_P): Change to handle the new encoding.
11491 (ARC_FAST_INTERRUPT_P): Likewise.
11492 (ARC_NORMAL_P): Define.
11493 (ARC_NAKED_P): Likewise.
11494 (arc_compute_function_type): Delete prototype.
11495 * config/arc/arc.md (in_ret_delay_slot): Use
11496 arc_return_address_register function.
11497 (simple_return): Likewise.
11498 (p_return_i): Likewise.
11500 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11502 PR tree-optimization/81428
11503 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
11504 can't be built for those types.
11506 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
11508 Remove stuff dead since r239246.
11510 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
11511 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
11512 (avr_inform_devices): Remove dead stuff.
11514 2017-07-17 Tamar Christina <tamar.christina@arm.com>
11516 * config/arm/arm_neon.h: Fix softp typo.
11518 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11520 PR tree-optimization/81365
11521 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
11522 aggregate moves onto bb predecessor edges, make sure there are no
11523 loads that could alias the lhs in between the start of bb and the
11526 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
11529 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
11530 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
11531 [LSHIFTRT, outer_code = TRUNCATE]: Same.
11533 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11535 PR tree-optimization/81396
11536 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
11537 (init_symbolic_number): Initialize it to 1.
11538 (perform_symbolic_merge): Add n_ops from both operands into the new
11540 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
11541 without base_addr as useless if they need more than one operation.
11542 (bswap_replace): Handle !bswap case for NULL base_addr.
11544 2017-07-17 Tom de Vries <tom@codesourcery.com>
11547 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
11550 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
11552 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
11553 conditional builtin define __FIX_LEON3FT_B2BST.
11555 2017-07-17 Daniel Cederman <cederman@gaisler.com>
11557 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
11558 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
11561 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
11563 PR rtl-optimization/81424
11564 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
11565 to remove potential trapping from operands if -fnon-call-exceptions.
11567 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11569 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
11570 profile_proability for scalling.
11571 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
11573 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11575 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
11577 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11579 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
11580 fixpoint arithmetics.
11582 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11584 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
11585 fixpoint arithmetics.
11587 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11589 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
11590 fixpoint arithmetics.
11592 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11594 * profile-count.h (profile_probability::from_reg_br_prob_note,
11595 profile_probability::to_reg_br_prob_note): New functions.
11596 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
11597 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
11598 * predict.c (probability_reliable_p): Update.
11599 (edge_probability_reliable_p): Update.
11600 (br_prob_note_reliable_p): Update.
11601 (invert_br_probabilities): Update.
11602 (add_reg_br_prob_note): New function.
11603 (combine_predictions_for_insn): Update.
11604 * asan.c (asan_clear_shadow): Update.
11605 * cfgbuild.c (compute_outgoing_frequencies): Update.
11606 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
11607 (update_br_prob_note): Update.
11608 (rtl_verify_edges): Update.
11609 (purge_dead_edges): Update.
11610 (fixup_reorder_chain): Update.
11611 * emit-rtl.c (try_split): Update.
11612 * ifcvt.c (cond_exec_process_insns): Update.
11613 (cond_exec_process_if_block): Update.
11614 (dead_or_predicable): Update.
11615 * internal-fn.c (expand_addsub_overflow): Update.
11616 (expand_neg_overflow): Update.
11617 (expand_mul_overflow): Update.
11618 * loop-doloop.c (doloop_modify): Update.
11619 * loop-unroll.c (compare_and_jump_seq): Update.
11620 * optabs.c (emit_cmp_and_jump_insn_1): Update.
11621 * predict.h: Update.
11622 * reorg.c (mostly_true_jump): Update.
11624 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
11625 * config/alpha/alpha.c (emit_unlikely_jump): Update.
11626 * config/arc/arc.c: (emit_unlikely_jump): Update.
11627 * config/arm/arm.c: (emit_unlikely_jump): Update.
11628 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
11629 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
11630 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
11631 (ix86_print_operand): Update.
11632 (ix86_split_fp_branch): Update.
11633 (predict_jump): Update.
11634 * config/ia64/ia64.c (ia64_print_operand): Update.
11635 * config/mmix/mmix.c (mmix_print_operand): Update.
11636 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
11637 (rs6000_expand_split_stack_prologue): Update.
11638 * config/rs6000/rs6000.c: Update.
11639 * config/s390/s390.c (s390_expand_vec_strlen): Update.
11640 (s390_expand_vec_movstr): Update.
11641 (s390_expand_cs_tdsi): Update.
11642 (s390_expand_split_stack_prologue): Update.
11643 * config/sh/sh.c (sh_print_operand): Update.
11644 (expand_cbranchsi4): Update.
11645 (expand_cbranchdi4): Update.
11646 * config/sparc/sparc.c (output_v9branch): Update.
11647 * config/spu/spu.c (get_branch_target): Update.
11648 (ea_load_store_inline): Update.
11649 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
11650 * config/tilepro/tilepro.c: Update.
11652 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
11654 * gimplify.c (mostly_copy_tree_r): Revert latest change.
11655 (gimplify_save_expr): Likewise.
11657 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11659 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
11661 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11663 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
11665 * timevar.def (TV_IPA_FNSUMMARY): Define.
11667 2017-07-16 Daniel Cederman <cederman@gaisler.com>
11669 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
11670 to back store errata sensitive sequence from being generated.
11671 (sqrtdf2_fix): Likewise.
11673 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11675 * tree-ssa-threadupdate.c (compute_path_counts,
11676 update_joiner_offpath_counts): Use profile_probability.
11678 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
11681 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11683 * config/arm/arm-c.c (arm_cpu_builtins): Define
11684 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
11686 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
11688 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11689 array entries to represent __ieee128 versions of the
11690 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
11691 scalar_extract_sig, and scalar_insert_exp built-in functions.
11692 (altivec_resolve_overloaded_builtin): Add special case handling
11693 for the __builtin_scalar_insert_exp function, as represented by
11694 the P9V_BUILTIN_VEC_VSIEDP constant.
11695 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
11696 exponent support for __ieee128 argument.
11697 (VSESQP): Add scalar extract signature support for __ieee128
11699 (VSTDCNQP): Add scalar test negative support for __ieee128
11701 (VSIEQP): Add scalar insert exponent support for __int128 argument
11702 with __ieee128 result.
11703 (VSIEQPF): Add scalar insert exponent support for __ieee128
11704 argument with __ieee128 result.
11705 (VSTDCQP): Add scalar test data class support for __ieee128
11707 (VSTDCNQP): Add overload support for scalar test negative with
11708 __ieee128 argument.
11709 (VSTDCQP): Add overload support for scalar test data class
11710 __ieee128 argument.
11711 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
11712 UNSPEC_VSX_SXSIGDP.
11713 (UNSPEC_VSX_SIEXPQP): New constant.
11714 (xsxexpqp): New insn for VSX scalar extract exponent quad
11716 (xsxsigqp): New insn for VSX scalar extract significand quad
11718 (xsiexpqpf): New insn for VSX scalar insert exponent quad
11719 precision with floating point argument.
11720 (xststdcqp): New expand for VSX scalar test data class quad
11722 (xststdcnegqp): New expand for VSX scalar test negative quad
11724 (xststdcqp): New insn to match expansions for VSX scalar test data
11725 class quad precision and VSX scalar test negative quad precision.
11726 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
11727 special case operand checking to enforce that second operand of
11728 VSX scalar test data class with quad precision argument is a 7-bit
11730 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
11731 prototypes and descriptions of __ieee128 versions of
11732 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
11733 scalar_test_data_class, and scalar_test_neg built-in functions.
11735 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11737 PR tree-optimization/81162
11738 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
11739 replace a negate with an add.
11741 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
11743 * doc/invoke.texi (arm/-mcpu): Document +crypto.
11745 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11747 * config/arm/arm-c.c (arm_cpu_builtins): Define
11748 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
11750 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11752 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
11753 (armv8-r): Set ARM Cortex-R52 as default CPU.
11754 * config/arm/arm-tables.opt: Regenerate.
11755 * config/arm/arm-tune.md: Regenerate.
11756 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
11758 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
11759 extension for -mcpu=cortex-r52.
11761 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11763 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
11764 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
11765 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
11766 (fp-armv8): Define it as FP_ARMv8 only.
11767 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
11768 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
11770 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
11771 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
11772 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
11773 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
11774 than TARGET_FPU_ARMV8.
11775 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
11776 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
11777 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
11779 * config/arm/neon.md (neon_vrint): Likewise.
11780 (neon_vcvt): Likewise.
11781 (neon_<fmaxmin_op><mode>): Likewise.
11782 (<fmaxmin><mode>3): Likewise.
11783 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
11784 * config/arm/predicates.md (arm_cond_move_operator): Check against
11785 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
11787 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
11789 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
11790 to top of function.
11792 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11794 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
11795 loop in comment with memset.
11797 2017-07-14 Martin Liska <mliska@suse.cz>
11799 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
11800 * dwarf2out.c (is_java): Remove the function.
11801 (output_pubname): Remove usage of the function.
11802 (lower_bound_default): Remove usage of DW_LANG_Java.
11803 (gen_compile_unit_die): Likewise.
11804 * gcc.c: Remove compiler defaults for .java and .zip files.
11805 * gimple-expr.c (remove_suffix): Change as there's no longer
11806 extension than 4-letter one.
11807 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
11808 (gimplify_save_expr): Likewise.
11809 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
11810 as it's possible even for other languages than Java.
11811 * langhooks.h (struct lang_hooks): Remove Java from a comment.
11812 * lto-opts.c (lto_write_options): Remove reference to Java.
11813 * opts.c (strip_off_ending): Update file extension handling.
11814 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
11815 * tree-eh.c (lower_resx): Likewise.
11816 * tree.c (free_lang_data_in_type): Remove dead code.
11817 (find_decls_types_r): Likewise.
11818 (build_common_builtin_nodes): Remove Java from a comment.
11819 (verify_type): Remove dead code.
11820 * varasm.c (assemble_external): Remove Java from a comment.
11822 2017-07-14 Martin Liska <mliska@suse.cz>
11824 * opts.c (finish_options): Add quotes.
11825 (common_handle_option): Likewise.
11827 2017-07-14 Martin Liska <mliska@suse.cz>
11829 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
11830 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
11831 Remove N_SO_PASCAL.
11832 * dwarf2out.c (lower_bound_default): Do not handle
11834 (gen_compile_unit_die): Likewise.
11835 * gcc.c: Remove default extension binding for GNU Pascal.
11836 * stmt.c: Remove Pascal language from a comment.
11837 * xcoffout.c: Likewise.
11839 2017-07-13 David Malcolm <dmalcolm@redhat.com>
11842 * diagnostic-show-locus.c (fixit_cmp): New function.
11843 (layout::layout): Sort m_fixit_hints.
11844 (column_range::column_range): Assert that the values are valid.
11845 (struct char_span): New struct.
11846 (correction::overwrite): New method.
11847 (struct source_line): New struct.
11848 (line_corrections::add_hint): Add assertions. Reimplement memcpy
11849 calls in terms of classes source_line and char_span, and
11850 correction::overwrite.
11851 (selftest::test_overlapped_fixit_printing_2): New function.
11852 (selftest::diagnostic_show_locus_c_tests): Call it.
11854 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
11856 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
11857 early if there is no lhs.
11859 2017-07-13 Martin Liska <mliska@suse.cz>
11861 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
11862 (gen_reference_type_die): Likewise.
11863 * stor-layout.c: Remove Pascal-related comment.
11865 2017-07-13 Martin Liska <mliska@suse.cz>
11867 * opts.c (finish_options): Add quotes to error messages.
11868 (parse_sanitizer_options): Likewise.
11870 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
11872 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
11874 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
11876 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
11878 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
11880 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
11882 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
11884 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11887 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
11888 provides the hardware capability bits, define the macro
11889 __BUILTIN_CPU_SUPPORTS__.
11890 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
11891 if GLIBC does not provide the hardware capability bits. Add a
11892 gcc_unreachable call if the built-in cpu function is neither
11893 __builtin_cpu_is nor __builtin_cpu_supports.
11894 (rs6000_get_function_versions_dispatcher): Change the warning
11895 that an old GLIBC is used which does not export the capability
11896 bits to be an error.
11897 * doc/extend.texi (target_clones attribute): Document the
11898 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
11899 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
11900 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
11901 the macros defined by GCC if the newer GLIBC is available.
11903 2017-07-12 Jeff Law <law@redhat.com>
11905 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
11906 remaining includes slightly.
11907 * config/riscv/riscv-builtins.c: Include profile-count.h.
11909 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
11912 * config/avr/avr.c (avr_set_current_function): In diagnostic
11913 messages: Quote keywords and (parts of) identifiers.
11914 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
11917 2017-07-12 Carl Love <cel@us.ibm.com>
11919 * config/rs6000/rs6000-c.c: Add support for built-in functions
11920 vector bool char vec_revb (vector bool char);
11921 vector bool short vec_revb (vector short char);
11922 vector bool int vec_revb (vector bool int);
11923 vector bool long long vec_revb (vector bool long long);
11924 * doc/extend.texi: Update the built-in documentation file for the
11925 new built-in functions.
11927 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11929 * config/s390/s390.md: Remove movcc splitter.
11931 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11933 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
11934 load/store on condition.
11936 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
11939 * config/avr/avr.c (avr_encode_section_info)
11940 [progmem && !TREE_READONLY]: Error if progmem object needs
11943 2017-07-11 Michael Collison <michael.collison@arm.com>
11945 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
11948 2017-07-11 Carl Love <cel@us.ibm.com>
11950 * config/rs6000/rs6000-c.c: Add support for builtins
11951 vector unsigned int vec_parity_lsbb (vector signed int);
11952 vector unsigned int vec_parity_lsbb (vector unsigned int);
11953 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
11954 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
11955 vector unsigned long long vec_parity_lsbb (vector signed long long);
11956 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
11957 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
11958 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
11959 * doc/extend.texi: Update the built-in documentation file for the
11960 new built-in functions.
11962 2017-07-11 David Malcolm <dmalcolm@redhat.com>
11964 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
11965 (layout::m_primary_loc): New field.
11966 (layout::layout): Initialize new field. Move location filtering
11967 logic from here to...
11968 (layout::maybe_add_location_range): ...this new method. Add
11969 support for filtering to just the lines already specified by other
11971 (layout::will_show_line_p): New method.
11972 (gcc_rich_location::add_location_if_nearby): New method.
11973 (selftest::test_add_location_if_nearby): New test function.
11974 (selftest::diagnostic_show_locus_c_tests): Call it.
11975 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
11978 2017-07-11 Tom de Vries <tom@codesourcery.com>
11980 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
11981 (bb_first_real_insn): New function.
11982 (nvptx_single): Add extra initialization of broadcasted condition
11985 2017-07-11 Nathan Sidwell <nathan@acm.org>
11987 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
11989 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
11991 * doc/extend.texi (AVR Function Attributes): Remove weblink to
11992 Binutils doc as TEXI will mess them up.
11993 * doc/invoke.texi (AVR Options): Same here.
11995 2017-07-11 Daniel Cederman <cederman@gaisler.com>
11997 * config/sparc/sparc.opt (mfix-ut700): New option.
11998 (mfix-gr712rc): Likewise.
11999 (sparc_fix_b2bst): New variable.
12000 * doc/invoke.texi (SPARC options): Document them.
12001 (ARM options): Fix warnings.
12002 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
12003 instructions to prevent sequences that can trigger the store-store
12004 errata for certain LEON3FT processors.
12005 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
12006 (sparc_option_override): Set sparc_fix_b2bst appropriately.
12007 * config/sparc/sparc.md (fix_b2bst): New attribute.
12008 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
12010 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
12013 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
12015 (*rsqrtsf2_sse): Ditto.
12017 (div<mode>3): Macroize insn from divdf3 and divsf3
12018 using MODEF mode iterator.
12020 2017-07-10 Martin Sebor <msebor@redhat.com>
12022 PR tree-optimization/80397
12023 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
12024 instead of testing for equality to INTEGER_TYPE.
12026 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
12028 * config.gcc: Remove uclibc from arc target spec.
12030 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
12032 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
12034 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12037 * lto-wrapper.c (remove_option): New function.
12038 (merge_and_complain): Merge PIC/PIE options more realistically.
12040 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
12042 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
12046 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
12047 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
12048 * config.in: Regenerate.
12049 * configure: Regenerate.
12050 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
12051 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
12052 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
12053 (TARGET_GASISR_PROLOGUES): ...target mask.
12054 * common/config/avr/avr-common.c
12055 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
12056 Set -mgas-isr-prologues.
12057 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
12058 INSERT_PASS_BEFORE for it.
12059 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
12060 * config/avr/avr.c (avr_option_override)
12061 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
12062 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
12063 (avr_attribute_table) <no_gccisr>: Add new function attribute.
12064 (avr_set_current_function) <is_no_gccisr>: Init machine field.
12065 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
12067 (make_avr_pass_pre_proep): New function.
12068 (emit_push_sfr) <treg>: Add argument to function and use it
12069 instead of TMP_REG.
12070 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
12071 and set machine->gasisr.yes.
12072 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
12073 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
12074 __gcc_isr.n_pushed to .L__stack_usage.
12075 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
12076 (avr_asm_final_postscan_insn): ...this new static function.
12077 * config/avr/avr.h (machine_function)
12078 <is_no_gccisr, use_L__stack_usage>: New fields.
12079 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
12080 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
12081 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
12082 (gasisr, *gasisr): New expander and insn.
12083 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
12084 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
12085 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
12087 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
12089 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
12092 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
12094 Move jump-tables out of .text again.
12097 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
12098 (ASM_OUTPUT_ADDR_VEC): New function.
12099 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
12100 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
12101 INSN_ADDRESSes as asm comment.
12102 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
12103 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
12104 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
12105 * config/avr/avr.md (*tablejump): Adjust comment.
12106 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
12107 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
12109 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
12110 (avr_output_addr_vec): New proto.
12111 (avr_log_t) <insn_addresses>: New field.
12113 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
12116 * config/i386/i386.c (ix86_function_arg_advance): Set
12117 outgoing_args_on_stack to true if there are outgoing arguments
12119 (ix86_function_arg): Likewise.
12120 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
12121 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
12122 * config/i386/i386.h (machine_function): Add
12123 outgoing_args_on_stack.
12125 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
12127 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
12128 supporting pthreds.
12129 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
12131 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
12133 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
12134 (REAL_H): Remove $(MACHMODE_H).
12135 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
12137 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
12138 $(MACHMODE_H) and double-int.h.
12139 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
12141 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
12144 2017-07-07 Andrew Pinski <apinski@cavium.com>
12146 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
12147 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
12149 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
12151 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
12152 Add warning if GCC was not configured to link against a GLIBC that
12153 exports the hardware capability bits.
12154 (make_resolver_func): Make resolver function private and not a
12155 COMDAT function. Create the name with clone_function_name instead
12156 of make_unique_name.
12159 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
12160 correct operand in doing the split.
12162 2017-07-07 Carl Love <cel@us.ibm.com>
12164 * config/rs6000/rs6000-c: Add support for built-in function
12165 vector unsigned short vec_pack_to_short_fp32 (vector float,
12167 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
12168 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
12169 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
12170 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
12171 (convert_4f32_8i16): Add define_expand.
12172 * doc/extend.texi: Update the built-in documentation file for the
12173 new built-in function.
12175 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12177 * config/sparc/m8.md: New file.
12178 * config/sparc/sparc.md: Include m8.md.
12180 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12182 * config/sparc/sparc.opt: New option -mvis4b.
12183 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
12184 (sparc_option_override): Handle VIS4B.
12185 (enum sparc_builtins): Define
12186 SPARC_BUILTIN_DICTUNPACK{8,16,32},
12187 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
12188 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
12189 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
12190 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
12191 (check_constant_argument): New function.
12192 (sparc_vis_init_builtins): Define builtins
12193 __builtin_vis_dictunpack{8,16,32},
12194 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
12195 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
12196 __builtin_vis_fpcmpde{8,16,32}shl and
12197 __builtin_vis_fpcmpur{8,16,32}shl.
12198 (sparc_expand_builtin): Check that the constant operands to
12199 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
12200 constant and in range.
12201 * config/sparc/sparc-c.c (sparc_target_macros): Handle
12203 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
12204 (SPARC_IMM5_P): Likewise.
12205 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
12206 (enabled): Handle vis4b.
12207 (UNSPEC_DICTUNPACK): New unspec.
12208 (UNSPEC_FPCMPSHL): Likewise.
12209 (UNSPEC_FPUCMPSHL): Likewise.
12210 (UNSPEC_FPCMPDESHL): Likewise.
12211 (UNSPEC_FPCMPURSHL): Likewise.
12212 (cpu_feature): New CPU feature `vis4b'.
12213 (dictunpack{8,16,32}): New insns.
12214 (FPCSMODE): New mode iterator.
12215 (fpcscond): New code iterator.
12216 (fpcsucond): Likewise.
12217 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
12218 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
12219 (fpcmpde{8,16,32}{si,di}shl): Likewise.
12220 (fpcmpur{8,16,32}{si,di}shl): Likewise.
12221 * config/sparc/constraints.md: Define constraints `q' for unsigned
12222 2-bit integer constants and `t' for unsigned 5-bit integer
12224 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
12226 (imm5_operand_dictunpack16): Likewise.
12227 (imm5_operand_dictunpack32): Likewise.
12228 (imm2_operand): Likewise.
12229 * doc/invoke.texi (SPARC Options): Document -mvis4b.
12230 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
12231 ditunpack* and fpcmp*shl builtins.
12233 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12235 * config.gcc: Handle m8 in --with-{cpu,tune} options.
12236 * config.in: Add HAVE_AS_SPARC6 define.
12237 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
12239 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
12241 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
12242 (CPP_CPU_SPEC): Handle m8.
12243 (ASM_CPU_SPEC): Likewise.
12244 * config/sparc/sparc-opts.h (enum processor_type): Add
12246 * config/sparc/sparc.c (m8_costs): New struct.
12247 (sparc_option_override): Handle TARGET_CPU_m8.
12248 (sparc32_initialize_trampoline): Likewise.
12249 (sparc64_initialize_trampoline): Likewise.
12250 (sparc_issue_rate): Likewise.
12251 (sparc_register_move_cost): Likewise.
12252 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
12253 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
12254 (ASM_CPU64_DEFAULT_SPEC): Likewise.
12255 (CPP_CPU_SPEC): Handle M8.
12256 (ASM_CPU_SPEC): Likewise.
12257 (AS_M8_FLAG): Define.
12258 * config/sparc/sparc.md: Add m8 to the cpu attribute.
12259 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
12260 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
12262 * configure: Regenerate.
12263 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
12266 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12268 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
12270 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
12271 ("*movdi_insn_sp32"): Do not set v3pipe.
12272 ("*movsi_insn"): Likewise.
12273 ("*movdi_insn_sp64"): Likewise.
12274 ("*movsf_insn"): Likewise.
12275 ("*movdf_insn_sp32"): Likewise.
12276 ("*movdf_insn_sp64"): Likewise.
12277 ("*zero_extendsidi2_insn_sp64"): Likewise.
12278 ("*sign_extendsidi2_insn"): Likewise.
12279 ("*mov<VM32:mode>_insn"): Likewise.
12280 ("*mov<VM64:mode>_insn_sp64"): Likewise.
12281 ("*mov<VM64:mode>_insn_sp32"): Likewise.
12282 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
12283 ("<vlop:code><VL:mode>3"): Likewise.
12284 ("*not_<vlop:code><VL:mode>3"): Likewise.
12285 ("*nand<VL:mode>_vis"): Likewise.
12286 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
12287 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
12288 ("one_cmpl<VL:mode>2"): Likewise.
12289 ("faligndata<VM64:mode>_vis"): Likewise.
12290 ("alignaddrsi_vis"): Likewise.
12291 ("alignaddrdi_vis"): Likweise.
12292 ("alignaddrlsi_vis"): Likewise.
12293 ("alignaddrldi_vis"): Likewise.
12294 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
12295 ("bmaskdi_vis"): Likewise.
12296 ("bmasksi_vis"): Likewise.
12297 ("bshuffle<VM64:mode>_vis"): Likewise.
12298 ("cmask8<P:mode>_vis"): Likewise.
12299 ("cmask16<P:mode>_vis"): Likewise.
12300 ("cmask32<P:mode>_vis"): Likewise.
12301 ("pdistn<P:mode>_vis"): Likewise.
12302 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
12304 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12306 * config/sparc/sparc.md ("subtype"): New insn attribute.
12307 ("*wrgsr_sp64"): Set insn subtype.
12308 ("*rdgsr_sp64"): Likewise.
12309 ("alignaddrsi_vis"): Likewise.
12310 ("alignaddrdi_vis"): Likewise.
12311 ("alignaddrlsi_vis"): Likewise.
12312 ("alignaddrldi_vis"): Likewise.
12313 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
12314 ("fexpand_vis"): Likewise.
12315 ("fpmerge_vis"): Likewise.
12316 ("faligndata<VM64:mode>_vis"): Likewise.
12317 ("bshuffle<VM64:mode>_vis"): Likewise.
12318 ("cmask8<P:mode>_vis"): Likewise.
12319 ("cmask16<P:mode>_vis"): Likewise.
12320 ("cmask32<P:mode>_vis"): Likewise.
12321 ("fchksm16_vis"): Likewise.
12322 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
12323 ("fmean16_vis"): Likewise.
12324 ("fp<plusminus_insn>64_vis"): Likewise.
12325 ("<plusminus_insn>v8qi3"): Likewise.
12326 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
12327 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
12328 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
12329 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
12330 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
12331 ("*movqi_insn"): Likewise.
12332 ("*movhi_insn"): Likewise.
12333 ("*movsi_insn"): Likewise.
12334 ("movsi_pic_gotdata_op"): Likewise.
12335 ("*movdi_insn_sp32"): Likewise.
12336 ("*movdi_insn_sp64"): Likewise.
12337 ("movdi_pic_gotdata_op"): Likewise.
12338 ("*movsf_insn"): Likewise.
12339 ("*movdf_insn_sp32"): Likewise.
12340 ("*movdf_insn_sp64"): Likewise.
12341 ("*zero_extendhisi2_insn"): Likewise.
12342 ("*zero_extendqihi2_insn"): Likewise.
12343 ("*zero_extendqisi2_insn"): Likewise.
12344 ("*zero_extendqidi2_insn"): Likewise.
12345 ("*zero_extendhidi2_insn"): Likewise.
12346 ("*zero_extendsidi2_insn_sp64"): Likewise.
12347 ("ldfsr"): Likewise.
12348 ("prefetch_64"): Likewise.
12349 ("prefetch_32"): Likewise.
12350 ("tie_ld32"): Likewise.
12351 ("tie_ld64"): Likewise.
12352 ("*tldo_ldub_sp32"): Likewise.
12353 ("*tldo_ldub1_sp32"): Likewise.
12354 ("*tldo_ldub2_sp32"): Likewise.
12355 ("*tldo_ldub_sp64"): Likewise.
12356 ("*tldo_ldub1_sp64"): Likewise.
12357 ("*tldo_ldub2_sp64"): Likewise.
12358 ("*tldo_ldub3_sp64"): Likewise.
12359 ("*tldo_lduh_sp32"): Likewise.
12360 ("*tldo_lduh1_sp32"): Likewise.
12361 ("*tldo_lduh_sp64"): Likewise.
12362 ("*tldo_lduh1_sp64"): Likewise.
12363 ("*tldo_lduh2_sp64"): Likewise.
12364 ("*tldo_lduw_sp32"): Likewise.
12365 ("*tldo_lduw_sp64"): Likewise.
12366 ("*tldo_lduw1_sp64"): Likewise.
12367 ("*tldo_ldx_sp64"): Likewise.
12368 ("*mov<VM32:mode>_insn"): Likewise.
12369 ("*mov<VM64:mode>_insn_sp64"): Likewise.
12370 ("*mov<VM64:mode>_insn_sp32"): Likewise.
12372 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12374 * config/sparc/sparc.md ("type"): New insn type viscmp.
12375 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
12377 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
12378 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
12379 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
12380 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
12382 ("n7_vis_logical_11cycle"): Likewise.
12383 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
12384 * config/sparc/niagara2.md ("niag3_vis": Likewise.
12385 * config/sparc/niagara.md ("niag_vis"): Likewise.
12386 * config/sparc/ultra3.md ("us3_fga"): Likewise.
12387 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
12389 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12391 * config/sparc/sparc.md: New instruction type `bmask'.
12392 (bmaskdi_vis): Use the `bmask' type.
12393 (bmasksi_vis): Likewise.
12394 * config/sparc/ultra3.md (us3_array): Likewise.
12395 * config/sparc/niagara7.md (n7_array): Likewise.
12396 * config/sparc/niagara4.md (n4_array): Likewise.
12397 * config/sparc/niagara2.md (niag2_vis): Likewise.
12398 (niag3_vis): Likewise.
12399 * config/sparc/niagara.md (niag_vis): Likewise.
12401 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12403 * ipa-comdats.c: Remove optimize check from gate.
12404 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
12405 for functions not optimized.
12406 (ipa_fn_summary_read): Skip optimize check.
12407 (ipa_fn_summary_write): Likewise.
12408 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
12410 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
12412 (can_inline_edge_p): Check flag_pcc_struct_return for match.
12413 (check_callers): Give up on caller which is not optimized.
12414 (inline_small_functions): Likewise.
12415 (ipa_inline): Do not give up when not optimizing.
12416 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
12417 away unoptimizes cdtors.
12418 (whole_program_function_and_variable_visibility): Do
12419 ipa_discover_readonly_nonaddressable_vars in LTO mode.
12420 * ipa.c (process_references): Do not check optimize.
12421 (symbol_table::remove_unreachable_nodes): Update optimize check.
12422 (set_writeonly_bit): Update optimize check.
12423 (pass_ipa_cdtor_merge::gate): Do not check optimize.
12424 (pass_ipa_single_use::gate): Remove.
12426 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12428 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
12429 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
12430 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
12431 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
12432 permute_load, permute_store, adjust_extract, adjust_splat,
12433 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
12434 replace_swap_with_copy, dump_swap_insn_table,
12435 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
12436 recombine_lvx_pattern, recombine_stvx_pattern,
12437 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
12438 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
12439 to file rs6000-p8swap.c.
12440 * config/rs6000/rs6000-p8swap.c: New file.
12441 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
12442 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
12443 and rs6000*-*-* targets.
12445 2017-07-06 David Malcolm <dmalcolm@redhat.com>
12447 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
12449 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12451 * lto-wrapper.c (merge_and_complain): Do not merge
12452 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
12453 fsigned_zeros, ftrapping_math, fwrapv.
12454 (append_compiler_options): Do not track these options.
12455 (append_linker_options): Likewie
12457 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12459 * cgraphunit.c (cgraph_node::finalize_function): When
12460 !flag_toplevel_reorde set no_reorder flag.
12461 (varpool_node::finalize_decl): Likewise.
12462 (symbol_table::compile): Drop no toplevel reorder path.
12464 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12466 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
12467 edges; zero probability is not better than uninitialized.
12469 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
12471 * asan.h (asan_sanitize_allocas_p): Declare.
12472 * asan.c (asan_sanitize_allocas_p): New function.
12473 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
12474 (handle_builtin_alloca): Likewise.
12475 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
12476 if !asan_sanitize_allocas_p.
12477 * params.def (asan-instrument-allocas): Add new option.
12478 * params.h (ASAN_PROTECT_ALLOCAS): Define.
12479 * opts.c (common_handle_option): Disable allocas sanitization for
12482 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
12484 * asan.c: Include gimple-fold.h.
12485 (get_last_alloca_addr): New function.
12486 (handle_builtin_stackrestore): Likewise.
12487 (handle_builtin_alloca): Likewise.
12488 (asan_emit_allocas_unpoison): Likewise.
12489 (get_mem_refs_of_builtin_call): Add new parameter, remove const
12490 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
12491 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
12492 (instrument_builtin_call): Pass gimple iterator to
12493 get_mem_refs_of_builtin_call.
12494 (last_alloca_addr): New global.
12495 * asan.h (asan_emit_allocas_unpoison): Declare.
12496 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
12497 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
12498 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
12499 if function calls alloca.
12500 * gimple-fold.c (replace_call_with_value): Remove static keyword.
12501 * gimple-fold.h (replace_call_with_value): Declare.
12502 * internal-fn.c: Include asan.h.
12503 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
12504 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
12506 2017-07-06 David Malcolm <dmalcolm@redhat.com>
12508 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
12509 (C_SELFTEST_FLAGS): New.
12510 (CPP_SELFTEST_FLAGS): New.
12511 (SELFTEST_DEPS): New, from deps of s-selftest.
12512 (C_SELFTEST_DEPS): New, from deps of s-selftest.
12513 (CPP_SELFTEST_DEPS): New.
12514 (selftest): Add dependency on s-selftest-c++.
12515 (s-selftest): Rename to...
12516 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
12517 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
12518 than SELFTEST_FLAGS.
12519 (selftest-gdb): Rename to...
12520 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
12522 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
12523 (selftest-valgrind): Rename to...
12524 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
12526 (selftest-valgrind): Reintroduce as an alias for
12527 selftest-c-valgrind.
12528 (s-selftest-c++): New.
12529 (selftest-c++-gdb): New.
12530 (selftest-c++-valgrind): New.
12532 2017-07-06 Olivier Hainque <hainque@adacore.com>
12534 * gcc.c (process_command): When deciding if undefined variables
12535 should be ignored when processing specs, accept "gcc -v" as well.
12537 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12539 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
12540 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
12542 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
12544 * config/arm/arm-cpus.in (armv8-r): Add new entry.
12545 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
12546 * config/arm/arm-tables.opt: Regenerate.
12547 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
12549 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
12551 2017-07-06 Carl Love <cel@us.ibm.com>
12553 * ChangeLog: Clean up from mid air collision
12555 2017-07-06 Carl Love <cel@us.ibm.com>
12557 * config/rs6000/rs6000-c.c: Add support for built-in functions
12558 vector signed int vec_subc (vector signed int, vector signed int);
12559 vector signed __int128 vec_subc (vector signed __int128,
12560 vector signed __int128);
12561 vector unsigned __int128 vec_subc (vector unsigned __int128,
12562 vector unsigned __int128);
12563 vector signed int vec_sube (vector signed int, vector signed int,
12564 vector signed int);
12565 vector unsigned int vec_sube (vector unsigned int,
12566 vector unsigned int,
12567 vector unsigned int);
12568 vector signed __int128 vec_sube (vector signed __int128,
12569 vector signed __int128,
12570 vector signed__int128);
12571 vector unsigned __int128 vec_sube (vector unsigned __int128,
12572 vector unsigned __int128,
12573 vector unsigned __int128);
12574 vector signed int vec_subec (vector signed int, vector signed int,
12575 vector signed int);
12576 vector unsigned int vec_subec (vector unsigned int,
12577 vector unsigned int,
12578 vector unsigned int);
12579 vector signed __int128 vec_subec (vector signed __int128,
12580 vector signed __int128,
12581 vector signed__int128);
12582 vector unsigned __int128 vec_subec (vector unsigned __int128,
12583 vector unsigned __int128,
12584 vector unsigned __int128);
12585 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
12586 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
12587 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
12588 BU_ALTIVEC_OVERLOAD_X definitions.
12589 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
12590 * doc/extend.texi: Update the built-in documentation file for the new
12591 built-in functions.
12593 2017-07-06 David Malcolm <dmalcolm@redhat.com>
12596 * diagnostic-show-locus.c (layout::layout): Use start and finish
12597 spelling location for the start and finish of each range.
12598 * genmatch.c (linemap_client_expand_location_to_spelling_point):
12599 Add unused aspect param.
12600 * input.c (expand_location_1): Add "aspect" param, and use it
12601 to access the correct part of the location.
12602 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
12604 (expand_location_to_spelling_point): Likewise.
12605 (linemap_client_expand_location_to_spelling_point): Add "aspect"
12606 param, and pass it to expand_location_1.
12608 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
12610 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
12611 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
12612 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
12613 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
12614 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
12615 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
12616 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
12617 _mm_maskz_getmant_ss): New intrinsics.
12618 (__builtin_ia32_getexpss128_mask): Changed to ...
12619 __builtin_ia32_getexpss128_round ... this.
12620 (__builtin_ia32_getexpsd128_mask): Changed to ...
12621 __builtin_ia32_getexpsd128_round ... this.
12622 * config/i386/i386-builtin-types.def
12623 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
12624 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
12625 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
12626 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
12627 __builtin_ia32_getmantss_mask_round): New builtins.
12628 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
12629 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
12630 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
12631 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
12632 * config/i386/sse.md
12633 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
12634 avx512f_sgetexp<mode><mask_scalar_name>
12635 <round_saeonly_scalar_name> ... this.
12636 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
12637 %0, %1, %2<round_saeonly_op3>}): Changed to ...
12638 vgetexp<ssescalarmodesuffix>
12639 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12640 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
12641 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
12642 avx512f_vgetmant<mode><mask_scalar_name>
12643 <round_saeonly_scalar_name> ... this.
12644 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
12645 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
12646 vgetmant<ssescalarmodesuffix>
12647 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
12648 %0<mask_scalar_operand4>, %1, %2
12649 <round_saeonly_scalar_mask_op4>, %3} ... this.
12650 * config/i386/subst.md (mask_scalar_operand4,
12651 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
12652 round_saeonly_scalar_nimm_predicate): New subst attributes.
12654 2017-07-06 Julia Koval <julia.koval@intel.com>
12656 * config/i386/i386.c (ix86_erase_embedded_rounding):
12657 Remove code for old rounding pattern.
12659 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
12661 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
12663 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
12665 * doc/sourcebuild.texi (Test Directives, Variants of
12666 dg-require-support): Add documentation for dg-require-stack-check.
12668 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
12670 * config/i386/subst.md (mask_scalar, round_scalar,
12671 round_saeonly_scalar): New meta-templates.
12672 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
12673 round_scalar_mask_operand3, round_scalar_mask_op3,
12674 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
12675 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
12676 round_saeonly_scalar_constraint,
12677 round_saeonly_scalar_prefix): New subst attribute.
12678 * config/i386/sse.md
12679 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
12680 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
12681 <round_scalar_name> ... this.
12682 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
12683 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
12684 <round_scalar_name> ... this.
12685 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
12686 <sse>_vm<code><mode>3<mask_scalar_name>
12687 <round_saeonly_scalar_name> ... this.
12688 (v<plusminus_mnemonic><ssescalarmodesuffix>
12689 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
12690 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
12691 v<plusminus_mnemonic><ssescalarmodesuffix>
12692 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12693 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
12694 (v<multdiv_mnemonic><ssescalarmodesuffix>
12695 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
12696 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
12697 v<multdiv_mnemonic><ssescalarmodesuffix>
12698 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12699 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
12700 (v<maxmin_float><ssescalarmodesuffix>
12701 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
12702 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
12703 v<maxmin_float><ssescalarmodesuffix>
12704 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12705 %0<mask_scalar_operand3>, %1, %<iptr>2
12706 <round_saeonly_scalar_mask_op3>} ... this.
12708 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
12710 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
12711 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
12713 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
12714 Alan Hayward <alan.hayward@arm.com>
12715 David Sherwood <david.sherwood@arm.com>
12717 * combine.c (simplify_if_then_else): Remove "enum" before
12719 * compare-elim.c (can_eliminate_compare): Likewise.
12720 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
12722 (aarch64_lookup_simd_builtin_type): Likewise.
12723 (aarch64_simd_builtin_type): Likewise.
12724 (aarch64_init_simd_builtin_types): Likewise.
12725 (aarch64_simd_expand_args): Likewise.
12726 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
12728 (aarch64_reverse_mask): Likewise.
12729 (aarch64_simd_emit_reg_reg_move): Likewise.
12730 (aarch64_gen_adjusted_ldpstp): Likewise.
12731 (aarch64_ccmp_mode_to_code): Likewise.
12732 (aarch64_operands_ok_for_ldpstp): Likewise.
12733 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12734 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
12736 (aarch64_min_divisions_for_recip_mul): Likewise.
12737 (aarch64_reassociation_width): Likewise.
12738 (aarch64_get_condition_code_1): Likewise.
12739 (aarch64_simd_emit_reg_reg_move): Likewise.
12740 (aarch64_simd_attr_length_rglist): Likewise.
12741 (aarch64_reverse_mask): Likewise.
12742 (aarch64_operands_ok_for_ldpstp): Likewise.
12743 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12744 (aarch64_gen_adjusted_ldpstp): Likewise.
12745 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
12747 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
12748 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
12749 (arm_lookup_simd_builtin_type): Likewise.
12750 (arm_simd_builtin_type): Likewise.
12751 (arm_init_simd_builtin_types): Likewise.
12752 (arm_expand_builtin_args): Likewise.
12753 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
12754 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
12755 (ft32_setup_incoming_varargs): Likewise.
12756 (ft32_function_arg): Likewise.
12757 (ft32_function_arg_advance): Likewise.
12758 (ft32_pass_by_reference): Likewise.
12759 (ft32_arg_partial_bytes): Likewise.
12760 (ft32_valid_pointer_mode): Likewise.
12761 (ft32_addr_space_pointer_mode): Likewise.
12762 (ft32_addr_space_legitimate_address_p): Likewise.
12763 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
12765 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
12766 (ix86_emit_outlined_ms2sysv_restore): Likewise.
12767 (iamcu_alignment): Likewise.
12768 (canonicalize_vector_int_perm): Likewise.
12769 (ix86_noce_conversion_profitable_p): Likewise.
12770 (ix86_mpx_bound_mode): Likewise.
12771 (ix86_operands_ok_for_move_multiple): Likewise.
12772 * config/microblaze/microblaze-protos.h
12773 (microblaze_expand_conditional_branch_reg): Likewise.
12774 * config/microblaze/microblaze.c
12775 (microblaze_expand_conditional_branch_reg): Likewise.
12776 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12778 (rs6000_reassociation_width): Likewise.
12779 (rs6000_invalid_binary_op): Likewise.
12780 (fusion_p9_p): Likewise.
12781 (emit_fusion_p9_load): Likewise.
12782 (emit_fusion_p9_store): Likewise.
12783 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
12785 (riscv_hard_regno_mode_ok_p): Likewise.
12786 (riscv_address_insns): Likewise.
12787 (riscv_split_symbol): Likewise.
12788 (riscv_legitimize_move): Likewise.
12789 (riscv_function_value): Likewise.
12790 (riscv_hard_regno_nregs): Likewise.
12791 (riscv_expand_builtin): Likewise.
12792 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
12793 (riscv_build_integer): Likewise.
12794 (riscv_split_integer): Likewise.
12795 (riscv_legitimate_constant_p): Likewise.
12796 (riscv_cannot_force_const_mem): Likewise.
12797 (riscv_regno_mode_ok_for_base_p): Likewise.
12798 (riscv_valid_base_register_p): Likewise.
12799 (riscv_valid_offset_p): Likewise.
12800 (riscv_valid_lo_sum_p): Likewise.
12801 (riscv_classify_address): Likewise.
12802 (riscv_legitimate_address_p): Likewise.
12803 (riscv_address_insns): Likewise.
12804 (riscv_load_store_insns): Likewise.
12805 (riscv_force_binary): Likewise.
12806 (riscv_split_symbol): Likewise.
12807 (riscv_force_address): Likewise.
12808 (riscv_legitimize_address): Likewise.
12809 (riscv_move_integer): Likewise.
12810 (riscv_legitimize_const_move): Likewise.
12811 (riscv_legitimize_move): Likewise.
12812 (riscv_address_cost): Likewise.
12813 (riscv_subword): Likewise.
12814 (riscv_output_move): Likewise.
12815 (riscv_canonicalize_int_order_test): Likewise.
12816 (riscv_emit_int_order_test): Likewise.
12817 (riscv_function_arg_boundary): Likewise.
12818 (riscv_pass_mode_in_fpr_p): Likewise.
12819 (riscv_pass_fpr_single): Likewise.
12820 (riscv_pass_fpr_pair): Likewise.
12821 (riscv_get_arg_info): Likewise.
12822 (riscv_function_arg): Likewise.
12823 (riscv_function_arg_advance): Likewise.
12824 (riscv_arg_partial_bytes): Likewise.
12825 (riscv_function_value): Likewise.
12826 (riscv_pass_by_reference): Likewise.
12827 (riscv_setup_incoming_varargs): Likewise.
12828 (riscv_print_operand): Likewise.
12829 (riscv_elf_select_rtx_section): Likewise.
12830 (riscv_save_restore_reg): Likewise.
12831 (riscv_for_each_saved_reg): Likewise.
12832 (riscv_register_move_cost): Likewise.
12833 (riscv_hard_regno_mode_ok_p): Likewise.
12834 (riscv_hard_regno_nregs): Likewise.
12835 (riscv_class_max_nregs): Likewise.
12836 (riscv_memory_move_cost): Likewise.
12837 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
12838 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
12839 (rl78_addr_space_address_mode): Likewise.
12840 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12842 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12843 (rs6000_reassociation_width): Likewise.
12844 (rs6000_invalid_binary_op): Likewise.
12845 (fusion_p9_p): Likewise.
12846 (emit_fusion_p9_load): Likewise.
12847 (emit_fusion_p9_store): Likewise.
12848 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
12849 (ok_for_simple_move_operands): Likewise.
12850 (ok_for_simple_move_strict_operands): Likewise.
12851 (ok_for_simple_arith_logic_operands): Likewise.
12852 (visium_legitimize_reload_address): Likewise.
12853 (visium_select_cc_mode): Likewise.
12854 (output_cbranch): Likewise.
12855 (visium_split_double_move): Likewise.
12856 (visium_expand_copysign): Likewise.
12857 (visium_expand_int_cstore): Likewise.
12858 (visium_expand_fp_cstore): Likewise.
12859 * config/visium/visium.c (visium_pass_by_reference): Likewise.
12860 (visium_function_arg): Likewise.
12861 (visium_function_arg_advance): Likewise.
12862 (visium_libcall_value): Likewise.
12863 (visium_setup_incoming_varargs): Likewise.
12864 (visium_legitimate_constant_p): Likewise.
12865 (visium_legitimate_address_p): Likewise.
12866 (visium_legitimize_address): Likewise.
12867 (visium_secondary_reload): Likewise.
12868 (visium_register_move_cost): Likewise.
12869 (visium_memory_move_cost): Likewise.
12870 (prepare_move_operands): Likewise.
12871 (ok_for_simple_move_operands): Likewise.
12872 (ok_for_simple_move_strict_operands): Likewise.
12873 (ok_for_simple_arith_logic_operands): Likewise.
12874 (visium_function_value_1): Likewise.
12875 (rtx_ok_for_offset_p): Likewise.
12876 (visium_legitimize_reload_address): Likewise.
12877 (visium_split_double_move): Likewise.
12878 (visium_expand_copysign): Likewise.
12879 (visium_expand_int_cstore): Likewise.
12880 (visium_expand_fp_cstore): Likewise.
12881 (visium_split_cstore): Likewise.
12882 (visium_select_cc_mode): Likewise.
12883 (visium_split_cbranch): Likewise.
12884 (output_cbranch): Likewise.
12885 (visium_print_operand_address): Likewise.
12886 * expmed.c (flip_storage_order): Likewise.
12887 * expmed.h (emit_cstore): Likewise.
12888 (flip_storage_order): Likewise.
12889 * genrecog.c (validate_pattern): Likewise.
12890 * hsa-gen.c (gen_hsa_addr): Likewise.
12891 * internal-fn.c (expand_arith_overflow): Likewise.
12892 * ira-color.c (allocno_copy_cost_saving): Likewise.
12893 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12894 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
12895 (process_invariant_for_inheritance): Likewise.
12896 * lra-eliminations.c (move_plus_up): Likewise.
12897 * omp-low.c (lower_oacc_reductions): Likewise.
12898 * simplify-rtx.c (simplify_subreg): Likewise.
12899 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
12900 (TARGET_CHKP_BOUND_MODE): Likewise..
12901 * targhooks.c (default_chkp_bound_mode): Likewise.
12902 (default_setup_incoming_vararg_bounds): Likewise.
12903 * targhooks.h (default_chkp_bound_mode): Likewise.
12904 (default_setup_incoming_vararg_bounds): Likewise.
12905 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
12906 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
12907 (have_whole_vector_shift): Likewise.
12908 * tree-vect-stmts.c (vectorizable_load): Likewise.
12909 * doc/tm.texi: Regenerate.
12911 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
12913 Graceful degrade if Binutils PR21472 is not available.
12916 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
12917 .rodata in flash test fails.
12918 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
12919 * confgure: Regenerate.
12920 * config.in: Regenerate.
12921 * config/avr/avr.c (avr_asm_named_section)
12922 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
12923 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
12924 (avr_asm_init_sections): Same.
12926 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12928 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
12929 (fma<VH:mode>4_intrinsic): Likewise.
12930 (*fmsub<VCVTF:mode>4): Likewise.
12931 (*fmsub<VH:mode>4_intrinsic): Likewise.
12933 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
12936 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
12937 Don't depend on "optimize > 0".
12938 (out_movhi_r_mr, out_movqi_mr_r): Same.
12939 (out_movhi_mr_r, out_movqi_r_mr): Same.
12940 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
12941 io_address_operand on "optimize > 0".
12943 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12945 * tree-loop-distribution.c: Add general explanantion on the pass.
12946 (generate_loops_for_partition): Mark distributed loop.
12947 (pg_add_dependence_edges): New parameter. Handle alias data
12948 dependence specially and record it in the parameter if asked.
12949 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
12950 (init_partition_graph_vertices, add_partition_graph_edge): New.
12951 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
12952 (free_partition_graph_vdata, build_partition_graph): New.
12953 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
12954 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
12955 (data_ref_segment_size, latch_dominated_by_data_ref): New.
12956 (compute_alias_check_pairs, version_loop_by_alias_check): New.
12957 (version_for_distribution_p, finalize_partitions): New.
12958 (distribute_loop): Handle alias data dependence specially. Factor
12959 out loop fusion code as functions and call these functions.
12961 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12963 * tree-loop-distribution.c (classify_partition): New parameter and
12964 better handle reduction statement.
12965 (rdg_build_partitions): Revise comment.
12966 (distribute_loop): Compute statements in all partitions and pass it
12967 to classify_partition.
12969 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12971 * tree-loop-distribution.c (enum partition_type): New.
12972 (struct partition): New field type.
12973 (partition_merge_into): Add parameter. Update partition type.
12974 (data_dep_in_cycle_p, update_type_for_merge): New functions.
12975 (build_rdg_partition_for_vertex): Compute partition type.
12976 (rdg_build_partitions): Dump partition type.
12977 (distribute_loop): Update calls to partition_merge_into.
12979 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12981 * tree-loop-distribution.c (struct ddr_hasher): New.
12982 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
12984 (classify_partition): Call get_data_dependence.
12985 (pg_add_dependence_edges): Ditto.
12986 (distribute_loop): Release data dependence hash table.
12988 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12990 * tree-loop-distribution.c (ref_base_address): Delete.
12991 (similar_memory_accesses): Rename ...
12992 (share_memory_accesses): ... to this. Check if partitions access
12993 the same memory reference.
12994 (distribute_loop): Call share_memory_accesses.
12996 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12998 * tree-loop-distribution.c (struct partition): New field recording
12999 its data reference.
13000 (partition_alloc, partition_free): Init and release data refs.
13001 (partition_merge_into): Merge data refs.
13002 (build_rdg_partition_for_vertex): Collect data refs for partition.
13003 (pg_add_dependence_edges): Change parameters from vector to bitmap.
13005 (distribute_loop): Remve data refs from vertice data of partition
13008 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13010 * tree-loop-distribution.c (params.h): Include header file.
13011 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
13012 (datarefs_vec): New global var.
13013 (create_rdg_vertices): Use datarefs_vec directly.
13014 (free_rdg): Don't free data references.
13015 (build_rdg): Update use. Don't free data references.
13016 (distribute_loop): Compute global variable for data references.
13017 Bail out if there are too many data references.
13019 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13021 * tree-loop-distribution.c (loop_nest): New global var.
13022 (build_rdg): Use loop directly, rather than loop nest.
13023 (pg_add_dependence_edges): Remove loop nest parameter. Use global
13025 (distribute_loop): Compute global variable loop nest. Update use.
13027 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13029 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
13030 (partition_merge_into): New parameter. Dump reason for fusion.
13031 (distribute_loop): Update use of partition_merge_into.
13033 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13035 * tree-loop-distribution.c (bb_top_order_index): New.
13036 (bb_top_order_index_size, bb_top_order_cmp): New.
13037 (stmts_from_loop): Use topological order.
13038 (pass_loop_distribution::execute): Compute and release topological
13039 order for basic blocks.
13041 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13043 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
13046 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13048 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
13049 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
13050 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
13051 * internal-fn.def (LOOP_DIST_ALIAS): New.
13052 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
13053 (fold_loop_internal_call): ... this.
13054 (vect_loop_dist_alias_call): New function.
13055 (set_uid_loop_bbs): Call fold_loop_internal_call.
13056 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
13059 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
13062 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
13063 Require dead FLAGS_REG at the beginning of a peephole.
13065 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
13068 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
13069 arguments in the call to __builtin_ia32_sbb_u32.
13070 (_subborrow_u64): Swap _X and _Y arguments in the call to
13071 __builtin_ia32_sbb_u64.
13073 2017-07-04 Jakub Jelinek <jakub@redhat.com>
13076 * tree-vrp.c (compare_assert_loc): Turn into a function template
13077 with stable template parameter. Only test if a->e is NULL,
13078 !a->e == !b->e has been verified already. Use e == NULL or
13079 e != NULL instead of e or ! e tests. If stable is true, don't use
13080 iterative_hash_expr, on the other side allow a or b or both NULL
13081 and sort the NULLs last.
13082 (process_assert_insertions): Sort using compare_assert_loc<false>
13083 instead of compare_assert_loc, later sort using
13084 compare_assert_loc<true> before calling process_assert_insertions_for
13085 in a loop. Use break instead of continue once seen NULL pointer.
13087 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
13089 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
13090 Cortex-R7 and Cortex-R8 processors.
13092 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
13094 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
13095 uninitialized while src is not.
13097 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
13099 * common/config/arm/arm-common.c: Adjust include path for
13101 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
13102 (arm-cpu.h): Create in build directory. Adjust dependency rules.
13103 (arm-cpu-data.h): Likewise.
13104 (arm-cpu-cdata.h): Likewise.
13105 * config/arm/arm-cpu.h: Delete.
13106 * config/arm/arm-cpu-cdata.h: Delete.
13107 * config/arm/arm-cpu-data.h: Delete.
13109 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
13111 * config/arm/arm-cpus.in (cortex-a55): New.
13112 (cortex-a75): Likewise.
13113 (cortex-a75.cortex-a55): Likewise.
13114 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
13116 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
13117 * config/arm/arm-cpu-cdata.h: Regenerate.
13118 * config/arm/arm-cpu-data.h: Regenerate.
13119 * config/arm/arm-cpu.h: Regenerate.
13120 * config/arm/arm-tables.opt: Regenerate.
13121 * config/arm/arm-tune.md: Regenerate.
13123 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
13125 * haifa-sched.c (sched_create_recovery_edges): Update profile.
13127 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
13129 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
13132 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
13134 PR tree-optimization/81292
13135 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
13136 full_string_p, also call adjust_related_strinfos if the adjustment
13137 is simple, otherwise invalidate related strinfos.
13139 2017-07-04 Martin Liska <mliska@suse.cz>
13142 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
13143 newly created variable as DECL_IGNORED_P.
13145 2017-07-04 Martin Liska <mliska@suse.cz>
13148 * ipa-inline.c (inline_small_functions):
13149 Use xstrdup_for_dump.
13151 2017-07-04 Tom de Vries <tom@codesourcery.com>
13153 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
13155 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
13158 * config/darwin.c (darwin_function_switched_text_sections):
13161 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
13163 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
13165 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
13167 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
13169 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13171 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
13172 min_profitable_iters, and th as inclusive lower bounds.
13173 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
13174 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
13175 for min_profitable_iters and min_profitable_estimate.
13176 (vect_transform_loop): Treat th as an inclusive lower bound.
13177 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
13179 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
13182 * config/darwin.c (darwin_function_switched_text_sections):
13183 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
13184 in two pieces, and suppress the use of buf.
13186 2017-07-03 Nathan Sidwell <nathan@acm.org>
13188 * hash-table.h (hash_table_mod1): Fix indentation.
13190 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13192 PR middle-end/81290
13193 * predict.c (force_edge_cold): Be more careful about propagation
13195 * profile-count.h (profile_probability::guessed,
13196 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
13198 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
13200 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
13202 * doc/invoke.texi (rcpc architecture extension): Document it.
13204 2017-07-03 Richard Biener <rguenther@suse.de>
13206 PR tree-optimization/60510
13207 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
13208 the scalar reduction PHI and use it.
13209 (vectorizable_reduction): Properly guard the single_defuse_cycle
13210 path for non-SLP reduction chains where we cannot use it.
13211 Rework reduc_def/index and vector type deduction. Rework
13212 vector operand gathering during reduction op code-gen.
13213 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
13214 chains dissolve the chain and leave it to non-SLP reduction
13217 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13219 * tree-data-ref.h (dr_alignment): Declare.
13220 * tree-data-ref.c (dr_alignment): New function.
13221 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
13222 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
13224 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
13226 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13228 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
13229 and base_misalignment fields.
13230 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
13231 * tree-data-ref.c: Include builtins.h.
13232 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
13233 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
13234 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
13235 * tree-vect-data-refs.c: Include tree-cfg.h.
13236 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
13237 fields instead of calculating an alignment here.
13238 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
13239 innermost_loop_behavior fields.
13241 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13243 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
13245 (DR_STEP_ALIGNMENT): New macro.
13246 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
13247 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
13248 (create_data_ref): Print it.
13249 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
13250 to tell whether the step preserves vector (mis)alignment.
13251 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
13252 Move the check for an integer step and generalise to all INTEGER_CST.
13253 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
13254 Print the outer step alignment.
13256 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13258 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
13259 with offset_alignment.
13260 (DR_ALIGNED_TO): Delete.
13261 (DR_OFFSET_ALIGNMENT): New macro.
13262 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
13263 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
13264 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
13265 (create_data_ref): Likewise.
13266 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
13267 (vect_analyze_data_refs): Likewise.
13268 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
13269 creating dummy innermost behavior.
13271 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13273 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
13274 with a "innermost_loop_behavior *" and refeence tree.
13275 * tree-data-ref.c (dr_analyze_innermost): Likewise.
13276 (create_data_ref): Update call accordingly.
13277 * tree-predcom.c (find_looparound_phi): Likewise.
13279 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13281 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
13282 fields with dr_wrt_vec_loop.
13283 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
13284 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
13285 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
13286 (vect_dr_behavior): New function.
13287 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
13288 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
13289 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
13290 track whether the step preserves the misalignment.
13291 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
13292 Use vect_dr_behavior.
13293 (vect_setup_realignment): Update call accordingly.
13294 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
13295 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
13296 call to vect_create_addr_base_for_vector_ref.
13297 (vect_create_cond_for_align_checks): Likewise.
13298 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
13299 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
13300 (vect_recog_mask_conversion_pattern): Likewise.
13301 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
13302 (new_stmt_vec_info): Remove redundant zeroing.
13304 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
13306 * common/config/arm/arm-common.c (arm_be8_option): New function.
13307 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
13308 (ISA_ARMv6): Add isa_bit_be8.
13309 * config/arm/arm.h (arm_be8_option): Add prototype.
13310 (BE8_SPEC_FUNCTION): New define.
13311 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
13312 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
13313 (mlittle-endian): Similarly.
13314 (mbe8, mbe32): New options.
13315 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
13316 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
13318 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13320 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
13322 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13324 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
13325 (cleanup_tree_cfg_bb): Use it.
13326 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
13328 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
13330 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13333 * loop-doloop.c (add_test): Update profile.
13335 2017-07-03 Martin Liska <mliska@suse.cz>
13338 * sanopt.c (rewrite_usage_of_param): New function.
13339 (sanitize_rewrite_addressable_params): Likewise.
13340 (pass_sanopt::execute): Call rewrite_usage_of_param.
13342 2017-07-03 Richard Biener <rguenther@suse.de>
13344 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
13345 back to using VIEW_CONVERT_EXPR.
13347 2017-07-03 Martin Liska <mliska@suse.cz>
13350 * doc/extend.texi: Document when a resolver function is
13351 generated for target_clones.
13353 2017-07-03 Martin Liska <mliska@suse.cz>
13355 * asan.c (asan_emit_stack_protection): Unpoison just red zones
13356 and shadow memory of auto variables which are subject of
13357 use-after-scope sanitization.
13358 (asan_expand_mark_ifn): Add do set only when is_poison.
13360 2016-07-03 Richard Biener <rguenther@suse.de>
13362 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
13364 (vect_force_simple_reduction): Record reduction def -> phi mapping.
13365 (vectorizable_reduction): Perform reduction PHI creation when
13366 visiting a reduction PHI and adjust and simplify code generation
13367 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
13368 (vect_transform_loop): Visit reduction PHIs.
13369 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
13370 defs into the SLP tree.
13371 (vect_build_slp_tree): Reduction defs terminate the recursion.
13372 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
13374 (vect_get_vec_defs_for_stmt_copy): Export.
13375 (vect_get_vec_defs): Likewise.
13376 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
13378 (vect_get_vec_defs_for_stmt_copy): Declare.
13379 (vect_get_vec_defs): Likewise.
13381 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13383 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
13384 parameter with a "loop" parameter and use it instead of the
13385 loop containing DR_STMT. Don't check simple_iv when doing
13386 BB analysis. Describe the two analysis modes in the comment.
13388 2017-07-03 Tom de Vries <tom@codesourcery.com>
13390 PR tree-optimization/69468
13391 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
13392 (find_same_succ_bb): Handle ignore_edge_flags.
13394 2017-07-03 Tom de Vries <tom@codesourcery.com>
13396 PR tree-optimization/81192
13397 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
13399 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
13401 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
13403 2017-07-03 Tom de Vries <tom@codesourcery.com>
13405 PR tree-optimization/81192
13406 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
13407 BB_SAME_SUCC (bb) == NULL.
13409 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13411 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
13414 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13416 * dumpfile.c: Include profile-count.h
13417 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
13419 (insert_cond_bb): Update profile.
13420 * tree-cfg.h (insert_cond_bb): Update prototype.
13421 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
13422 * tree-dump.c: Do not include tree-cfg.
13424 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13426 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
13428 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13430 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
13433 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13435 * tree-complex.c (expand_complex_div_wide): update profile.
13437 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13438 Alan Hayward <alan.hayward@arm.com>
13439 David Sherwood <david.sherwood@arm.com>
13441 * Makefile.in (MACHMODE_H): Remove insn-modes.h
13442 (CORETYPES_H): New define.
13443 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
13444 (insn-modes-inline.h, s-modes-inline-h): New rules.
13445 (generated_files): Add insn-modes-inline.h.
13446 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
13447 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
13448 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
13449 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
13450 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
13451 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
13452 (build/gencodes.o, build/genconditions.o): Likewise.
13453 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
13454 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
13455 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
13456 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
13457 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
13458 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
13459 * coretypes.h: Include everything up to real.h for generators.
13460 Include insn-modes.h first. Include wide-int-print.h after
13461 wide-int.h. Include insn-modes-inline.h and then machmode.h.
13462 * machmode.h: Don't include insn-modes.h here.
13463 * function-tests.c: Remove includes of signop.h, machmode.h,
13464 double-int.h and wide-int.h.
13466 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
13468 * optc-save-gen.awk: Likewise.
13469 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
13470 * godump.c: Remove include of wide-int-print.h.
13471 * pretty-print.h: Likewise.
13472 * wide-int-print.cc: Likewise.
13473 * wide-int.cc: Likewise.
13474 * hash-map-tests.c: Remove include of signop.h.
13475 * hash-set-tests.c: Likewise.
13476 * rtl-tests.c: Likewise.
13477 * mkconfig.sh: Remove include of machmode.h.
13478 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
13480 (emit_insn_modes_inline_h): ...this new function. Emit the code
13481 into an insn-modes-inline.h header file, adding appropriate
13482 include guards and end comments.
13483 (emit_insn_modes_c_header): Remove include of machmode.h.
13484 (emit_min_insn_modes_c_header): Include coretypes.h rather than
13486 (main): Handle -i flag and call emit_insn_modes_inline_h when
13489 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13491 * tree-ssa-strlen.c (strinfo): Rename the length field to
13492 nonzero_chars. Add a full_string_p field.
13493 (compare_nonzero_chars, zero_length_string_p): New functions.
13494 (get_addr_stridx): Add an offset_out parameter.
13495 Use compare_nonzero_chars.
13496 (get_stridx): Update accordingly. Use compare_nonzero_chars.
13497 (new_strinfo): Update after above changes to strinfo.
13498 (set_endptr_and_length): Set full_string_p.
13499 (get_string_length): Update after above changes to strinfo.
13500 (unshare_strinfo): Update call to new_strinfo.
13501 (maybe_invalidate): Likewise.
13502 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
13503 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
13504 as a uhwi instead of an shwi. Update after above changes to
13505 strinfo and new_strinfo.
13506 (zero_length_string): Assert that chainsi contains full strings.
13507 Use zero_length_string_p. Update call to new_strinfo.
13508 (adjust_related_strinfos): Update after above changes to strinfo.
13509 Copy full_string_p from origsi.
13510 (adjust_last_stmt): Use zero_length_string_p.
13511 (handle_builtin_strlen): Update after above changes to strinfo and
13512 new_strinfo. Install the lhs as the string length if the previous
13513 entry didn't describe a full string.
13514 (handle_builtin_strchr): Update after above changes to strinfo
13516 (handle_builtin_strcpy): Likewise.
13517 (handle_builtin_strcat): Likewise.
13518 (handle_builtin_malloc): Likewise.
13519 (handle_pointer_plus): Likewise.
13520 (handle_builtin_memcpy): Likewise. Track nonzero characters
13521 that aren't necessarily followed by a nul terminator.
13522 (handle_char_store): Likewise.
13524 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13526 PR tree-optimization/80769
13527 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
13528 for malloc and calloc. Document the new invariant that all related
13529 strinfos have delayed lengths or none do.
13530 (verify_related_strinfos): Move earlier in file.
13531 (set_endptr_and_length): New function, split out from...
13532 (get_string_length): ...here. Also set the lengths of related
13534 (zero_length_string): Assert that chainsi has known (rather than
13536 (adjust_related_strinfos): Likewise.
13538 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13540 PR tree-optimization/81136
13541 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
13542 assert that two references with the same misalignment have the same
13543 compile-time misalignment if those compile-time misalignments
13546 2017-07-01 Andi Kleen <ak@linux.intel.com>
13548 * print-tree.c (print_node): Print all attributes.
13550 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13552 * cfg.c (scale_bbs_frequencies): New function.
13553 * cfg.h (scale_bbs_frequencies): Declare it.
13554 * cfgloopanal.c (single_likely_exit): Cleanup.
13555 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
13557 (scale_loop_profile): Likewise.
13558 (loop_version): Likewise.
13559 (create_empty_loop_on_edge): Update.
13560 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
13561 scale_loop_frequencies, scale_loop_profile, loopify,
13562 loop_version): Update prototypes.
13563 * modulo-sched.c (sms_schedule): Update.
13564 * predict.c (unlikely_executed_edge_p): Also check probability.
13565 (probably_never_executed_edge_p): Fix typo.
13566 * tree-if-conv.c (version_loop_for_if_conversion): Update.
13567 * tree-parloops.c (gen_parallel_loop): Update.
13568 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
13569 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13570 * tree-ssa-loop-split.c (split_loop): Update.
13571 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
13572 * tree-vect-loop-manip.c (vect_do_peeling): Update.
13573 (vect_loop_versioning): Update.
13574 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13576 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13578 * trans-mem.c (split_bb_make_tm_edge): Update profile.
13580 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13582 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
13583 to keep profile consistent.
13585 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13587 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
13588 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
13589 * profile-count.h (max_safe_multiplier): Make unsigned.
13590 (profile_count::guessed_zero): New.
13592 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13594 * bb-reorder.c (fix_up_crossing_landing_pad,
13595 fix_crossing_conditional_branches): Use make_single_succ_edge
13596 to keep profile consistent.
13598 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13600 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
13603 2017-07-01 Jakub Jelinek <jakub@redhat.com>
13606 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
13607 the right scopes, make sure cond_jump isn't preserved between multiple
13608 iterations. Search for fallthru edge whenever there are 3+ edges and
13609 use find_fallthru_edge for it.
13611 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13613 Patch by Alexander Monakov <amonakov@ispras.ru>
13614 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
13615 probabilities consistently.
13617 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13619 * pa.c (pa_expand_compare_and_swap_loop): Update call of
13620 emit_cmp_and_jump_insns.
13622 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13625 * tree-inline.c (expand_call_inline): Combine profile statuses.
13627 2017-06-30 Andrew Pinski <apinski@cavium.com>
13629 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
13630 fold_stmt returned true.
13632 2017-06-30 Nathan Sidwell <nathan@acm.org>
13634 * ggc.h (empty_string): Delete.
13635 * cfgexpand.c (expand_asm_stmt): Use plain "".
13636 * optabs.c (expand_asm_memory_barrier): Likewise.
13637 * stringpool.c (empty_string): Delete.
13638 (digit_vector, digit_string): Delete.
13639 (ggc_alloc_string): Use plain "", don't optimize single digit
13640 strings. Use ggc_alloc_atomic.
13642 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
13644 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
13645 comparison set and one other set, use the cost of the non-comparison
13648 2017-06-30 Nathan Sidwell <nathan@acm.org>
13650 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
13653 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
13655 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
13656 loops. Remove now unneeded calls to gimple_switch_set_label() that
13657 just set removed labels to NULL_TREE.
13659 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
13661 * tree-ssanames.c (set_range_info_raw): Abstract from ...
13662 (set_range_info): ...here. Only call set_range_info_raw if domain
13664 (set_nonzero_bits): Call set_range_info_raw.
13665 * tree-ssanames.h (set_range_info_raw): New.
13667 2017-06-30 Jakub Jelinek <jakub@redhat.com>
13670 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
13671 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
13672 of nonimmediate_operand and <store_mask_constraint> instead of m
13673 for the input operand. For V8FI iterator, always split if input
13674 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
13675 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
13676 <store_mask_predicate> instead of register_operand and
13677 <store_mask_constraint> instead of v for the input operand. Make
13678 sure both operands aren't MEMs for if not <mask_applied>.
13680 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
13682 * lto-wrapper.c (copy_file) Close both file descriptors before
13685 2017-06-30 Martin Liska <mliska@suse.cz>
13688 * multiple_target.c (create_dispatcher_calls): Make ifunc
13689 also for function that don't have calls or are not referenced.
13691 2017-06-30 Richard Biener <rguenther@suse.de>
13693 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
13694 analyze the first scalar stmt. Move vector type computation
13695 for the BB case here from ...
13696 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
13697 live operation processing in the SLP case properly.
13699 2017-06-30 Richard Biener <rguenther@suse.de>
13701 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
13703 2017-06-30 Martin Liska <mliska@suse.cz>
13706 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
13707 before BUILT_IN_UNWIND_RESUME when ASAN is used.
13709 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
13711 * doc/invoke.texi (AArch64): Add missing options and remove redundant
13714 2017-06-30 Richard Biener <rguenther@suse.de>
13716 PR tree-optimization/81249
13717 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
13718 condition reduction result to original scalar type.
13720 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13722 * profile-count.h (enum profile_quality): Fix typos and whitespace
13725 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13727 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
13728 type for branch probabilities.
13730 2017-06-29 Julian Brown <julian@codesourcery.com>
13731 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13733 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
13734 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
13735 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
13736 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
13738 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13740 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
13741 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
13742 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
13743 CC usage from generic code to here.
13744 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
13745 CC usage into the target macros.
13747 2017-06-29 Maya Rashish <coypu@sdf.org>
13749 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
13752 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13754 * arm/arm-builtins.c: Include profile-count.h
13755 * except.c (sjlj_emit_function_enter): Use
13756 profile_probability::unlikely.
13758 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13760 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
13761 and tocrel_offset be pointer args rather than implicitly using
13763 (legitimate_constant_pool_address_p, rs6000_emit_move,
13764 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
13765 tocrel_offset and use in toc_relative_expr_p call.
13766 (print_operand, print_operand_address): Use static tocrel_base_oac
13767 and tocrel_offset_oac.
13768 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
13771 2017-06-29 Maya Rashish <coypu@sdf.org>
13773 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
13775 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
13777 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
13778 objects, take into account only the alignment of 'op0' and 'mode1' if
13781 2017-06-29 Steve Ellcey <sellcey@cavium.com>
13783 * ccmp.c (ccmp_tree_comparison_p): New function.
13784 (ccmp_candidate_p): Update to use above function.
13785 (get_compare_parts): New function.
13786 (expand_ccmp_next): Update to use new functions.
13787 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
13789 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
13790 take mode as argument.
13791 * ccmp.h (expand_ccmp_expr): Add mode as argument.
13792 * expr.c (expand_expr_real_1): Pass mode as argument.
13794 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
13796 * combine.c (combine_instructions): Print insns to dump_file, together
13799 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13801 * asan.c (asan_emit_stack_protection): Update.
13802 (create_cond_insert_point): Update.
13803 * auto-profile.c (afdo_propagate_circuit): Update.
13804 * basic-block.h (struct edge_def): Turn probability to
13805 profile_probability.
13806 (EDGE_FREQUENCY): Update.
13807 * bb-reorder.c (find_traces_1_round): Update.
13808 (better_edge_p): Update.
13809 (sanitize_hot_paths): Update.
13810 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
13811 (make_single_succ_edge): Update.
13812 (check_bb_profile): Update.
13813 (dump_edge_info): Update.
13814 (update_bb_profile_for_threading): Update.
13815 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
13816 probabilitycount to 0.
13817 * cfgbuild.c (compute_outgoing_frequencies): Update.
13818 * cfgcleanup.c (try_forward_edges): Update.
13819 (outgoing_edges_match): Update.
13820 (try_crossjump_to_edge): Update.
13821 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
13822 (expand_gimple_tailcall): Update.
13823 (construct_init_block): Use make_single_succ_edge.
13824 (construct_exit_block): Use make_single_succ_edge.
13825 * cfghooks.c (verify_flow_info): Update.
13826 (redirect_edge_succ_nodup): Update.
13827 (split_edge): Update.
13828 (account_profile_record): Update.
13829 * cfgloopanal.c (single_likely_exit): Update.
13830 * cfgloopmanip.c (scale_loop_profile): Update.
13831 (set_zero_probability): Remove.
13832 (duplicate_loop_to_header_edge): Update.
13833 * cfgloopmanip.h (loop_version): Update prototype.
13834 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
13835 (force_nonfallthru_and_redirect): Update.
13836 (update_br_prob_note): Update.
13837 (rtl_verify_edges): Update.
13838 (purge_dead_edges): Update.
13839 (rtl_lv_add_condition_to_bb): Update.
13840 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
13841 * cgraphunit.c (init_lowered_empty_function): Update.
13842 (cgraph_node::expand_thunk): Update.
13843 * cilk-common.c: Include profile-count.h
13844 * dojump.c (inv): Remove.
13845 (jumpifnot): Update.
13846 (jumpifnot_1): Update.
13847 (do_jump_1): Update.
13849 (do_jump_by_parts_greater_rtx): Update.
13850 (do_compare_rtx_and_jump): Update.
13851 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
13852 do_jump_1. do_compare_rtx_and_jump): Update prototype.
13853 * dwarf2cfi.c: Include profile-count.h
13854 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
13855 (sjlj_emit_dispatch_table): Likewise.
13856 * explow.c: Include profile-count.h
13857 * expmed.c (emit_store_flag_force): Update.
13858 (do_cmp_and_jump): Update.
13859 * expr.c (compare_by_pieces_d::generate): Update.
13860 (compare_by_pieces_d::finish_mode): Update.
13861 (emit_block_move_via_loop): Update.
13862 (store_expr_with_bounds): Update.
13863 (store_constructor): Update.
13864 (expand_expr_real_2): Update.
13865 (expand_expr_real_1): Update.
13866 * expr.h (try_casesi, try_tablejump): Update prototypes.
13867 * gimple-pretty-print.c (dump_probability): Update.
13868 (dump_profile): New.
13869 (dump_gimple_label): Update.
13870 (dump_gimple_bb_header): Update.
13871 * graph.c (draw_cfg_node_succ_edges): Update.
13872 * hsa-gen.c (convert_switch_statements): Update.
13873 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
13874 (find_if_case_1): Update.
13875 (find_if_case_2): Update.
13876 * internal-fn.c (expand_arith_overflow_result_store): Update.
13877 (expand_addsub_overflow): Update.
13878 (expand_neg_overflow): Update.
13879 (expand_mul_overflow): Update.
13880 (expand_vector_ubsan_overflow): Update.
13881 * ipa-cp.c (good_cloning_opportunity_p): Update.
13882 * ipa-split.c (split_function): Use make_single_succ_edge.
13883 * ipa-utils.c (ipa_merge_profiles): Update.
13884 * loop-doloop.c (add_test): Update.
13885 (doloop_modify): Update.
13886 * loop-unroll.c (compare_and_jump_seq): Update.
13887 (unroll_loop_runtime_iterations): Update.
13888 * lra-constraints.c (lra_inheritance): Update.
13889 * lto-streamer-in.c (input_cfg): Update.
13890 * lto-streamer-out.c (output_cfg): Update.
13891 * mcf.c (adjust_cfg_counts): Update.
13892 * modulo-sched.c (sms_schedule): Update.
13893 * omp-expand.c (expand_omp_for_init_counts): Update.
13894 (extract_omp_for_update_vars): Update.
13895 (expand_omp_ordered_sink): Update.
13896 (expand_omp_for_ordered_loops): Update.
13897 (expand_omp_for_generic): Update.
13898 (expand_omp_for_static_nochunk): Update.
13899 (expand_omp_for_static_chunk): Update.
13900 (expand_cilk_for): Update.
13901 (expand_omp_simd): Update.
13902 (expand_omp_taskloop_for_outer): Update.
13903 (expand_omp_taskloop_for_inner): Update.
13904 * omp-simd-clone.c (simd_clone_adjust): Update.
13905 * optabs.c (expand_doubleword_shift): Update.
13906 (expand_abs): Update.
13907 (emit_cmp_and_jump_insn_1): Update.
13908 (expand_compare_and_swap_loop): Update.
13909 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
13910 * predict.c (predictable_edge_p): Update.
13911 (edge_probability_reliable_p): Update.
13912 (set_even_probabilities): Update.
13913 (combine_predictions_for_insn): Update.
13914 (combine_predictions_for_bb): Update.
13915 (propagate_freq): Update.
13916 (estimate_bb_frequencies): Update.
13917 (force_edge_cold): Update.
13918 * profile-count.c (profile_count::dump): Add missing space into dump.
13919 (profile_count::debug): Add newline.
13920 (profile_count::differs_from_p): Explicitly convert to unsigned.
13921 (profile_count::stream_in): Update.
13922 (profile_probability::dump): New member function.
13923 (profile_probability::debug): New member function.
13924 (profile_probability::differs_from_p): New member function.
13925 (profile_probability::differs_lot_from_p): New member function.
13926 (profile_probability::stream_in): New member function.
13927 (profile_probability::stream_out): New member function.
13928 * profile-count.h (profile_count_quality): Rename to ...
13929 (profile_quality): ... this one.
13930 (profile_probability): New.
13931 (profile_count): Update.
13932 * profile.c (compute_branch_probabilities): Update.
13933 * recog.c (peep2_attempt): Update.
13934 * sched-ebb.c (schedule_ebbs): Update.
13935 * sched-rgn.c (find_single_block_region): Update.
13936 (compute_dom_prob_ps): Update.
13937 (schedule_region): Update.
13938 * sel-sched-ir.c (compute_succs_info): Update.
13939 * stmt.c (struct case_node): Update.
13940 (do_jump_if_equal): Update.
13941 (get_outgoing_edge_probs): Update.
13942 (conditional_probability): Update.
13943 (emit_case_dispatch_table): Update.
13944 (expand_case): Update.
13945 (expand_sjlj_dispatch_table): Update.
13946 (emit_case_nodes): Update.
13947 * targhooks.c: Update.
13948 * tracer.c (better_p): Update.
13949 (find_best_successor): Update.
13950 * trans-mem.c (expand_transaction): Update.
13951 * tree-call-cdce.c: Update.
13952 * tree-cfg.c (gimple_split_edge): Upate.
13953 (move_sese_region_to_fn): Upate.
13954 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
13955 * tree-eh.c (lower_resx): Upate.
13956 (cleanup_empty_eh_move_lp): Upate.
13957 * tree-if-conv.c (version_loop_for_if_conversion): Update.
13958 * tree-inline.c (copy_edges_for_bb): Update.
13959 (copy_cfg_body): Update.
13960 * tree-parloops.c (gen_parallel_loop): Update.
13961 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
13962 (gimple_gen_time_profiler): Update.
13963 * tree-ssa-dce.c (remove_dead_stmt): Update.
13964 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
13965 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
13966 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
13967 (unloop_loops): Update.
13968 (try_peel_loop): Update.
13969 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13970 * tree-ssa-loop-split.c (connect_loops): Update.
13971 (split_loop): Update.
13972 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
13973 (hoist_guard): Update.
13974 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
13975 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
13976 (value_replacement): Update.
13977 * tree-ssa-reassoc.c (branch_fixup): Update.
13978 * tree-ssa-tail-merge.c (replace_block_by): Update.
13979 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
13980 (create_edge_and_update_destination_phis): Update.
13981 (compute_path_counts): Update.
13982 (recompute_probabilities): Update.
13983 (update_joiner_offpath_counts): Update.
13984 (freqs_to_counts_path): Update.
13985 (duplicate_thread_path): Update.
13986 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
13987 (struct switch_conv_info): Update.
13988 (gen_inbound_check): Update.
13989 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
13990 (vect_do_peeling): Update.
13991 (vect_loop_versioning): Update.
13992 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13993 (optimize_mask_stores): Update.
13994 * ubsan.c (ubsan_expand_null_ifn): Update.
13995 * value-prof.c (gimple_divmod_fixed_value): Update.
13996 (gimple_divmod_fixed_value_transform): Update.
13997 (gimple_mod_pow2): Update.
13998 (gimple_mod_pow2_value_transform): Update.
13999 (gimple_mod_subtract): Update.
14000 (gimple_mod_subtract_transform): Update.
14001 (gimple_ic): Update.
14002 (gimple_stringop_fixed_value): Update.
14003 (gimple_stringops_transform): Update.
14004 * value-prof.h: Update.
14006 2017-06-29 Carl Love <cel@us.ibm.com>
14008 * config/rs6000/rs6000-c.c: Add support for built-in functions
14009 vector signed int vec_signed (vector float);
14010 vector signed long long vec_signed (vector double);
14011 vector signed int vec_signed2 (vector double, vector double);
14012 vector signed int vec_signede (vector double);
14013 vector signed int vec_signedo (vector double);
14014 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
14015 instruction generator.
14016 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
14017 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
14018 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
14020 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
14021 vunsignede_v2df): Add define_expands.
14022 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
14023 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
14024 VEC_UNSIGNEDO): Add definitions.
14025 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
14026 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
14027 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
14028 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
14029 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
14030 * config/rs6000/altivec.h (vec_signed, vec_signed2,
14031 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
14032 vec_unsignede, vec_unsignedo): Add builtin defines.
14033 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
14035 * doc/extend.texi: Update the built-in documentation file for the
14036 new built-in functions.
14038 2017-06-29 Richard Biener <rguenther@suse.de>
14040 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
14041 reduction chains to LOOP_VINFO_REDUCTIONS.
14042 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
14043 SLP reductions after processing reduction chains.
14045 2017-06-29 Nathan Sidwell <nathan@acm.org>
14047 * builtins.c (fold_builtin_FUNCTION): Use
14048 lang_hooks.decl_printable_name.
14050 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
14052 PR middle-end/81194
14053 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
14054 with only one label.
14055 * stmt.c (expand_case): Assert NCASES is greater than one.
14057 2017-06-29 Richard Biener <rguenther@suse.de>
14059 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
14061 (group_case_labels): Likewise.
14062 (find_taken_edge): Push sanity checking on val to workers...
14063 (find_taken_edge_cond_expr): ... here
14064 (find_taken_edge_switch_expr): ... and here, handle cases
14065 with just a default label.
14066 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
14067 (group_case_labels): Likewise.
14068 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
14069 group_case_labels does anything cleanup the CFG again.
14071 2017-06-29 Bin Cheng <bin.cheng@arm.com>
14073 PR tree-optimization/81196
14074 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
14075 exit condition comparing two IVs.
14077 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
14079 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
14080 profile to the dummy entry at the end of the list of architectures.
14081 * config/arm/arm-cpu-cdata.h: Regenerated.
14083 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14084 Michael Collison <michael.collison@arm.com>
14087 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
14089 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
14090 (*aarch64_reg_<mode>3_minus_mask): New pattern.
14091 (*aarch64_<optab>_reg_di3_mask2): New pattern.
14092 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
14093 of shift when the shift amount is masked with constant equal to
14094 the size of the mode.
14095 * config/aarch64/predicates.md (subreg_lowpart_operator): New
14098 2017-06-29 Martin Liska <mliska@suse.cz>
14100 * config/i386/i386.opt: Change range from [1,5] to [0,5].
14102 2017-06-29 Yury Gribov <tetra2005@gmail.com>
14105 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
14108 (edge_growth_cache_entry::edge_growth_cache_entry): New
14110 (reset_edge_growth_cache): Update to use constructor.
14112 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14114 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
14115 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
14116 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
14118 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
14120 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
14121 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
14123 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
14125 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
14126 (*-linux-uclibc*): Add t-uclibc tmake_file.
14127 * config/t-musl: New.
14128 * config/t-uclibc: New.
14130 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
14132 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
14134 (gen_comm_data): Emit architectural setting of arch_prof.
14135 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
14137 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
14138 (armv8-m.base, armv8-m.main): Likewise.
14139 * arm-protos.h (arm_build_target): Add profile field.
14140 (arch_option): Likewise.
14141 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
14143 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
14144 arm_active_target.profile.
14146 2017-06-28 Richard Biener <rguenther@suse.de>
14148 PR middle-end/81227
14149 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
14150 TYPE_OVERFLOW_WRAPS.
14151 * match.pd (negate_expr_p): Likewise.
14152 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
14153 fold_build2, not fold_binary.
14155 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
14157 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
14158 Convert memory address to Pmode.
14159 (aarch64_print_operand): Assert MEM operands are always Pmode.
14161 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
14164 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
14165 Remove redundant if.
14166 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
14167 * config/arm/aarch-common-protos.h
14168 (aarch_forward_to_shift_is_not_shifted_re): Remove.
14169 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
14171 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
14174 * multiple_target.c (create_dispatcher_calls): Set the default
14175 clone to be static, not public.
14177 2017-06-28 Richard Biener <rguenther@suse.de>
14179 * tree-vect-loop.c (vectorizable_reduction): Move special
14180 cond reduction IV var creation ...
14181 (vect_create_epilog_for_reduction): ... here. Remove induction_index
14182 parameter. Use STMT_VINFO_VECTYPE.
14183 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
14186 2017-06-28 Martin Liska <mliska@suse.cz>
14189 * ipa-visibility.c (non_local_p): Handle visibility.
14191 2017-06-28 Martin Liska <mliska@suse.cz>
14194 * common.opt: Add IntegerRange to various options.
14195 * opt-functions.awk (integer_range_info): New function.
14196 * optc-gen.awk: Add integer_range_info to cl_options struct.
14197 * opts-common.c (decode_cmdline_option): Handle
14198 CL_ERR_INT_RANGE_ARG.
14199 (cmdline_handle_error): Likewise.
14200 * opts.c (print_filtered_help): Show valid interval in
14201 when --help is provided.
14202 * opts.h (struct cl_option): Add range_min and range_max fields.
14203 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
14205 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
14207 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
14208 (x * C EQ/NE y * C): New transformation.
14210 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
14212 * genmultilib (combination_space): Accept '+' in option names.
14214 2017-06-28 Martin Liska <mliska@suse.cz>
14217 * asan.c (instrument_derefs): Bail out inner references
14218 that are hard register variables.
14220 2017-06-28 Jakub Jelinek <jakub@redhat.com>
14223 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
14224 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
14226 2017-06-28 Richard Biener <rguenther@suse.de>
14228 * tree-vectorizer.h (vect_get_vec_defs): Remove.
14229 (vect_get_slp_defs): Adjust.
14230 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
14232 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
14234 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
14235 get_initial_defs_for_reduction instead of vect_get_vec_defs.
14236 (vectorizable_reduction): Adjust.
14237 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
14239 (vect_get_slp_defs): Likewise.
14240 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
14241 (vectorizable_bswap): Adjust.
14242 (vectorizable_call): Likewise.
14243 (vectorizable_conversion): Likewise.
14244 (vectorizable_assignment): Likewise.
14245 (vectorizable_shift): Likewise.
14246 (vectorizable_operation): Likewise.
14247 (vectorizable_store): Likewise.
14248 (vectorizable_condition): Likewise.
14249 (vectorizable_comparison): Likewise.
14251 2017-06-28 Michael Collison <michael.collison@arm.com>
14254 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
14256 (arm_gen_movmemqi): Removed unused variable 'i'.
14257 Convert 'for' loop into 'while' loop.
14258 (arm_expand_prologue): Remove last unnecessary set of insn.
14259 (thumb_pop): Remove unused variable 'pushed_words'.
14260 (thumb_exit): Remove last unnecessary set of regs_to_pop.
14262 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14264 * config/s390/predicates.md: Use s390_rel_address_ok_p.
14265 * config/s390/s390-protos.h: Add prototype of
14266 s390_rel_address_ok_p.
14267 * config/s390/s390.c (s390_got_symbol): New function.
14268 (s390_rel_address_ok_p): New function.
14269 (legitimize_pic_address): Use s390_rel_address_ok_p.
14270 (s390_load_got): Use s390_got_symbol.
14271 (s390_option_override): Issue error if
14272 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
14273 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
14275 * config/s390/s390.opt: New option mpic-data-is-text-relative.
14277 2017-06-27 Andrew Pinski <apinski@cavium.com>
14279 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
14280 (X * copysign (1.0, X)): New pattern.
14281 (X * copysign (1.0, -X)): New pattern.
14282 (copysign (-1.0, CST)): New pattern.
14284 2017-06-27 Joseph Myers <joseph@codesourcery.com>
14286 * genmultilib (combination_space): Remove variable.
14287 Validate reuse rules against regular expression for any sequence
14288 of multilib options in any order.
14290 2017-06-27 Michael Collison <michael.collison@arm.com>
14292 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
14293 call aarch64_split_simd_combine.
14294 * (aarch64_combine_internal<mode>): Delete pattern.
14295 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
14296 Allow register and subreg operands.
14298 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14300 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
14301 specific need, just fallback on defaults.
14302 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
14304 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14305 Olivier Hainque <hainque@adacore.com>
14307 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
14309 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
14310 targets. Pick a default if no particular attempt applied.
14311 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
14314 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14316 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
14317 (x86_64-wrs-vxworks7): Likewise.
14319 2017-06-27 Marek Polacek <polacek@redhat.com>
14322 * ubsan.c (instrument_null): Check get_base_address's result for null.
14324 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
14326 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
14328 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
14330 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
14331 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
14332 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
14333 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
14334 New function types.
14335 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
14336 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
14337 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
14338 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
14339 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
14340 BUILT_IN_FEUPDATEENV): New builtins.
14341 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
14342 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
14343 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
14344 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
14346 (builtin_structptr_types): Adjust size.
14347 * tree.c (builtin_structptr_types): Add four entries.
14349 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14350 Olivier Hainque <hainque@adacore.com>
14352 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
14353 (TLS_SYM): New local macro, forcing reference to __tls__ on
14354 link command lines for VxWorks 7 RTPs, triggering initialization
14356 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
14357 OS features TLS support, true for RTPs on VxWorks 7.
14358 * config/vxworks.c (vxworks_override_options): Setup emutls
14361 2017-06-27 Jakub Jelinek <jakub@redhat.com>
14363 * predict.c (test_prediction_value_range): Use -1U instead of -1
14364 to avoid narrowing conversion warning.
14365 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
14366 to avoid narrowing conversion warning.
14367 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
14369 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
14371 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14373 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
14374 64bit configurations.
14375 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
14376 (SIZE_TYPE): Likewise.
14377 * config/vxworks.c (vxworks_emutls_var_fields): Use
14378 long_unsigned_type_node instead of unsigned_type_node as the offset
14379 field type, which is "pointer" mode in emutls.c.
14381 2017-06-27 Jakub Jelinek <jakub@redhat.com>
14384 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
14386 PR middle-end/81207
14387 * gimple-fold.c (replace_call_with_call_and_fold): Handle
14388 gimple_vuse copying separately from gimple_vdef copying.
14390 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14392 * value-prof.c (free_hist): Remove call to memset and the enclosing if
14395 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
14396 Olivier Hainque <hainque@adacore.com>
14398 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
14399 for all vxworks7 targets.
14400 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
14401 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
14402 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
14403 variations for VX6/VX7 and 32/64bits later on in ...
14404 (VXWORKS_LIB_SPEC): Leverage new macros.
14405 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
14406 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
14408 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
14410 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
14411 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
14413 2017-06-26 Carl Love <cel@us.ibm.com>
14415 * config/rs6000/rs6000-c.c: Add support for built-in functions
14416 vector bool char vec_reve (vector bool char);
14417 vector signed char vec_reve (vector signed char);
14418 vector unsigned char vec_reve (vector unsigned char);
14419 vector bool int vec_reve (vector bool int);
14420 vector signed int vec_reve (vector signed int);
14421 vector unsigned int vec_reve (vector unsigned int);
14422 vector bool long long vec_reve (vector bool long long);
14423 vector signed long long vec_reve (vector signed long long);
14424 vector unsigned long long vec_reve (vector unsigned long long);
14425 vector bool short vec_reve (vector bool short);
14426 vector signed short vec_reve (vector signed short);
14427 vector double vec_reve (vector double);
14428 vector float vec_reve (vector float);
14429 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
14430 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
14431 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
14432 (altivec_vreve): New pattern.
14433 * config/rs6000/altivec.h (vec_reve): New define.
14434 * doc/extend.texi (vec_rev): Update the built-in documentation file
14435 for the new built-in functions.
14437 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14439 PR tree-optimization/71815
14440 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
14442 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
14444 (slsr_process_phi): Likewise.
14445 (replace_uncond_cands_and_profitable_phis): Don't replace a
14446 multiply candidate with a stride of 1 (copy or cast).
14447 (phi_incr_cost): Call uses_consumed_by_stmt rather than
14449 (lowest_cost_path): Likewise.
14450 (total_savings): Likewise.
14452 2017-06-26 Richard Biener <rguenther@suse.de>
14455 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
14456 Use def_builtin_pure for all gather builtins.
14458 2017-06-26 Richard Biener <rguenther@suse.de>
14460 PR tree-optimization/81203
14461 * tree-tailcall.c (find_tail_calls): Do not move stmts into
14462 non-dominating BBs.
14464 2017-06-26 Marek Polacek <polacek@redhat.com>
14467 * doc/invoke.texi: Document -Wmultistatement-macros.
14469 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
14471 * doc/sourcebuild.texi (ARM-specific attributes): Document new
14472 arm_neon_ok_no_float_abi effective target.
14474 2017-06-26 Richard Biener <rguenther@suse.de>
14476 PR tree-optimization/80928
14477 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
14478 (copy_bbs): Set BB_DUPLICATED flag early.
14479 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
14481 (execute_on_shrinking_pred): Likewise.
14482 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
14483 BB_DUPLICATED blocks.
14484 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
14485 iterate over all PHIs considering removal of *gsi.
14487 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
14489 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
14492 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14494 * config/rs6000/rs6000-string.c: (expand_block_clear,
14495 do_load_for_compare, select_block_compare_mode,
14496 compute_current_alignment, expand_block_compare,
14497 expand_strncmp_align_check, expand_strn_compare,
14498 expand_block_move, rs6000_output_load_multiple)
14499 Move functions related to string/block move/compare
14500 to a separate file.
14501 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
14502 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
14503 for this function which is now used in two files.
14504 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
14505 * config.gcc: Add rs6000-string.o to extra_objs for
14506 targets powerpc*-*-* and rs6000*-*-*.
14508 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
14511 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
14512 32-bit, since indexed is not valid for DImode.
14513 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
14514 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
14515 (define_peephole2 for Altivec d-form load): Add 32-bit support.
14516 (define_peephole2 for Altivec d-form store): Likewise.
14519 * multiple_target.c (create_dispatcher_calls): Only create the
14520 dispatcher call if the function is the default clone of a
14521 versioned function.
14523 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
14525 PR middle-end/80902
14526 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
14527 a call, force the call to not be a tail call.
14529 2017-06-23 Jeff Law <law@redhat.com>
14531 * doc/contrib.texi: Add entry for Steven Pemberton's work on
14534 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
14536 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
14537 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
14538 handling for early expansion of vector shifts (sl,sr,sra,rl).
14539 (builtin_function_type): Add vector shift right instructions
14540 to the unsigned argument list.
14542 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
14544 rtl-optimizatoin/79286
14545 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
14546 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
14547 trap. PIC register plus a const unspec without offset can never trap.
14549 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
14551 * tree.h (builtin_structptr_type): New type.
14552 (builtin_structptr_types): Declare new array.
14553 * tree.c (builtin_structptr_types): New array.
14554 (free_lang_data, build_common_tree_nodes): Use it.
14556 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
14559 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
14562 2017-06-22 Matt Turner <mattst88@gmail.com>
14564 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
14565 Lake models to skylake case. Assume skylake for unknown
14566 models with clflushopt.
14568 2017-06-22 Jeff Law <law@redhat.com>
14570 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
14571 frame sizes that do not satisfy aarch64_uimm12_shift.
14573 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
14575 * profile-count.h (apply_probability,
14576 apply_scale, probability_in): Fix checks for zero.
14578 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14580 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
14581 * doc/cppdiropts.texi (-I @var{dir}): Document it.
14583 2016-06-22 Richard Biener <rguenther@suse.de>
14585 * tree-vect-loop.c (vect_model_reduction_cost): Handle
14586 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
14587 REDUC_MAX_EXPR support.
14588 (vectorizable_reduction): Likewise.
14589 (vect_create_epilog_for_reduction): Likewise.
14591 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
14593 * match.pd (A / (1 << B) -> A >> B): New.
14594 * generic-match-head.c: Include optabs-tree.h.
14595 * gimple-match-head.c: Likewise.
14596 * optabs-tree.h (target_supports_op_p): New.
14597 * optabs-tree.c (target_supports_op_p): New.
14599 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14601 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
14602 $gcc_cv_ld --help output.
14603 (gcc_cv_ld_demangle): Likewise.
14604 (gcc_cv_ld_eh_frame_hdr): Likewise.
14605 (gcc_cv_ld_pie): Likewise.
14606 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
14607 (gcc_cv_ld_buildid): Likewise.
14608 (gcc_cv_ld_sysroot): Likewise.
14609 (ld_bndplt_support): Likewise.
14610 (ld_pushpopstate_support): Likewise.
14611 * configure: Regenerate.
14612 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
14614 2017-06-21 Jakub Jelinek <jakub@redhat.com>
14617 * config/i386/sse.md (round<mode>2): Renumber match_dup and
14618 operands indexes to avoid gap between operands and match_dups.
14620 2017-06-21 Andrew Pinski <apinski@cavium.com>
14622 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
14623 Increment Arith_shift and Arith_shift_reg by 1.
14624 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
14626 * config/aarch64/aarch64.c (thunderx_tunings): Enable
14627 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
14628 (aarch64_strip_extend): Add new argument and test for it.
14629 (aarch64_cheap_mult_shift_p): New function.
14630 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
14631 add a cost if it is true.
14632 Update calls to aarch64_strip_extend.
14633 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
14635 2017-06-21 Andrew Pinski <apinski@cavium.com>
14637 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
14639 (thunderxt88): Likewise.
14640 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
14641 (thunderx_prefetch_tune): New variable.
14642 (thunderx2t99_prefetch_tune): Update for the correct values.
14643 (thunderxt88_tunings): New variable.
14644 (thunderx_tunings): Use thunderx_prefetch_tune instead of
14645 generic_prefetch_tune.
14646 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
14648 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14650 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
14651 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
14652 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
14653 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
14654 (aarch64_atomic_cas<mode>, GPI): Likewise.
14656 2017-06-21 Martin Liska <mliska@suse.cz>
14658 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
14659 statements on cold and hot labels.
14660 * predict.c (tree_estimate_probability_bb): Remove the
14661 prediction from this place.
14663 2017-06-21 Martin Liska <mliska@suse.cz>
14665 PR tree-optimization/79489
14666 * gimplify.c (maybe_add_early_return_predict_stmt): New
14668 (gimplify_return_expr): Call the function.
14669 * predict.c (tree_estimate_probability_bb): Remove handling
14671 * predict.def: Update comment about early return predictor.
14672 * gimple-predict.h (is_gimple_predict): New function.
14673 * predict.def: Change default value of early return to 66.
14674 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
14676 * passes.def: Put pass_strip_predict_hints to the beginning of
14679 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
14681 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
14682 FUNCTION_DECL declarations.
14683 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
14685 (dwaf2out_decl): Likewise.
14686 * godump.c (go_early_global_decl): Skip call to the real debug hook
14687 for FUNCTION_DECL declarations.
14688 * passes.c (rest_of_decl_compilation): Skip call to the
14689 early_global_decl debug hook for FUNCTION_DECL declarations, unless
14690 -fdump-go-spec is passed.
14692 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
14694 * config/i386/i386.c (struct builtin_isa): New field pure_p.
14695 Reorder for compactness.
14696 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
14697 (def_builtin_pure, def_builtin_pure2): New functions.
14698 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
14700 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
14702 * match.pd (nop_convert): New predicate.
14703 ((A +- CST1) +- CST2): Allow some NOP conversions.
14705 2017-06-21 Jakub Jelinek <jakub@redhat.com>
14708 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
14709 with ctors/dtors if GOVD_SHARED is set.
14711 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14713 * config/aarch64/aarch64.md (movti_aarch64):
14714 Emit mov rather than orr.
14715 (movtf_aarch64): Likewise.
14716 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
14717 Emit mov rather than orr.
14719 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14721 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
14722 Swap alternatives, make integer dup more expensive.
14724 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14726 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
14727 Return true for non-tls symbols.
14729 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
14731 * config/aarch64/aarch64-cores.def (cortex-a55): New.
14732 (cortex-a75): Likewise.
14733 (cortex-a75.cortex-a55): Likewise.
14734 * config/aarch64/aarch64-tune.md: Regenerate.
14735 * doc/invoke.texi (-mtune): Document new values for -mtune.
14737 2017-06-21 Tom de Vries <tom@codesourcery.com>
14739 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
14740 stack_size feature.
14741 (Effective-Target Keywords, Other attributes): Suggest using
14742 dg-add-options stack_size feature to get stack limit in stack_size
14743 effective target documentation.
14745 2017-06-21 Julian Brown <julian@codesourcery.com>
14746 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14748 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
14749 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
14750 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
14752 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
14753 attribute type list for neon_multiply.
14754 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
14755 attribute type list for neon_multiply.
14756 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
14757 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
14758 attribute type list for neon_multiply.
14759 * config/arm/types.md (crypto_pmull): Add.
14760 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
14761 attribute type list.
14763 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
14765 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
14768 2017-06-20 Jakub Jelinek <jakub@redhat.com>
14770 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
14771 to make sure not to dereference a NULL cost_classes_ptr pointer.
14773 2017-06-20 Carl Love <cel@us.ibm.com>
14775 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14776 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14777 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14778 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14779 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14780 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
14781 VMULOSW): New enum "unspec" values.
14782 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
14783 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
14784 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
14785 altivec_vmulosw): New patterns.
14786 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14787 VMULOSW): Add definitions.
14789 2017-06-20 Julia Koval <julia.koval@intel.com>
14791 * config/i386/i386.c: Fix rounding expand for new pattern.
14792 * config/i386/subst.md: Fix pattern (parallel -> unspec).
14794 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14796 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
14797 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
14799 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14801 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
14804 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14806 * config/aarch64/aarch64-cores.def: Rearrange to sort by
14807 architecture, then by implementer ID.
14808 * config/aarch64/aarch64-tune.md: Regenerate.
14810 2017-06-20 Richard Biener <rguenther@suse.de>
14812 PR middle-end/81097
14813 * fold-const.c (split_tree): Fold to type before negating.
14815 2017-06-20 David Malcolm <dmalcolm@redhat.com>
14817 * diagnostic-show-locus.c
14818 (selftest::test_fixit_deletion_affecting_newline): New function.
14819 (selftest::diagnostic_show_locus_c_tests): Call it.
14821 2017-06-20 Andreas Schwab <schwab@suse.de>
14824 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
14827 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14829 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
14830 __ARM_FEATURE_COPROC according to support.
14832 2017-06-20 Jakub Jelinek <jakub@redhat.com>
14834 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
14835 Rewritten to avoid overflow for > 32-bit pointers.
14838 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
14839 by removing enum keyword.
14840 (ubsan_type_descriptor): Likewise. Formatting fix.
14843 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
14844 splitter): Require TARGET_SSE2 in the condition.
14846 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
14849 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
14850 for doing vector set of SFmode on ISA 3.0.
14851 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
14852 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
14854 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
14855 SFmode value into a V4SF variable that was extracted from another
14856 V4SF variable without converting the element to double precision
14857 and back to single precision vector format.
14858 (vsx_insert_extract_v4sf_p9_2): Likewise.
14860 2017-06-19 Jakub Jelinek <jakub@redhat.com>
14862 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
14863 in UWHI to avoid undefined overflow.
14866 * ubsan.h (enum ubsan_encode_value_phase): New.
14867 (ubsan_encode_value): Change second argument to
14868 enum ubsan_encode_value_phase with default value of
14869 UBSAN_ENCODE_VALUE_GENERIC.
14870 * ubsan.c (ubsan_encode_value): Change second argument to
14871 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
14872 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
14873 create_tmp_var_raw instead of create_tmp_var and use a
14875 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
14876 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
14877 ubsan_encode_value callers.
14880 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
14881 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
14882 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
14884 2017-06-19 Richard Biener <rguenther@suse.de>
14886 PR middle-end/81118
14887 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
14888 estimates if we changed anything.
14890 2017-06-19 Richard Biener <rguenther@suse.de>
14892 PR tree-optimization/80887
14893 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
14894 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
14895 simplified lookups, then reset mprts_hook.
14896 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
14898 (try_to_simplify): Likewise.
14900 2017-06-19 Martin Liska <mliska@suse.cz>
14903 * gimplify.c (gimplify_switch_expr):
14904 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
14906 2017-06-19 Martin Liska <mliska@suse.cz>
14908 * doc/install.texi: Document that PGO runs in 4 stages.
14910 2017-06-19 Martin Liska <mliska@suse.cz>
14913 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
14914 to dispatcher function name.
14915 * multiple_target.c (replace_function_decl): New function.
14916 (create_dispatcher_calls): Redirect both edges and references.
14918 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
14920 * profile-count.c (profile_count::dump): Dump quality.
14921 (profile_count::differs_from_p): Update for unsigned val.
14922 * profile-count.h (profile_count_quality): New enum.
14923 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
14925 2017-06-19 Richard Biener <rguenther@suse.de>
14927 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
14928 struct function as arg.
14929 (estimate_numbers_of_iterations): Export overload with loop arg.
14930 (free_numbers_of_iterations_estimates_loop): Use an overload of
14931 free_numbers_of_iterations_estimates instead.
14932 * tree-cfg.c (remove_bb): Adjust.
14933 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
14934 * tree-parloops.c (gen_parallel_loop): Likewise.
14935 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
14937 (tree_unroll_loops_completely): Likewise.
14938 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
14939 Use an overload instead and export.
14940 (estimated_loop_iterations): Adjust.
14941 (max_loop_iterations): Likewise.
14942 (likely_max_loop_iterations): Likewise.
14943 (estimate_numbers_of_iterations): Take struct function as arg
14945 (loop_exits_before_overflow): Adjust.
14946 (free_numbers_of_iterations_estimates_loop): Use an overload.
14947 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
14948 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
14950 2017-06-19 Richard Biener <rguenther@suse.de>
14953 * ipa-prop.c (find_constructor_constant_at_offset): Handle
14954 RANGE_EXPR conservatively.
14956 2017-06-16 Carl Love <cel@us.ibm.com>
14958 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14959 definitions for vec_float, vec_float2, vec_floato,
14960 vec_floate built-ins.
14961 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
14962 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
14964 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
14965 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
14966 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
14967 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
14968 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
14969 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
14970 vec_floato): Add builtin defines.
14971 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
14972 Update the built-in documentation file for the new built-in
14975 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14977 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
14978 (mthumb): Mark as the negative of -marm.
14980 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14982 * doc/invoke.texi (ARM Options, -mcpu): Document supported
14984 (ARM Options, -mtune): Document that this accepts the same
14985 extension options as -mcpu.
14986 (ARM Options, -mfpu): Document addition of -mfpu=auto.
14988 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14990 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
14991 permitted extensions.
14993 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14995 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
14996 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
14997 (armv8-m.main): Add option +nodsp.
14998 * config/arm/arm-cpu-cdata.h: Regenerated.
15000 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15002 * config/arm/t-fuchsia: New file.
15003 * config.gcc (arm*-*-fuchsia*): Use it.
15005 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15007 * config/arm/t-symbian: Rewrite for new option infrastructure.
15009 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15011 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
15012 (MULTILIB_REQUIRED): Likewise.
15014 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15016 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
15017 (MULTILIB_RESUE): Likewise.
15018 (MULTILIB_MATCHES): Likewise.
15019 (MULTLIB_REQUIRED): Likewise.
15021 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15023 * config/arm/t-rtems: Rewrite for new option framework.
15025 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15027 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
15028 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
15029 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
15030 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
15031 * config/arm/t-multilib: ... here.
15032 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
15033 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
15034 armv7-a and armv8*-a when A-profile libraries have not been built.
15035 * config/arm/t-rmprofile: Rewrite.
15037 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15039 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
15040 with a backslash. Remove the backslash after substituting unescaped
15042 * doc/fragments.texi (MULTILIB_REUSE): Document it.
15044 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15046 * config.gcc: (arm*-*-*): When building a-profile libraries, force
15047 the driver to pass through the default setting of -mfloat-abi.
15048 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
15050 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
15051 (all_feat_combs): New rule.
15052 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
15054 * config/arm/t-aprofile: Rewrite.
15056 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15058 * config/arm/arm.h (FPUTYPE_AUTO): Define.
15059 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
15060 fpu is not specified by the user/command-line.
15061 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
15062 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
15063 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
15064 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
15065 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
15066 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
15068 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15070 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
15071 * config/arm/t-arm-elf: Rewritten.
15073 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15075 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
15076 have some floating-point instructions.
15077 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
15078 (TARGET_MAYBE_HARD_FLOAT): New macro.
15079 * config/arm/arm-builtins.c (arm_init_builtins): Use
15080 TARGET_MAYBE_HARD_FLOAT.
15081 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
15083 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15085 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
15086 (configargs.h): Include it.
15087 (arm_print_hint_for_fpu_option): New function.
15088 (arm_parse_fpu_option): New function.
15089 (candidate_extension): New class.
15090 (arm_canon_for_multilib): New function.
15091 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
15092 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
15093 (ARCH_CANONICAL_SPECS): New macro.
15094 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
15096 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15098 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
15099 are set after handling multilib fragments. Set target_cpu_default2
15102 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15104 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
15106 (arm*-*-*): Set target_cpu_default2 to a quoted string.
15107 * config/arm/parsecpu.awk (check_cpu): Validate any extension
15109 (check_arch): Likewise.
15110 * config/arm/arm.c (arm_configure_build_target): Handle
15111 TARGET_CPU_DEFAULT being a string constant. Scan any feature
15112 options in the default.
15114 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15116 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
15117 when an option is an alias of another.
15118 * config/arm/parsecpu.awk (optalias): New parser token.
15119 (gen_comm_data): Mark non-alias options as such. Emit entries
15120 for extension aliases.
15121 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
15122 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
15123 (armv6kz, armv6zk, armv6t2): Likewise.
15124 (armv7): Make vfpv3-d16 an alias.
15125 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
15127 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
15128 Sort in canonical order.
15129 (armv8-a): Sort in canonical order.
15130 (armv8.1-a, armv8.2-a): Likewise.
15131 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
15133 (cortex-a9): Sort in canonical order.
15134 * config/arm/arm.c (selftests.h): Include it.
15135 (arm_test_cpu_arch_data): New function.
15136 (arm_run_self_tests): New function.
15137 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
15138 (targetm): Move declaration to the end of the file.
15139 * arm-cpu-cdata.h: Regenerated.
15141 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15143 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
15144 call to target_mode_check describing the type of option passed.
15145 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
15146 (arm_target_thumb_only): Use arm_parse_arch_option_name or
15147 arm_parse_cpu_option_name to match parameters against list of
15149 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
15150 arm_arch_core_flags data structure.
15151 * config/arm/arm-cpu_cdata.h: Regenerated.
15153 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15155 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
15157 (arm_print_hint_for_cpu_option): Likewise.
15158 (arm_print_hint_for_arch_option): Likewise.
15159 (arm_parse_cpu_option_name): Likewise.
15160 (arm_parse_arch_option_name): Likewise.
15161 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
15162 of entries in the all_fpus list.
15163 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
15164 (arm_parse_cpu_option_name): Declare.
15165 (arm_parse_arch_option_name): Declare.
15166 (arm_parse_option_features): Declare.
15167 (arm_intialize_isa): Declare.
15168 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
15170 (gen_comm_data): ... here. Make definitions non-static.
15171 * config/arm/arm-cpu-data.h: Regenerated.
15172 * config/arm/arm-cpu-cdata.h: Regenerated.
15174 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15176 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
15177 (cpu_arch_extension): New structure.
15178 (cpu_arch_option, arch_option, cpu_option): New structures.
15179 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
15180 architecture types.
15181 (gen_data): Generate new format data tables.
15182 * config/arm/arm.c (cpu_tune): New structure.
15183 (cpu_option, processors): Delete.
15184 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
15185 (arm_print_hint_for_cpu_option): ... this and ...
15186 (arm_print_hint_for_arch_option): ... this.
15187 (arm_parse_arch_cpu_name): Delete. Replace with ...
15188 (arm_parse_cpu_option_name): ... this and ...
15189 (arm_parse_arch_option_name): ... this.
15190 (arm_unrecognized_feature): Change type of target parameter to
15192 (arm_parse_arch_cpu_features): Delete. Replace with ...
15193 (arm_parse_option_features): ... this.
15194 (arm_configure_build_target): Rework to use new configuration data
15196 (arm_print_tune_info): Rework for new configuration data tables.
15197 * config/arm/arm-cpu-data.h: Regenerated.
15198 * config/arm/arm-cpu.h: Regenerated.
15200 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15202 * Makefile.in (OBJS): Move sbitmap.o from here ...
15203 (OBJS-libcommon): ... to here.
15205 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15207 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
15208 (ISA_ALL_CRYPTO): New macro.
15209 (ISA_ALL_SIMD): New macro
15210 (ISA_ALL_FP): New macro.
15211 * config/arm/arm.c (fpu_bitlist): Update initializer.
15212 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
15214 (arm9e): Add fpu. Add option for nofp
15215 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
15216 (arm926ej-s, arm1026ej-s): Likewise.
15217 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
15218 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
15219 neon-fp16, neon-vfpv4, nofp and nosimd.
15220 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
15221 (cortex-a8): Add fpu. Add option for nofp.
15222 (cortex-a9): Add fpu. Add options for nosimd and nofp.
15223 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
15224 (cortex-r4f): Add fpu.
15225 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
15226 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
15228 (cortex-r8): Likewise.
15229 (cortex-m4): Add fpu. Add option for nofp.
15230 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
15231 (cortex-a17.cortex-a7): Likewise.
15232 (cortex-a32): Add fpu. Add options for crypto and nofp.
15233 (cortex-a35, cortex-a53): Likewise.
15234 (cortex-a57): Add fpu. Add option for crypto.
15235 (cortex-a72, cortex-a73): Likewise.
15236 (exynos-m1): Likewise.
15237 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
15238 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
15239 (cortex-m33): Add fpu. Add option for nofp.
15240 * config/arm/arm-cpu-cdata.h: Regenerated
15241 * config/arm/arm-cpu-data.h: Regenerated.
15243 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15245 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
15246 (armv5te, armv5tej): Likewise.
15247 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
15248 (armv7): Add options fp and vfpv3-d16.
15249 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
15250 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
15252 (armv7ve): Likewise.
15253 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
15254 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
15255 (armv8-a): Add nocrypto option.
15256 (armv8.1-a, armv8.2-a): Likewise.
15257 (armv8-m.main): add options fp, fp.dp and nofp.
15259 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15261 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
15263 (armv8-a+crc): Delete.
15264 (armv8.1-a): Add options simd, crypto and nofp.
15265 (armv8.2-a): Add options fp16, simd, crypto and nofp.
15266 (armv8.2-a+fp16): Delete.
15267 (armv8-m.main): Add option dsp.
15268 (armv8-m.main+dsp): Delete.
15269 (cortex-a8): Add fpu. Add nofp option.
15270 (cortex-a9): Add fpu. Add nofp and nosimd options.
15271 * config/arm/parsecpu.awk (gen_data): Generate option tables and
15272 link to main cpu and architecture data structures.
15273 (gen_comm_data): Only put isa attributes from the main architecture
15275 (option): New statement for architecture and CPU entries.
15276 * arm.c (struct cpu_option): New structure.
15277 (struct processors): Add entry for options.
15278 (arm_unrecognized_feature): New function.
15279 (arm_parse_arch_cpu_name): Ignore any characters after the first
15281 (arm_parse_arch_cpu_feature): New function.
15282 (arm_configure_build_target): Separate out any CPU and architecture
15283 features and parse separately. Don't error out if -mfpu=auto is
15284 used with only an architecture string.
15285 (arm_print_asm_arch_directives): New function.
15286 (arm_file_start): Call it.
15287 * config/arm/arm-cpu-cdata.h: Regenerated.
15288 * config/arm/arm-cpu-data.h: Likewise.
15289 * config/arm/arm-tables.opt: Likewise.
15291 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15293 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
15294 assembler when it is not -mfpu=auto.
15296 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15298 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
15299 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
15300 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
15301 (ASM_CPU_SPEC): Rewrite.
15302 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
15303 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
15304 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
15305 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
15306 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
15307 copied string is NUL-terminated. Also strip any characters prefixed
15309 (arm_rewrite_selected_arch): New function.
15310 (arm_rewrite_march): New function.
15312 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15314 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
15315 (x_arm_cpu_string, x_arm_tune_string): Likewise.
15316 (march, mcpu, mtune): Convert to string-based options.
15317 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
15318 (arm_parse_arch_cpu_name): New function.
15319 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
15320 identify selected architecture or CPU.
15321 (arm_option_save): New function.
15322 (TARGET_OPTION_SAVE): Redefine.
15323 (arm_option_restore): Restore string options.
15324 (arm_option_print): Print string options.
15326 2017-06-16 Martin Sebor <msebor@redhat.com>
15328 PR tree-optimization/80933
15329 PR tree-optimization/80934
15330 * builtins.c (fold_builtin_3): Do not handle bcmp here.
15331 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
15332 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
15333 (gimple_fold_builtin): Call them.
15335 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15337 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
15338 as unlikely; update profile.
15340 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15342 * predict.c (force_edge_cold): Handle declaring edges impossible
15345 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15347 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
15349 (try_unroll_loop_completely): Fix reporting.
15351 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15353 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
15355 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
15358 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
15359 if given a non-constant argument for an intrinsic which requires a
15362 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15364 * profile.c (compare_freqs): New function.
15365 (branch_prob): Sort edge list.
15366 (find_spanning_tree): Assume that the list is priority sorted.
15368 2017-06-16 Richard Biener <rguenther@suse.de>
15370 PR tree-optimization/81090
15371 * passes.def (pass_record_bounds): Remove.
15372 * tree-pass.h (make_pass_record_bounds): Likewise.
15373 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
15374 make_pass_record_bounds): Likewise.
15375 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
15376 not free niter estimates at the beginning but at the end.
15377 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
15379 2017-06-16 Richard Biener <rguenther@suse.de>
15381 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
15382 initializer to workaround ICE in host GCC 4.8.
15384 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15386 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
15388 (clone_inlined_nodes): Update.
15390 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15392 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
15393 prefetch settings, and enable prefetching by default at -O3.
15395 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15397 * config/aarch64/aarch64.c (aarch64_override_options_internal):
15398 Set flag_prefetch_loop_arrays according to tuning data.
15400 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15402 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
15403 New tune structure.
15404 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
15405 [Unrelated to main purpose of the patch] Place the pointer field last
15406 to enable type checking errors when tune structure are wrongly merged.
15407 * config/aarch64/aarch64.c (generic_prefetch_tune,)
15408 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
15409 (thunderx2t99_prefetch_tune): New tune constants.
15410 (tune_params *_tunings): Update all tunings (no functional change).
15411 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
15412 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
15413 from tunings structures.
15415 2017-06-16 Jakub Jelinek <jakub@redhat.com>
15418 * ubsan.c (instrument_null): Add T argument, use it instead
15419 of computing it based on IS_LHS.
15420 (instrument_object_size): Likewise.
15421 (pass_ubsan::execute): Adjust instrument_null and
15422 instrument_object_size callers to pass gimple_get_lhs or
15423 gimple_assign_rhs1 result to it. Use instrument_null instead of
15424 calling get_base_address and instrument_mem_ref. Handle
15425 aggregate call arguments for object-size sanitization.
15427 2017-06-16 Yury Gribov <tetra2005@gmail.com>
15429 PR tree-optimization/81089
15430 * tree-vrp.c (is_masked_range_test): Validate operands of
15433 2017-06-15 Martin Sebor <msebor@redhat.com>
15436 * dumpfile.c (dump_register): Avoid calling memset to initialize
15437 a class with a default ctor.
15438 * gcc.c (struct compiler): Remove const qualification.
15439 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
15440 * hash-table.h: Ditto.
15441 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
15443 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
15444 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
15446 * params.h (struct param_info): Make struct members non-const.
15447 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
15448 with default initialization.
15449 * vec.h (vec_copy_construct, vec_default_construct): New helper
15451 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
15452 with vec_copy_construct.
15453 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
15454 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
15455 * doc/invoke.texi (-Wclass-memaccess): Document.
15457 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15459 * emit-rtl.h (is_leaf): Update comment about local
15460 register allocator.
15462 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15465 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
15466 for a variable to have a section before checking if the section has a
15468 Set section to.persistent if persistent attribute is set.
15469 Warn if .persistent attribute is used on an automatic variable.
15471 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
15473 PR rtl-optimization/80474
15474 * reorg.c (update_block): Do not ignore instructions in a delay slot.
15476 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
15478 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
15481 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
15483 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
15484 (casesi): Emit bounds checking as RTL.
15485 (casesi_internal_mips16_<mode>): Remove bounds checking.
15487 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
15489 * config/xtensa/xtensa.c (xtensa_option_override): Append
15490 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
15491 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
15492 xtensa_doloop_hooks): Define unconditionally.
15493 (xtensa_reorg_loops): Only call reorg_loops in the presence of
15495 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
15496 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
15497 for it in xtensa_option_override.
15498 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
15499 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
15501 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
15503 * doc/cppopts.texi: Document '-' special value to -MF.
15505 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
15507 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
15508 (cortex_a53_fconst): Likewise.
15509 (cortex_a53_fpmul): Likewise.
15510 (cortex_a53_f_load_64): Likewise.
15511 (cortex_a53_f_load_many): Likewise.
15512 (cortex_a53_advsimd_alu): Likewise.
15513 (cortex_a53_advsimd_alu_q): Likewise.
15514 (cortex_a53_advsimd_mul): Likewise.
15515 (cortex_a53_advsimd_mul_q): Likewise.
15516 (fpmac bypass): Add new bypass for fpmac-fpmac case.
15517 Add missing fmul, r2f_cvt and fconst cases.
15519 2017-06-14 Richard Biener <rguenther@suse.de>
15521 PR middle-end/81088
15522 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
15524 (fold_binary_loc): When associating do not treat pre-existing
15525 TREE_OVERFLOW on literal constants as a reason to allow
15526 TREE_OVERFLOW on associated literal constants.
15528 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15530 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
15531 (MASK_FEATURES): New macro.
15532 * config/sparc/sparc.c (sparc_option_override): Remove the special
15533 handling of -mfpu and generalize it to all MASK_FEATURES switches.
15535 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15537 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
15538 a division of 0 if non-call exceptions are enabled.
15540 2017-06-14 Andrew Pinski <apinski@cavium.com>
15541 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15544 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
15545 Improve vector initialization code gen for only variable case.
15547 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15549 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
15551 2017-06-14 Richard Biener <rguenther@suse.de>
15553 PR tree-optimization/81083
15554 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
15557 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15559 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
15560 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
15561 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
15562 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
15563 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
15564 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
15566 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15568 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
15569 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
15571 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15573 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
15575 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15577 * config/rs6000/t-rtems: Don't handle SPE.
15579 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15581 * config/rs6000/t-linux: Don't handle SPE.
15583 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15585 * config/rs6000/eabispe.h: Delete file.
15587 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15589 * config/rs6000/t-spe: Delete file.
15591 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15593 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
15594 (rs6000_legitimate_offset_address_p): Return false for anything in
15595 V2SImode or V2SFmode.
15597 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15599 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
15600 except V2SF and V2SI. Rearrange the vector modes, and add comments.
15601 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
15603 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
15604 (rs6000_legitimate_offset_address_p): Ditto.
15605 (rs6000_emit_move): Ditto.
15606 (rs6000_init_builtins): Remove V4HI_type_node.
15608 2017-06-13 Martin Liska <mliska@suse.cz>
15611 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
15612 (gate_asan): Likewise.
15613 * asan.h (asan_no_sanitize_address_p): Remove the function.
15614 (sanitize_flags_p): New function.
15615 * builtins.def: Fix coding style.
15616 * common.opt: Use renamed enum value.
15617 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
15618 * doc/extend.texi: Document no_sanitize attribute.
15619 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
15620 to SANITIZE_UNDEFINED_NONDEFAULT.
15621 * gcc.c (sanitize_spec_function): Use the renamed enum value.
15622 * gimple-fold.c (optimize_atomic_compare_exchange_p):
15623 Use sanitize_flags_p.
15624 * gimplify.c (gimplify_function_tree): Likewise.
15625 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
15626 * opts.c (parse_no_sanitize_attribute): New function.
15627 (common_handle_option): Use renamed enum value.
15628 * opts.h (parse_no_sanitize_attribute): Declare.
15629 * tree.c (sanitize_flags_p): New function.
15630 * tree.h: Declared here.
15631 * tsan.c: Use sanitize_flags_p.
15632 * ubsan.c (ubsan_expand_null_ifn): Likewise.
15633 (instrument_mem_ref): Likewise.
15634 (instrument_bool_enum_load): Likewise.
15635 (do_ubsan_in_current_function): Remove the function.
15636 (pass_ubsan::execute): Use sanitize_flags_p.
15637 * ubsan.h: Remove do_ubsan_in_current_function
15638 * tree-cfg.c (print_no_sanitize_attr_value): New function.
15639 (dump_function_to_file): Use it here.
15641 2017-06-13 Martin Jambor <mjambor@suse.cz>
15643 PR tree-optimization/80803
15644 PR tree-optimization/81063
15645 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
15646 (propagate_subaccesses_across_link): Enqueue subtree whenever
15647 necessary instead of relying on the caller.
15649 2017-06-13 Martin Jambor <mjambor@suse.cz>
15651 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
15652 that have a first_link.
15653 (sort_and_splice_var_accesses): Do not check first_link before
15655 (subtree_mark_written_and_enqueue): Likewise.
15656 (propagate_all_subaccesses): Likewise and do not stop at first
15657 parent with a first_link.
15659 2017-06-13 Martin Jambor <mjambor@suse.cz>
15661 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
15664 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15666 * match.pd: New pattern.
15668 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15670 * tree-vrp.c (is_masked_range_test): New function.
15671 (register_edge_assert_for): Determine ranges for
15674 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15676 PR tree-optimization/67328
15677 * fold-const.c (maskable_range_p): New function.
15678 (build_range_check): Generate bittests if possible.
15680 2017-06-13 Martin Liska <mliska@suse.cz>
15682 * gimple-pretty-print.c (dump_probability): Add new argument.
15683 (dump_edge_probability): Dump both probability and count.
15684 (dump_gimple_label): Likewise.
15685 (dump_gimple_bb_header): Likewise.
15687 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
15690 * config/avr/avr-devices.c: Fix indentation.
15691 * config/avr/gen-avr-mmcu-specs.c: Dito.
15693 2017-06-13 Richard Biener <rguenther@suse.de>
15695 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
15696 instead get vector type from stmt_info.
15697 (vectorizable_reduction): Adjust. Remove dead code.
15699 2017-06-13 Richard Biener <rguenther@suse.de>
15701 PR middle-end/81065
15702 * fold-const.c (extract_muldiv_1): Remove bogus distribution
15703 case of C * (x * C2 + C3).
15704 (fold_addr_of_array_ref_difference): Properly fold index difference.
15706 2017-06-12 David S. Miller <davem@davemloft.net>
15709 * config/sparc/sparc.md (return expander): Emit frame blockage if
15710 function uses alloca.
15712 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
15714 * combine.c (make_field_assignment): Check len rather than the mode
15715 precision when calling force_to_mode.
15717 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
15719 Support multilibs and devices that see flash in RAM address range.
15722 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
15723 (avr_mcu_t) <flash_pm_offset>: New field.
15724 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
15725 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
15726 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
15727 (AVR_TINY_PM_OFFSET): Remove macro.
15728 * config/avr/avr.opt (-mshort-calls): New option.
15729 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
15730 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
15731 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
15732 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
15733 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
15734 instead of avr_arch->have_jmp_call.
15735 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
15736 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
15737 avr_arch->flash_pm_offset to define.
15738 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
15739 new field flash_pm_offset. Add entry for avrxmega3.
15740 (avr_texinfo): Add entry for avrxmega3.
15741 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
15742 attiny212, attiny214,
15743 attiny412, attiny414, attiny416, attiny417,
15744 attiny814, attiny816, attiny817,
15745 attiny1614, attiny1616, attiny1617,
15746 attiny3214, attiny3216, attiny3217.
15747 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
15748 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
15749 (avr_print_operand_address) [AVR_TINY]: Same.
15750 (avr_asm_init_sections) <readonly_data_section>: Only patch
15751 callback if avr_arch->flash_pm_offset = 0.
15752 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
15753 for rodata if avr_arch->flash_pm_offset != 0.
15754 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
15755 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
15756 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
15757 (m_options): Append opt_rcall.
15758 (m_dirnames): Append dir_rcall.
15759 * config/avr/t-multilib: Regenerate.
15761 * configure.ac [target=avr]: Check whether avrxmega3 default
15762 linker description file works as needed.
15763 * configure: Regenerate.
15764 * doc/avr-mmcu.texi: Regenerate.
15765 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
15766 <__AVR_ARCH__>: Document avrxmega3 and 103.
15767 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
15768 <__AVR_SHORT_CALLS__>: Document it.
15769 <__AVR_PM_BASE_ADDRESS__>: Document it.
15770 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
15771 (AVR Variable Attributes) <progmem>: Document this is
15772 not needed for avrxmega3.
15773 (AVR Named Address Spaces) <__flash>: Dito.
15775 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
15777 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
15779 2017-06-12 Doug Rupp <rupp@adacore.com>
15781 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
15782 Append vxworks-stdint.h to the tm_file list.
15783 * config/vxworks-stdint.h: New file.
15785 2017-06-12 Martin Liska <mliska@suse.cz>
15787 PR tree-optimization/81041
15788 * tree-profile.c (gimple_gen_ic_func_profiler):
15789 Create an extra BB in profile-generate
15790 (gimple_gen_time_profiler): Likewise.
15792 2017-06-12 Jakub Jelinek <jakub@redhat.com>
15794 PR tree-optimization/81003
15795 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
15796 (update_range_test): Use it instead of force_gimple_operand_gsi.
15798 2017-06-12 Richard Biener <rguenther@suse.de>
15800 PR tree-optimization/81053
15801 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
15802 with backedge value not defined in loop. Simplify def stmt
15805 2017-06-11 Tom de Vries <tom@codesourcery.com>
15808 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
15810 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
15811 nvptx_cannot_force_const_mem.
15813 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15815 * opts.c (finish_options): Move test for flag_split_stack after
15816 it has been initialized.
15818 2017-06-11 Jason Merrill <jason@redhat.com>
15820 * tree.h (id_equal): New.
15821 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
15822 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
15823 instead of strcmp of IDENTIFIER_POINTER.
15825 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15827 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
15828 (mark_all_inlined_calls_cdtor): Fix formating.
15829 (inline_transform): Rescale profile before inlining.
15831 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15833 * cgraph.h (cgraph_edge::clone): Update prototype.
15834 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
15835 (cgraph_node::create_clone): Update.
15836 (cgraph_node::create_version_clone): Update.
15837 * tree-inline.c (copy_bb): Update.
15838 (expand_call_inline): Update.
15840 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
15842 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
15843 factored out from ...
15844 (rs6000_emit_prologue): ... here.
15846 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
15848 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
15849 factored out from ...
15850 (rs6000_emit_prologue): ... here.
15852 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15854 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
15856 (handle_missing_profiles): Fix computation of tp_first_run.
15857 (counts_to_freqs): Do not touch freqs when count is 0.
15859 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15861 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
15864 2017-06-10 Tom de Vries <tom@codesourcery.com>
15866 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
15867 attributes): Document signal effective target.
15869 2017-06-10 Tom de Vries <tom@codesourcery.com>
15871 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
15872 Document effective target stack_size.
15874 2017-06-09 David Malcolm <dmalcolm@redhat.com>
15876 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
15877 to the edit_context if they can be auto-applied.
15879 2017-06-9 Ian Lance Taylor <iant@golang.org>
15881 * opts.c (finish_options): If -fsplit-stack, disable implicit
15882 -forder-blocks-and-partition.
15883 * doc/invoke.texi (Optimize Options): Document that when using
15884 -fsplit-stack -forder-blocks-and-partition is not implicitly
15887 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15889 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
15890 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
15891 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
15892 * builtins.def (abort, trap, unreachable): Declare cold.
15893 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
15894 * tree-core.h (ECF_COLD): New.
15895 * tree.c (set_call_expr_flags): Handle ECF_COLD.
15896 (build_common_builtin_nodes): Mark unreachable and abort as cold.
15898 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15900 * predict.c (unlikely_executed_stmt_p): Cleanup.
15902 2017-06-09 Richard Biener <rguenther@suse.de>
15904 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
15905 model if the ref is always written to.
15907 2017-06-09 Tamar Christina <tamar.christina@arm.com>
15909 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
15911 2017-06-09 Tamar Christina <tamar.christina@arm.com>
15913 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
15916 2017-06-09 Tom de Vries <tom@codesourcery.com>
15919 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
15920 "target cannot support label values" when encountering LABEL_REF.
15922 2017-06-09 Martin Liska <mliska@suse.cz>
15924 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
15925 (gimple_gen_ic_func_profiler): Emit direct comparison
15926 of __gcov_indirect_call_callee with NULL.
15927 (gimple_gen_time_profiler): Change probability from
15928 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
15930 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15932 * profile.c (edge_gcov_counts): Turn to pointer.
15933 (compute_branch_probabilities, compute_branch_probabilities): Update.
15934 (branch_prob): Do not clear edge_gcov_count.
15935 * profile.h (edge_gcov_counts): Turn to pointer.
15936 (edge_gcov_count): Update.
15938 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15940 * gimple.h (gimple_check_failed): Mark cold.
15942 2017-06-09 Richard Biener <rguenther@suse.de>
15944 PR tree-optimization/66623
15945 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
15946 refactor check_reduction into two parts, properly computing
15947 whether we have to check reduction validity for outer loop
15950 2017-06-09 Richard Biener <rguenther@suse.de>
15952 PR tree-optimization/79483
15953 * graphite-scop-detection.c (order): New global.
15954 (get_order): Compute bb to order mapping that satisfies code
15955 generation constraints.
15956 (cmp_pbbs): New helper.
15957 (build_scops): Start domwalk at entry block, sort generated
15960 2017-06-09 Richard Biener <rguenther@suse.de>
15962 PR middle-end/81007
15963 * ipa-polymorphic-call.c
15964 (ipa_polymorphic_call_context::restrict_to_inner_class):
15965 Skip FIELD_DECLs with error_mark_node type.
15966 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
15969 2017-06-09 Martin Liska <mliska@suse.cz>
15971 * predict.c (struct branch_predictor): New struct.
15972 (test_prediction_value_range): New test.
15973 (predict_c_tests): New function.
15974 * selftest-run-tests.c (selftest::run_tests): Run the function.
15975 * selftest.h: Declare new tests.
15977 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
15980 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
15981 gen_add3_insn did not fail.
15982 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
15983 r0, construct that number in a temporary reg and add that reg to r0.
15984 If asked to put the result in r0 as well, fail.
15986 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
15988 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15989 for early expansion of vec_eqv.
15991 2017-06-08 Jakub Jelinek <jakub@redhat.com>
15993 PR middle-end/81005
15994 * ubsan.c (instrument_null): Avoid pointless code temporary.
15995 (pass_ubsan::execute): Instrument aggregate arguments of calls.
15997 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
16001 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
16004 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
16007 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16009 * predict.c (unlikely_executed_edge_p): Move ahead.
16010 (probably_never_executed_edge_p): Use it.
16012 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
16014 PR middle-end/79988
16015 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
16016 gimple_call_builtin_p call.
16018 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16020 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
16021 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
16022 rtl_check_failed_type2, rtl_check_failed_code1,
16023 rtl_check_failed_code2, rtl_check_failed_code_mode,
16024 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
16025 rtvec_check_failed_bounds, rtl_check_failed_flag,
16026 _fatal_insn_not_found, _fatal_insn): Likewise.
16027 * tree.h (tree_contains_struct_check_failed,
16028 tree_check_failed, tree_not_check_failed,
16029 tree_class_check_failed, tree_range_check_failed,
16030 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
16031 tree_vec_elt_check_failed, phi_node_elt_check_failed,
16032 tree_operand_check_failed, omp_clause_check_failed,
16033 omp_clause_operand_check_failed, omp_clause_range_check_failed):
16036 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16038 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
16039 flag_branch_probabilities.
16040 * ipa-inline.c (edge_badness): Likewise.
16041 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
16042 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
16043 * predict.c (maybe_hot_frequency_p): Likewise.
16044 (probably_never_executed): Likewise.
16045 * sched-ebb.c (schedule_ebbs): Likewise.
16046 * sched-rgn.c (find_single_block_region): Likewise.
16047 * tracer.c (tail_duplicate): Likewise.
16049 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16051 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
16052 longer requires x_flag_profile_use.
16054 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16056 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
16057 instead of flag_reorder_blocks_and_partition.
16058 * dbxout.c (dbxout_function_end): Likewise.
16059 * dwarf2out.c (gen_subprogram_die): Likewise.
16060 * haifa-sched.c (sched_create_recovery_edges): Likewise.
16061 * hw-doloop.c (reorg_loops): Likewise.
16062 * varasm.c (assemble_start_function,
16063 assemble_end_function): Likewise.
16064 (decide_function_section): Do not check for
16065 flag_reorder_blocks_and_partition.
16067 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
16069 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
16071 (chkp_get_hard_register_fake_addr_expr): Ditto.
16072 (chkp_build_addr_expr): Add check for hard reg case.
16073 (chkp_parse_array_and_component_ref): Ditto.
16074 (chkp_find_bounds_1): Ditto.
16075 (chkp_process_stmt): Don't generate bounds store for
16078 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16080 * predict.c (maybe_hot_bb_p): Do not check profile status.
16081 (maybe_hot_edge_p): Likewise.
16082 (probably_never_executed): Check for zero counts even if profile
16084 (unlikely_executed_edge_p): New function.
16085 (unlikely_executed_stmt_p): New function.
16086 (unlikely_executed_bb_p): New function.
16087 (set_even_probabilities): Use unlikely predicates.
16088 (combine_predictions_for_bb): Likewise.
16089 (predict_paths_for_bb): Likewise.
16090 (predict_paths_leading_to_edge): Likewise.
16091 (determine_unlikely_bbs): New function.
16092 (estimate_bb_frequencies): Use it.
16093 (compute_function_frequency): Use zero counts even if profile is
16095 * profile-count.h: Fix typo.
16097 2017-08-08 Julia Koval <julia.koval@intel.com>
16099 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
16100 _mm512_mask_cvtsepi16_storeu_epi8,
16101 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
16102 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
16103 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
16104 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
16105 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
16106 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
16107 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
16108 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
16109 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
16110 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
16111 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
16112 __builtin_ia32_pmovuswb256mem_mask,
16113 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
16114 __builtin_ia32_pmovwb512mem_mask): New builtins.
16116 2017-08-08 Julia Koval <julia.koval@intel.com>
16118 PR target/73350,80862
16119 * config/i386/subst.md (round): Fix round pattern.
16120 * config/i386/i386.c (ix86_erase_embedded_rounding):
16121 Fix erasing rounding for the fixed pattern.
16123 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16125 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
16127 2017-06-08 Martin Liska <mliska@suse.cz>
16129 PR gcov-profile/80911
16130 * gcov.c (block_info::block_info): New constructor.
16132 2017-06-07 Carl Love <cel@us.ibm.com>
16134 * config/rs6000/rs6000-c: The return type of the following
16135 built-in functions was implemented as int not long long. Fix sign
16136 of return value for the unsigned version of vec_mulo and vec_mule.
16137 vector unsigned long long vec_bperm (vector unsigned long long,
16138 vector unsigned char)
16139 vector signed long long vec_mule (vector signed int,
16141 vector unsigned long long vec_mule (vector unsigned int,
16142 vector unsigned int)
16143 vector signed long long vec_mulo (vector signed int,
16145 vector unsigned long long vec_mulo (vector unsigned int,
16146 vector unsigned int)
16147 * doc/extend.texi: Fix the documentation for the built-in
16150 2017-06-07 Carl Love <cel@us.ibm.com>
16153 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
16156 2017-06-07 Carl Love <cel@us.ibm.com>
16158 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
16159 support, Generate doublehv for signed int/float for BE case only.
16161 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
16163 * doc/invoke.texi (mcx16): Rewrite.
16165 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16167 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
16168 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
16169 *mov<mode>_softfloat, and an anonymous splitter): Use
16170 nonimmediate_operand instead of rs6000_nonimmediate_operand.
16172 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16174 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
16176 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
16177 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
16178 (rs6000_debug_reg_global): Adjust.
16179 (rs6000_init_hard_regno_mode_ok): Adjust.
16180 (rs6000_dbx_register_number): Adjust.
16181 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
16182 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
16183 Remove SPE_ACC and SPEFSCR.
16184 (REG_ALLOC_ORDER): Ditto.
16185 (FRAME_POINTER_REGNUM): Change to 111.
16186 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
16187 (REG_CLASS_NAMES): Ditto.
16188 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
16189 (REGISTER_NAMES): Ditto.
16190 (ADDITIONAL_REG_NAMES): Ditto.
16191 (rs6000_reg_names): Ditto.
16192 * config/rs6000/rs6000.md: Renumber some register number
16195 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16197 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
16199 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
16200 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
16202 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
16203 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
16204 Delete the SPE high registers.
16205 (REG_ALLOC_ORDER): Ditto.
16206 (enum reg_class): Remove SPE_HIGH_REGS.
16207 (REG_CLASS_NAMES): Ditto.
16208 (REG_CLASS_CONTENTS): Delete the SPE high registers.
16209 (REGISTER_NAMES): Ditto.
16210 (rs6000_reg_names): Ditto.
16211 * doc/tm.texi.in: Remove SPE as example.
16212 * doc/tm.texi: Regenerate.
16214 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16216 * config/rs6000/8540.md (ppc8540_brinc): Delete.
16217 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
16218 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
16219 * config/rs6000/rs6000.md (type): Remove "brinc".
16221 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16223 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
16224 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
16225 * config/rs6000/linuxspe.h: Delete file.
16226 * config/rs6000/rs6000.md: Don't include spe.md.
16227 * config/rs6000/spe.h: Delete file.
16228 * config/rs6000/spe.md: Delete file.
16229 * config/rs6000/t-rs6000: Remove spe.md.
16231 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16233 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
16234 (reg_or_none500mem_operand): Delete.
16235 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
16236 instead of reg_or_none500mem_operand.
16238 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16240 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
16241 handling of SPE flags.
16242 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
16244 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16246 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
16248 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
16249 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
16250 paired_divv2sf3): Similar.
16251 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
16252 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
16253 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
16255 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
16256 Rename the paired_* instruction patterns.
16257 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
16259 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
16260 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
16261 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
16262 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
16263 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
16264 PAIRED_VECTOR_MODE.
16265 (struct machine_function): Delete field spe_insn_chain_scanned_p.
16266 (spe_func_has_64bit_regs_p): Delete.
16267 (spe_expand_predicate_builtin): Delete.
16268 (spe_expand_evsel_builtin): Delete.
16269 (TARGET_DWARF_REGISTER_SPAN): Do not define.
16270 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
16271 (invalid_e500_subreg): Delete.
16272 (rs6000_legitimize_address): Always force_reg op2 as well, for
16273 paired single memory accesses.
16274 (rs6000_member_type_forces_blk): Delete.
16275 (rs6000_spe_function_arg): Delete.
16276 (rs6000_expand_unop_builtin): Delete SPE handling.
16277 (rs6000_expand_binop_builtin): Ditto.
16278 (spe_expand_stv_builtin): Delete.
16279 (bdesc_2arg_spe): Delete.
16280 (spe_expand_builtin): Delete.
16281 (spe_expand_predicate_builtin): Delete.
16282 (spe_expand_evsel_builtin): Delete.
16283 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
16284 (spe_init_builtins): Delete.
16285 (spe_func_has_64bit_regs_p): Delete.
16286 (savres_routine_name): Delete "info" parameter. Adjust callers.
16287 (rs6000_emit_stack_reset): Ditto.
16288 (rs6000_dwarf_register_span): Delete.
16289 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
16290 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
16291 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
16293 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
16295 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
16296 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
16297 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
16298 mulv2sf3, divv2sf3): Delete expanders.
16300 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16302 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
16304 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16306 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
16307 * config/rs6000/rs6000.c: Ditto.
16309 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16311 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
16312 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
16313 comparison_operator.
16315 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16317 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
16318 * config/rs6000/rs6000.opt: Ditto.
16319 * config/rs6000/t-rtems: Ditto.
16321 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16323 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
16324 TARGET_E500_SINGLE by 0, simplify.
16325 * config/rs6000/rs6000.c: Ditto.
16326 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
16327 (spe_build_register_parallel): Delete.
16328 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
16329 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
16330 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
16331 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
16332 (E500_CONVERT): Delete.
16333 * config/rs6000/spe.md: Remove many patterns and all define_constants.
16335 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16337 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
16338 * config/rs6000/dfp.md: Ditto.
16339 (negdd2, *negdd2_fpr): Merge.
16340 (absdd2, *absdd2_fpr): Merge.
16341 (negtd2, *negtd2_fpr): Merge.
16342 (abstd2, *abstd2_fpr): Merge.
16343 * config/rs6000/e500.h: Delete file.
16344 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
16345 TARGET_FPRS by 1 and simplify.
16346 * config/rs6000/rs6000-c.c: Ditto.
16347 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
16348 TARGET_DF_SPE by 0.
16349 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
16351 * config/rs6000/rs6000.md: Ditto.
16352 (floatdidf2, *floatdidf2_fpr): Merge.
16353 (move_from_CR_gt_bit): Delete.
16354 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
16355 (E500_CR_IOR_COMPARE): Delete.
16356 (All patterns that require !TARGET_FPRS): Delete.
16357 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
16359 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16361 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
16363 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16365 * graphds.c (add_edge): Intitialize edge's attached data.
16366 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
16367 pointer parameter. Call pointed function on each edge during
16368 graph traversing. Skip traversing the edge when the function
16370 (graphds_dfs, graphds_scc): Ditto.
16371 (for_each_edge): New parameter. Pass the new parameter to callback
16373 * graphds.h (skip_edge_callback): New function pointer type.
16374 (graphds_dfs, graphds_scc): New function pointer parameter.
16375 (graphds_edge_callback, for_each_edge): New parameter.
16377 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16379 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
16380 out code checking if runtime alias check is possible to below ...
16381 Call the new function.
16382 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
16383 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
16385 2017-06-07 Marek Polacek <polacek@redhat.com>
16388 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
16389 TYPE_OVERFLOW_WRAPS check.
16391 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16393 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
16394 if versioning is required.
16395 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
16396 peeling with the check for versioning.
16398 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16400 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
16401 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
16402 Set true to new parameter if new ssa variable is defined.
16403 (vect_gen_vector_loop_niters): Refactor. Set range information
16404 for the new vector loop bound variable.
16405 (vect_do_peeling): Ditto.
16407 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16409 * tree-affine.c (ssa.h): Include header file.
16410 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
16411 has wrapping overflow behavior.
16413 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16415 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
16417 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16419 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
16420 (tree_to_aff_combination): ... here.
16422 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16424 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
16425 reg_pressure model function.
16426 (ivopts_global_cost_for_size): Delete.
16427 (determine_set_costs, iv_ca_recount_cost): Call new model function
16428 ivopts_estimate_reg_pressure.
16430 2017-06-07 Tamar Christina <tamar.christina@arm.com>
16432 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
16433 expensive than udiv. Remove floating point cases from mod.
16435 2017-06-07 Tamar Christina <tamar.christina@arm.com>
16437 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
16438 Increase idiv cost.
16440 2017-06-07 Tamar Christina <tamar.christina@arm.com>
16442 * config/aarch64/aarch64.md
16443 (copysignsf3): Fix mask generation.
16445 2017-06-07 Jakub Jelinek <jakub@redhat.com>
16447 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
16449 (class dump_manager): Add register_dumps method.
16450 * dumpfile.c: Include langhooks.h.
16451 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
16452 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
16453 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
16454 (dump_manager::dump_register): Start with 512 entries instead of 32.
16455 (dump_manager::register_dumps): New method.
16456 * toplev.c (general_init): Instead of invoking register_dumps
16457 langhook, invoke register_dumps method on the dump manager.
16458 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
16461 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
16463 * doc/md.texi: Clarify the restrictions on a define_insn condition.
16464 Say that # requires an associated define_split to exist, and that
16465 the define_split must be suitable for use after register allocation.
16467 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16469 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
16470 (compute_outgoing_frequencies): Also initialize zero counts.
16471 (find_many_sub_basic_blocks): Do not produce uninitialized profile
16472 around loops; preserve more of profile when nothing changes.
16474 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
16476 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
16478 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
16479 * config/arm/arm-cpu-cdata.h: Regenerate.
16480 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
16481 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
16482 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
16483 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
16484 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
16486 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
16488 * config/arm/t-rmprofile: Likewise.
16489 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
16491 2017-06-06 David S. Miller <davem@davemloft.net>
16494 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
16495 blockage if function uses alloca.
16497 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16499 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
16500 New "uid" fields to hold pretty-print IDs of group and ref.
16501 Memory references are now identified as <group_id>:<ref_id>
16502 instead of using [random] addresses.
16503 (dump_mem_details): Simplify, no functional change.
16504 (dump_mem_ref): Simplify and make output more concise.
16505 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
16506 (find_or_create_group): Initialize group uid.
16507 (record_ref): Initialize ref uid. Improve debug output.
16508 (prune_group_by_reuse, should_issue_prefetch_p,)
16509 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
16510 (mark_nontemporal_store, determine_loop_nest_reuse):
16511 Improve debug output.
16513 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16515 * dbgcnt.def (prefetch): New debug counter.
16516 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
16517 (schedule_prefetches): Stop issueing prefetches if debug counter
16520 2017-06-06 Tom de Vries <tom@codesourcery.com>
16522 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
16523 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
16525 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16527 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
16528 Use aarch64_reg_or_zero predicate for operand 4.
16529 (aarch64_compare_and_swap<mode> define_insn_and_split):
16530 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
16531 (aarch64_store_exclusive<mode>): Likewise for operand 2.
16533 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
16535 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
16536 (arm_compute_save_core_reg_mask): This.
16537 (thumb1_compute_save_reg_mask): Rename into ...
16538 (thumb1_compute_save_core_reg_mask): This.
16539 (arm_compute_save_reg0_reg12_mask): Adapt comment.
16540 (arm_compute_frame_layout): Likewise.
16542 2017-06-06 Richard Biener <rguenther@suse.de>
16544 PR tree-optimization/80974
16545 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
16546 keep or clear leaders SSA info.
16548 2017-06-06 Tom de Vries <tom@codesourcery.com>
16550 * config/nvptx/nvptx.c (split_mode_p): New function.
16551 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
16553 2017-06-06 Tom de Vries <tom@codesourcery.com>
16555 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
16557 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16560 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
16563 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16565 * shrink-wrap.c (handle_simple_exit): Update profile.
16566 (try_shrink_wrapping): Upate profile.
16568 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16570 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
16571 (tree_guess_outgoing_edge_probabilities): New.
16572 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
16573 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
16575 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16577 * ipa-split.c (split_function): Initialize return bb profile.
16579 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16581 * profile.c (compute_branch_probabilities): Also initialize
16582 EXIT_BLOCK profile.
16584 2017-06-06 Richard Biener <rguenther@suse.de>
16586 PR tree-optimization/80928
16587 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
16588 (vect_analyze_loop_operations): Properly guard analysis for
16590 (vect_transform_loop): Likewise.
16591 (vect_analyze_loop_2): Also reset SLP type on PHIs.
16592 (vect_model_induction_cost): Do not cost for pure SLP.
16593 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
16594 of induction in inner loop vectorization.
16595 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
16596 (vect_get_and_check_slp_defs): Handle vect_induction_def.
16597 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
16599 (vect_analyze_slp_cost_1): Cost induction.
16600 (vect_detect_hybrid_slp_stmts): Handle PHIs.
16601 (vect_get_slp_vect_defs): Likewise.
16602 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
16603 (vect_transform_stmt): Handle SLP reductions.
16604 * tree-vectorizer.h (vectorizable_induction): Adjust.
16606 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16608 * config/rs6000/rs6000.c (make_resolver_func): Update
16609 init_lowered_empty_function call.
16611 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
16613 * doc/invoke.texi: Document the -fprofile-abs-path option.
16614 * common.opt (fprofile-abs-path): New option.
16615 * gcov-io.h (gcov_write_filename): Declare.
16616 * gcov-io.c (gcov_write_filename): New function.
16617 * coverage.c (coverage_begin_function): Use gcov_write_filename.
16618 * profile.c (output_location): Likewise.
16620 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16622 * shring-wrap.c: Revert accidental commit.
16624 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
16626 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
16628 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16630 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
16632 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
16633 profile in callgraph edge.
16634 * profile-count.h (apply_probability): If THIS is 0, then result is 0
16635 (apply_scale): Likewise.
16636 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
16637 Also scale profile when inlining function with zero profile.
16638 (initialize_cfun): Update exit block profile even when it is zero.
16639 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
16640 when profile is read.
16642 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16644 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
16645 (CLONE_*): New constants to define the processors we can generate
16646 code for with the target_clone attribute.
16647 (rs6000_clone_map): New array to identify which clone processors
16648 the current program is running on.
16649 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
16650 target_clone attribute.
16651 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
16652 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
16653 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
16654 (cpu_expand_builtin): Add support for target_clone attribute.
16655 (rs6000_valid_attribute_p): Allow "default" attribute.
16656 (get_decl_name): New debug function to simplify printing the
16657 current function name in debugging statements.
16658 (rs6000_clone_priority): New functions to support the target_clone
16659 attribute, and be able to generate code to switch between ISA 2.05
16660 through ISA 3.0 (power6 through power9).
16661 (rs6000_compare_version_priority): Likewise.
16662 (rs6000_get_function_versions_dispatcher): Likewise.
16663 (make_resolver_func): Likewise.
16664 (add_condition_to_bb): Likewise.
16665 (dispatch_function_versions): Likewise.
16666 (rs6000_generate_version_dispatcher_body): Likewise.
16667 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
16668 (fusion_gpr_load_p): Fix a spacing issue.
16669 * doc/extend.texi (Common Function Attributes): Document that the
16670 PowerPC supports the target_clone attribute.
16672 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
16674 * config/arm/arm.h: explain F symbol found in description of ARM
16675 register allocation in its legend.
16677 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16679 * config/mips/frame-header-opt.c: Include profile-count.h.
16680 * config/riscv/riscv.c: Include profile-count.h
16682 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16684 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
16686 (sm_set_flag_if_changed): Add bbs field.
16687 (execute_sm_if_changed_flag_set): Pass BBS.
16688 (execute_sm): Update.
16690 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16692 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
16695 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16697 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
16698 (peephole2): New peephole2 to emit the above.
16699 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
16701 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16703 * config/aarch64/aarch64.c (define_peephole2 above
16704 *sub_<shift>_<mode>): New peephole.
16706 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16708 * config/i386/i386.c (make_resolver_func): Update.
16709 * Makefile.in: Add profile-count.h and profile-count.o
16710 * auto-profile.c (afdo_indirect_call): Update to new API.
16711 (afdo_set_bb_count): Update.
16712 (afdo_propagate_edge): Update.
16713 (afdo_propagate_circuit): Update.
16714 (afdo_calculate_branch_prob): Update.
16715 (afdo_annotate_cfg): Update.
16716 * basic-block.h: Include profile-count.h
16717 (struct edge_def): Turn count to profile_count.
16718 (struct basic_block_def): Likewie.
16719 (REG_BR_PROB_BASE): Move to profile-count.h
16720 (RDIV): Move to profile-count.h
16721 * bb-reorder.c (max_entry_count): Turn to profile_count.
16722 (find_traces): Update.
16723 (rotate_loop):Update.
16724 (connect_traces):Update.
16725 (sanitize_hot_paths):Update.
16726 * bt-load.c (migrate_btr_defs): Update.
16727 * cfg.c (RDIV): Remove.
16728 (init_flow): Use alloc_block.
16729 (alloc_block): Uninitialize count.
16730 (unchecked_make_edge): Uninitialize count.
16731 (check_bb_profile): Update.
16732 (dump_edge_info): Update.
16733 (dump_bb_info): Update.
16734 (update_bb_profile_for_threading): Update.
16735 (scale_bbs_frequencies_int): Update.
16736 (scale_bbs_frequencies_gcov_type): Update.
16737 (scale_bbs_frequencies_profile_count): New.
16738 * cfg.h (update_bb_profile_for_threading): Update.
16739 (scale_bbs_frequencies_profile_count): Declare.
16740 * cfgbuild.c (compute_outgoing_frequencies): Update.
16741 (find_many_sub_basic_blocks): Update.
16742 * cfgcleanup.c (try_forward_edges): Update.
16743 (try_crossjump_to_edge): Update.
16744 * cfgexpand.c (expand_gimple_tailcall): Update.
16745 (construct_exit_block): Update.
16746 * cfghooks.c (verify_flow_info): Update.
16747 (dump_bb_for_graph): Update.
16748 (split_edge): Update.
16749 (make_forwarder_block): Update.
16750 (duplicate_block): Update.
16751 (account_profile_record): Update.
16752 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
16753 (get_estimated_loop_iterations): Update.
16754 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
16755 (single_likely_exit): Update.
16756 * cfgloopmanip.c (scale_loop_profile): Update.
16758 (set_zero_probability): Update.
16759 (lv_adjust_loop_entry_edge): Update.
16760 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16761 (purge_dead_edges): Update.
16762 (rtl_account_profile_record): Update.
16763 * cgraph.c (cgraph_node::create): Uninitialize count.
16764 (symbol_table::create_edge): Uninitialize count.
16765 (cgraph_update_edges_for_call_stmt_node): Update.
16766 (cgraph_edge::dump_edge_flags): Update.
16767 (cgraph_node::dump): Update.
16768 (cgraph_edge::maybe_hot_p): Update.
16769 * cgraph.h: Include profile-count.h
16770 (create_clone), create_edge, create_indirect_edge): Update.
16771 (cgraph_node): Turn count to profile_count.
16772 (cgraph_edge0: Likewise.
16773 (make_speculative, clone): Update.
16774 (create_edge): Update.
16775 (init_lowered_empty_function): Update.
16776 * cgraphclones.c (cgraph_edge::clone): Update.
16777 (duplicate_thunk_for_node): Update.
16778 (cgraph_node::create_clone): Update.
16779 * cgraphunit.c (cgraph_node::analyze): Update.
16780 (cgraph_node::expand_thunk): Update.
16781 * final.c (dump_basic_block_info): Update.
16782 * gimple-streamer-in.c (input_bb): Update.
16783 * gimple-streamer-out.c (output_bb): Update.
16784 * graphite.c (print_global_statistics): Update.
16785 (print_graphite_scop_statistics): Update.
16786 * hsa-brig.c: Include basic-block.h.
16787 * hsa-dump.c: Include basic-block.h.
16788 * hsa-gen.c (T sum_slice): Update.
16789 (convert_switch_statements):Update.
16790 * hsa-regalloc.c: Include basic-block.h.
16791 * ipa-chkp.c (chkp_produce_thunks): Update.
16792 * ipa-cp.c (struct caller_statistics): Update.
16793 (init_caller_stats): Update.
16794 (gather_caller_stats): Update.
16795 (ipcp_cloning_candidate_p): Update.
16796 (good_cloning_opportunity_p): Update.
16797 (get_info_about_necessary_edges): Update.
16798 (dump_profile_updates): Update.
16799 (update_profiling_info): Update.
16800 (update_specialized_profile): Update.
16801 (perhaps_add_new_callers): Update.
16802 (decide_about_value): Update.
16803 (ipa_cp_c_finalize): Update.
16804 * ipa-devirt.c (struct odr_type_warn_count): Update.
16805 (struct decl_warn_count): Update.
16806 (struct final_warning_record): Update.
16807 (possible_polymorphic_call_targets): Update.
16808 (ipa_devirt): Update.
16809 * ipa-fnsummary.c (redirect_to_unreachable): Update.
16810 * ipa-icf.c (sem_function::merge): Update.
16811 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
16812 * ipa-inline.c (compute_uninlined_call_time): Update.
16813 (compute_inlined_call_time): Update.
16814 (want_inline_small_function_p): Update.
16815 (want_inline_self_recursive_call_p): Update.
16816 (edge_badness): Update.
16817 (lookup_recursive_calls): Update.
16818 (recursive_inlining): Update.
16819 (inline_small_functions): Update.
16820 (dump_overall_stats): Update.
16821 (dump_inline_stats): Update.
16822 * ipa-profile.c (ipa_profile_generate_summary): Update.
16823 (ipa_propagate_frequency): Update.
16824 (ipa_profile): Update.
16825 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
16826 * ipa-utils.c (ipa_merge_profiles): Update.
16827 * loop-doloop.c (doloop_modify): Update.
16828 * loop-unroll.c (report_unroll): Update.
16829 (unroll_loop_runtime_iterations): Update.
16830 * lto-cgraph.c (lto_output_edge): Update.
16831 (lto_output_node): Update.
16832 (input_node): Update.
16833 (input_edge): Update.
16834 (merge_profile_summaries): Update.
16835 * lto-streamer-in.c (input_cfg): Update.
16836 * lto-streamer-out.c (output_cfg): Update.
16837 * mcf.c (create_fixup_graph): Update.
16838 (adjust_cfg_counts): Update.
16839 (sum_edge_counts): Update.
16840 * modulo-sched.c (sms_schedule): Update.
16841 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
16842 * predict.c (maybe_hot_count_p): Update.
16843 (probably_never_executed): Update.
16844 (dump_prediction): Update.
16845 (combine_predictions_for_bb): Update.
16846 (propagate_freq): Update.
16847 (handle_missing_profiles): Update.
16848 (counts_to_freqs): Update.
16849 (rebuild_frequencies): Update.
16850 (force_edge_cold): Update.
16851 * predict.h: Include profile-count.h
16852 (maybe_hot_count_p, counts_to_freqs): UPdate.
16853 * print-rtl-function.c: Do not include cfg.h
16854 * print-rtl.c: Include basic-block.h
16855 * profile-count.c: New file.
16856 * profile-count.h: New file.
16857 * profile.c (is_edge_inconsistent): Update.
16858 (correct_negative_edge_counts): Update.
16859 (is_inconsistent): Update.
16860 (set_bb_counts): Update.
16861 (read_profile_edge_counts): Update.
16862 (compute_frequency_overlap): Update.
16863 (compute_branch_probabilities): Update; Initialize and deinitialize
16865 (branch_prob): Update.
16866 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
16867 (edge_gcov_count): New.
16868 (bb_gcov_count): New.
16869 * shrink-wrap.c (try_shrink_wrapping): Update.
16870 * tracer.c (better_p): Update.
16871 * trans-mem.c (expand_transaction): Update.
16872 (ipa_tm_insert_irr_call): Update.
16873 (ipa_tm_insert_gettmclone_call): Update.
16874 * tree-call-cdce.c: Update.
16875 * tree-cfg.c (gimple_duplicate_sese_region): Update.
16876 (gimple_duplicate_sese_tail): Update.
16877 (gimple_account_profile_record): Update.
16878 (execute_fixup_cfg): Update.
16879 * tree-inline.c (copy_bb): Update.
16880 (copy_edges_for_bb): Update.
16881 (initialize_cfun): Update.
16882 (freqs_to_counts): Update.
16883 (copy_cfg_body): Update.
16884 (expand_call_inline): Update.
16885 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
16886 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
16887 (try_unroll_loop_completely): Update.
16888 (try_peel_loop): Update.
16889 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
16890 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
16891 * tree-ssa-loop-split.c (connect_loops): Update.
16892 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
16893 * tree-ssa-reassoc.c (branch_fixup): Update.
16894 * tree-ssa-tail-merge.c (replace_block_by): Update.
16895 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
16896 (compute_path_counts): Update.
16897 (update_profile): Update.
16898 (recompute_probabilities): Update.
16899 (update_joiner_offpath_counts): Update.
16900 (estimated_freqs_path): Update.
16901 (freqs_to_counts_path): Update.
16902 (clear_counts_path): Update.
16903 (ssa_fix_duplicate_block_edges): Update.
16904 (duplicate_thread_path): Update.
16905 * tree-switch-conversion.c (case_bit_test_cmp): Update.
16906 (struct switch_conv_info): Update.
16907 * tree-tailcall.c (decrease_profile): Update.
16908 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
16909 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
16910 * value-prof.c (check_counter): Update.
16911 (gimple_divmod_fixed_value): Update.
16912 (gimple_mod_pow2): Update.
16913 (gimple_mod_subtract): Update.
16914 (gimple_ic_transform): Update.
16915 (gimple_stringop_fixed_value): Update.
16916 * value-prof.h (gimple_ic): Update.
16918 2017-06-02 Carl Love <cel@us.ibm.com>
16920 * config/rs6000/rs6000-c: Add support for built-in functions
16921 vector double vec_doublee (vector signed int);
16922 vector double vec_doublee (vector unsigned int);
16923 vector double vec_doublee (vector float);
16924 vector double vec_doubleh (vector signed int);
16925 vector double vec_doubleh (vector unsigned int);
16926 vector double vec_doubleh (vector float);
16927 vector double vec_doublel (vector signed int);
16928 vector double vec_doublel (vector unsigned int);
16929 vector double vec_doublel (vector float);
16930 vector double vec_doubleo (vector signed int);
16931 vector double vec_doubleo (vector unsigned int);
16932 vector double vec_doubleo (vector float);.
16933 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
16934 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
16936 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
16937 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
16938 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
16940 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
16941 vec_doublel, vec_doubleh.
16942 * doc/extend.texi: Update the built-in documentation file for the
16943 new built-in functions.
16945 2017-06-02 David Malcolm <dmalcolm@redhat.com>
16948 * ipa-inline-analysis.c (free_growth_caches): Set
16949 edge_removal_hook_holder to NULL after removing it.
16951 2017-06-02 Sudakshina Das <sudi.das@arm.com>
16953 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
16954 comparision with zero.
16956 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
16957 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16958 for early expansion of vec_min and vec_max builtins.
16959 (builtin_function_type): Add min/max unsigned variants to those
16960 identified as having unsigned arguments.
16962 2017-06-02 Olivier Hainque <hainque@adacore.com>
16964 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
16966 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16968 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
16969 Use VALL_F16 iterator rather than VALL.
16971 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16973 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
16974 Emit CBNZ inside loop when doing a strong exchange and comparing
16975 against zero. Generate the CC flags after the loop.
16977 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
16979 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
16980 (dl_section_ref): New.
16981 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
16982 On AIX, append an expression to subtract the size of the
16983 section length to dl_section_ref.
16985 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
16987 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16988 for early expansion of vector absolute builtins.
16990 2017-06-02 Richard Biener <rguenther@suse.de>
16992 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
16993 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
16995 2017-06-02 Richard Biener <rguenther@suse.de>
16997 PR tree-optimization/80948
16998 * tree-tailcall.c (find_tail_calls): Track stmts to move in
16999 stmt order as well.
17001 2017-06-02 Richard Biener <rguenther@suse.de>
17003 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
17005 * tree-vect-stmts.c (process_use): Do not mark backedge defs
17006 for inductions as relevant.
17008 2017-06-02 Richard Biener <rguenther@suse.de>
17010 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
17011 (vectorizable_induction): ... this. Remove dead code.
17013 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
17015 * builtins. (expand_builtin_alloca): Remove second parameter and
17016 infer its value from the first parameter instead.
17017 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
17019 2017-06-02 Jakub Jelinek <jakub@redhat.com>
17021 PR rtl-optimization/80903
17022 * loop-doloop.c (add_test): Unshare sequence.
17024 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
17026 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
17028 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
17030 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
17032 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
17033 xlogue_layout::get_instance, logue_layout::xlogue_layout,
17034 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
17035 (xlogue_layout::get_stub_rtx): Make static.
17036 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
17037 (xlogue_layout::compute_stub_managed_regs): Rename to...
17038 (xlogue_layout::count_stub_managed_regs): ...this.
17039 (xlogue_layout::is_stub_managed_reg): New function.
17040 (xlogue_layout::m_stub_names): Rename to...
17041 (xlogue_layout::s_stub_names): ...this, make static.
17042 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
17043 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
17044 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
17045 xlogue_layout::s_stub_names): Instantiate statics.
17046 (stub_managed_regs): Remove.
17047 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
17048 (disable_call_ms2sysv_xlogues): Rename to...
17049 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
17050 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
17052 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
17053 change after reload_completed.
17054 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
17056 (ix86_expand_prologue): Likewise.
17057 (ix86_expand_epilogue): Likewise.
17058 (ix86_expand_split_stack_prologue): Likewise.
17059 (ix86_compute_frame_layout): Remove frame parameter ...
17060 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
17061 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
17063 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
17064 (ix86_frame): Move from here ...
17065 * config/i386/i386.h (ix86_frame): ... to here.
17066 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
17067 complete ix86_frame data structure instead. Remove some_ld_name.
17069 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
17071 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
17072 symbols that hold a DECL_VALUE_EXPR.
17074 2017-06-01 Martin Jambor <mjambor@suse.cz>
17076 PR tree-optimization/80898
17077 * tree-sra.c (process_subtree_disqualification): Removed.
17078 (disqualify_candidate): Do not acll
17079 process_subtree_disqualification.
17080 (subtree_mark_written_and_enqueue): New function.
17081 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
17082 RHS has been disqualified and re-queue LHS if necessary. Apart
17083 from that, ignore disqualified RHS.
17085 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17087 * config/s390/s390.c (s390_emit_epilogue): Disable early return
17088 address fetch for z10 or later.
17090 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17092 * config/arc/arc.md (tst_movb): Add guard when splitting.
17094 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17096 * config/arc/arc.c (arc_can_eliminate): Test against
17097 arc_frame_pointer_needed.
17099 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17101 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
17102 to prevent store reordering.
17103 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
17104 (type): Add block type.
17105 (stack_tie): Define special instruction to be used in
17108 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17110 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
17111 constraint. It is not valid for the pattern.
17112 (noncommutative_binary_comparison): Likewise.
17114 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17116 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
17119 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17121 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
17122 be used by the reg-alloc.
17124 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17126 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
17127 reg-alloc when having mul64 or mul32x16 instructions.
17128 (mulsidi3): Likewise.
17129 (umulsidi3): Likewise.
17130 (mulsi32x16): New pattern.
17131 (mulsi64): Likewise.
17132 (mulsidi64): Likewise.
17133 (umulsidi64): Likewise.
17134 (MUL32x16_REG): Define.
17135 (mul64_600): Use MUL32x16_REG.
17136 (mac64_600): Likewise.
17137 (umul64_600): Likewise.
17138 (umac64_600): Likewise.
17140 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17142 * config/arc/arc.md (mulsi3_700): Make it commutative.
17144 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
17146 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
17148 (*sign_extendsidi2_insn): Likewise for movstosw.
17150 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
17152 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
17153 the type of the input discriminant value. Convert the
17154 discriminant value of signedness vary.
17156 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
17158 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
17159 Add to -Wall section.
17161 2017-06-01 Richard Biener <rguenther@suse.de>
17163 PR middle-end/66313
17164 * fold-const.c (fold_plusminus_mult_expr): If the factored
17165 factor may be zero use a wrapping type for the inner operation.
17166 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
17167 and handle moved defs.
17168 (process_assignment): Properly guard the unary op case. Return a
17169 tri-state indicating that moving the stmt before the call may allow
17170 to continue. Pass through to_move.
17171 (find_tail_calls): Handle moving unrelated defs before
17174 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
17177 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
17178 splitter result in the canonical way.
17180 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
17182 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
17183 also for 32bit target. Update insn attributes.
17184 (zero-extendsidi2 splitter): Allow all registers for operand 1.
17186 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
17188 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
17189 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
17190 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
17191 (_mm_maskz_min_ss): New intrinsics.
17193 2017-05-31 Martin Liska <mliska@suse.cz>
17195 * tree-vect-loop.c (vect_create_epilog_for_reduction):
17196 Change comment style to one we normally use.
17197 (vectorizable_reduction): Likewise.
17198 (vectorizable_induction): Likewise.
17199 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
17200 (vectorizable_call): Likewise.
17201 (vectorizable_simd_clone_call): Likewise.
17202 (vectorizable_conversion): Likewise.
17203 (vectorizable_assignment): Likewise.
17204 (vectorizable_shift): Likewise.
17205 (vectorizable_operation): Likewise.
17206 (vectorizable_store): Likewise.
17207 (vectorizable_load): Likewise.
17208 * tree-vectorizer.h: Likewise.
17210 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
17212 * passes.c (emergency_dump_function): New.
17213 * tree-pass.h (emergency_dump_function): Declare.
17214 * plugin.c (plugins_internal_error_function): Remove.
17215 * plugin.h (plugins_internal_error_function): Remove declaration.
17216 * toplev.c (internal_error_function): New static function. Use it...
17217 (general_init): ...here.
17219 2017-05-31 Graham Markall <graham.markall@embecosm.com>
17221 * config/arc/arc.c (arc_print_operand): Handle constant operands.
17222 (arc_rtx_costs): Add costs for new patterns.
17223 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
17224 * config/arc/predicates.md: Add _1_2_3_operand predicate.
17226 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
17228 * tree-ssa-strlen.c (get_next_strinfo): New function.
17229 (get_stridx_plus_constant): Use it.
17230 (zero_length_string): Likewise.
17231 (adjust_related_strinfos): Likewise.
17232 (adjust_last_stmt): Likewise.
17234 2017-05-31 Richard Biener <rguenther@suse.de>
17237 * config/i386/i386.c (ix86_expand_builtin): Remove assert
17238 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
17240 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
17242 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
17243 loop_vinfo argument and use of dependence distance vectors.
17244 Check instead whether the two references differ only in their
17245 initial value and assume that they have the same alignment if the
17246 difference is a multiple of the vector alignment.
17247 (vect_analyze_data_refs_alignment): Update call accordingly.
17249 2017-05-31 Martin Liska <mliska@suse.cz>
17252 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
17254 2017-05-31 Bin Cheng <bin.cheng@arm.com>
17256 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
17257 (create_intersect_range_checks): Move from ...
17258 * tree-data-ref.c (create_intersect_range_checks_index)
17259 (create_intersect_range_checks): ... to here.
17260 (create_runtime_alias_checks): New function factored from ...
17261 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
17262 here. Call above function.
17263 * tree-data-ref.h (create_runtime_alias_checks): New function.
17265 2017-05-31 Bin Cheng <bin.cheng@arm.com>
17267 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
17268 segment length for dr_b and compute it in wide_int.
17270 2017-05-31 Richard Biener <rguenther@suse.de>
17272 PR tree-optimization/80906
17273 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
17274 and pass through iv_map.
17275 (copy_bb_and_scalar_dependences): Adjust.
17276 (translate_pending_phi_nodes): Likewise.
17277 (copy_loop_close_phi_args): Handle code-generating IVs instead
17280 2017-05-30 David Malcolm <dmalcolm@redhat.com>
17282 * diagnostic-color.c (color_dict): Add "type-diff".
17283 (parse_gcc_colors): Update comment.
17284 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
17285 -fdiagnostics-show-template-tree and -fno-elide-type.
17286 (GCC_COLORS): Add type-diff to example.
17288 (-fdiagnostics-show-template-tree): New.
17289 (-fno-elide-type): New.
17290 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
17291 the pp_format_decoder callback. Call any m_format_postprocessor's
17293 (pretty_printer::pretty_printer): Initialize
17294 m_format_postprocessor.
17295 (pretty_printer::~pretty_printer): Delete any
17296 m_format_postprocessor.
17297 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
17298 (class format_postprocessor): New class.
17299 (struct pretty_printer::format_decoder): Document the new parameters.
17300 (struct pretty_printer::m_format_postprocessor): New field.
17301 * tree-diagnostic.c (default_tree_printer): Update for new
17302 bool and const char ** params.
17303 * tree-diagnostic.h (default_tree_printer): Likewise.
17305 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
17307 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
17308 (lwa_operand): Delete rs6000_gen_cell_microcode test.
17309 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
17310 rs6000_gen_cell_microcode code.
17311 (rs6000_final_prescan_insn): Delete.
17312 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
17313 "warn-cell-microcode" entries.
17314 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
17315 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
17316 throughout. Change cc_reg_not_micro_cr0_operand to
17317 cc_reg_not_cr0_operand throughout.
17318 (*extendhi<mode>2_noload): Delete.
17319 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
17320 (mwarn-cell-microcode): Delete.
17321 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
17322 -mgen-cell-microcode and -mwarn-cell-microcode.
17324 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
17327 * config/i386/constraints.md (Yd): New constraint.
17329 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
17330 and (?Yd, r) alternatives. Update insn attributes.
17331 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
17332 and (?*Yd, r) alternatives. Update insn attributes.
17333 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
17335 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
17337 * gimplify.c (gimplify_modify_expr): Don't create a
17338 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
17341 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
17343 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
17345 2017-05-30 Richard Biener <rguenther@suse.de>
17347 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
17348 and reduc_def fields.
17349 (STMT_VINFO_REDUC_TYPE): New define.
17350 (STMT_VINFO_REDUC_DEF): Likewise.
17351 (vect_force_simple_reduction): Adjust prototype.
17352 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
17353 (vect_is_simple_reduction): Remove check_reduction argument.
17354 (vect_force_simple_reduction): Adjust and set
17355 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
17356 (vectorizable_reduction): Do not re-do reduction analysis
17357 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
17358 * tree-parloops.c (gather_scalar_reductions): Adjust.
17360 2017-05-30 Richard Biener <rguenther@suse.de>
17362 PR middle-end/80901
17363 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
17366 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
17368 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
17369 Introduce unknown_misalignment parameter and remove vf.
17370 (vect_peeling_hash_get_lowest_cost):
17371 Pass unknown_misalignment parameter.
17372 (vect_enhance_data_refs_alignment):
17373 Fix unsupportable data ref treatment.
17375 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17377 * tree-vect-data-refs.c (vect_get_data_access_cost):
17378 Workaround for SLP handling.
17379 (vect_enhance_data_refs_alignment):
17380 Compute costs for doing no peeling at all, compare to the best
17381 peeling costs so far and avoid peeling if cheaper.
17383 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17385 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
17386 Return peeling info and set costs to zero for unlimited cost
17388 (vect_enhance_data_refs_alignment): Also inspect all datarefs
17389 with unknown misalignment. Compute and costs for unknown
17390 misalignment, compare them to the costs for known misalignment
17391 and choose the cheapest for peeling.
17393 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17395 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
17396 (vect_get_peeling_costs_all_drs): Create function.
17397 (vect_peeling_hash_get_lowest_cost):
17398 Use vect_get_peeling_costs_all_drs.
17399 (vect_peeling_supportable): Create function.
17400 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
17402 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17404 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
17405 DR_HAS_NEGATIVE_STEP.
17406 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
17407 (vect_enhance_data_refs_alignment): Use.
17408 (vect_duplicate_ssa_name_ptr_info): Use.
17409 * tree-vectorizer.h (dr_misalignment): Use.
17410 (known_alignment_for_access_p): Use.
17412 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17415 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
17417 (has_section_name): New function.
17419 2017-05-30 Martin Liska <mliska@suse.cz>
17422 * auto-profile.c (get_function_decl_from_block): Fix
17425 2017-05-30 Richard Biener <rguenther@suse.de>
17427 PR middle-end/80876
17428 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
17430 2017-05-30 Martin Liska <mliska@suse.cz>
17432 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
17433 * dumpfile.h (struct dump_file_info): Remove ctors.
17435 2017-05-30 Martin Liska <mliska@suse.cz>
17437 * predict.def: Fix GNU coding style.
17439 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
17441 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
17442 Mark 'to' argument with ATTRIBUTE_UNUSED.
17444 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
17446 * config/xtensa/xtensa.c (xtensa_emit_call): Use
17447 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
17448 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
17451 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
17453 * doc/install.texi (Options specification): Restore entry of
17454 --enable-sjlj-exceptions.
17456 2017-05-27 Michael Eager <eager@eagercon.com>
17459 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
17461 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
17463 * config/microblaze/microblaze.h
17464 (FIXED_REGISTERS): Update in macro.
17465 (CALL_USED_REGISTERS): Update in macro.
17467 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
17469 * doc/install.texi: Add links to macOS binary distributions.
17471 2017-05-27 Jakub Jelinek <jakub@redhat.com>
17475 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17477 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
17479 2017-05-26 Martin Liska <mliska@suse.cz>
17481 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
17483 2017-05-26 Martin Liska <mliska@suse.cz>
17485 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
17486 always leading ';; '.
17487 (dump_bb_info): Likewise.
17488 (brief_dump_cfg): Likewise.
17489 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
17490 * dumpfile.c: Remove usage of TDF_VERBOSE.
17491 * dumpfile.h (enum dump_kind): Likewise.
17492 (dump_gimple_bb_header): Do not use TDF_COMMENT.
17493 * print-tree.c (debug_verbose): Remove.
17494 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
17495 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
17496 * tree-diagnostic.c (default_tree_printer): Replace
17497 TDF_DIAGNOSTIC with TDF_SLIM.
17499 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17501 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
17502 in parameter loop, rather than loop_vinfo.
17503 (create_intersect_range_checks): Ditto.
17504 (vect_create_cond_for_alias_checks): Update call to above functions.
17506 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17508 PR tree-optimization/80815
17509 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
17510 for merging runtime alias checks. Handle negative DR_STEPs.
17512 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17514 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
17516 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
17517 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
17518 out code pruning runtime alias checks.
17519 * tree-data-ref.c (prune_runtime_alias_test_list): New function
17520 factored out from above.
17521 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
17523 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
17525 (prune_runtime_alias_test_list): New decalaration.
17527 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17529 * tree-vect-data-refs.c (compare_tree): Rename and move ...
17530 * tree-data-ref.c (data_ref_compare_tree): ... to here.
17531 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
17532 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
17533 (operator==, comp_dr_with_seg_len_pair): Ditto.
17534 (vect_prune_runtime_alias_test_list): Ditto.
17536 2017-05-26 Martin Liska <mliska@suse.cz>
17539 * params.def: Bound partial-inlining-entry-probability param.
17541 2017-05-26 Marek Polacek <polacek@redhat.com>
17544 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
17547 2017-05-26 Richard Biener <rguenther@suse.de>
17549 PR tree-optimization/80842
17550 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
17553 2017-05-26 Richard Biener <rguenther@suse.de>
17555 PR tree-optimization/80844
17556 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
17558 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
17560 * doc/md.texi (Machine Constraints): Update x86 family
17561 machine constraints section to match 'config/i386/constraints.md'.
17563 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
17565 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
17567 2017-05-25 Nathan Sidwell <nathan@acm.org>
17569 * doc/invoke.texi (--enable-languages): Update documentation.
17571 2017-05-25 Martin Liska <mliska@suse.cz>
17573 * dumpfile.c: Add TDF_FOLDING.
17574 * dumpfile.h (enum dump_kind): Likewise.
17575 * genmatch.c (dt_simplify::gen_1): Use it.
17577 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17579 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
17581 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17583 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
17584 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
17586 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17588 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
17589 * match.pd (X == C): Rewrite it here.
17590 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
17591 with_certain_nonzero_bits2): New predicates.
17592 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
17594 2017-05-24 Nathan Sidwell <nathan@acm.org>
17596 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
17599 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
17602 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
17604 * config/powerpcspe: New port. Files are copied from the rs6000
17605 port, with "rs6000" in filenames replaced by "powerpcspe".
17607 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
17609 PR rtl-optimization/80754
17610 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
17612 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
17614 * config/sparc/sparc.md (length): Return the correct value for -mflat
17615 sibcalls to match output_sibcall.
17617 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
17621 * config/rs6000/rs6000.c (struct machine_function): Add new field
17623 (rs6000_get_separate_components): Init that field, use it.
17624 (rs6000_components_for_bb): Use the field.
17626 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
17628 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
17630 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
17632 PR middle-end/80823
17633 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
17635 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17638 * config/s390/s390.c (s390_check_qrst_address): Check incoming
17639 address against address_operand predicate.
17640 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
17642 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
17644 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
17645 parameters passed indirectly.
17647 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
17649 * config/i386/i386.md (*movdi_internal): Remove SSE4
17650 alternative 18 (?r, *v). Update insn attributes.
17651 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
17652 Update insn attributes.
17653 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
17654 Update insn attributes.
17655 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
17656 alternative 1 (r, v). Remove isa attribute.
17657 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
17658 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
17659 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
17661 2017-05-23 Tom de Vries <tom@codesourcery.com>
17663 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
17666 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17668 * cgraphunit.c (symbol_table::process_new_functions): Update.
17669 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
17670 (inline_generate_summary): Rename to ...
17671 (ipa_fn_summary_generate): ... this one.
17672 (inline_read_summary): Rename to ...
17673 (ipa_fn_summary_read): ... this one.
17674 (inline_write_summary): Rename to ...
17675 (ipa_fn_summary_write): ... this one.
17676 (inline_free_summary): Rename to ...
17677 (ipa_free_fn_summary): ... this one.
17678 (pass_data_local_fn_summary, pass_local_fn_summary,
17679 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
17680 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
17681 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
17682 make_pass_ipa_fn_summary): New.
17683 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
17684 inline_write_summary, inline_free_summary): Remove.
17685 (ipa_free_fn_summary) : New.
17686 * ipa-inline.c (ipa_inline): Update.
17687 (pass_ipa_inline): Do not generate summaries.
17688 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
17690 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
17691 and add pass_ipa_fn_summary.
17692 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
17694 (make_pass_inline_parameters): Remove.
17696 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
17698 * omp-low.c (struct omp_context): Remove "default_kind" member.
17701 * omp-offload.c (execute_oacc_device_lower): Remove the
17702 parallelism dimensions function attributes for unparallelized
17703 OpenACC kernels constructs.
17705 2017-05-23 Martin Liska <mliska@suse.cz>
17707 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
17709 (cgraph_edge::make_speculative): Likewise.
17710 (cgraph_edge::resolve_speculation): Likewise.
17711 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
17712 (cgraph_node::dump): Likewise.
17713 * cgraph.h: Likewise.
17714 * cgraphunit.c (analyze_functions): Likewise.
17715 (symbol_table::compile): Likewise.
17716 * ipa-cp.c (print_all_lattices): Likewise.
17717 (determine_versionability): Likewise.
17718 (initialize_node_lattices): Likewise.
17719 (ipcp_verify_propagated_values): Likewise.
17720 (estimate_local_effects): Likewise.
17721 (update_profiling_info): Likewise.
17722 (create_specialized_node): Likewise.
17723 (perhaps_add_new_callers): Likewise.
17724 (decide_about_value): Likewise.
17725 (decide_whether_version_node): Likewise.
17726 (identify_dead_nodes): Likewise.
17727 (ipcp_store_bits_results): Likewise.
17728 * ipa-devirt.c (dump_targets): Likewise.
17729 (ipa_devirt): Likewise.
17730 * ipa-icf.c (sem_item::dump): Likewise.
17731 (sem_function::equals): Likewise.
17732 (sem_variable::equals): Likewise.
17733 (sem_item_optimizer::read_section): Likewise.
17734 (sem_item_optimizer::execute): Likewise.
17735 (congruence_class::dump): Likewise.
17736 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
17737 (dump_inline_summary): Likewise.
17738 (estimate_node_size_and_time): Likewise.
17739 (inline_analyze_function): Likewise.
17740 * ipa-inline-transform.c (inline_call): Likewise.
17741 * ipa-inline.c (report_inline_failed_reason): Likewise.
17742 (want_early_inline_function_p): Likewise.
17743 (edge_badness): Likewise.
17744 (update_edge_key): Likewise.
17745 (inline_small_functions): Likewise.
17746 * ipa-profile.c (ipa_profile): Likewise.
17747 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
17748 (ipa_make_edge_direct_to_target): Likewise.
17749 (remove_described_reference): Likewise.
17750 (ipa_impossible_devirt_target): Likewise.
17751 (propagate_controlled_uses): Likewise.
17752 (ipa_print_node_params): Likewise.
17753 (ipcp_transform_function): Likewise.
17754 * ipa-pure-const.c (pure_const_read_summary): Likewise.
17755 (propagate_pure_const): Likewise.
17756 * ipa-reference.c (generate_summary): Likewise.
17757 (read_write_all_from_decl): Likewise.
17758 (propagate): Likewise.
17759 (ipa_reference_read_optimization_summary): Likewise.
17760 * ipa-utils.c (ipa_merge_profiles): Likewise.
17761 * ipa.c (walk_polymorphic_call_targets): Likewise.
17762 (symbol_table::remove_unreachable_nodes): Likewise.
17763 (ipa_single_use): Likewise.
17764 * passes.c (execute_todo): Likewise.
17765 * predict.c (drop_profile): Likewise.
17766 * symtab.c (symtab_node::get_dump_name): New function.
17767 (symtab_node::dump_name): Likewise.
17768 (symtab_node::dump_asm_name): Likewise.
17769 (symtab_node::dump_references): Likewise.
17770 (symtab_node::dump_referring): Likewise.
17771 (symtab_node::dump_base): Likewise.
17772 (symtab_node::debug_symtab): Likewise.
17773 * tree-sra.c (convert_callers_for_node): Likewise.
17774 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
17775 * value-prof.c (init_node_map): Likewise.
17777 2017-05-23 Martin Liska <mliska@suse.cz>
17779 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
17780 and symtab_node::debug_symtab to symbol_table::debug.
17781 * cgraphunit.c (analyze_functions): Use the renamed function.
17782 (symbol_table::compile): Likewise.
17783 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
17784 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
17785 * passes.c (execute_todo): Likewise.
17786 * symtab.c (symbol_table::dump): New function.
17787 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
17789 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17791 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
17792 that nonconst implies exec.
17794 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17796 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
17797 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
17798 (inline_edge_summary_vec): Turn into ...
17799 (ipa_call_summaries): ... this one.
17800 (redirect_to_unreachable, edge_set_predicate,
17801 evaluate_properties_for_edge, inline_summary_alloc,
17802 reset_ipa_call_summary, reset_inline_summary,
17803 inline_summary_t::duplicate): Update.
17804 (inline_edge_duplication_hook): Turn to ...
17805 (ipa_call_summary_t::duplicate): ... this one.
17806 (inline_edge_removal_hook): Turn to ...
17807 (ipa_call_summary_t::remove): ... this one.
17808 (dump_inline_edge_summary): Turn to ...
17809 (dump_ipa_call_summary): ... this one.
17810 (estimate_function_body_sizes): Update.
17811 (inline_update_callee_summaries): Update.
17812 (remap_edge_change_prob): Update.
17813 (remap_edge_summaries): Update.
17814 (inline_merge_summary): Update.
17815 (do_estimate_edge_time): Update.
17816 (inline_generate_summary): Update.
17817 (inline_read_section): Update.
17818 (inline_read_summary): Update.
17819 (inline_free_summary): Update.
17820 * ipa-inline.c (can_inline_edge_p): Update.
17821 (compute_inlined_call_time): Update.
17822 (want_inline_small_function_p): Update.
17823 (edge_badness): Update.
17824 (early_inliner): Update.
17825 * ipa-inline.h (inline_edge_summary): Turn to ...
17826 (ipa_call_summary): ... this one.
17827 (ipa_call_summary_t): New class.
17828 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
17829 (ipa_call_summaries): New.
17830 (inline_edge_summary): Remove.
17831 (estimate_edge_growth): Update.
17832 * ipa-profile.c (ipa_propagate_frequency_1): Update.
17833 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
17834 * ipa-split.c (execute_split_functions): Update.
17835 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
17837 2017-05-23 Tom de Vries <tom@codesourcery.com>
17839 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
17840 attributes): Document rdrand effective target.
17842 2017-05-23 Tom de Vries <tom@codesourcery.com>
17844 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
17845 attributes): Sort alphabetically.
17847 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
17849 * config/avr/genmultilib.awk: Use gsub instead of gensub.
17851 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
17854 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
17855 V2DF/V2DI splat into two separate patterns, one that handles
17856 registers, and the other that only handles memory. Drop support
17857 for splatting from a GPR on ISA 2.07 and then splitting the
17858 splat into direct move and splat.
17859 (vsx_splat_<mode>_reg): Likewise.
17860 (vsx_splat_<mode>_mem): Likewise.
17862 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
17864 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
17866 2017-05-22 Jakub Jelinek <jakub@redhat.com>
17868 PR middle-end/80809
17869 * omp-low.c (finish_taskreg_remap): New function.
17870 (finish_taskreg_scan): If unit size of ctx->record_type
17871 is non-constant, unshare the size expression and replace
17872 decls in it with possible outer var refs.
17874 PR middle-end/80809
17875 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
17876 GOVD_SHARED rather than GOVD_PRIVATE with it.
17877 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
17878 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
17880 PR middle-end/80853
17881 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
17882 as last argument to build_outer_var_ref for pointer bases of array
17883 section reductions.
17885 2017-05-19 Martin Sebor <msebor@redhat.com>
17887 * print-tree.c (print_node): Print DECL_READ_P flag.
17889 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17891 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
17892 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
17893 * cgraph.c: Likewise.
17894 * cgraphunit.c: Likewise.
17895 * gengtype.c: Likewise.
17896 * ipa-cp.c: Likewise.
17897 * ipa-devirt.c: Likewise.
17898 * ipa-icf.c: Likewise.
17899 * ipa-predicate.c: Likewise.
17900 * ipa-profile.c: Likewise.
17901 * ipa-prop.c: Likewise.
17902 * ipa-split.c: Likewise.
17904 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
17905 edge_predicate_pool, dump_inline_hints,
17906 inline_summary::account_size_time, redirect_to_unreachable,
17907 edge_set_predicate, set_hint_predicate,
17908 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
17909 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
17910 inline_summary_t::remove, remap_hint_predicate_after_duplication,
17911 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
17912 ipa_call_summary_t::remove, initialize_growth_caches,
17913 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
17914 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
17915 mark_modified, unmodified_parm_1, unmodified_parm,
17916 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
17917 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
17918 compute_bb_predicates, will_be_nonconstant_expr_predicate,
17919 will_be_nonconstant_predicate, record_modified_bb_info,
17920 get_minimal_bb, record_modified, param_change_prob,
17921 phi_result_unknown_predicate, predicate_for_phi_result,
17922 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
17923 estimate_function_body_sizes, compute_inline_parameters,
17924 compute_inline_parameters_for_curren, pass_data_inline_parameters,
17925 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
17926 inline_update_callee_summaries, remap_edge_change_prob,
17927 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
17928 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
17929 inline_analyze_function, inline_summary_t::insert,
17930 inline_generate_summary, read_ipa_call_summary, inline_read_section,
17931 inline_read_summary, write_ipa_call_summary, inline_write_summary,
17932 inline_free_summary): Move to ipa-fnsummary.h
17933 (predicate_t): Remove.
17934 * ipa-fnsummary.c: New file.
17935 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
17936 (enum inline_hints_vals, inline_hints, agg_position_info,
17937 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
17938 inline_summaries, ipa_call_summary, ipa_call_summary_t,
17939 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
17940 dump_inline_summary, dump_inline_hints, inline_generate_summary,
17941 inline_read_summary, inline_write_summary, inline_free_summary,
17942 inline_analyze_function, initialize_inline_failed,
17943 inline_merge_summary, inline_update_overall_summary,
17944 compute_inline_parameters): Move to ipa-fnsummary.h
17945 * ipa-fnsummary.h: New file.
17946 * ipa-inline-transform.h: Include ipa-inline.h.
17947 * ipa-inline.c: LIkewise.
17949 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17951 * ipa-inline.c (edge_badness): Use inlined_time instead of
17952 inline_summaries->get.
17954 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17956 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
17958 2017-05-22 Nathan Sidwell <nathan@acm.org>
17960 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
17961 (fdump-lang): Document 'raw' option.
17962 * dumpfile.h (TDI_tu): Delete.
17963 * dumpfile.c (dump_files): Remove translation-unit.
17964 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
17966 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
17968 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
17969 command option from $(AWK) call.
17970 * config/avr/genmultilib.awk: Simplify and rewrite so that it
17971 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
17972 [FORMAT]: Remove handling of variable.
17973 * config/avr/t-multilib: Regenerate.
17975 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17977 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
17979 (dump_inline_summary): Do not print self_time.
17980 (estimate_function_body_sizes): Do not set self_time.
17981 (compute_inline_parameters): Likewise.
17982 (inline_read_section, inline_write_summary): Do not stream self_time.
17983 * ipa-inline.h (inline_summary): Drop self_time.
17985 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17987 * ipa-inline-analysis.c (account_size_time): Rename to ...
17988 (inline_summary::account_size_time): ... this one.
17989 (reset_ipa_call_summary): Turn to ...
17990 (ipa_call_summary::reset): ... this one.
17991 (reset_inline_summary): Turn to ...
17992 (inline_summary::reset): ... this one.
17993 (inline_summary_t::remove): Update.
17994 (inline_summary_t::duplicate): Update.
17995 (ipa_call_summary_t::remove): Update.
17996 (dump_inline_summary): Update.
17997 (estimate_function_body_sizes): Update.
17998 (compute_inline_parameters): Update.
17999 (estimate_node_size_and_time): Update.
18000 (inline_merge_summary): Update.
18001 (inline_update_overall_summary): Update.
18002 (inline_read_section): Update.
18003 (inline_write_summary): Update.
18004 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
18005 add account_size_time and reset member functions.
18006 (ipa_call_summary): Add reset function.
18007 * ipa-predicate.h (predicate::operator &): Constify.
18009 2017-05-22 Richard Biener <rguenther@suse.de>
18011 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
18013 2017-05-19 Jason Merrill <jason@redhat.com>
18015 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
18017 2017-05-19 Marek Polacek <polacek@redhat.com>
18020 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
18021 TYPE_OVERFLOW_WRAPS checks.
18023 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
18025 * tree-core.h (enum omp_clause_default_kind): Add
18026 "OMP_CLAUSE_DEFAULT_PRESENT".
18027 * tree-pretty-print.c (dump_omp_clause): Handle it.
18028 * gimplify.c (enum gimplify_omp_var_data): Add
18029 "GOVD_MAP_FORCE_PRESENT".
18030 (gimplify_adjust_omp_clauses_1): Map it to
18031 "GOMP_MAP_FORCE_PRESENT".
18032 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
18034 * gimplify.c (oacc_default_clause): Clarify.
18036 2017-05-19 Nathan Sidwell <nathan@acm.org>
18038 LANG_HOOK_REGISTER_DUMPS
18039 * toplev.c (general_init): Call register dump lang hook.
18040 * doc/invoke.texi: Document -fdump-lang option family.
18041 * dumpfile.c (dump_files): Remove class dump here.
18042 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
18043 * dumpfile.h (tree_dump_index): Remove TDI_class.
18044 * langhooks-def.h (lhd_register_dumps): Declare.
18045 (LANG_HOOKS_REGISTER_DUMPS): Define.
18046 (LANG_HOOKS_INITIALIZER): Add it.
18047 * langhooks.c (lhd_register_dumps): Define.
18048 * langhooks.h (struct lang_hooks): Add register_dumps.
18050 2017-05-19 Nathan Sidwell <nathan@acm.org>
18052 * context.h (context::set_passes): New.
18053 * context.c (context::context): Do not create pass manager.
18054 * toplev.c (general_init): Create pass manager here.
18056 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
18058 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
18059 use this splitter if two add or or instructions would also work for
18060 the constant we want to generate.
18062 2017-05-19 Richard Biener <rguenther@suse.de>
18065 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
18066 predicate evaluation.
18068 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
18070 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
18072 * ipa-inline.c (want_inline_small_function_p): Do not cast to
18075 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
18077 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
18078 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
18079 (inline_edge_summary_vec): Turn into ...
18080 (ipa_call_summaries): ... this one.
18081 (redirect_to_unreachable, edge_set_predicate,
18082 evaluate_properties_for_edge, inline_summary_alloc,
18083 reset_ipa_call_summary, reset_inline_summary,
18084 inline_summary_t::duplicate): Update.
18085 (inline_edge_duplication_hook): Turn to ...
18086 (ipa_call_summary_t::duplicate): ... this one.
18087 (inline_edge_removal_hook): Turn to ...
18088 (ipa_call_summary_t::remove): ... this one.
18089 (dump_inline_edge_summary): Turn to ...
18090 (dump_ipa_call_summary): ... this one.
18091 (estimate_function_body_sizes): Update.
18092 (inline_update_callee_summaries): Update.
18093 (remap_edge_change_prob): Update.
18094 (remap_edge_summaries): Update.
18095 (inline_merge_summary): Update.
18096 (do_estimate_edge_time): Update.
18097 (inline_generate_summary): Update.
18098 (inline_read_section): Update.
18099 (inline_read_summary): Update.
18100 (inline_free_summary): Update.
18101 * ipa-inline.c (can_inline_edge_p): Update.
18102 (compute_inlined_call_time): Update.
18103 (want_inline_small_function_p): Update.
18104 (edge_badness): Update.
18105 (early_inliner): Update.
18106 * ipa-inline.h (inline_edge_summary): Turn to ...
18107 (ipa_call_summary): ... this one.
18108 (ipa_call_summary_t): New class.
18109 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
18110 (ipa_call_summaries): New.
18111 (inline_edge_summary): Remove.
18112 (estimate_edge_growth): Update.
18113 * ipa-profile.c (ipa_propagate_frequency_1): Update.
18114 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
18115 * ipa-split.c (execute_split_functions): Update.
18116 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
18118 2017-05-19 Richard Biener <rguenther@suse.de>
18120 PR middle-end/80764
18121 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
18123 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
18125 * config/rs6000/rs6000.c (struct machine_function): Add field
18126 fpr_is_wrapped_separately.
18127 (rs6000_get_separate_components): Use 64 components. Handle the
18128 new FPR components.
18129 (rs6000_components_for_bb): Handle the FPR components.
18130 (rs6000_emit_prologue_components): Handle the FPR components.
18131 (rs6000_emit_epilogue_components): Handle the FPR components.
18132 (rs6000_set_handled_components): Handle the FPR components.
18133 (rs6000_emit_prologue): Don't output prologue code for those FPRs
18134 that are already separately shrink-wrapped.
18135 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
18136 that are already separately shrink-wrapped.
18138 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
18141 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
18144 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
18145 (define_peephole2 for Altivec d-form load): Add peepholes to catch
18146 cases where the register allocator uses a move and an offsettable
18147 memory operation to/from a FPR register on ISA 2.06/2.07.
18148 (define_peephole2 for Altivec d-form store): Likewise.
18150 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
18153 * config/i386/mmx.md (*mov<mode>_internal): Enable
18154 alternatives 11, 12, 13 and 14 also for 32bit targets.
18155 Remove alternatives 15, 16, 17 and 18.
18156 * config/i386/sse.md (vec_concatv2di): Change
18157 alternative (!x, *y) to (x, ?!*Yn).
18159 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
18161 * dumpfile.h (enum dump_kind): Remove stray comma.
18163 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
18165 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
18166 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
18167 predicate::num_conditions
18168 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
18169 (CHANGED): turn into predicate::changed.
18170 (agg_position_info): Move to ipa-predicate.h
18171 (add_condition, predicate::add_clause, predicate::operator &=,
18172 predicate::or_with, predicate::evaluate, predicate::probability,
18173 dump_condition, dump_clause, predicate::dump,
18174 predicate::remap_after_duplication, predicate::remap_after_inlining,
18175 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
18176 (evaluate_conditions_for_known_args): Update.
18177 (set_cond_stmt_execution_predicate): Update.
18178 * ipa-inline.h: Include ipa-predicate.h
18179 (condition, inline_param_summary, conditions, agg_position_info,
18180 predicate): Move to ipa-predicate.h
18181 * ipa-predicate.c: New file.
18182 * ipa-predicate.h: New file.
18184 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
18186 * final.c (leaf_function_p): Check we are not in a sequence.
18188 2017-05-18 Martin Liska <mliska@suse.cz>
18190 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
18191 * dumpfile.c (dump_register): Use new enum dump_kind.
18192 (get_dump_file_name): Likewise.
18193 (dump_enable_all): Likewise.
18194 (dump_switch_p_1): Likewise.
18195 (enable_rtl_dump_file): Remove usage of TDF_RTL.
18196 * dumpfile.h (enum dump_kind): New enum type.
18197 (struct dump_file_info): Create constructor and
18198 format fields and comments.
18199 * passes.c (pass_manager::register_one_dump_file):
18201 * statistics.c (statistics_early_init): Likewise.
18202 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
18203 TDF_TREE with TDF_SLIM.
18204 (gather_memory_references_ref): Likewise.
18206 2017-05-18 Martin Liska <mliska@suse.cz>
18208 * vec.h (struct vnull): Use it.
18210 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
18212 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
18213 (true_predicate, false_predicate, true_predicate_p,
18214 false_predicate_p): Remove.
18215 (single_cond_predicate, not_inlined_predicate): Turn to member function
18217 (add_condition): Update.
18218 (add_clause): Turn to...
18219 (predicate::add_clause): ... this one; update; allow passing NULL
18221 (and_predicates): Turn to ...
18222 (predicate::operator &=): ... this one.
18223 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
18224 (or_predicates): Turn to ...
18225 (predicate::or_with): ... this one.
18226 (evaluate_predicate): Turn to ...
18227 (predicate::evaluate): ... this one.
18228 (predicate_probability): Turn to ...
18229 (predicate::probability): ... this one.
18230 (dump_condition): Update.
18231 (dump_predicate): Turn to ...
18232 (predicate::dump): ... this one.
18233 (account_size_time): Update.
18234 (edge_set_predicate): Update.
18235 (set_hint_predicate): UPdate.
18236 (evaluate_conditions_for_known_args): Update.
18237 (evaluate_properties_for_edge): Update.
18238 (remap_predicate_after_duplication): Turn to...
18239 (predicate::remap_after_duplication): ... this one.
18240 (remap_hint_predicate_after_duplication): Update.
18241 (inline_summary_t::duplicate): UPdate.
18242 (dump_inline_edge_summary): Update.
18243 (dump_inline_summary): Update.
18244 (set_cond_stmt_execution_predicate): Update.
18245 (set_switch_stmt_execution_predicate): Update.
18246 (compute_bb_predicates): Update.
18247 (will_be_nonconstant_expr_predicate): Update.
18248 (will_be_nonconstant_predicate): Update.
18249 (phi_result_unknown_predicate): Update.
18250 (predicate_for_phi_result): Update.
18251 (array_index_predicate): Update.
18252 (estimate_function_body_sizes): Update.
18253 (estimate_node_size_and_time): Update.
18254 (estimate_ipcp_clone_size_and_time): Update.
18255 (remap_predicate): Rename to ...
18256 (predicate::remap_after_inlining): ... this one.
18257 (remap_hint_predicate): Update.
18258 (inline_merge_summary): Update.
18259 (inline_update_overall_summary): Update.
18260 (estimate_size_after_inlining): Update.
18261 (read_predicate): Rename to ...
18262 (predicate::stream_in): ... this one.
18263 (read_inline_edge_summary): Update.
18264 (write_predicate): Rename to ...
18265 (predicate::stream_out): ... this one.
18266 (write_inline_edge_summary): Update.
18267 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
18268 (clause_t): Turn to uint32_t
18269 (predicate): Turn to class; implement constructor and operators
18271 (size_time_entry): Update.
18272 (inline_summary): Update.
18273 (inline_edge_summary): Update.
18275 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
18277 * fold-const.c (fold_binary_loc): Move transformation...
18278 * match.pd (C - X CMP X): ... here.
18280 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
18282 * config/sparc/sparc.c (sparc_option_override): Set function
18283 alignment for -mcpu=niagara7 to 64 to match the I$ line.
18284 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
18286 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
18288 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
18290 2017-05-18 Marek Polacek <polacek@redhat.com>
18293 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
18294 (pass_ubsan::execute): Call gimple_assign_single_p instead of
18295 gimple_assign_load_p.
18297 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
18299 PR middle-end/80692
18300 * real.c (do_compare): Give decimal_do_compare preference over
18301 comparing just the signs.
18303 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
18305 * doc/md.texi (Canonicalization of Instructions): Describe the
18306 canonical form of instructions that inherently set a condition
18309 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
18311 PR middle-end/80775
18312 * tree-cfg.c: Move deletion of unreachable case statements to after
18313 the merging of consecutive case labels.
18315 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
18317 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
18318 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
18319 restoring of callee-saved registers.
18321 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
18323 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
18324 * config/visium/visium.c (single_set_and_flags): Likewise.
18325 * config/visium/visium.md (Substitutions): Likewise.
18327 2017-05-17 Martin Liska <mliska@suse.cz>
18329 * cfg.c: Introduce dump_flags_t type and
18330 use it instead of int type.
18332 * cfghooks.c: Likewise.
18333 * cfghooks.h (struct cfg_hooks): Likewise.
18334 * cfgrtl.c: Likewise.
18335 * cfgrtl.h: Likewise.
18336 * cgraph.c (cgraph_node::get_body): Likewise.
18337 * coretypes.h: Likewise.
18338 * domwalk.c: Likewise.
18339 * domwalk.h: Likewise.
18340 * dumpfile.c (struct dump_option_value_info): Likewise.
18341 (dump_enable_all): Likewise.
18342 (dump_switch_p_1): Likewise.
18343 (opt_info_switch_p): Likewise.
18344 * dumpfile.h (enum tree_dump_index): Likewise.
18345 (struct dump_file_info): Likewise.
18346 * genemit.c: Likewise.
18347 * generic-match-head.c: Likewise.
18348 * gengtype.c (open_base_files): Likewise.
18349 * gimple-pretty-print.c: Likewise.
18350 * gimple-pretty-print.h: Likewise.
18351 * graph.c (print_graph_cfg): Likewise.
18352 * graphite-scop-detection.c (dot_all_sese): Likewise.
18353 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
18354 * loop-unroll.c (report_unroll): Likewise.
18355 * passes.c (pass_manager::register_one_dump_file): Likewise.
18356 * print-tree.c: Likewise.
18357 * statistics.c: Likewise.
18358 * tree-cfg.c: Likewise.
18359 * tree-cfg.h: Likewise.
18360 * tree-dfa.c: Likewise.
18361 * tree-dfa.h: Likewise.
18362 * tree-dump.c (dump_function): Likewise.
18363 * tree-dump.h (struct dump_info): Likewise.
18364 * tree-pretty-print.c: Likewise.
18365 * tree-pretty-print.h: Likewise.
18366 * tree-ssa-live.c: Likewise.
18367 * tree-ssa-live.h: Likewise.
18368 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
18369 * tree-vect-loop.c: Likewise.
18370 * tree-vect-slp.c: Likewise.
18372 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
18373 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18375 PR tree-optimization/80457
18376 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
18377 of all arguments to a statement as scalar_to_vec operations.
18378 (vectorizable_call): Adjust call to vect_model_simple_cost for
18380 (vectorizable_conversion): Likewise.
18381 (vectorizable_assignment): Likewise.
18382 (vectorizable_shift): Likewise.
18383 (vectorizable_operation): Likewise.
18384 (vectorizable_comparison): Likewise.
18385 (vect_is_simple_cond): Record the def types for operands.
18386 (vectorizable_condition): Likewise, call vect_model_simple_cost.
18387 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
18388 for statement argument count.
18390 2017-05-16 Carl Love <cel@us.ibm.com>
18392 * config/rs6000/rs6000-c: Add support for built-in functions
18393 vector unsigned long long vec_bperm (vector unsigned long long,
18394 vector unsigned char)
18395 vector signed long long vec_mule (vector signed int,
18397 vector unsigned long long vec_mule (vector unsigned int,
18398 vector unsigned int)
18399 vector signed long long vec_mulo (vector signed int,
18401 vector unsigned long long vec_mulo (vector unsigned int,
18402 vector unsigned int)
18403 vector signed char vec_sldw (vector signed char,
18404 vector signed char,
18406 vector unsigned char vec_sldw (vector unsigned char,
18407 vector unsigned char,
18409 vector signed short vec_sldw (vector signed short,
18410 vector signed short,
18412 vector unsigned short vec_sldw (vector unsigned short,
18413 vector unsigned short,
18415 vector signed int vec_sldw (vector signed int,
18418 vector unsigned int vec_sldw (vector unsigned int,
18419 vector unsigned int,
18421 vector signed long long vec_sldw (vector signed long long,
18422 vector signed long long,
18424 vector unsigned long long vec_sldw (vector unsigned long long,
18425 vector unsigned long long,
18427 * config/rs6000/rs6000-c: Add support for built-in functions
18428 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
18429 * config/rs6000/altivec.h: Add defintion for vec_sldw.
18430 * doc/extend.texi: Update the built-in documentation for the
18431 new built-in functions.
18433 2017-05-16 Marek Polacek <polacek@redhat.com>
18437 * tree.c (save_expr): Don't fold the expression.
18439 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
18441 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
18442 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
18443 and (?*y,m). Update insn attributes.
18445 2017-05-16 Martin Liska <mliska@suse.cz>
18447 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
18448 flags argument of print_gimple_stmt, print_gimple_expr,
18449 print_generic_stmt and print_generic_expr.
18450 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
18451 * coretypes.h: Likewise.
18452 * except.c (dump_eh_tree): Likewise.
18453 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
18454 * gimple-pretty-print.h: Likewise.
18455 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
18456 (backprop::push_to_worklist): Likewise.
18457 (backprop::pop_from_worklist): Likewise.
18458 (backprop::process_use): Likewise.
18459 (backprop::intersect_uses): Likewise.
18460 (note_replacement): Likewise.
18461 * gimple-ssa-store-merging.c
18462 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
18463 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
18464 (pass_store_merging::execute): Likewise.
18465 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
18466 (ssa_base_cand_dump_callback): Likewise.
18467 (dump_incr_vec): Likewise.
18468 (replace_refs): Likewise.
18469 (replace_mult_candidate): Likewise.
18470 (create_add_on_incoming_edge): Likewise.
18471 (create_phi_basis): Likewise.
18472 (insert_initializers): Likewise.
18473 (all_phi_incrs_profitable): Likewise.
18474 (introduce_cast_before_cand): Likewise.
18475 (replace_one_candidate): Likewise.
18476 * gimplify.c (gimplify_expr): Likewise.
18477 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
18478 (set_rename): Likewise.
18479 (rename_uses): Likewise.
18480 (copy_loop_phi_nodes): Likewise.
18481 (add_close_phis_to_merge_points): Likewise.
18482 (copy_loop_close_phi_args): Likewise.
18483 (copy_cond_phi_args): Likewise.
18484 (graphite_copy_stmts_from_block): Likewise.
18485 (translate_pending_phi_nodes): Likewise.
18486 * graphite-poly.c (print_pdr): Likewise.
18487 (dump_gbb_cases): Likewise.
18488 (dump_gbb_conditions): Likewise.
18489 (print_scop_params): Likewise.
18490 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
18491 (build_cross_bb_scalars_use): Likewise.
18492 (gather_bbs::before_dom_children): Likewise.
18493 * hsa-dump.c (dump_hsa_immed): Likewise.
18494 * ipa-cp.c (print_ipcp_constant_value): Likewise.
18495 (get_replacement_map): Likewise.
18496 * ipa-inline-analysis.c (dump_condition): Likewise.
18497 (estimate_function_body_sizes): Likewise.
18498 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
18499 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
18500 * ipa-prop.c (ipa_dump_param): Likewise.
18501 (ipa_print_node_jump_functions_for_edge): Likewise.
18502 (ipa_modify_call_arguments): Likewise.
18503 (ipa_modify_expr): Likewise.
18504 (ipa_dump_param_adjustments): Likewise.
18505 (ipa_dump_agg_replacement_values): Likewise.
18506 (ipcp_modif_dom_walker::before_dom_children): Likewise.
18507 * ipa-pure-const.c (check_stmt): Likewise.
18508 (pass_nothrow::execute): Likewise.
18509 * ipa-split.c (execute_split_functions): Likewise.
18510 * omp-offload.c (dump_oacc_loop_part): Likewise.
18511 (dump_oacc_loop): Likewise.
18512 * trans-mem.c (tm_log_emit): Likewise.
18513 (tm_memopt_accumulate_memops): Likewise.
18514 (dump_tm_memopt_set): Likewise.
18515 (dump_tm_memopt_transform): Likewise.
18516 * tree-cfg.c (gimple_verify_flow_info): Likewise.
18517 (print_loop): Likewise.
18518 * tree-chkp-opt.c (chkp_print_addr): Likewise.
18519 (chkp_gather_checks_info): Likewise.
18520 (chkp_get_check_result): Likewise.
18521 (chkp_remove_check_if_pass): Likewise.
18522 (chkp_use_outer_bounds_if_possible): Likewise.
18523 (chkp_reduce_bounds_lifetime): Likewise.
18524 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
18525 (chkp_mark_completed_bounds): Likewise.
18526 (chkp_register_incomplete_bounds): Likewise.
18527 (chkp_mark_invalid_bounds): Likewise.
18528 (chkp_maybe_copy_and_register_bounds): Likewise.
18529 (chkp_build_returned_bound): Likewise.
18530 (chkp_get_bound_for_parm): Likewise.
18531 (chkp_build_bndldx): Likewise.
18532 (chkp_get_bounds_by_definition): Likewise.
18533 (chkp_generate_extern_var_bounds): Likewise.
18534 (chkp_get_bounds_for_decl_addr): Likewise.
18535 * tree-chrec.c (chrec_apply): Likewise.
18536 * tree-data-ref.c (dump_data_reference): Likewise.
18537 (dump_subscript): Likewise.
18538 (dump_data_dependence_relation): Likewise.
18539 (analyze_overlapping_iterations): Likewise.
18540 * tree-inline.c (expand_call_inline): Likewise.
18541 (tree_function_versioning): Likewise.
18542 * tree-into-ssa.c (dump_defs_stack): Likewise.
18543 (dump_currdefs): Likewise.
18544 (dump_names_replaced_by): Likewise.
18545 (dump_update_ssa): Likewise.
18546 (update_ssa): Likewise.
18547 * tree-object-size.c (pass_object_sizes::execute): Likewise.
18548 * tree-parloops.c (build_new_reduction): Likewise.
18549 (try_create_reduction_list): Likewise.
18550 (ref_conflicts_with_region): Likewise.
18551 (oacc_entry_exit_ok_1): Likewise.
18552 (oacc_entry_exit_single_gang): Likewise.
18553 * tree-pretty-print.h: Likewise.
18554 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
18555 (get_scalar_evolution): Likewise.
18556 (add_to_evolution): Likewise.
18557 (get_loop_exit_condition): Likewise.
18558 (analyze_evolution_in_loop): Likewise.
18559 (analyze_initial_condition): Likewise.
18560 (analyze_scalar_evolution): Likewise.
18561 (instantiate_scev): Likewise.
18562 (number_of_latch_executions): Likewise.
18563 (gather_chrec_stats): Likewise.
18564 (final_value_replacement_loop): Likewise.
18565 (scev_const_prop): Likewise.
18566 * tree-sra.c (dump_access): Likewise.
18567 (disqualify_candidate): Likewise.
18568 (create_access): Likewise.
18569 (reject): Likewise.
18570 (maybe_add_sra_candidate): Likewise.
18571 (create_access_replacement): Likewise.
18572 (analyze_access_subtree): Likewise.
18573 (analyze_all_variable_accesses): Likewise.
18574 (sra_modify_assign): Likewise.
18575 (initialize_constant_pool_replacements): Likewise.
18576 (find_param_candidates): Likewise.
18577 (decide_one_param_reduction): Likewise.
18578 (replace_removed_params_ssa_names): Likewise.
18579 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
18580 * tree-ssa-copy.c (dump_copy_of): Likewise.
18581 (copy_prop_visit_cond_stmt): Likewise.
18582 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
18583 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
18584 (record_equivalences_from_stmt): Likewise.
18585 * tree-ssa-dse.c (compute_trims): Likewise.
18586 (delete_dead_call): Likewise.
18587 (delete_dead_assignment): Likewise.
18588 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
18589 (forward_propagate_into_cond): Likewise.
18590 (pass_forwprop::execute): Likewise.
18591 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
18592 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
18594 (move_computations_worker): Likewise.
18595 (execute_sm): Likewise.
18596 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
18597 (remove_exits_and_undefined_stmts): Likewise.
18598 (remove_redundant_iv_tests): Likewise.
18599 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
18600 (adjust_iv_update_pos): Likewise.
18601 * tree-ssa-math-opts.c (bswap_replace): Likewise.
18602 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
18603 (value_replacement): Likewise.
18604 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
18605 * tree-ssa-pre.c (print_pre_expr): Likewise.
18606 (get_representative_for): Likewise.
18607 (create_expression_by_pieces): Likewise.
18608 (insert_into_preds_of_block): Likewise.
18609 (eliminate_insert): Likewise.
18610 (eliminate_dom_walker::before_dom_children): Likewise.
18611 (eliminate): Likewise.
18612 (remove_dead_inserted_code): Likewise.
18613 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
18614 * tree-ssa-reassoc.c (get_rank): Likewise.
18615 (eliminate_duplicate_pair): Likewise.
18616 (eliminate_plus_minus_pair): Likewise.
18617 (eliminate_not_pairs): Likewise.
18618 (undistribute_ops_list): Likewise.
18619 (eliminate_redundant_comparison): Likewise.
18620 (update_range_test): Likewise.
18621 (optimize_range_tests_var_bound): Likewise.
18622 (optimize_vec_cond_expr): Likewise.
18623 (rewrite_expr_tree): Likewise.
18624 (rewrite_expr_tree_parallel): Likewise.
18625 (linearize_expr): Likewise.
18626 (break_up_subtract): Likewise.
18627 (linearize_expr_tree): Likewise.
18628 (attempt_builtin_powi): Likewise.
18629 (attempt_builtin_copysign): Likewise.
18630 (transform_stmt_to_copy): Likewise.
18631 (transform_stmt_to_multiply): Likewise.
18632 (dump_ops_vector): Likewise.
18633 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
18634 (print_scc): Likewise.
18635 (set_ssa_val_to): Likewise.
18636 (visit_reference_op_store): Likewise.
18637 (visit_use): Likewise.
18638 (sccvn_dom_walker::before_dom_children): Likewise.
18639 (run_scc_vn): Likewise.
18640 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
18642 (expr_hash_elt::print): Likewise.
18643 (const_and_copies::pop_to_marker): Likewise.
18644 (const_and_copies::record_const_or_copy_raw): Likewise.
18645 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
18646 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
18647 (dump_predicates): Likewise.
18648 (find_uninit_use): Likewise.
18649 (warn_uninitialized_phi): Likewise.
18650 (pass_late_warn_uninitialized::execute): Likewise.
18651 * tree-ssa.c (verify_vssa): Likewise.
18652 (verify_ssa): Likewise.
18653 (maybe_optimize_var): Likewise.
18654 * tree-vrp.c (dump_value_range): Likewise.
18655 (dump_all_value_ranges): Likewise.
18656 (dump_asserts_for): Likewise.
18657 (register_edge_assert_for_2): Likewise.
18658 (vrp_visit_cond_stmt): Likewise.
18659 (vrp_visit_switch_stmt): Likewise.
18660 (vrp_visit_stmt): Likewise.
18661 (vrp_visit_phi_node): Likewise.
18662 (simplify_cond_using_ranges_1): Likewise.
18663 (fold_predicate_in): Likewise.
18664 (evrp_dom_walker::before_dom_children): Likewise.
18665 (evrp_dom_walker::push_value_range): Likewise.
18666 (evrp_dom_walker::pop_value_range): Likewise.
18667 (execute_early_vrp): Likewise.
18669 2017-05-16 Richard Biener <rguenther@suse.de>
18671 * dwarf2out.c (loc_list_from_tree_1): Do not create
18672 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
18674 2017-05-16 Richard Biener <rguenther@suse.de>
18676 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
18678 (note_variable_value_in_expr): If we resolved the decl ref
18679 do not push to the stack.
18681 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
18683 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
18684 operations in fast-math mode.
18685 (vaddq_f16): Likewise.
18686 (vmul_f16): Likewise.
18687 (vmulq_f16): Likewise.
18688 (vsub_f16): Likewise.
18689 (vsubq_f16): Likewise.
18690 * config/arm/neon.md (add<mode>3): New.
18692 (fma:<VH:mode>3): New. Also remove outdated comment.
18695 2017-05-16 Martin Liska <mliska@suse.cz>
18699 * ipa-devirt.c (warn_types_mismatch): Fix typo.
18700 (odr_types_equivalent_p): Likewise.
18702 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
18704 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
18706 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
18709 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
18710 non-interunit SSE move alternatives with '?'.
18711 (zero-extendsidi peephole2): New peephole to skip intermediate
18712 general register in SSE zero-extend sequence.
18714 2017-05-15 Jeff Law <law@redhat.com>
18716 * reorg.c (relax_delay_slots): Create a new variable to hold
18717 the temporary target rather than clobbering TARGET_LABEL.
18719 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
18720 missing argument to extract_bit_field call.
18721 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
18723 2017-05-15 Martin Liska <mliska@suse.cz>
18726 * gcc.c (process_command): Do not allow empty argument of -o option.
18728 2017-05-15 Renlin Li <renlin.li@arm.com>
18730 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
18731 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
18732 * config/aarch64/constraints.md (Usf): Add long call check.
18733 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
18734 (call_value): Likewise.
18735 (sibcall): Likewise.
18736 (sibcall_value): Likewise.
18738 (call_value_insn): New.
18739 (sibcall_insn): Update rtx pattern.
18740 (sibcall_value_insn): Likewise.
18741 (call_internal): Remove.
18742 (call_value_internal): Likewise.
18743 (sibcall_internal): Likewise.
18744 (sibcall_value_internal): Likewise.
18745 (call_reg): Likewise.
18746 (call_symbol): Likewise.
18747 (call_value_reg): Likewise.
18748 (call_value_symbol): Likewise.
18750 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
18753 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
18755 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
18757 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
18758 compatible with CCGOCmode and with CCZmode.
18760 2017-05-14 Martin Sebor <msebor@redhat.com>
18762 PR middle-end/77671
18763 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
18764 (gimple_fold_builtin_snprintf): Same.
18765 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
18766 (gimple_fold_builtin_snprintf): Same.
18767 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
18768 of character types.
18769 (is_call_safe): New function.
18770 (try_substitute_return_value): Call it.
18771 (try_simplify_call): New function.
18772 (pass_sprintf_length::handle_gimple_call): Call it.
18774 2017-05-14 Martin Sebor <msebor@redhat.com>
18776 PR middle-end/80669
18777 * builtins.c (expand_builtin_stpncpy): Simplify.
18779 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
18781 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
18782 * config/i386/i386.h
18783 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
18784 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
18785 (struct machine_function): Add new members call_ms2sysv,
18786 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
18787 (struct machine_frame_state): New fields sp_realigned and
18788 sp_realigned_offset.
18789 * config/i386/i386.c
18790 (enum xlogue_stub): New enum.
18791 (enum xlogue_stub_sets): New enum.
18792 (class xlogue_layout): New class.
18793 (struct ix86_frame): New fields stack_realign_allocate_offset,
18794 stack_realign_offset and outlined_save_offset. Modify comments to
18795 detail stack layout when using out-of-line stubs.
18796 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
18797 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
18798 -mcall-ms2sysv-xlogues.
18799 (stub_managed_regs): New static variable.
18800 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
18801 registers managed by out-of-line stub.
18802 (disable_call_ms2sysv_xlogues): New function.
18803 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
18804 m->call_ms2sysv when appropriate and compute frame layout for
18806 (sp_valid_at, fp_valid_at): New inline functions.
18807 (choose_basereg): New function.
18808 (choose_baseaddr): Add align parameter, use choose_basereg and modify
18810 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
18811 Use align parameter of choose_baseaddr to generated aligned SSE movs
18813 (pro_epilogue_adjust_stack): Modify to track
18814 machine_frame_state::sp_realigned.
18815 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
18816 (ix86_nsaved_sseregs): Likewise.
18817 (ix86_emit_save_regs): Likewise.
18818 (ix86_emit_save_regs_using_mov): Likewise.
18819 (ix86_emit_save_sse_regs_using_mov): Likewise.
18820 (get_scratch_register_on_entry): Likewise.
18821 (gen_frame_set): New function.
18822 (gen_frame_load): Likewise.
18823 (gen_frame_store): Likewise.
18824 (emit_outlined_ms2sysv_save): Likewise.
18825 (emit_outlined_ms2sysv_restore): Likewise.
18826 (ix86_expand_prologue): Modify stack re-alignment code and call
18827 emit_outlined_ms2sysv_save when appropriate.
18828 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
18829 parameter rtx_insn *insn, which allows the function to be used to only
18830 generate the notes.
18831 (ix86_expand_epilogue): Modify validity checks of frame and stack
18832 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
18833 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
18834 * config/i386/predicates.md
18835 (save_multiple): New predicate.
18836 (restore_multiple): Likewise.
18837 * config/i386/sse.md
18838 (save_multiple<mode>): New pattern.
18839 (save_multiple_realign<mode>): Likewise.
18840 (restore_multiple<mode>): Likewise.
18841 (restore_multiple_and_return<mode>): Likewise.
18842 (restore_multiple_leave_return<mode>): Likewise.
18843 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
18845 2017-05-14 Julia Koval <julia.koval@intel.com>
18847 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
18848 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
18849 (__builtin_ia32_xsetbv): New builtins.
18850 * config/i386/i386.c (ix86_expand_special_args_builtin):
18852 (ix86_expand_builtin): Special expand for new intrinsics.
18853 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
18854 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
18855 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
18857 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18859 * cfganal.c (inverted_post_order_compute): Change argument type
18861 * cfganal.h (inverted_post_order_compute): Adjust prototype.
18862 * df-core.c (rest_of_handle_df_initialize): Adjust.
18863 (rest_of_handle_df_finish): Likewise.
18864 (df_analyze_1): Likewise.
18865 (df_analyze): Likewise.
18866 (loop_inverted_post_order_compute): Change argument to be a vec *.
18867 (df_analyze_loop): Adjust.
18868 (df_get_n_blocks): Likewise.
18869 (df_get_postorder): Likewise.
18870 * df.h (struct df_d): Change field to be a vec.
18871 * lcm.c (compute_laterin): Adjust.
18872 (compute_available): Likewise.
18873 * lra-lives.c (lra_create_live_ranges_1): Likewise.
18874 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
18875 * tree-ssa-pre.c (compute_antic): Likewise.
18877 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18879 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
18880 (depth_first_search::depth_first_search): Change structure init
18881 function to this constructor.
18882 (depth_first_search::add_bb): Rename function to this member.
18883 (depth_first_search::execute): Likewise.
18884 (flow_dfs_compute_reverse_finish): Adjust.
18886 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18888 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
18889 (longest_simple_path): Likewise.
18890 * shrink-wrap.c (spread_components): Likewise.
18891 (disqualify_problematic_components): Likewise.
18892 (emit_common_heads_for_components): Likewise.
18893 (emit_common_tails_for_components): Likewise.
18894 (insert_prologue_epilogue_for_components): Likewise.
18896 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18898 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
18901 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18903 * df-core.c (df_set_blocks): Start using auto_bitmap.
18904 (df_compact_blocks): Likewise.
18905 * df-problems.c (df_rd_confluence_n): Likewise.
18906 * df-scan.c (df_insn_rescan_all): Likewise.
18907 (df_process_deferred_rescans): Likewise.
18908 (df_update_entry_block_defs): Likewise.
18909 (df_update_exit_block_uses): Likewise.
18910 (df_entry_block_bitmap_verify): Likewise.
18911 (df_exit_block_bitmap_verify): Likewise.
18912 (df_scan_verify): Likewise.
18913 * lra-constraints.c (lra_constraints): Likewise.
18914 (undo_optional_reloads): Likewise.
18915 (lra_undo_inheritance): Likewise.
18916 * lra-remat.c (calculate_gen_cands): Likewise.
18917 (do_remat): Likewise.
18918 * lra-spills.c (assign_spill_hard_regs): Likewise.
18919 (spill_pseudos): Likewise.
18920 * tree-ssa-pre.c (bitmap_set_and): Likewise.
18921 (bitmap_set_subtract_values): Likewise.
18923 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18925 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
18926 management with auto_bitmap.
18927 (fix_inter_tick): Likewise.
18928 (fix_recovery_deps): Likewise.
18929 * ira.c (add_store_equivs): Likewise.
18930 (find_moveable_pseudos): Likewise.
18931 (split_live_ranges_for_shrink_wrap): Likewise.
18932 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
18933 (rtx_reuse_manager::seen_def_p): Likewise.
18934 (rtx_reuse_manager::set_seen_def): Likewise.
18935 * print-rtl.h (class rtx_reuse_manager): Likewise.
18937 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18939 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
18941 (migrate_btr_def): Likewise.
18942 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
18943 * df-core.c (loop_post_order_compute): Likewise.
18944 (loop_inverted_post_order_compute): Likewise.
18945 * hsa-common.h: Likewise.
18946 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
18947 * init-regs.c (initialize_uninitialized_regs): Likewise.
18948 * ipa-inline.c (resolve_noninline_speculation): Likewise.
18949 (inline_small_functions): Likewise.
18950 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
18951 * ira.c (combine_and_move_insns): Likewise.
18952 (build_insn_chain): Likewise.
18953 * loop-invariant.c (find_invariants): Likewise.
18954 * lower-subreg.c (propagate_pseudo_copies): Likewise.
18955 * predict.c (tree_predict_by_opcode): Likewise.
18956 (predict_paths_leading_to): Likewise.
18957 (predict_paths_leading_to_edge): Likewise.
18958 (estimate_loops_at_level): Likewise.
18959 (estimate_loops): Likewise.
18960 * shrink-wrap.c (try_shrink_wrapping): Likewise.
18961 (spread_components): Likewise.
18962 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
18963 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
18964 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
18965 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
18966 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
18967 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
18968 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
18969 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
18970 (mark_threaded_blocks): Likewise.
18971 (thread_through_all_blocks): Likewise.
18972 * tree-ssa.c (verify_ssa): Likewise.
18973 (execute_update_addresses_taken): Likewise.
18974 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
18976 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18978 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
18980 (post_order_compute): Likewise.
18981 (inverted_post_order_compute): Likewise.
18982 (pre_and_rev_post_order_compute_fn): Likewise.
18984 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18986 * genrecog.c (int_set::int_set): Explicitly construct our
18987 auto_vec base class.
18988 * vec.h (auto_vec::auto_vec): New constructor.
18990 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18992 * bitmap.h (class auto_bitmap): New constructor taking
18993 bitmap_obstack * argument.
18995 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18997 * bitmap.h (class auto_bitmap): Change type of m_bits to
18998 bitmap_head, and adjust ctor / dtor and member operators.
19000 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
19002 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
19003 when returned register mode doesn't match original mode.
19005 2017-05-12 Jeff Law <law@redhat.com>
19006 Jakub Jelinek <jakub@redhat.com>
19008 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
19009 we look for cc setter after the compare-elim changes.
19010 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
19011 within the vector to match what compare-elim now expects.
19012 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
19013 (xorsi3_flags, one_cmplsi2_flags): Likewise.
19015 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
19016 after the compare-elim changes.
19017 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
19018 the vector to match what compare-elim now expects.
19019 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
19020 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
19021 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
19022 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
19023 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
19025 * config/visium/visium.c (single_set_and_flags): Fix where
19026 we look for cc setter after the compare-elim changes.
19027 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
19028 with the vector to match what compare-elim now expects.
19029 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
19030 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
19031 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
19032 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
19033 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
19034 (neg<mode>2_insn_set_overflow): Likewise.
19036 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
19038 PR middle-end/79794
19039 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
19040 maybe_expand_insn call, set ops[0].target. If still set after call,
19041 set alt_rtl. Add extra arg to recursive calls.
19042 (extract_bit_field): Add alt_rtl argument. Pass to
19044 * expmed.h (extract_bit_field): Fix prototype.
19045 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
19046 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
19047 to extract_bit_field_calls.
19048 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
19049 Pass alt_rtl to extract_bit_field calls.
19050 * calls.c (store_unaligned_arguments_into_psuedos)
19051 load_register_parameters): Pass extra NULL to extract_bit_field calls.
19052 * optabs.c (maybe_legitimize_operand): Clear op->target when call
19054 * optabs.h (struct expand_operand): Add target bitfield.
19056 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
19058 * compare-elim.c (try_eliminate_compare): Canonicalize
19059 operation with embedded compare to
19060 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
19061 (set (reg) (operation)].
19063 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
19065 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
19068 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
19069 cost of adding a carry flag for ADC instruction.
19070 [case MINUS]: Ignore the cost of subtracting a carry flag
19071 for SBB instruction.
19073 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
19075 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
19077 * config/rs6000/bmiintrin.h: New file.
19078 * config/rs6000/bmi2intrin.h: New file.
19079 * config/rs6000/x86intrin.h: New file.
19081 2017-05-12 Jeff Law <law@redhat.com>
19083 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
19086 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
19088 PR middle-end/80707
19089 * tree-cfg.c: Remove cfg edges of unreachable case statements.
19091 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19093 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
19094 early expansion of vector divide builtins.
19095 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
19096 builtins identified as having unsigned arguments.
19098 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19100 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
19101 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
19102 expansion of vector logical operations (and, andc, or, xor,
19105 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19107 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
19108 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
19110 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19112 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
19113 early GIMPLE expansion of vector multiplies.
19115 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19117 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
19118 TARGET_HAVE_MOVT conditional.
19119 (movt splitter): Likewise.
19121 2017-05-12 Richard Biener <rguenther@suse.de>
19123 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
19124 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
19125 Fold all stmts not inplace.
19127 2017-05-12 Richard Biener <rguenther@suse.de>
19129 PR tree-optimization/80713
19130 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
19131 inserted_exprs bit for not removed stmts.
19133 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
19135 PR middle-end/69921
19136 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
19137 parallelized" attribute for parallelized OpenACC kernels.
19138 * omp-offload.c (execute_oacc_device_lower): Use it.
19140 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
19141 Set "oacc kernels" attribute.
19142 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
19143 parameter. Adjust all users.
19144 (oacc_fn_attrib_kernels_p): Remove function.
19145 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
19146 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
19147 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
19148 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
19149 assert "oacc kernels" attribute is set.
19151 2017-05-11 Carl Love <cel@us.ibm.com>
19153 * config/rs6000/rs6000-c: Add support for built-in functions
19154 vector unsigned char vec_popcnt (vector signed char)
19155 vector unsigned char vec_popcnt (vector unsigned char)
19156 vector unsigned short vec_popcnt (vector signed short)
19157 vector unsigned short vec_popcnt (vector unsigned short)
19158 vector unsigned int vec_popcnt (vector signed int)
19159 vector unsigned int vec_popcnt (vector unsigned int)
19160 vector unsigned long long vec_popcnt (vector signed long long)
19161 vector unsigned long long vec_popcnt (vector unsigned long long)
19162 vector signed long long vec_slo (vector signed long long,
19163 vector signed char)
19164 vector signed long long vec_slo (vector signed long long,
19165 vector unsigned char)
19166 vector unsigned long long vec_slo (vector unsigned long long,
19167 vector signed char)
19168 vector unsigned long long vec_slo (vector unsigned long long,
19169 vector unsigned char)
19170 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
19171 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
19172 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
19173 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
19174 * doc/extend.texi: Update the built-in documentation file for the
19175 new built-in functions.
19177 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
19179 * attribs.h (sorted_attr_string): Move machine independent
19180 functions for target clone support from the i386 port to common
19181 code. Rename ix86_function_versions to common_function_versions.
19182 Rename make_name to make_unique_name.
19183 (common_function_versions): Likewise.
19184 (make_unique_name): Likewise.
19185 (make_dispatcher_decl): Likewise.
19186 (is_function_default_version): Likewise.
19187 * attribs.c (attr_strcmp): Likewise.
19188 (sorted_attr_string): Likewise.
19189 (common_function_versions): Likewise.
19190 (make_unique_name): Likewise.
19191 (make_dispatcher_decl): Likewise.
19192 (is_function_default_version): Likewise.
19193 * config/i386/i386.c (attr_strcmp): Likewise.
19194 (sorted_attr_string): Likewise.
19195 (ix86_function_versions): Likewise.
19196 (make_name): Likewise.
19197 (make_dispatcher_decl): Likewise.
19198 (is_function_default_version): Likewise.
19199 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
19201 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19204 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
19205 Account for direct move costs for vec_construct of integer
19208 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
19211 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
19212 (UNSPEC_STX_ATOMIC): Ditto.
19213 (loaddi_via_sse): New insn.
19214 (storedi_via_sse): Ditto.
19215 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
19216 Update corresponding peephole2 patterns.
19217 (atomic_storedi_fpu): Ditto.
19219 2017-05-11 Julia Koval <julia.koval@intel.com>
19221 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
19222 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
19224 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
19225 (__builtin_ia32_rsqrt14ss_mask): New builtins.
19226 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
19228 2017-05-11 Nathan Sidwell <nathan@acm.org>
19230 * graphite-poly.c: Include dumpfile.h.
19232 * dumpfle.h (dump_function): Declare here ...
19233 * tree-dump.h (dump_function): ... not here.
19234 * dumpfile.c: #include tree-cfg.h.
19235 (dump_function): Move here from ...
19236 * tree-dump.c (dump_function): ... here.
19237 * gimplify.c: #include splay-tree.h, not tree-dump.h.
19238 * graphite-poly.c: Don't include tree-dump.h.
19239 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
19240 * print-tree.c: Likewise.
19241 * stor-layout.c: Likewise.
19242 * tree-nested.c: Likewise.
19244 * dumpfile.c (dump_start): Use TDF_FLAGS.
19245 (dump_enable_all): Fix TDF_KIND check thinko.
19247 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
19249 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19250 array entries to represent two legal parameterizations of the
19251 overloaded __builtin_cmpb function, as represented by the
19252 P6_OV_BUILTIN_CMPB constant.
19253 (altivec_resolve_overloaded_builtin): Add special case handling
19254 for the __builtin_cmpb function, as represented by the
19255 P6_OV_BUILTIN_CMPB constant.
19256 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
19257 (BU_P6_64BIT_2): New macro.
19258 (BU_P6_OVERLOAD_2): New macro
19259 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
19260 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
19261 (CMPB): Add overload support to represent both 32-bit and 64-bit
19262 compare-bytes function.
19263 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
19264 support for TARGET_CMPB.
19265 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
19266 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19267 documentation of the __builtin_cmpb overloaded built-in function.
19269 2017-05-11 Richard Biener <rguenther@suse.de>
19271 PR tree-optimization/80705
19272 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
19273 bases are not vectorizable.
19275 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19277 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
19278 when counting register pressure.
19280 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19282 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
19283 (iv_ca_more_deps): Renamed to ...
19284 (iv_ca_compare_deps): ... this.
19285 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
19287 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19289 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
19291 (determine_group_iv_costs): ... here.
19292 (find_inv_vars_cb): Record inv var if it's not recorded before.
19294 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19296 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
19297 (get_shiftadd_cost): Ditto.
19299 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19301 * tree-ssa-address.c: Include header file.
19302 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
19304 (add_to_parts): Refactor.
19305 (addr_to_parts): New parameter. Update use of move_hint_to_base.
19306 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
19309 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19311 PR tree-optimization/53090
19312 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
19314 (extract_cond_operands): Detect condition with IV on both sides
19315 and return COMP_IV_EXPR_2.
19316 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
19317 (rewrite_use_compare): Simplify by removing call to function
19318 extract_cond_operands.
19320 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19322 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
19323 (extract_cond_operands): Detect condition comparing against non-
19324 invariant bound and return appropriate enum value.
19325 (find_interesting_uses_cond): Update use of extract_cond_operands.
19326 Handle its return value accordingly.
19327 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
19329 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19331 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
19332 nonlinear iv_use computation in loop invariant sensitive way.
19334 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19336 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
19337 (find_iv_candidates): Call relate_compare_use_with_all_cands.
19339 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19341 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
19342 (dump_cand): Support iv_cand.inv_exprs.
19343 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
19345 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
19348 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19350 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
19352 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
19353 as local function. Include necessary header files.
19354 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
19356 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19358 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
19360 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19362 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
19363 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
19364 RSHIFT_EXPR and BIT_NOT_EXPR.
19366 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19368 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
19369 (adjust_setup_cost): New parameter supporting round up adjustment.
19370 (struct address_cost_data): Delete.
19371 (force_expr_to_var_cost): Don't bound cost with spill_cost.
19372 (split_address_cost, ptr_difference_cost): Delete.
19373 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
19374 (struct ainc_cost_data): New struct.
19375 (get_address_cost_ainc): New function.
19376 (get_address_cost, get_computation_cost): Reimplement.
19377 (determine_group_iv_cost_address): Record inv_expr for all uses of
19379 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
19380 (iv_ca_has_deps): Reimplemented to ...
19381 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
19383 (iv_ca_extend): Call iv_ca_more_deps.
19385 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19387 * tree-ssa-address.c (struct mem_address): Move to header file.
19388 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
19389 * tree-ssa-address.h (struct mem_address): Move from C file.
19390 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
19392 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19394 * tree-affine.h (aff_combination_type): New interface.
19395 (aff_combination_zero_p): Remove static.
19396 (aff_combination_const_p): New interface.
19397 (aff_combination_singleton_var_p): New interfaces.
19399 2017-05-11 Richard Biener <rguenther@suse.de>
19401 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
19402 Skip unreachable blocks and destinations.
19403 (eliminate): Move stmt removal and fixup ...
19404 (fini_eliminate): ... here. Skip inserted exprs.
19405 (pass_pre::execute): Move fini_pre after fini_eliminate.
19406 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
19407 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
19408 PRE to get rid of dead code that has invalid SSA form and
19409 split critical edges again.
19411 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19413 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19415 2017-05-11 Richard Biener <rguenther@suse.de>
19417 * passes.c (execute_function_todo): Verify loops if they are
19418 said to be up-to-date.
19419 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
19420 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
19422 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
19425 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
19426 handle calling assemble_external ourself.
19429 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
19430 modes with zero size. Enhance comment.
19432 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19434 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
19435 built-ins for vec_xl and vec_xst with short and char pointer
19438 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
19440 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
19441 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
19442 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
19443 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
19444 (_mm_maskz_min_round_ss): New intrinsics.
19445 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
19446 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
19447 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
19448 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
19449 (__builtin_ia32_minss_mask_round): New builtins.
19450 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19451 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
19452 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
19454 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
19455 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
19457 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
19460 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
19462 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
19463 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
19464 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
19465 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
19466 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
19467 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
19468 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
19469 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19470 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
19471 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
19472 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
19473 (__builtin_ia32_mulss_mask_round): New builtins.
19474 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19475 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
19476 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
19478 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
19479 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
19481 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
19484 2017-05-10 Julia Koval <julia.koval@intel.com>
19486 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
19487 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
19488 (_mm256_setr_m128i): New intrinsics.
19490 2017-05-10 Julia Koval <julia.koval@intel.com>
19492 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
19493 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
19494 (_mm_maskz_rcp14_ss): New intrinsics.
19495 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
19496 (__builtin_ia32_rcp14ss_mask): New builtins.
19497 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
19499 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
19501 PR tree-optimization/51513
19502 * tree-cfg.c (gimple_seq_unreachable_p): New function.
19503 (assert_unreachable_fallthru_edge_p): Use it.
19504 (group_case_labels_stmt): Likewise.
19505 * tree-cfg.h: Prototype it.
19506 * stmt.c: Include cfghooks.h and tree-cfg.h.
19507 (emit_case_dispatch_table) <gap_label>: New local variable.
19508 Use it to fill dispatch table gaps.
19509 Test for default_label before updating probabilities.
19510 (expand_case) <default_label>: Remove unneeded initialization.
19511 Test for unreachable default case statement and remove its edge.
19512 Set default_label accordingly.
19513 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
19515 2017-05-10 Carl Love <cel@us.ibm.com>
19517 * config/rs6000/rs6000-c: Add support for built-in functions
19518 vector signed char vec_neg (vector signed char)
19519 vector signed short int vec_neg (vector short int)
19520 vector signed int vec_neg (vector signed int)
19521 vector signed long long vec_neg (vector signed long long)
19522 vector float vec_neg (vector float)
19523 vector double vec_neg (vector double)
19524 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
19526 * config/rs6000/altivec.h: Add define for vec_neg
19527 * doc/extend.texi: Update the built-in documentation for the
19528 new built-in functions.
19530 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19532 PR tree-optimization/77644
19533 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
19535 2017-05-10 Nathan Sidwell <nathan@acm.org>
19537 * dumpfile.h (TDI_lang_all): New.
19538 (TDF_KIND): New. Renumber others
19539 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
19541 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
19543 (get_dump_file_name): Adjust suffix generation.
19544 (dump_enable_all): Use TDF_KIND.
19545 * doc/invoke.texi (-fdump-lang-all): Document.
19547 * dumpfile.h: Tabify.
19549 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
19552 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
19553 Move member access before delete.
19555 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
19557 * tree-inline.c (expand_call_inline): Split block at stmt
19560 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
19563 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
19564 are now unused after splitting mov{sf,sd}_hardfloat.
19565 (f32_lr2): Likewise.
19566 (f32_lm): Likewise.
19567 (f32_lm2): Likewise.
19568 (f32_li): Likewise.
19569 (f32_li2): Likewise.
19570 (f32_lv): Likewise.
19571 (f32_sr): Likewise.
19572 (f32_sr2): Likewise.
19573 (f32_sm): Likewise.
19574 (f32_sm2): Likewise.
19575 (f32_si): Likewise.
19576 (f32_si2): Likewise.
19577 (f32_sv): Likewise.
19578 (f32_dm): Likewise.
19579 (f32_vsx): Likewise.
19580 (f32_av): Likewise.
19581 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
19582 For movsf, order stores so the VSX stores occur before the GPR
19583 store which encourages the register allocator to use a traditional
19584 FPR instead of a GPR. For movsd, order the stores so that the GPR
19585 store comes before the VSX stores to allow the power6 to work.
19586 This is due to the power6 not having a 32-bit integer store
19587 instruction from a FPR.
19588 (movsf_hardfloat): Likewise.
19589 (movsd_hardfloat): Likewise.
19591 2017-05-09 Martin Sebor <msebor@redhat.com>
19593 PR translation/80280
19594 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
19597 PR translation/80280
19598 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
19599 data member added in r247778.
19600 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
19602 2017-05-09 Nathan Sidwell <nathan@acm.org>
19604 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
19606 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
19609 2017-05-09 Marek Polacek <polacek@redhat.com>
19611 * doc/invoke.texi: Fix typo.
19613 2017-05-09 Richard Biener <rguenther@suse.de>
19615 * tree-vrp.c (vrp_val_is_max): Adjust comment.
19616 (vrp_val_is_min): Likewise.
19617 (set_value_range_to_value): Likewise.
19618 (set_value_range_to_nonnegative): Likewise.
19619 (gimple_assign_nonzero_p): Likewise.
19620 (gimple_stmt_nonzero_p): Likewise.
19621 (vrp_int_const_binop): Likewise. Remove unreachable case.
19622 (adjust_range_with_scev): Adjust comments.
19623 (compare_range_with_value): Likewise.
19624 (extract_range_from_phi_node): Likewise.
19625 (test_for_singularity): Likewise.
19627 2017-05-09 Richard Biener <rguenther@suse.de>
19629 * tree-vrp.c (get_single_symbol): Add assert that we don't
19630 get overflowed constants as invariant part.
19631 (compare_values_warnv): Add comment before the TREE_NO_WARNING
19632 checks. Use wi::cmp instead of recursing for integer constants.
19633 (compare_values): Just ignore whether we assumed undefined
19634 overflow instead of failing the compare.
19635 (extract_range_for_var_from_comparison_expr): Add comment before the
19636 TREE_NO_WARNING sets.
19637 (test_for_singularity): Likewise.
19638 (extract_range_from_comparison): Do not disable optimization
19639 when we assumed undefined overflow.
19640 (extract_range_basic): Remove init of unused var.
19642 2017-05-09 Richard Biener <rguenther@suse.de>
19644 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
19645 (extract_range_from_multiplicative_op_1): Adjust.
19646 (extract_range_from_binary_expr_1): Use int_const_binop.
19648 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
19651 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
19652 rs6000_store_data_bypass_p in seven define_bypass directives and
19653 in several comments.
19654 * config/rs6000/rs6000-protos.h: Add prototype for
19655 rs6000_store_data_bypass_p function.
19656 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
19657 function implements slightly different (rs6000-specific) semantics
19658 than store_data_bypass_p, returning false rather than aborting
19659 with assertion error when arguments do not satisfy the
19660 requirements of store data bypass.
19661 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
19662 rs6000_store_data_bypass_p.
19664 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
19666 * config/xtensa/xtensa-protos.h
19667 (xtensa_initial_elimination_offset): New declaration.
19668 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
19669 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
19670 macro definition, add case for FRAME_POINTER_REGNUM when
19671 FRAME_GROWS_DOWNWARD.
19672 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
19673 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
19674 xtensa_initial_elimination_offset.
19676 2017-05-08 Nathan Sidwell <nathan@acm.org>
19678 * doc/invoke.texi: Alphabetize -fdump options.
19680 2017-05-08 Martin Sebor <msebor@redhat.com>
19682 PR translation/80280
19683 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
19685 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
19687 * target.def (compute_frame_layout): New optional target hook.
19688 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
19689 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
19690 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
19692 * reload1.c (verify_initial_elim_offsets): Likewise.
19693 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
19694 (use_simple_return_p): Call arm_compute_frame_layout if needed.
19695 (arm_get_frame_offsets): Split up into this ...
19696 (arm_compute_frame_layout): ... and this function.
19698 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
19700 * config/aarch64/constraints.md (Usa): New constraint.
19701 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
19703 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
19705 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
19706 with_multilib_list after it has been checked.
19708 2017-05-08 Richard Biener <rguenther@suse.de>
19710 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
19711 (bitmap_set_subtract_values): Likewise.
19713 2017-05-08 Richard Biener <rguenther@suse.de>
19715 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
19716 (gimple_assign_nonzero): ... this and remove strict_overflow_p
19718 (gimple_stmt_nonzero_warnv_p): Rename to ...
19719 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
19721 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
19722 (extract_range_basic): Adjust, do not disable propagation on
19723 strict overflow sensitive simplification.
19724 (vrp_visit_cond_stmt): Likewise.
19726 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
19728 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
19729 body size unconditionally.
19731 2017-05-07 Jeff Law <law@redhat.com>
19734 2017-05-06 Jeff Law <law@redhat.com>
19735 PR tree-optimization/78496
19736 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
19739 PR tree-optimization/78496
19740 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
19741 (simplify_stmt_using_ranges): Call it.
19742 (vrp_dom_walker::before_dom_children): Extract equivalences
19743 from an ASSERT_EXPR with an equality comparison against a
19746 2017-05-06 Jeff Law <law@redhat.com>
19748 PR tree-optimization/78496
19749 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
19752 PR tree-optimization/78496
19753 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
19754 (simplify_stmt_using_ranges): Call it.
19755 (vrp_dom_walker::before_dom_children): Extract equivalences
19756 from an ASSERT_EXPR with an equality comparison against a
19759 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
19761 * lra-constraints.c (lra_copy_reg_equiv): New function.
19762 (split_reg): Use it to copy equivalence information from the
19763 original register to the spill register.
19765 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
19767 PR rtl-optimization/75964
19768 * simplify-rtx.c (simplify_const_relational_operation): Remove
19769 invalid handling of comparisons of integer ABS.
19771 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
19773 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
19774 initialize to zero.
19775 (init_regs): Remove declaration.
19776 (function_arg_advance_32): Initialize error_p as boolean variable.
19778 2017-05-05 Nathan Sidwell <nathan@acm.org>
19780 * store-motion.c (remove_reachable_equiv_notes): Reformat long
19781 lines. Use for (;;).
19783 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19785 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
19786 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
19787 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
19788 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
19789 VF=2 that require versioning.
19791 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19793 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
19796 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19798 * diagnostic.h (diagnostic_override_option_index): Convert from
19799 macro to inline function.
19801 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19803 * diagnostic.c (last_module_changed_p): New function.
19804 (set_last_module): New function.
19805 (diagnostic_report_current_module): Convert macro usage to
19806 the above functions.
19807 * diagnostic.h (diagnostic_context::last_module): Strengthen
19808 from const line_map * to const line_map_ordinary *.
19809 (diagnostic_last_module_changed): Delete macro.
19810 (diagnostic_set_last_module): Delete macro.
19812 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19814 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
19815 with diagnostic_report_diagnostic.
19816 (diagnostic_n_impl_richloc): Likewise.
19817 * diagnostic.h (report_diagnostic): Delete macro.
19818 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
19819 with diagnostic_report_diagnostic.
19820 * substring-locations.c (format_warning_va): Likewise.
19822 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19824 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
19825 save/restor of format_spec. Move option-printing code to...
19826 (print_option_information): ...this new function, and
19827 reimplement by simply printing to the pretty_printer,
19828 rather than appending to the format string.
19830 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19832 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
19833 handling logic into...
19834 (update_effective_level_from_pragmas): ...this new function.
19836 2017-05-04 Andrew Waterman <andrew@sifive.com>
19838 * config/riscv/riscv.opt (mstrict-align): New option.
19839 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
19840 (SLOW_UNALIGNED_ACCESS): Define.
19841 (riscv_slow_unaligned_access): Declare.
19842 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
19844 (riscv_slow_unaligned_access): New variable.
19845 (rocket_tune_info): Set slow_unaligned_access to true.
19846 (optimize_size_tune_info): Set slow_unaligned_access to false.
19847 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
19848 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
19849 (riscv_option_override): Set riscv_slow_unaligned_access.
19850 * doc/invoke.texi: Add -mstrict-align to RISC-V.
19852 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
19854 * config/riscv/riscv.md: Unify indentation.
19856 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
19861 * config/rs6000/rs6000.md (u code attribute): Add FIX and
19863 (extendsi<mode>2): Add support for doing sign extension via
19864 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
19865 don't have ISA 3.0 instructions.
19866 (extendsi<mode>2 splitter): Likewise.
19867 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
19868 generate the normal insns since SImode can now go in vector
19869 registers. Disallow the special UNSPECs needed for previous
19870 machines to hide SImode being used. Add new insns
19871 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
19872 (fix_trunc<mode>si2_stfiwx): Likewise.
19873 (fix_trunc<mode>si2_internal): Likewise.
19874 (fixuns_trunc<mode>si2): Likewise.
19875 (fixuns_trunc<mode>si2_stfiwx): Likewise.
19876 (fctiw<u>z_<mode>_smallint): Likewise.
19877 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
19878 of floating point to 32-bit integer from doing a direct move to
19879 the GPR registers to do a store.
19880 (fctiwz_<mode>): Break long line.
19882 2017-05-05 Bin Cheng <bin.cheng@arm.com>
19884 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
19885 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
19886 (addr_list, addr_offset_valid_p): New.
19887 (split_address_groups): Check offset validity with above function.
19888 (gt-tree-ssa-loop-ivopts.h): Include header file.
19890 2017-05-05 Nathan Sidwell <nathan@acm.org>
19892 * config.gcc (arm*-*-*): Add missing 'fi'.
19894 2017-05-05 Steve Ellcey <sellcey@cavium.com>
19896 * doc/invoke.texi (-fopt-info): Explicitly say order of options
19897 included in -fopt-info does not matter.
19898 * doc/optinfo.texi (-fopt-info): Fix description of default
19899 behavour. Explicitly say order of options included in -fopt-info
19902 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
19904 * config.gcc: Allow combinations of aprofile and rmprofile values for
19905 --with-multilib-list.
19906 * config/arm/t-multilib: New file.
19907 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
19908 variables. Remove setting of ISA and floating-point ABI in
19909 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
19910 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
19911 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
19912 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
19914 * config/arm/t-rmprofile: Likewise except for the matches changes.
19915 * doc/install.texi (--with-multilib-list): Document the combination of
19916 aprofile and rmprofile values and warn about pitfalls in doing that.
19918 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
19920 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
19921 (movdi_aarch64): Likewise.
19923 2017-05-05 Jakub Jelinek <jakub@redhat.com>
19925 PR tree-optimization/80632
19926 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
19928 (build_arrays): Initialize it for virtual phis.
19929 (fix_phi_nodes): Use it for virtual phis.
19931 PR tree-optimization/80558
19932 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
19933 [x, y] op z into [x op, y op z] for op & or | if conditions
19936 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
19937 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19940 * config/arm/arm.md (use_literal_pool): Remove.
19941 (64-bit immediate split): No longer takes cost into consideration
19942 if arm_disable_literal_pool is enabled.
19943 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
19944 used when arm_disable_literal_pool is enabled.
19945 (arm_max_const_double_inline_cost): Remove use of
19946 arm_disable_literal_pool.
19947 (push_minipool_fix): Add assert.
19948 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
19949 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
19950 (no_literal_pool_sf_immediate): New.
19952 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19954 PR tree-optimization/80613
19955 * tree-ssa-dce.c (propagate_necessity): Remove cases for
19956 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
19958 2017-05-05 Richard Biener <rguenther@suse.de>
19960 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
19962 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
19964 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
19965 of this flag from insn conditions due to removal from r247495.
19967 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
19969 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
19971 (arm_early_store_addr_dep_ptr): Likewise.
19972 * config/arm/aarch-common-protos.h
19973 (arm_early_load_addr_dep_ptr): Add prototype.
19974 (arm_early_store_addr_dep_ptr): Likewise.
19975 * config/arm/cortex-a53.md: Add new bypasses.
19977 2017-05-05 Jakub Jelinek <jakub@redhat.com>
19979 * tree.c (next_type_uid): Change type to unsigned.
19980 (type_hash_canon): Decrement back next_type_uid if
19981 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
19982 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
19985 2017-05-04 Martin Sebor <msebor@redhat.com>
19987 * builtins.c: Fix a trivial typo in a comment.
19989 PR middle-end/79234
19990 * builtins.c (check_sizes): Adjust to handle reading past the end.
19991 Avoid printing excessive upper bound of ranges. Use %E to print
19992 tree nodes instead of converting them to %wu.
19993 (expand_builtin_memchr): New function.
19994 (compute_dest_size): Rename...
19995 (compute_objsize): ...to this.
19996 (expand_builtin_memcpy): Adjust.
19997 (expand_builtin_mempcpy): Adjust.
19998 (expand_builtin_strcat): Adjust.
19999 (expand_builtin_strcpy): Adjust.
20000 (check_strncat_sizes): Adjust.
20001 (expand_builtin_strncat): Adjust.
20002 (expand_builtin_strncpy): Adjust and simplify.
20003 (expand_builtin_memset): Adjust.
20004 (expand_builtin_bzero): Adjust.
20005 (expand_builtin_memcmp): Adjust.
20006 (expand_builtin): Handle memcmp.
20007 (maybe_emit_chk_warning): Check strncat just once.
20009 2017-05-04 Martin Sebor <msebor@redhat.com>
20011 PR preprocessor/79214
20012 PR middle-end/79222
20013 PR middle-end/79223
20014 * builtins.c (check_sizes): Add inlining context and issue
20015 warnings even when -Wno-system-headers is set.
20016 (check_strncat_sizes): Same.
20017 (expand_builtin_strncat): Same.
20018 (expand_builtin_memmove): New function.
20019 (expand_builtin_stpncpy): Same.
20020 (expand_builtin): Handle memmove and stpncpy.
20022 2017-05-04 Bin Cheng <bin.cheng@arm.com>
20024 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
20025 which is not used any more.
20027 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20029 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
20031 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20033 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
20034 (cortexa53_tunings): Likewise.
20035 (cortexa57_tunings): Likewise.
20036 (cortexa72_tunings): Likewise.
20037 (cortexa73_tunings): Likewise.
20039 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20041 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
20042 Set loop alignment to 8.
20044 2017-05-04 Martin Sebor <msebor@redhat.com>
20046 PR translation/80280
20047 * builtins.c (expand_builtin_object_size): Add missing quoting to
20048 %D and like directives.
20049 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
20050 (hsa_type_for_tree_type): Same.
20051 (verify_function_arguments): Same.
20052 * symtab.c (symbol_table::change_decl_assembler_name): Same.
20053 * varasm.c (get_section): Same.
20056 2017-05-04 Martin Sebor <msebor@redhat.com>
20058 PR translation/80280
20059 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
20061 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20063 * config/aarch64/aarch64.c (generic_addrcost_table):
20064 Change HI/TI mode setting.
20066 2017-05-04 Martin Jambor <mjambor@suse.cz>
20068 PR tree-optimization/80622
20069 * tree-sra.c (comes_initialized_p): New function.
20070 (build_accesses_from_assign): Only set write lazily when
20071 comes_initialized_p is false.
20072 (analyze_access_subtree): Use comes_initialized_p.
20073 (propagate_subaccesses_across_link): Assert !comes_initialized_p
20074 instead of testing for PARM_DECL.
20076 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20078 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
20079 constraint on operand 0 to allow more general addressing modes.
20080 Adjust output template.
20081 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
20083 * config/aarch64/aarch64-protos.h
20084 (aarch64_address_valid_for_prefetch_p): Declare prototype.
20085 * config/aarch64/constraints.md (Dp): New address constraint.
20086 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
20089 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
20091 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
20092 update use of estimate_ipcp_clone_size_and_time.
20093 (estimate_local_effects): Update use of
20094 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
20095 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
20096 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
20097 Return nonspecialized time.
20099 2017-05-04 Richard Biener <rguenther@suse.de>
20101 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
20102 for the last VUSE which def dominates the PHI. Directly call
20104 (get_continuation_for_phi_1): Remove.
20106 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
20108 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
20109 to explain the use of truncating division. Cap the number of
20110 iterations to the maximum given by nb_iterations_upper_bound,
20113 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20115 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
20116 * configure: Regenerate.
20117 * config.in: Regenerate.
20118 * config/i386/driver-mingw32.c: new file.
20119 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
20120 * config.host: Link driver-mingw32.o on MinGW host.
20121 * doc/install.texi: Document new --enable-mingw-wildcard configure
20124 2017-05-04 Marek Polacek <polacek@redhat.com>
20126 PR tree-optimization/80612
20127 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
20129 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20130 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
20132 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
20133 (movt splitter): Likewise.
20134 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
20135 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
20136 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
20137 block for Thumb-1 with MOVT.
20138 (thumb2_legitimate_address_p): Move code block ...
20139 (can_avoid_literal_pool_for_label_p): ... into this new function.
20140 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
20142 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
20143 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
20144 "M-profile targets with the MOVT instruction".
20146 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20148 * config/arm/arm-builtins.c (arm_init_builtins): Rename
20149 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
20150 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
20152 2017-05-04 Martin Liska <mliska@suse.cz>
20154 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
20155 variable cond_code.
20157 2017-05-04 Richard Biener <rguenther@suse.de>
20159 * tree.c (array_at_struct_end_p): Handle arrays at struct
20160 end with flexarrays more conservatively. Refactor and treat
20161 arrays of arrays or aggregates more strict. Fix
20162 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
20163 * tree.c (array_at_struct_end_p): Adjust prototype.
20164 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
20165 * gimple-fold.c (get_range_strlen): Likewise.
20166 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
20168 2017-05-04 Richard Biener <rguenther@suse.de>
20170 PR tree-optimization/31130
20171 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
20173 (supports_overflow_infinity): Likewise.
20174 (is_negative_overflow_infinity): Likewise.
20175 (is_positive_overflow_infinity): Likewise.
20176 (is_overflow_infinity): Likewise.
20177 (stmt_overflow_infinity): Likewise.
20178 (overflow_infinity_range_p): Likewise.
20179 (usable_range_p): Remove as always returning true.
20180 (make_overflow_infinity): Remove.
20181 (negative_overflow_infinity): Likewise.
20182 (positive_overflow_infinity): Likewise.
20183 (avoid_overflow_infinity): Likewise.
20184 (set_value_range): Adjust accordingly.
20185 (set_value_range_to_nonnegative): Likewise, remove now unused
20186 overflow_infinity arg.
20187 (vrp_operand_equal_p): Adjust.
20188 (update_value_range): Likewise.
20189 (range_int_cst_singleton_p): Likewise.
20190 (operand_less_p): Likewise.
20191 (compare_values_warnv): Likewise.
20192 (extract_range_for_var_from_comparison_expr): Likewise.
20193 (vrp_int_const_binop): Likewise.
20194 (zero_nonzero_bits_from_vr): Likewise.
20195 (extract_range_from_multiplicative_op_1): Likewise.
20196 (extract_range_from_binary_expr_1): Likewise.
20197 (extract_range_from_unary_expr): Likewise.
20198 (extract_range_from_comparison): Likewise.
20199 (extract_range_basic): Likewise.
20200 (adjust_range_with_scev): Likewise.
20201 (compare_ranges): Likewise.
20202 (compare_range_with_value): Likewise.
20203 (dump_value_range): Likewise.
20204 (test_for_singularity): Likewise, remove strict_overflow_p parameter
20206 (simplify_cond_using_ranges): Adjust.
20208 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
20210 * brig-builtins.def: Added a builtin for class_f64.
20211 * builtin-types.def: Added a builtin type needed by class_f64.
20213 2017-05-03 Jason Merrill <jason@redhat.com>
20215 * timevar.def: Add TV_CONSTEXPR.
20217 2017-05-03 David Malcolm <dmalcolm@redhat.com>
20219 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
20221 2017-05-03 Martin Jambor <mjambor@suse.cz>
20223 * ipa-prop.c (ipa_update_after_lto_read): Removed.
20224 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
20225 * ipa-cp.c (ipcp_propagate_stage): Do not call
20226 ipa_update_after_lto_read.
20227 * ipa-inline.c (ipa_inline): Likewise.
20229 2017-05-03 Martin Jambor <mjambor@suse.cz>
20231 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
20232 tag. Added a default constructor and a destructor.
20233 (ipa_edge_args_sum_t): New class;
20234 (ipa_edge_args_sum): Declare.
20235 (ipa_edge_args_vector): Remove declaration.
20236 (IPA_EDGE_REF): Use ipa_edge_args_sum.
20237 (ipa_free_edge_args_substructures): Remove declaration.
20238 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
20239 (ipa_edge_args_info_available_for_edge_p): Likewise.
20240 * ipa-prop.c (ipa_edge_args_vector): Removed.
20241 (edge_removal_hook_holder): Likewise.
20242 (edge_duplication_hook_holder): Likewise.
20243 (ipa_edge_args_sum): New variable.
20244 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
20245 ipa_edge_args_vector.
20246 (ipa_free_edge_args_substructures): Likewise.
20247 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
20248 ipa_edge_args_vector.
20249 (ipa_edge_removal_hook): Turned into method
20250 ipa_edge_args_sum_t::remove.
20251 (ipa_edge_duplication_hook): Turned into method
20252 ipa_edge_args_sum_t::duplicate.
20253 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
20254 registering edge hooks.
20255 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
20256 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
20257 ipa_edge_args_sum instead of ipa_edge_args_vector.
20258 * ipa-profile.c (ipa_profile): Likewise.
20260 2017-05-03 Martin Jambor <mjambor@suse.cz>
20262 * symbol-summary.h (function_summary): New method exists.
20263 (function_summary::symtab_removal): Deallocate through release.
20264 (call_summary): New class.
20265 (gt_ggc_mx): New overload.
20266 (gt_pch_nx): Likewise.
20267 (gt_pch_nx): Likewise.
20269 2017-05-03 Jeff Law <law@redhat.com>
20271 PR tree-optimization/78496
20272 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
20273 from simplify_cond_using_ranges. Split off code to walk
20274 backwards through casts into ...
20275 (simplify_cond_using_ranges_2): New function.
20276 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
20277 (execute_vrp): After identifying jump threads, call
20278 simplify_cond_using_ranges_2.
20280 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
20283 * ipa-inline.h (inline_summary): Add ctor.
20284 (create_ggc): Do not use ggc_cleared_alloc.
20286 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
20287 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20289 * gcc.c (handle_braces): Support escaping in switch matching
20291 * doc/invoke.texi (Spec Files): Document it.
20292 Remove superfluous @code markup in items.
20294 2017-05-03 David Malcolm <dmalcolm@redhat.com>
20296 * diagnostic-show-locus.c (struct column_range): New struct.
20297 (get_affected_columns): New function.
20298 (get_printed_columns): New function.
20299 (struct correction): New struct.
20300 (correction::ensure_capacity): New function.
20301 (correction::ensure_terminated): New function.
20302 (struct line_corrections): New struct.
20303 (line_corrections::~line_corrections): New dtor.
20304 (line_corrections::add_hint): New function.
20305 (layout::print_trailing_fixits): Reimplement in terms of the new
20307 (selftest::test_overlapped_fixit_printing): New function.
20308 (selftest::diagnostic_show_locus_c_tests): Call it.
20310 2017-05-03 Nathan Sidwell <nathan@acm.org>
20312 Canonicalize canonical type hashing
20313 * tree.h (type_hash_canon_hash): Declare.
20314 * tree.c (type_hash_list, attribute_hash_list): Move into
20315 type_hash_canon_hash.
20316 (build_type_attribute_qual_variant): Break out hash code calc into
20317 type_hash_canon_hash.
20318 (type_hash_canon_hash): New. Generic type hash computation.
20319 (build_range_type_1, build_array_type_1, build_function_type,
20320 build_method_type_directly, build_offset_type, build_complex_type,
20321 make_vector_type): Call it.
20323 2017-05-03 Richard Biener <rguenther@suse.de>
20325 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
20326 When all DRs have unknown misaligned do not always peel
20327 when there is a store but apply the same costing model as if
20328 there were only loads.
20330 2017-05-03 Richard Biener <rguenther@suse.de>
20333 PR tree-optimization/80492
20334 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
20335 compare_base_decls returning dont-know properly.
20337 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
20339 * config/arm/iterators.md (CCSI): New mode iterator.
20340 (arch): New mode attribute.
20341 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
20342 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
20343 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
20344 code iterator for success result mode.
20345 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
20346 the corresponding new insn generators.
20348 2017-05-03 Bin Cheng <bin.cheng@arm.com>
20351 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20352 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
20354 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
20356 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
20357 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
20358 (DDR_A): Wrap DDR argument in brackets.
20359 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
20360 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
20361 (DDR_REVERSED_P): Likewise.
20363 2017-05-03 Jakub Jelinek <jakub@redhat.com>
20365 PR tree-optimization/79472
20366 * tree-switch-conversion.c (struct switch_conv_info): Add
20367 contiguous_range and default_case_nonstandard fields.
20368 (collect_switch_conv_info): Compute contiguous_range and
20369 default_case_nonstandard fields, don't clear final_bb if
20370 contiguous_range and only the default case doesn't have the required
20372 (check_all_empty_except_final): Set default_case_nonstandard instead
20373 of failing if contiguous_range and the default case doesn't have empty
20375 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
20376 and only the default case doesn't have the required constants. Skip
20378 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
20379 if default_case_nonstandard.
20380 (build_constructors): Build constant 1 just once. Assert that default
20381 values aren't inserted in between cases if contiguous_range. Skip
20383 (build_arrays): Skip virtual phis.
20384 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
20385 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
20386 Handle virtual phis.
20387 (gen_inbound_check): Handle default_case_nonstandard case.
20388 (process_switch): Adjust check_final_bb caller. Call
20389 gather_default_values with the first non-default case instead of
20390 default case if default_case_nonstandard.
20392 2017-05-02 Nathan Sidwell <nathan@acm.org>
20394 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
20395 check. Fix formatting.
20397 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
20399 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
20400 errors when comparing specialized and unspecialized times.
20402 2017-05-02 David Malcolm <dmalcolm@redhat.com>
20404 * diagnostic-show-locus.c
20405 (layout::should_print_annotation_line_p): Make private.
20406 (layout::print_annotation_line): Make private.
20407 (layout::annotation_line_showed_range_p): Make private.
20408 (layout::show_ruler): Make private.
20409 (layout::print_source_line): Make private. Pass in line and
20410 line_width, rather than calling location_get_source_line. Drop
20412 (layout::print_leading_fixits): New method.
20413 (layout::print_any_fixits): Rename to...
20414 (layout::print_trailing_fixits): ...this, and make private.
20415 Don't print newline fixits.
20416 (diagnostic_show_locus): Move logic for printing one row into...
20417 (layout::print_line): ...this new function. Move the
20418 location_get_source_line call and error-handling from
20419 print_source_line to here. Call print_leading_fixits, and rename
20420 print_any_fixits to print_trailing_fixits.
20421 (selftest::test_fixit_insert_containing_newline): Update now that
20422 newlines are partially supported.
20423 (selftest::test_fixit_insert_containing_newline_2): New test.
20424 (selftest::test_fixit_replace_containing_newline): Update comments.
20425 (selftest::diagnostic_show_locus_c_tests): Call the new test.
20426 * edit-context.c (class added_line): New class.
20427 (class edited_line): Describe newline handling in comment.
20428 (edited_line::actually_edited_p): New method.
20429 (edited_line::print_content): Delete redundant decl.
20430 (edited_line::m_predecessors): New field.
20431 (edited_file::print_content): Call edited_line::print_content.
20432 (edited_file::print_diff): Update to support newlines.
20433 (edited_file::print_diff_hunk): Likewise.
20434 (edited_file::print_run_of_changed_lines): New function.
20435 (edited_file::print_diff_line): Convert to...
20436 (print_diff_line): ...this.
20437 (edited_file::get_effective_line_count): New function.
20438 (edited_line::edited_line): Initialize new field m_predecessors.
20439 (edited_line::~edited_line): Clean up m_predecessors.
20440 (edited_line::apply_fixit): Handle newlines.
20441 (edited_line::get_effective_line_count): New function.
20442 (edited_line::print_content): New function.
20443 (edited_line::print_diff_lines): New function.
20444 (selftest::test_applying_fixits_insert_containing_newline): New
20446 (selftest::test_applying_fixits_replace_containing_newline): New
20448 (selftest::insert_line): New function.
20449 (selftest::test_applying_fixits_multiple_lines): Add example of
20451 (selftest::edit_context_c_tests): Call the new tests.
20453 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20455 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
20456 parameter cand. Update dump information.
20457 (get_computation_cost): Update uses.
20459 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20461 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
20462 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
20463 (get_computation_at, rewrite_use_address): Update use of
20464 get_computation_aff.
20466 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20468 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
20469 (get_computation): Delete.
20470 (get_computation_cost): Implement like get_computation_cost_at.
20471 Use get_computation_at.
20472 (get_computation_cost_at): Delete.
20473 (rewrite_use_nonlinear_expr): Use get_computation_at.
20474 (rewrite_use_compare, remove_unused_ivs): Ditto.
20476 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20478 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
20480 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20482 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
20483 (ivopts_global_cost_for_size): Rename parameter and update uses.
20484 (iv_ca_recount_cost): Update uses.
20485 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
20486 candidates seperately in n_invs and n_cands.
20487 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
20489 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20491 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
20492 (find_inv_vars_cb): New.
20493 (find_depends): Renamed to ...
20494 (find_inv_vars): ... this.
20495 (add_candidate_1, force_var_cost): Call find_inv_vars.
20496 (split_address_cost, determine_group_iv_cost_cond): Ditto.
20498 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20500 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
20501 inv_vars. Add inv_exprs.
20502 (struct iv_cand): Rename depends_on to inv_vars.
20503 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
20504 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
20505 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
20506 (dump_cand): Dump inv_vars.
20507 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
20508 (record_invariant, find_depends, add_candidate_1): Ditto.
20509 (set_group_iv_cost, force_var_cost): Ditto.
20510 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
20511 (get_computation_cost_at, get_computation_cost): Ditto.
20512 (determine_group_iv_cost_generic): Ditto.
20513 (determine_group_iv_cost_address): Ditto.
20514 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
20515 (determine_group_iv_costs): Ditto.
20516 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
20517 (iv_ca_set_remove_invariants): Renamed to ...
20518 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
20519 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
20520 (iv_ca_set_add_invariants): Renamed to ...
20521 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
20522 (iv_ca_set_cp): Use iv_ca_set_add_invs.
20523 (iv_ca_has_deps): Support inv_vars and inv_exprs.
20524 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
20525 (create_new_ivs): Remove useless dump.
20527 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20529 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
20531 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
20532 (iv_ca_set_no_cp, create_new_iv): Ditto.
20534 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20536 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
20538 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
20540 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
20541 function tree_check2.
20543 2017-05-02 Martin Liska <mliska@suse.cz>
20545 * doc/gcov.texi: Add missing preposition.
20546 * gcov.c (function_info::function_info): Properly fill up
20547 all member variables.
20549 2017-05-02 Tamar Christina <tamar.christina@arm.com>
20551 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
20553 2017-05-02 Tamar Christina <tamar.christina@arm.com>
20555 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
20557 2017-05-02 Martin Liska <mliska@suse.cz>
20560 * lto-streamer-in.c (lto_read_tree_1): Remove
20561 LTO_STREAMER_DEBUG.
20562 * lto-streamer.c (struct tree_hash_entry): Likewise.
20563 (struct tree_entry_hasher): Likewise.
20564 (tree_entry_hasher::hash): Likewise.
20565 (tree_entry_hasher::equal): Likewise.
20566 (lto_streamer_init): Likewise.
20567 (lto_orig_address_map): Likewise.
20568 (lto_orig_address_get): Likewise.
20569 (lto_orig_address_remove): Likewise.
20570 * lto-streamer.h: Likewise.
20571 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
20572 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
20574 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
20576 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
20577 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
20578 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
20579 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
20580 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
20581 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
20582 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
20583 (mm_maskz_sub_ss): New intrinsics.
20584 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20585 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
20586 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
20587 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
20588 (__builtin_ia32_subss_mask_round): New builtins.
20589 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20590 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20591 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
20593 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
20594 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
20596 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
20599 2017-05-02 Martin Jambor <mjambor@suse.cz>
20601 PR tree-optimization/78687
20602 * tree-sra.c (access): New field parent.
20603 (process_subtree_disqualification): New function.
20604 (disqualify_candidate): Call it.
20605 (build_accesses_from_assign): Reset write flag if creating an
20607 (build_access_subtree): Fill in parent field and also prpagate
20608 down grp_write flag.
20609 (create_artificial_child_access): New parameter set_grp_write, set
20610 grp_write to its value.
20611 (propagate_subaccesses_across_link): Also propagate grp_write flag
20613 (propagate_all_subaccesses): Push the closest parent back to work
20614 queue if add_access_to_work_queue returned true.
20616 2017-05-02 Richard Biener <rguenther@suse.de>
20618 * common.opt (fstrict-overflow): Alias negative to fwrapv.
20619 * doc/invoke.texi (fstrict-overflow): Remove all traces of
20620 -fstrict-overflow documentation.
20621 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
20622 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
20623 flag_strict_overflow.
20624 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
20625 * lto-opts.c (lto_write_options): Do not stream it.
20626 * lto-wrapper.c (merge_and_complain): Do not handle it.
20627 * opts.c (default_options_table): Do not set -fstrict-overflow.
20628 (finish_options): Likewise do not clear it when sanitizing.
20629 * simplify-rtx.c (simplify_const_relational_operation): Do not
20630 test flag_strict_overflow.
20632 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
20634 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
20635 using enabled attribute.
20636 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
20637 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
20638 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
20639 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
20640 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
20641 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
20642 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
20643 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
20644 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
20645 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
20647 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
20649 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
20651 2017-05-02 Richard Biener <rguenther@suse.de>
20653 PR tree-optimization/80591
20655 2017-04-10 Richard Biener <rguenther@suse.de>
20657 * tree-ssa-structalias.c (find_func_aliases): Properly handle
20660 2017-05-02 Richard Biener <rguenther@suse.de>
20662 PR tree-optimization/80549
20663 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
20664 (cleanup_tree_cfg_noloop): Create forwarders to known loop
20665 headers if they do not have a preheader.
20667 2017-05-02 Martin Liska <mliska@suse.cz>
20670 * common.opt: Fix typo.
20671 * doc/invoke.texi: Likewise.
20673 2017-05-01 Jan Beulich <jbeulich@suse.com>
20675 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
20676 swapping, add (x,x,m,x,n) alternative.
20678 2017-05-01 Nathan Sidwell <nathan@acm.org>
20680 * calls.c (combine_pending_stack_adjustment_and_call): Remove
20681 unnecessary unadjusted_alignment check.
20683 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
20686 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
20688 * gimplify.c (gimplify_cilk_detach): New function.
20689 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
20690 * tree-core.h: Document EXPR_CILK_SPAWN.
20691 * tree.h (EXPR_CILK_SPAWN): Define.
20693 2017-05-01 David Malcolm <dmalcolm@redhat.com>
20695 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
20696 to use new fixit_hint representation, using the "replace" logic.
20697 (get_line_span_for_fixit_hint): Likewise.
20698 (layout::print_any_fixits): Likewise.
20699 (selftest::test_one_liner_many_fixits): Rename to...
20700 (selftest::test_one_liner_many_fixits_1): ...this, and update
20701 comment and expected output to reflect that the multiple fix-it
20702 hints are now consolidated into one insertion.
20703 (selftest::test_one_liner_many_fixits_2): New test.
20704 (selftest::test_diagnostic_show_locus_one_liner): Update for
20706 (selftest::test_fixit_consolidation): Update for fix-it API
20708 * diagnostic.c (print_parseable_fixits): Likewise.
20709 * edit-context.c (edited_line::m_line_events): Convert from
20710 auto_vec <line_event *> to auto_vec <line_event>.
20711 (class line_event): Convert from abstract base class to a concrete
20712 class, taking over the role of replace_event.
20713 (class insert_event): Delete.
20714 (class replace_event): Rename to class line_event. Convert to
20716 (edit_context::add_fixits): Reimplement.
20717 (edit_context::apply_insert): Delete.
20718 (edit_context::apply_replace): Rename to...
20719 (edit_context::apply_fixit): ...this. Convert to half-open range.
20720 (edited_file::apply_insert): Delete.
20721 (edited_file::apply_replace): Rename to...
20722 (edited_file::apply_fixit): ...this.
20723 (edited_line::~edited_line): Drop deletion of events.
20724 (edited_line::apply_insert): Delete.
20725 (edited_line::apply_replace): Rename to...
20726 (edited_line::apply_fixit): ...this. Convert to half-open range.
20727 Update for change to type of m_line_events.
20728 * edit-context.h (edit_context::apply_insert): Delete.
20729 (edit_context::apply_replace): Rename to...
20730 (edit_context::apply_fixit): ...this.
20732 2017-05-01 Martin Sebor <msebor@redhat.com>
20734 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
20737 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
20740 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
20741 __get_cpuid_max returns 0.
20742 (__get_cpuid_count): Ditto.
20744 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
20746 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
20747 replacement expression is another instance of one of its arguments.
20749 2017-05-01 Jakub Jelinek <jakub@redhat.com>
20752 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
20753 check for stack push/pop autoinc.
20754 * config/i386/i386.c (ix86_agi_dependent): Return false
20755 if the only reason why modified_in_p returned true is that
20756 addr is SP based and set_insn is a push or pop.
20758 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
20760 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
20763 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
20766 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
20767 (account_size_time): Use two predicates - exec_pred and
20769 (evaluate_conditions_for_known_args): Compute both clause and
20771 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
20772 (inline_summary_t::duplicate): Update.
20773 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
20775 (compute_inline_parameters): Likewise.
20776 (estimate_edge_size_and_time): Update caluclation of time.
20777 (estimate_node_size_and_time): Compute both time and nonspecialized
20779 (estimate_ipcp_clone_size_and_time): Update.
20780 (inline_merge_summary): Update.
20781 (do_estimate_edge_time): Update.
20782 (do_estimate_edge_size): Update.
20783 (do_estimate_edge_hints): Update.
20784 (inline_read_section, inline_write_summary): Stream both new predicates.
20785 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
20787 (compute_inlined_call_time): Cleanup.
20788 (big_speedup_p): Update.
20789 (edge_badness): Update.
20790 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
20791 (size_time_entry): Replace predicate by exec_predicate and
20792 nonconst_predicate.
20793 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
20794 (estimate_edge_time): Return also nonspec_time.
20795 (reset_edge_growth_cache): Update.
20797 2017-04-29 Jakub Jelinek <jakub@redhat.com>
20799 PR rtl-optimization/80491
20800 * ifcvt.c (noce_process_if_block): When looking for x setter
20801 with missing else_bb, don't check only the insn right before
20802 cond_earliest, but look for the last insn that x is modified in
20803 within the same bb.
20805 PR rtl-optimization/80491
20806 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
20808 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
20810 PR tree-optimization/80487
20811 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
20813 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20815 PR tree-optimization/79697
20816 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
20817 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
20818 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
20820 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
20821 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
20823 2017-04-28 Martin Sebor <msebor@redhat.com>
20825 PR tree-optimization/80523
20826 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
20827 (init_target_to_host_charmap, target_to_host, target_strtol10): New
20829 (maybe_warn, format_directive, parse_directive): Use new functions.
20830 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
20832 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
20834 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
20836 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
20838 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
20839 target_header_dir): Set correctly.
20840 * configure: Regenerated.
20841 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
20842 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
20843 instead of SYSTEM_HEADER_DIR.
20845 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
20847 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
20848 (estimate_local_effects): Likewise.
20849 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
20850 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
20851 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
20852 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
20853 do_estimate_edge_time, estimate_edge_time): Likewise.
20854 * ipa-inline-analysis.c (estimate_node_size_and_time,
20855 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
20856 (estimate_time_after_inlining): Remove.
20858 2017-04-28 Martin Liska <mliska@suse.cz>
20860 * doc/gcov.texi: Enhance documentation of gcov.
20862 2017-04-28 Martin Liska <mliska@suse.cz>
20864 * doc/gcov.texi: Sort options in alphabetic order.
20865 * doc/gcov-dump.texi: Likewise.
20866 * doc/gcov-tool.texi: Likewise.
20867 * gcov.c (print_usage): Likewise.
20868 * gcov-dump.c (print_usage): Likewise.
20869 * gcov-tool.c (print_merge_usage_message): Likewise.
20870 (print_rewrite_usage_message): Likewise.
20871 (print_overlap_usage_message): Likewise.
20873 2017-04-28 Martin Liska <mliska@suse.cz>
20875 PR gcov-profile/53915
20876 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
20878 2017-04-28 Martin Liska <mliska@suse.cz>
20880 PR gcov-profile/79891
20881 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
20882 is marked by compiler as living on a line.
20883 (get_cycles_count): Remove usage of the union.
20884 (output_intermediate_file): Likewise.
20885 (find_source): Fix GNU coding style.
20886 (accumulate_line_counts): Remove old non-all block mode.
20887 (output_lines): Remove usage of the union.
20888 * profile.c (output_location): Include all BBs, even if
20889 belonging to a same line (and file) as a previous BB.
20891 2017-04-28 Martin Liska <mliska@suse.cz>
20893 * gcov.c (process_args): Handle new argument 'w'.
20894 (read_graph_file): Assign ID to BBs.
20895 (output_branch_count): Display BB # if verbose flag is set.
20896 (output_lines): Likewise for arcs.
20897 (print_usage): Add '--verbose' option help.
20898 * doc/gcov.texi: Document --verbose (-w) option.
20900 2017-04-28 Martin Liska <mliska@suse.cz>
20902 * gcov.c (struct block_location_info): New struct.
20903 (process_file): Fill up the new structure.
20904 (read_graph_file): Replace usage of encoding by the newly added
20906 (add_line_counts): Likewise.
20907 (accumulate_line_counts): Remove usage of the union.
20908 (function_info::function_info): New function.
20909 (function_info::~function_info): Likewise.
20910 (process_file): Call delete instead of release_function.
20911 (release_function): Release the function.
20912 (release_structures): Call delete instead of release_function.
20913 (solve_flow_graph): Replace usage of num_blocks.
20914 (find_exception_blocks): Likewise.
20915 (output_lines): Fix GNU coding style.
20917 2017-04-28 Martin Liska <mliska@suse.cz>
20920 * coverage.c (coverage_remove_note_file): New function.
20921 * coverage.h: Declare the function.
20922 * toplev.c (finalize): Clean if an error has been seen.
20924 2017-04-28 Martin Liska <mliska@suse.cz>
20926 PR gcov-profile/80031
20927 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
20928 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
20929 * gcov.c (read_graph_file): Read just number of blocks.
20930 * profile.c (branch_prob): Do not stream 0 flags per a basic
20933 2017-04-28 Martin Liska <mliska@suse.cz>
20935 * gcov-dump.c (tag_*): Add new argument to declarations.
20936 (dump_gcov_file): Likewise.
20937 (tag_blocks): Add and use new argument depth.
20938 (tag_arcs): Likewise.
20939 (tag_lines): Likewise.
20940 (tag_counters): Likewise.
20941 (tag_summary): Likewise.
20942 (dump_working_sets): Use depth to do a proper indentation.
20944 2017-04-28 Jakub Jelinek <jakub@redhat.com>
20947 * cgraph.h (symtab_node::debug_symtab): No longer inline.
20948 * symtab.c (symtab_node::debug_symtab): Move definition here.
20950 2017-04-28 Richard Biener <rguenther@suse.de>
20952 * lto-streamer.h (LTO_major_version): Bump to 7.
20954 2017-04-28 Richard Biener <rguenther@suse.de>
20956 * tree-vrp.c (assert_info): New struct.
20957 (add_assert_info): New helper.
20958 (register_edge_assert_for_2): Refactor to add asserts to a vector
20960 (register_edge_assert_for_1): Likewise.
20961 (register_edge_assert_for): Likewise.
20962 (finish_register_edge_assert_for): New helper actually registering
20963 asserts where live on edge.
20964 (find_conditional_asserts): Adjust.
20965 (find_switch_asserts): Likewise.
20966 (evrp_dom_walker::try_find_new_range): Generalize.
20967 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
20969 2017-04-27 Marek Polacek <polacek@redhat.com>
20972 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
20973 arg10 and arg11 to itype.
20975 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
20977 * doc/extend.texi (Object Size Checking): Improve grammar.
20979 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
20982 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
20983 that the logic for permitting reciprocal estimates matches that
20986 2017-04-27 Jakub Jelinek <jakub@redhat.com>
20989 * tree.c (type_cache_hasher::equal): Only compare
20990 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
20991 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
20992 non-aggregate element types.
20993 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
20994 about the flag on ARRAY_TYPEs in the comment, formatting fix.
20996 2017-04-27 Richard Biener <rguenther@suse.de>
20998 PR middle-end/80533
20999 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
21000 stripping ARRAY_REFs from MEM_EXPR make sure we're not
21001 keeping a reference to a trailing array.
21003 2017-04-27 Richard Biener <rguenther@suse.de>
21005 PR middle-end/80539
21006 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
21007 being in loop-closed SSA form conservatively.
21008 (chrec_fold_multiply_poly_poly): Likewise.
21010 2017-04-27 Tamar Christina <tamar.christina@arm.com>
21012 PR middle-end/79665
21013 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
21014 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
21016 2017-04-27 Jakub Jelinek <jakub@redhat.com>
21019 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
21020 (aarch64_function_arg_alignment): Return unsigned int again, but still
21021 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
21022 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
21023 Don't emit -Wpsabi note.
21024 (aarch64_function_arg_boundary): Likewise.
21025 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
21028 2017-04-26 Nathan Sidwell <nathan@acm.org>
21030 * tree.h (crc32_unsigned_n): Declare.
21031 (crc32_unsigned, crc32_unsigned): Make inline.
21032 * tree.c (crc32_unsigned_bits): Replace with ...
21033 (crc32_unsigned_n): ... this.
21034 (crc32_unsigned, crc32_byte): Remove.
21035 (crc32_string): Remove unnecessary braces.
21037 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
21039 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
21040 * ipa-inline-analysis.c (MAX_TIME): Remove.
21041 (account_size_time): Use sreal for time.
21042 (dump_inline_summary): Update.
21043 (estimate_function_body_sizes): Update.
21044 (estimate_edge_size_and_time): Update.
21045 (estimate_calls_size_and_time): Update.
21046 (estimate_node_size_and_time): Update.
21047 (inline_merge_summary): Update.
21048 (inline_update_overall_summary): Update.
21049 (estimate_time_after_inlining): Update.
21050 (inline_read_section): Update.
21051 (inline_write_summary): Update.
21052 * ipa-inline.c (compute_uninlined_call_time): Update.
21053 (compute_inlined_call_time): Update.
21054 (recursive_inlining): Update.
21055 (inline_small_functions): Update.
21056 (dump_overall_stats): Update.
21057 * ipa-inline.h: Include sreal.h.
21058 (size_time_entry): Turn time to sreal.
21059 (inline_summary): Turn self_time nad time to sreal.
21061 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
21063 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
21065 (sreal::stream_out, sreal::stream_in): New.
21066 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
21068 2017-04-25 Jakub Jelinek <jakub@redhat.com>
21070 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
21073 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
21076 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
21078 (dimode_scalar_chain::compute_convert_gain): Ditto.
21079 (dimode_scalar_chain::make_vector_copies): Ditto.
21080 (dimode_scalar_chain::convert_reg): Ditto.
21081 (dimode_scalar_chain::convert_insn): Ditto.
21082 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
21083 (VI248_AVX512BW_1): New mode iterator.
21084 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
21085 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
21088 2017-04-25 Martin Sebor <msebor@redhat.com>
21090 PR tree-optimization/80497
21091 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
21092 constants are representable in HOST_WIDE_INT.
21093 (parse_directive): Ditto.
21095 2017-04-25 Martin Sebor <msebor@redhat.com>
21098 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
21099 (new_zero_array): Adjust signature.
21100 (dom_info::dom_init): Used unsigned rather that size_t.
21101 (dom_info::dom_info): Same.
21103 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21104 Jakub Jelinek <jakub@redhat.com>
21107 * config/arm/arm.c: Include gimple.h.
21108 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
21109 returns negative, increment ncrn only if it returned positive.
21110 (arm_needs_doubleword_align): Return int instead of bool,
21111 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
21112 members, but if there is any such non-FIELD_DECL
21113 > PARM_BOUNDARY aligned decl, return -1 instead of false.
21114 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
21115 returns negative, increment nregs only if it returned positive.
21116 (arm_setup_incoming_varargs): Likewise.
21117 (arm_function_arg_boundary): Emit -Wpsabi note if
21118 arm_needs_doubleword_align returns negative, return
21119 DOUBLEWORD_ALIGNMENT only if it returned positive.
21121 2017-04-25 Marek Polacek <polacek@redhat.com>
21124 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
21125 first argument to type.
21127 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
21130 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
21131 type checks to test for compatibility instead of equality.
21133 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21134 Jakub Jelinek <jakub@redhat.com>
21137 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
21139 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
21140 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
21141 the alignment computation, but return their maximum in warn_alignment.
21142 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
21143 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
21145 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
21146 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
21149 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21151 * config/arc/simdext.md (dmpyh): Fix typo.
21153 2017-04-25 Richard Biener <rguenther@suse.de>
21155 PR tree-optimization/80492
21156 * alias.c (compare_base_decls): Handle registers with asm
21157 specification conservatively.
21158 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
21159 compare_base_decls returning dont-know properly.
21161 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21163 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
21164 (legitimate_offset_address_p): New function.
21165 (arc_legitimate_address_p): Use above function.
21167 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21169 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
21171 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21173 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
21174 ACCH registers whenever they are available.
21176 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21178 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
21179 double regs fix when not used.
21181 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21183 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
21185 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
21186 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
21188 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21190 * config/arc/arc.c (arc_output_addsi): Check for h-register class
21191 when emitting short ADD instructions.
21193 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21195 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
21197 (cmpsi_cc_c_insn): Likewise.
21198 (cbranchsi4_scratch): Compute proper instruction length using
21199 compact_hreg_operand.
21200 * config/arc/predicates.md (compact_hreg_operand): New predicate.
21202 2017-04-25 Richard Biener <rguenther@suse.de>
21204 PR middle-end/80509
21205 * passes.c (pass_manager::pass_manager): Initialize
21206 m_name_to_pass_map.
21208 2017-04-25 Richard Biener <rguenther@suse.de>
21210 PR tree-optimization/79201
21211 * tree-ssa-sink.c (statement_sink_location): Handle calls.
21213 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21216 * config/s390/vector.md: Split MEM->GPR vector moves for
21217 non-s_operand addresses.
21219 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21222 * config/s390/predicates.md (reload_const_wide_int_operand): New
21224 * config/s390/s390.md ("movti"): Remove d/P alternative.
21225 ("movti_bigconst"): New pattern definition.
21227 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
21230 * s390-protos.h (s390_expand_cs_hqi): Removed.
21231 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
21232 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
21233 modes as well as CCZ1mode and CCZmode.
21234 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
21235 signature of s390_emit_compare_and_swap.
21236 (s390_expand_cs_hqi): Likewise, make static.
21237 (s390_expand_cs_tdsi): Generate an explicit compare before trying
21238 compare-and-swap, in some cases.
21239 (s390_expand_cs): Wrapper function.
21240 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
21242 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
21243 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
21244 patterns for small and large integers. Forbid symref memory operands.
21245 Move expander to s390.c. Require cc register.
21246 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
21247 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
21248 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
21249 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
21250 symref memory operands. Remove CC mode and call s390_match_ccmode
21252 ("atomic_exchange<mode>"): Allow and implement all integer modes.
21254 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
21256 * config/s390/s390.md (define_peephole2): New peephole to help
21257 combining the load-and-test pattern with volatile memory.
21259 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
21261 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
21262 with CCZmode for TARGET_Z196.
21264 2017-04-25 Jakub Jelinek <jakub@redhat.com>
21266 PR rtl-optimization/80501
21267 * combine.c (make_compound_operation_int): Set subreg_code to SET
21268 even for AND with mask of the sign bit of mode.
21270 PR rtl-optimization/80500
21271 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
21272 sum's initial value.
21274 2017-04-25 Julian Brown <julian@codesourcery.com>
21275 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21277 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
21279 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
21281 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
21283 2017-04-25 Julian Brown <julian@codesourcery.com>
21284 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21286 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
21287 (thunderx2t99_sha): New Reservation.
21289 2017-04-25 Julian Brown <julian@codesourcery.com>
21290 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21292 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
21293 type for 1-element load.
21295 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
21297 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
21299 2017-04-24 Martin Jambor <mjambor@suse.cz>
21301 PR tree-optimization/80293
21302 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
21303 char arrays not totally scalarizable if it is false.
21304 (analyze_all_variable_accesses): Pass correct value in the new
21305 parameter. Add a statistics counter.
21307 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
21309 PR middle-end/79931
21310 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
21312 2017-04-24 Richard Biener <rguenther@suse.de>
21314 PR tree-optimization/80494
21315 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
21316 out for complex types.
21318 2017-04-24 Richard Biener <rguenther@suse.de>
21320 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
21321 * tree-ssa-sccvn.c (print_scc): Print SCC size.
21322 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
21323 (DFS): Adjust and never fail.
21324 (sccvn_dom_walker::fail): Remove.
21325 (sccvn_dom_walker::before_dom_children): Adjust.
21326 (run_scc_vn): Likewise and never fail.
21327 * tree-ssa-pre.c (pass_pre::execute): Adjust.
21328 (pass_fre::execute): Likewise.
21330 2017-04-24 Richard Biener <rguenther@suse.de>
21332 PR tree-optimization/79725
21333 * tree-ssa-sink.c (statement_sink_location): Return whether
21334 failure reason was zero uses. Move that check later.
21335 (sink_code_in_bb): Deal with zero uses by removing the stmt
21338 2017-04-24 Richard Biener <rguenther@suse.de>
21341 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
21342 pointer-based references.
21344 2017-04-24 Richard Biener <rguenther@suse.de>
21347 * pass_manager.h (pass_manager::operator new): Remove.
21348 (pass_manager::operator delete): Likewise.
21349 * passes.c (pass_manager::operator new): Remove.
21350 (pass_manager::operator delete): Likewise.
21351 (pass_manager::pass_manager): Zero individual pass members.
21353 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
21356 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
21357 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
21358 Check "XEXP (src, 1)" operand here.
21359 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
21360 Check "XEXP (src, 1)" operand here.
21361 (dimode_scalar_chain::make_vector_copies): Detect count register
21362 of a shift instruction. Zero extend count register from QImode
21363 to DImode to satisfy vector shift pattern count operand predicate.
21364 Substitute vector shift count operand with a DImode copy.
21365 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
21368 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
21370 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
21371 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
21372 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
21373 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
21374 (UNSPEC_NOREX_MEM): Remove definition.
21376 2017-04-21 Richard Biener <rguenther@suse.de>
21378 PR tree-optimization/79547
21379 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
21380 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
21381 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
21382 without any constraints.
21384 2017-04-21 Richard Biener <rguenther@suse.de>
21386 PR tree-optimization/78847
21387 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
21389 2017-04-21 Richard Biener <rguenther@suse.de>
21391 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
21392 (build_distinct_type_copy): Likewise.
21393 (build_variant_type_copy): Likewise.
21394 * tree.c (build_qualified_type): Pass down mem-stat info.
21395 (build_distinct_type_copy): Likewise.
21396 (build_variant_type_copy): Likewise.
21398 2017-04-21 Richard Biener <rguenther@suse.de>
21400 PR tree-optimization/80237
21401 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
21403 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
21404 for a simplified result.
21406 2016-04-21 Richard Biener <rguenther@suse.de>
21408 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
21409 sth as strict as a simple_iv but a chrec without symbols and an
21410 operand defined in the loop we are peeling (and not some subloop).
21411 (propagate_constants_for_unrolling): Propagate all constants.
21413 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
21416 * config/i386/i386.c (print_reg): Remove assert for disalowed
21417 regno values, call output_operand_lossage instead.
21419 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
21422 * config/i386/constraints.md (Yc): New register constraint.
21423 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
21424 Use Yc constraint for alternative 2 of operand 0. Remove
21425 preferred_for_speed attribute.
21427 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
21429 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
21430 lastprivate clauses in SIMT case.
21432 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
21434 * doc/invoke.texi (-Wextra-semi): Document new warning option.
21436 2017-04-20 Richard Biener <rguenther@suse.de>
21438 PR tree-optimization/57796
21439 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
21440 as N scalar stores.
21441 (vect_model_load_cost): Cost gathers as N scalar loads.
21443 2017-04-20 Richard Biener <rguenther@suse.de>
21445 * ggc-page.c (ggc_allocated_p): Rename to ...
21446 (safe_lookup_page_table_entry): ... this and return the lookup
21448 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
21450 2017-04-20 Richard Biener <rguenther@suse.de>
21452 PR tree-optimization/80453
21453 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
21454 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
21455 from the conditions.
21456 (vn_phi_eq): Pass them down.
21457 (vn_phi_lookup): Record them.
21458 (vn_phi_insert): Likewise.
21460 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
21462 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
21463 uninitialized variable warning to avoid buffer overrun.
21465 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
21468 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
21469 is suppressed for '{ 0 }' in C.
21471 2017-04-20 Jakub Jelinek <jakub@redhat.com>
21473 * BASE-VER: Set to 8.0.0.
21475 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
21477 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
21478 priority .init_array and .fini_array section with SECTION_NOTYPE
21481 2017-04-20 Jakub Jelinek <jakub@redhat.com>
21483 PR middle-end/80423
21484 * tree.h (build_array_type): Add typeless_storage default argument.
21485 * tree.c (type_cache_hasher::equal): Also compare
21486 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
21487 (build_array_type): Add typeless_storage argument, set
21488 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
21490 (build_nonshared_array_type): Adjust build_array_type_1 caller.
21491 (build_array_type): Likewise. Add typeless_storage argument.
21493 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
21494 Jakub Jelinek <jakub@redhat.com>
21496 PR tree-optimization/80426
21497 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
21498 operation on symbolic operands, also compute the overflow for the
21499 invariant part when the operation degenerates into a negation.
21501 2017-04-19 Jakub Jelinek <jakub@redhat.com>
21504 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
21505 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
21508 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
21510 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
21513 * config/avr/avr.c (tree.h): Include it.
21514 (cgraph.h): Include it.
21515 (avr_encode_section_info): Don't warn for uninitialized progmem
21516 variable if it's just an alias.
21518 2017-04-19 Richard Biener <rguenther@suse.de>
21521 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
21522 when needed by AutoPGO.
21524 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
21527 * doc/lto.texi: Remove an extra 'that'.
21529 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
21531 PR rtl-optimization/80429
21532 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
21533 are only used in debug insns.
21535 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
21536 Vladimir Makarov <vmakarov@redhat.com>
21538 * config/sparc/predicates.md (input_operand): Add comment. Return
21539 true for any memory operand when LRA is in progress.
21540 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
21542 2017-04-18 Jeff Law <law@redhat.com>
21545 * mips.md ({return,simple_return}_internal): Do not overwrite
21548 2017-04-18 Jakub Jelinek <jakub@redhat.com>
21550 PR tree-optimization/80443
21551 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
21552 instead of adding 1, subtract -1 and similarly instead of subtracting
21555 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
21557 PR rtl-optimization/80357
21558 * haifa-sched.c (tmp_bitmap): New variable.
21559 (model_recompute): Handle duplicate use records.
21560 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
21561 (free_global_sched_pressure_data): Free it.
21563 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
21566 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
21567 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21568 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21569 instead of SYSTEM_HEADER_DIR.
21571 2017-04-18 Jeff Law <law@redhat.com>
21573 PR middle-end/80422
21574 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
21575 predecessors after walking up the insn chain.
21577 2017-04-18 Jakub Jelinek <jakub@redhat.com>
21580 * dwarf2out.c (modified_type_die): Try harder not to emit internal
21581 sizetype type into debug info.
21583 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
21586 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
21587 unneeded test for TARGET_UPPER_REGS_SF.
21588 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
21590 2017-04-18 Jakub Jelinek <jakub@redhat.com>
21593 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
21594 instead of gsi_after_labels.
21596 2017-04-18 Jeff Law <law@redhat.com>
21598 * regcprop.c (maybe_mode_change): Avoid creating copies of the
21602 2017-04-13 Jeff Law <law@redhat.com>
21603 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
21604 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
21606 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
21609 * config/avr/avr.c (intl.h): Include it.
21610 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
21612 2017-04-18 Martin Liska <mliska@suse.cz>
21614 PR gcov-profile/78783
21615 * gcov-tool.c (gcov_output_files): Validate that destination
21616 file is either removed by the tool or by a user.
21618 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
21619 Guy Benyei <guybe@mellanox.com>
21621 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
21622 block, and do not negate it, the stored id is already negative.
21624 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
21626 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
21628 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
21631 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
21632 masks of options that should be turned off if the VSX vector
21633 options are turned off.
21634 (OTHER_P8_VECTOR_MASKS): Likewise.
21635 (OTHER_VSX_VECTOR_MASKS): Likewise.
21636 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
21637 rs6000_disable_incompatible_switches to validate no type switches
21639 (rs6000_incompatible_switch): New function to disallow turning on
21640 other vector options if -mno-vsx, -mno-power8-vector, or
21641 -mno-power9-vector are specified.
21643 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21645 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
21647 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21649 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
21650 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
21651 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
21652 (ARG_POINTER_CFA_OFFSET): Likewise.
21654 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21656 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
21657 conditions to take advantage of various optimizations.
21659 2017-04-13 Jeff Law <law@redhat.com>
21661 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
21662 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
21663 (zero_extendsidi2_dext): Likewise.
21665 2017-04-13 Jakub Jelinek <jakub@redhat.com>
21668 * fold-const.c (fold_ternary_loc): Revert
21669 use op0 instead of fold_convert_loc (loc, type, arg0) part of
21672 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
21674 PR rtl-optimization/80343
21675 * lra-remat.c (update_scratch_ops): Assign original hard reg to
21676 new scratch pseudo.
21678 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
21681 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
21682 to ubsan_encode_value.
21684 2017-04-13 Jeff Law <law@redhat.com>
21686 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
21687 appearing in DEBUG_INSNs.
21689 2017-04-13 Martin Liska <mliska@suse.cz>
21691 PR gcov-profile/80413
21692 * gcov-io.c (gcov_write_string): Copy to buffer just when
21693 allocated size is greater than zero.
21695 2017-04-13 Jakub Jelinek <jakub@redhat.com>
21698 * dwarf2out.c (decls_for_scope): Ignore declarations of
21699 current_function_decl in BLOCK_NONLOCALIZED_VARS.
21701 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
21704 * ipa-visibility.c (non_local_p): Fix typos.
21705 (localize_node): When localizing symbol in same comdat group,
21706 dissolve the group only when we know external symbols are going
21708 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
21710 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21712 PR tree-optimization/79390
21713 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
21714 order does not result in usable sequence, retry with reversed operand
21720 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
21721 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
21722 op0 instead of fold_convert_loc (loc, type, arg0).
21724 2017-04-12 Jeff Law <law@redhat.com>
21726 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
21727 has a delay slot in the generated code.
21729 * config/cris/cris.md (cris_preferred_reload_class): Return
21730 GENNONACR_REGS rather than GENERAL_REGS.
21732 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21735 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
21736 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
21737 signedness of the result type.
21739 2017-04-12 Richard Biener <rguenther@suse.de>
21740 Jeff Law <law@redhat.com>
21742 PR tree-optimization/80359
21743 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
21744 trim stores to TARGET_MEM_REFs.
21746 2017-04-12 Richard Biener <rguenther@suse.de>
21748 PR tree-optimization/79390
21749 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
21750 threading case even more.
21752 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
21755 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
21756 for quad_address_p for TImode, instead of just not indexed_address.
21758 2017-04-12 Richard Biener <rguenther@suse.de>
21759 Bernd Edlinger <bernd.edlinger@hotmail.de>
21761 PR middle-end/79671
21762 * alias.c (component_uses_parent_alias_set_from): Handle
21763 TYPE_TYPELESS_STORAGE.
21764 (get_alias_set): Likewise.
21765 * tree-core.h (tree_type_common): Add typeless_storage flag.
21766 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
21767 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
21768 for types containing members with TYPE_TYPELESS_STORAGE.
21769 (place_field): Likewise.
21770 (layout_type): Likewise for ARRAY_TYPE.
21771 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
21772 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
21773 TYPE_TYPELESS_STORAGE.
21774 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
21776 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21779 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
21780 first argument to type.
21782 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21786 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
21787 CONST0_RTX (mode) rather than const0_rtx where appropriate.
21788 (rs6000_expand_binop_builtin): Likewise.
21789 (rs6000_expand_ternop_builtin): Likewise; also add missing
21790 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
21791 vshasigma built-ins.
21792 * doc/extend.texi: Document that vec_xxpermdi's third argument
21793 must be a constant.
21795 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
21797 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
21798 Use shift_const cost parameter when calculating gain of STV shifts.
21800 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
21802 PR rtl-optimization/70478
21803 * lra-constraints.c (process_alt_operands): Check memory for
21804 disfavoring memory insn operand.
21806 2017-04-11 Jakub Jelinek <jakub@redhat.com>
21808 PR middle-end/80100
21809 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
21810 left shift in unsigned HOST_WIDE_INT type.
21812 PR rtl-optimization/80385
21813 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
21814 (not (neg X)) into (plus X -1) for complex or non-integral modes.
21817 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
21818 if they have any depend clauses.
21820 2017-04-11 Martin Liska <mliska@suse.cz>
21823 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
21824 * ipa-split.c (split_function): Create a local comdat symbol
21825 if caller is in a comdat group.
21827 2017-04-11 Martin Liska <mliska@suse.cz>
21830 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
21833 2017-04-11 Martin Sebor <msebor@redhat.com>
21835 PR middle-end/80364
21836 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
21837 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
21839 (directive::set_width, directive::set_precision, format_character):
21841 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
21844 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
21847 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
21848 conflict, set target->arch_name instead of target->cpu_name.
21850 2017-04-11 Richard Biener <rguenther@suse.de>
21852 PR tree-optimization/80374
21853 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
21854 build_zero_cst, remove fold_convertible_p check again.
21856 2017-04-11 Martin Liska <mliska@suse.cz>
21859 * ubsan.c (instrument_object_size): Do not instrument register
21862 2017-04-11 Jakub Jelinek <jakub@redhat.com>
21865 * config/i386/i386-builtin-types.def
21866 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
21867 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
21868 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
21869 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
21870 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
21871 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
21872 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
21873 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
21874 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
21875 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
21876 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
21877 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
21878 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
21879 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
21880 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
21881 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
21882 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
21883 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
21884 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
21885 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
21886 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
21887 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
21888 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
21889 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
21890 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
21891 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
21892 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
21893 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
21894 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
21895 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
21896 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
21897 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
21898 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
21899 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
21900 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
21901 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
21902 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
21903 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
21904 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
21905 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
21906 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
21907 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
21908 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
21909 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
21910 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
21911 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
21913 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
21914 flag to second_arg_count, handle 4 argument function type _COUNT
21915 aliases, handle second_arg_count on second argument rather than last.
21917 2017-04-10 Jeff Law <law@redhat.com>
21919 PR tree-optimization/80374
21920 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
21921 record anything if we can not convert integer_zero_node to the
21924 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21927 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21928 Enhance special handling given to the TARGET_P9_MINMAX option in
21929 relation to certain other options.
21931 2017-04-10 Bin Cheng <bin.cheng@arm.com>
21933 PR tree-optimization/80153
21934 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
21935 remove POINTER_PLUS_EXPR's base part directly, rather than through
21938 2017-04-10 Richard Biener <rguenther@suse.de>
21939 Bin Cheng <bin.cheng@arm.com>
21941 PR tree-optimization/80153
21942 * tree-affine.c (aff_combination_to_tree): Get base pointer from
21943 the first element of pointer type aff_tree. Build result expr in
21945 (add_elt_to_tree): Convert to type unconditionally. Remove other
21946 fold_convert calls.
21947 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
21948 (rewrite_use_nonlinear_expr): Check invariant using iv information.
21950 2017-04-10 Richard Biener <rguenther@suse.de>
21952 * tree-ssa-structalias.c (find_func_aliases): Properly handle
21955 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
21957 PR rtl-optimization/70478
21958 * lra-constraints.c (curr_small_class_check): New.
21959 (update_and_check_small_class_inputs): New.
21960 (process_alt_operands): Update curr_small_class_check. Disfavor
21961 alternative insn memory operands. Check available regs for small
21964 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
21967 * config/mips/mips.opt (-mvirt): Update description.
21968 * doc/invoke.texi (-mvirt): Likewise.
21970 2017-04-10 Richard Biener <rguenther@suse.de>
21972 PR middle-end/80362
21973 * fold-const.c (fold_binary_loc): Look at unstripped ops when
21974 looking for NEGATE_EXPR in -A / -B to A / B folding.
21976 2017-04-10 Martin Liska <mliska@suse.cz>
21978 PR gcov-profile/80224
21979 * gcov.c (print_usage): Fix usage string.
21980 (get_gcov_intermediate_filename): Remove.
21981 (output_gcov_file): Use both for normal and intermediate format.
21982 (generate_results): Do not initialize special file for
21983 intermediate format.
21985 2017-04-10 Richard Biener <rguenther@suse.de>
21987 PR tree-optimization/80304
21988 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
21991 2017-04-10 Nathan Sidwell <nathan@acm.org>
21994 * config/rs6000/rs6000.c (rs6000_vector_type): New.
21995 (rs6000_init_builtins): Use it.
21997 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21999 * config/arm/arm.md (<mrc>): Add mode to SET source.
22000 (<mrrc>): Likewise.
22002 2017-04-10 Richard Biener <rguenther@suse.de>
22004 PR middle-end/80344
22005 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
22007 2017-04-10 Jakub Jelinek <jakub@redhat.com>
22010 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
22011 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
22012 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
22013 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
22014 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
22015 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
22016 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
22017 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
22018 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
22019 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
22020 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
22021 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
22022 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
22023 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
22024 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
22025 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
22026 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
22027 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
22028 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
22029 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
22030 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
22031 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
22032 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
22034 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
22036 PR rtl-optimization/70478
22037 * lra-constraints.c: Reverse the last patch.
22039 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
22041 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
22042 Add comment for WCHAR_T.
22044 2017-04-08 Martin Liska <mliska@suse.cz>
22047 2017-04-07 Martin Liska <mliska@suse.cz>
22050 * ipa-split.c (split_function): Add function part to a same comdat
22053 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22056 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
22058 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
22060 * rs6000/rs6000.c (vec_load_pendulum): Rename...
22061 (vec_pairing): ...to this.
22062 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
22063 (rs6000_sched_init): Adjust for name change.
22064 (struct rs6000_sched_context): Likewise.
22065 (rs6000_init_sched_context): Likewise.
22066 (rs6000_set_sched_context): Likewise.
22068 2017-04-07 Jakub Jelinek <jakub@redhat.com>
22074 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
22076 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
22077 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
22078 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
22080 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
22082 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
22084 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
22086 PR rtl-optimization/70703
22087 * ira-color.c (update_conflict_hard_regno_costs): Use
22088 int64_t instead of HOST_WIDE_INT.
22090 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
22092 PR rtl-optimization/70478
22093 * lra-constraints.c (process_alt_operands): Disfavor alternative
22094 insn memory operands.
22096 2017-04-07 Jeff Law <law@redhat.com>
22098 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
22099 CALL and NOTE_INSN_CALL_ARG_LOCATION.
22101 2017-04-07 Martin Liska <mliska@suse.cz>
22104 * config/aarch64/aarch64.c (aarch64_process_target_attr):
22105 Show error message instead of an ICE.
22107 2017-04-07 Martin Liska <mliska@suse.cz>
22110 * ipa-split.c (split_function): Add function part to a same comdat
22113 2017-04-07 Richard Biener <rguenther@suse.de>
22115 PR middle-end/80341
22116 * tree.c (get_unwidened): Also handle ! for_type case for
22118 * convert.c (do_narrow): Split out from ...
22119 (convert_to_integer_1): ... here. Do not pass final truncation
22120 type to get_unwidened for TRUNC_DIV_EXPR.
22122 2017-04-07 Richard Biener <rguenther@suse.de>
22124 * tree-affine.c (wide_int_ext_for_comb): Take type rather
22126 (aff_combination_const): Adjust.
22127 (aff_combination_scale): Likewise.
22128 (aff_combination_add_elt): Likewise.
22129 (aff_combination_add_cst): Likewise.
22130 (aff_combination_convert): Likewise.
22131 (add_elt_to_tree): Likewise. Remove unused argument.
22132 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
22134 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
22136 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
22138 * config/arm/arm.c (arm_default_short_enums): Use
22139 ARM_DEFAULT_SHORT_ENUMS.
22140 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
22142 2017-04-06 Jakub Jelinek <jakub@redhat.com>
22145 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
22146 members with redundant out-of-class redeclaration.
22148 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
22151 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
22152 * config/i386/i386.md (*zero_extendsidi2):
22153 Add (?*x,*x) and (?*v,*v) alternatives.
22155 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
22158 * config/i386/i386.c (ix86_expand_builtin)
22159 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
22160 mode from insn data. Convert operands to insn operand mode.
22161 Copy operands that don't satisfy insn predicate to a register.
22163 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
22165 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
22168 2017-04-06 Richard Biener <rguenther@suse.de>
22170 PR tree-optimization/80334
22171 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
22172 preserve alignment of accesses.
22174 2017-04-06 Richard Biener <rguenther@suse.de>
22176 PR tree-optimization/80262
22177 * tree-sra.c (build_ref_for_offset): Preserve address-space
22179 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
22180 Drop useless address-space information on MEM_REF offsets.
22182 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
22184 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
22186 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
22188 PR rtl-optimization/70703
22189 * ira-color.c (update_conflict_hard_regno_costs): Use
22190 HOST_WIDE_INT instead of long.
22192 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
22195 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
22196 not defined for x86_64 target. Add -mmmx target option when __SSE2__
22198 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
22199 for x86_64 target. Handle -m3dnowa option.
22201 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
22203 PR rtl-optimization/70703
22204 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
22205 (update_conflict_hard_regno_costs): Use long instead of unsigned
22206 arithmetic for cost calculation.
22208 2017-04-05 Jakub Jelinek <jakub@redhat.com>
22209 Bernd Edlinger <bernd.edlinger@hotmail.de>
22212 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
22215 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
22218 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
22219 ptr_mode with Pmode throughout.
22220 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
22221 into probe_stack_range and use DImode.
22223 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
22226 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
22227 call_eh_return is true.
22229 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22231 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
22232 Initialize last_match_fntype_index.
22234 2017-04-05 Jakub Jelinek <jakub@redhat.com>
22237 * tree-nvr.c: Include internal-fn.h.
22238 (pass_return_slot::execute): Ignore internal calls without
22241 2017-04-04 Jakub Jelinek <jakub@redhat.com>
22242 Richard Biener <rguenther@suse.de>
22245 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
22246 captures used multiple times, except for the last use.
22247 * generic-match-head.c: Include gimplify.h.
22249 2017-04-04 Jakub Jelinek <jakub@redhat.com>
22251 PR tree-optimization/79390
22252 * target.h (struct noce_if_info): Declare.
22253 * targhooks.h (default_noce_conversion_profitable_p): Declare.
22254 * target.def (noce_conversion_profitable_p): New target hook.
22255 * ifcvt.h (struct noce_if_info): New type, moved from ...
22256 * ifcvt.c (struct noce_if_info): ... here.
22257 (noce_conversion_profitable_p): Renamed to ...
22258 (default_noce_conversion_profitable_p): ... this. No longer
22260 (noce_try_store_flag_constants, noce_try_addcc,
22261 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
22262 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
22263 instead of noce_conversion_profitable_p.
22264 * config/i386/i386.c: Include ifcvt.h.
22265 (ix86_option_override_internal): Don't override
22266 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
22267 (ix86_noce_conversion_profitable_p): New function.
22268 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
22269 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
22270 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
22271 * doc/tm.texi: Regenerated.
22273 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22275 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
22278 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22281 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
22282 instructions for small multiply cores.
22284 2017-04-04 Jeff Law <law@redhat.com>
22286 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
22288 (mips_expand_vec_perm_const): Initialize elements in orig_perm
22289 that are not set by the loop over the elements.
22291 2017-04-04 Jakub Jelinek <jakub@redhat.com>
22294 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
22295 int mode, convert_modes it to mode as unsigned, otherwise use
22296 lowpart_subreg to mode rather than SImode.
22297 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
22298 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
22299 Use DImode instead of SImode for the shift count operand.
22300 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
22303 2017-04-04 Richard Biener <rguenther@suse.de>
22305 PR middle-end/80281
22306 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
22307 arithmetic done for the negate or the plus. Simplify.
22308 (A - (-B) -> A + B): Likewise.
22309 * fold-const.c (split_tree): Make sure to not negate pointers.
22311 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
22313 PR rtl-optimization/60818
22314 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
22315 a compare of comparisons with the thing compared if this results
22316 in a different machine mode.
22318 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
22320 * alias.c (base_alias_check): Fix typo in comment.
22321 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
22322 * cgraphunit.c (symbol_table::compile): Likewise.
22323 * collect2.c (maybe_run_lto_and_relink): Likewise.
22324 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
22325 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
22326 * config/avr/avr.c (avr_map_op_t): Likewise.
22327 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
22328 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
22329 * config/epiphany/epiphany.md (movcc): Likewise.
22330 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
22331 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
22333 * config/mips/mips.c (mips_save_restore_reg): Likewise.
22334 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
22335 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
22336 * config/sh/sh.c (sh_rtx_costs): Likewise.
22337 * fold-const.c (fold_truth_andor): Likewise.
22338 * genautomata.c (collapse_flag): Likewise.
22339 * gengtype.h (struct type::u::s): Likewise.
22340 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
22341 * input.c (FORMAT_AMOUNT): Likewise.
22342 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
22343 (known_aggs_to_agg_replacement_list): Likewise.
22344 * ipa-inline-analysis.c: Likewise.
22345 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
22346 * ipa-polymorphic-call.c
22347 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
22348 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
22349 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
22351 * modulo-sched.c (apply_reg_moves): Likewise.
22352 * omp-expand.c (build_omp_regions_1): Likewise.
22353 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
22354 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
22355 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
22356 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
22357 * value-prof.c: Likewise.
22358 * var-tracking.c (val_reset): Likewise.
22360 2017-04-03 Richard Biener <rguenther@suse.de>
22362 PR tree-optimization/80275
22363 * fold-const.c (split_address_to_core_and_offset): Handle
22366 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
22368 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
22369 descriptors is at least equal to that of functions.
22371 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
22373 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
22375 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
22378 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
22379 (mov<IMOD4:mode>): New expander.
22380 (*mov<IMOD4:mode>_internal): New insn and split pattern.
22382 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
22384 PR rtl-optimization/79405
22385 * fwprop.c (propagations_left): New variable.
22386 (forward_propagate_into): Decrement it.
22387 (fwprop_init): Initialize it.
22388 (fw_prop): If the variable has reached zero, stop propagating.
22389 (fwprop_addr): Ditto.
22391 2017-03-31 Jakub Jelinek <jakub@redhat.com>
22394 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
22395 a FUNCTION_DECL, pass it as decl instead of origin to
22398 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
22400 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
22403 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
22406 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
22407 TARGET_VSX_SMALL_INTEGER.
22409 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22411 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
22412 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
22414 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
22416 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
22417 extraction from odd-numbered MSA register.
22419 2017-03-31 Jakub Jelinek <jakub@redhat.com>
22421 PR middle-end/80173
22422 * expmed.c (store_bit_field_1): Don't attempt to create
22423 a word subreg out of hard registers wider than word if they
22424 have HARD_REGNO_NREGS of 1 for their mode.
22426 PR middle-end/80163
22427 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
22428 conversions to integer types wider than word and pointer.
22431 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
22432 (rtx_equal_for_cselib_p): Pass 0 to it.
22433 * cselib.c (cselib_hasher::equal): Likewise.
22434 (rtx_equal_for_cselib_1): Add depth argument. If depth
22435 is 128, don't look up VALUE locs and punt. Increment
22436 depth in recursive calls when walking VALUE locs.
22438 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
22440 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
22441 (make_gcov_file_name): Use the canonical path name for generating
22443 (read_line): Fix handling of files with ascii null bytes.
22445 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
22447 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
22448 to initialise a vector register instead
22449 of using a const_int.
22451 2017-03-30 Jakub Jelinek <jakub@redhat.com>
22453 PR translation/80189
22454 * gimplify.c (omp_default_clause): Use %qs instead of %s in
22455 diagnostic messages.
22457 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
22460 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
22461 (dfp_diex_<mode>): Update mode of operand 1.
22462 * doc/extend.texi (dxex, dxexq): Document change to return type.
22463 (diex, diexq): Document change to argument type.
22465 2017-03-30 Martin Jambor <mjambor@suse.cz>
22468 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
22469 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
22470 it reflects the signature changes performed at the callee side.
22471 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
22472 to cgraph_build_function_type_skip_args.
22473 (build_function_decl_skip_args): Adjust call to the above function.
22475 2017-03-30 Jakub Jelinek <jakub@redhat.com>
22478 * config/i386/sse.md
22479 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
22480 register as dest whenever it is a MEM not rtx_equal_p to the
22481 corresponding dup operand, and when forcing into reg move the
22482 reg into the memory afterwards.
22483 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
22484 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
22485 for the force_reg mode.
22486 (avx512vl_vextractf128<mode>): Use register as dest either
22487 always when a MEM, or when it is a MEM not rtx_equal_p to the
22488 corresponding dup operand, or even not when it is a CONST_VECTOR
22489 depending on the mode and lo vs. hi.
22490 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
22492 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
22493 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
22494 Likewise. Require that operands[2] is even.
22495 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
22496 Remove extraneous parens. Require that operands[2] is a multiple
22498 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
22499 operands[0] is a MEM if <mask_applied>, the predicates/constraints
22500 disallow memory then.
22502 2017-03-30 Richard Biener <rguenther@suse.de>
22504 PR tree-optimization/77498
22505 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
22506 to non-constants over backedges.
22508 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
22510 PR rtl-optimization/80233
22511 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
22512 as last_combined_insn. Do not test for BARRIER_P separately.
22514 2017-03-29 Andreas Schwab <schwab@suse.de>
22517 * calls.c (prepare_call_address): Convert funexp to Pmode before
22518 copying to temp reg.
22520 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22522 PR tree-optimization/80158
22523 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
22524 Handle possible future case of more than one alternate
22526 (replace_rhs_if_not_dup): Likewise.
22527 (replace_one_candidate): Likewise.
22529 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
22531 PR rtl-optimization/80193
22532 * ira.c (ira): Do not check allocation for LRA.
22534 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
22536 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
22537 (nvptx_output_simt_exit): Declare.
22538 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
22539 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
22540 (init_softstack_frame): Move initialization of crtl->is_leaf to...
22541 (nvptx_declare_function_name): ...here. Emit declaration of local
22542 memory space buffer for omp_simt_enter insn.
22543 (nvptx_output_unisimt_switch): New.
22544 (nvptx_output_softstack_switch): New.
22545 (nvptx_output_simt_enter): New.
22546 (nvptx_output_simt_exit): New.
22547 * config/nvptx/nvptx.h (struct machine_function): New fields
22548 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
22549 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
22550 (UNSPECV_SIMT_EXIT): Ditto.
22551 (omp_simt_enter_insn): New insn.
22552 (omp_simt_enter): New expansion.
22553 (omp_simt_exit): New insn.
22554 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
22556 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
22557 (expand_GOMP_SIMT_ENTER_ALLOC): New.
22558 (expand_GOMP_SIMT_EXIT): New.
22559 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
22560 (GOMP_SIMT_ENTER_ALLOC): Ditto.
22561 (GOMP_SIMT_EXIT): Ditto.
22562 * target-insns.def (omp_simt_enter): New insn.
22563 (omp_simt_exit): Ditto.
22564 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
22566 (lower_rec_simd_input_clauses): Implement SIMT privatization.
22567 (lower_rec_input_clauses): Likewise.
22568 (lower_lastprivate_clauses): Handle SIMT privatization.
22570 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
22571 (ompdevlow_adjust_simt_enter): New.
22572 (find_simtpriv_var_op): New.
22573 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
22574 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
22576 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
22577 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
22578 (copy_decl_for_dup_finish): Ditto.
22580 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
22582 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
22585 * config/i386/i386.c (ix86_option_override_internal): Always
22586 allow -mpreferred-stack-boundary=3 for 64-bit targets.
22588 2017-03-28 Bin Cheng <bin.cheng@arm.com>
22590 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
22592 2017-03-28 Bin Cheng <bin.cheng@arm.com>
22594 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
22595 mark new edge's irreducible flag accordign to it.
22596 (vect_do_peeling): Check loop preheader edge's irreducible flag
22597 and pass it to function slpeel_add_loop_guard.
22599 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
22601 PR tree-optimization/80218
22602 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
22603 Update block frequencies and counts.
22605 2017-03-28 Richard Biener <rguenther@suse.de>
22607 PR tree-optimization/78644
22608 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
22609 of a simplification result we may not use it at all.
22611 2017-03-28 Richard Biener <rguenther@suse.de>
22614 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
22615 without arguments, generate default definition of a SSA name.
22617 2017-03-28 Richard Biener <rguenther@suse.de>
22619 PR middle-end/80222
22620 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
22621 TYPE_REF_CAN_ALIAS_ALL references.
22622 * fold-const.c (fold_indirect_ref_1): Likewise.
22624 2017-03-28 Martin Liska <mliska@suse.cz>
22627 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
22628 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
22630 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
22631 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
22633 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
22634 (EXTRA_SPECS): Define.
22635 (SUBTARGET_EXTRA_SPECS): Likewise.
22636 (SUBTARGET_CPP_SPEC): Likewise.
22637 * config/arc/elf.h (EXTRA_SPECS): Renamed to
22638 SUBTARGET_EXTRA_SPECS.
22639 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
22641 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
22643 * config/arc/simdext.md (vst64_insn): Update pattern.
22644 (vld32wh_insn): Likewise.
22645 (vld32wl_insn): Likewise.
22646 (vld64_insn): Likewise.
22647 (vld32_insn): Likewise.
22649 2017-03-28 Marek Polacek <polacek@redhat.com>
22652 * fold-const.c (fold_comparison): Use protected_set_expr_location
22653 instead of SET_EXPR_LOCATION.
22655 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
22657 * tree.c (add_expr): Avoid name lookup warning.
22659 2017-03-27 Jeff Law <law@redhat.com>
22661 PR tree-optimization/80216
22662 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
22663 function name. Limit recursion depth.
22664 (record_temporary_equivalences): Corresponding changes.
22666 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
22668 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
22671 2017-03-27 Jakub Jelinek <jakub@redhat.com>
22674 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
22676 * cfgcleanup.c (reg_note_cfa_p): New array.
22677 (insns_have_identical_cfa_notes): New function.
22678 (old_insns_match_p): Don't cross-jump in between /f
22679 and non-/f instructions. If both i1 and i2 are frame related,
22680 verify all CFA notes, their order and content.
22682 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
22685 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
22686 HImode and SImode with zero extend to DImode to one insn.
22687 (bswap<mode>2_extenddi): Likewise.
22688 (bswapsi2_extenddi): Likewise.
22689 (bswaphi2_extendsi): Likewise.
22690 (bswaphi2): Combine bswap HImode and SImode into one insn.
22691 Separate memory insns from swapping register.
22692 (bswapsi2): Likewise.
22693 (bswap<mode>2): Likewise.
22694 (bswaphi2_internal): Delete, no longer used.
22695 (bswapsi2_internal): Likewise.
22696 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
22697 store, and gpr<-gpr swap insns.
22698 (bswap<mode>2_store): Likewise.
22699 (bswaphi2_reg): Register only splitter, combine with the splitter.
22700 (bswaphi2 splitter): Likewise.
22701 (bswapsi2_reg): Likewise.
22702 (bswapsi2 splitter): Likewise.
22703 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
22704 the insns into load, store, and register/register insns.
22705 (bswapdi2_ldbrx): Likewise.
22706 (bswapdi2_load): Likewise.
22707 (bswapdi2_store): Likewise.
22708 (bswapdi2_reg): Likewise.
22710 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
22712 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
22713 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
22715 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
22718 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
22720 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
22721 special handling for target option conflicts between dform
22722 options (-mpower9-dform, -mpower9-dform-vector,
22723 -mpower9-dform-scalar) and -mno-direct-move.
22725 2017-03-27 Richard Biener <rguenther@suse.de>
22727 PR tree-optimization/80181
22728 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
22730 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22732 * config/arc/predicates.md (move_double_src_operand): Replace the
22733 call to move_double_src_operand with a call to address_operand.
22735 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22737 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
22738 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
22739 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
22741 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22743 * config/arc/predicates.md (long_immediate_loadstore_operand):
22744 Consider scaled addresses cases.
22746 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22748 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
22749 restored when in interrupt.
22750 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
22751 doesn't have delay slot.
22753 2017-03-27 Richard Biener <rguenther@suse.de>
22756 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
22757 inlined thunk clones.
22759 2017-03-27 Jakub Jelinek <jakub@redhat.com>
22762 * asan.c (instrument_derefs): Copy over last operand from
22763 original COMPONENT_REF to the new COMPONENT_REF with
22764 DECL_BIT_FIELD_REPRESENTATIVE.
22765 * ubsan.c (instrument_object_size): Likewise.
22767 2017-03-27 Richard Biener <rguenther@suse.de>
22769 PR tree-optimization/80170
22770 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
22771 sure DR/SCEV didnt fold in constants we do not see when looking
22772 at the reference base alignment.
22774 2017-03-27 Richard Biener <rguenther@suse.de>
22776 PR middle-end/80171
22777 * gimple-fold.c (fold_ctor_reference): Properly guard against
22778 NULL return value from canonicalize_constructor_val.
22780 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
22783 * config/i386/i386.c (ix86_expand_builtin)
22784 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
22785 flags reg setting and flags reg using instructions.
22786 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
22787 clobbering instructions to zero extend op2.
22789 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
22791 * doc/install.texi (Configuration) <--with-aix-soname>:
22792 Update link to AIX ld.
22794 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
22796 PR rtl-optimization/80160
22797 PR rtl-optimization/80159
22798 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
22799 reg_alternate_class into account.
22801 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
22804 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
22805 to consider in curr_insn_transform.
22807 2017-03-24 Jakub Jelinek <jakub@redhat.com>
22809 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
22810 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
22811 and emit_mode_inner.
22813 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22815 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
22816 argument to the overloaded builtin variants. Use the new flag to
22817 deprecate certain builtin variants.
22818 * config/s390/s390-builtin-types.def: Add new builtin types.
22819 * config/s390/s390-builtins.h: Support new flags field for
22820 overloaded builtins.
22821 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
22822 (s390_macro_to_expand): Enable vector float data type.
22823 (s390_cpu_cpp_builtins_internal): Indicate support of the new
22824 builtins by incrementing the __VEC__ version number.
22825 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
22827 (s390_resolve_overloaded_builtin): Emit error messages depending
22828 on the builtin flags.
22829 * config/s390/s390.c (s390_expand_builtin): Support additional
22830 flags argument. Change error message to match the messages
22831 emitted in s390-c.c.
22832 * config/s390/s390.md: New UNSPEC_* constants.
22833 (op_type): Add new instruction types.
22834 * config/s390/vecintrin.h: Add new builtins and test data class
22836 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
22837 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
22838 (VEC_INEXACT, VEC_NOINEXACT): New constants.
22839 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
22840 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
22841 ("vec_mergel<mode>"): V_HW -> VEC_HW.
22843 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
22844 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
22845 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
22846 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
22848 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
22849 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
22850 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
22851 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
22853 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
22854 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
22855 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
22856 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
22857 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
22858 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
22859 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
22861 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
22862 ("vec_scatter_element<V_HW_4:mode>_DI")
22863 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
22864 ("vec_fpint<mode>", "vflls")
22865 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
22866 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
22867 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
22868 ("*vec_cmphe<mode>_cc"): ... these.
22870 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
22871 mode constant instead of magic value.
22873 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22875 * config/s390/s390.c (s390_expand_vec_compare): Support other
22876 vector floating point modes than just V2DF.
22877 (s390_expand_vcond): Likewise.
22878 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
22879 (s390_cannot_change_mode_class): Prevent mode changes between TF
22880 and V1TF in vector registers.
22881 * config/s390/s390.md (DF, SF): New mode attributes.
22882 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
22883 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
22884 SFmode support for VRs.
22885 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
22887 (VFT, VF_HW): New mode iterators.
22888 (vw, sdx): New mode attributes.
22889 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
22890 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
22891 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
22892 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
22893 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
22894 also the new vector floating point modes. Renaming to ...
22896 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
22897 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
22898 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
22899 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
22900 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
22901 ("vec_unordered<mode>"): ... these.
22903 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
22904 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
22905 ("*vec_extendv2df"): New insn definitions.
22907 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22909 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
22910 ("mulditi3_2", "*muldi3_sign"): New patterns.
22911 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
22912 rename the pattern definition.
22914 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22916 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
22918 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
22920 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22922 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
22923 instruction if possible.
22924 * config/s390/vector.md (vec_halfnumelts): New mode
22926 ("*vec_vllezlf<mode>"): New pattern.
22928 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22930 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
22931 ("popcountv4si2", "popcountv2di2"): Rename to ...
22932 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
22933 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
22935 ("popcount<mode>2_vxe"): New pattern.
22937 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22939 * common/config/s390/s390-common.c (processor_flags_table): Add
22941 * config.gcc: Add arch12.
22942 * config/s390/driver-native.c (s390_host_detect_local_cpu):
22943 Default to arch12 for unknown CPU model numbers.
22944 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
22945 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
22946 PROCESSOR_max sanity check.
22947 * config/s390/s390-opts.h (enum processor_type): Add
22949 * config/s390/s390.c (processor_table): Add arch12.
22950 (s390_expand_builtin): Add check for B_VXE flag.
22951 (s390_issue_rate): Add PROCESSOR_ARCH12.
22952 (s390_get_sched_attrmask): Likewise.
22953 (s390_get_unit_mask): Likewise.
22954 (s390_sched_score): Enable z13 scheduling for arch12.
22955 (s390_sched_reorder): Likewise.
22956 (s390_sched_variable_issue): Likewise.
22957 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
22959 (s390_tune_attr): Use z13 scheduling also for arch12.
22960 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
22961 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
22962 (TARGET_VXE_P): New macros.
22963 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
22964 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
22965 * config/s390/s390.opt: Add arch12 as processor_type.
22967 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22969 * config/s390/s390.md
22970 ("fixuns_truncdddi2", "fixuns_trunctddi2")
22971 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
22972 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
22974 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
22975 Rename expanders to ...
22977 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
22978 ("fixuns_truncdddi2_emu"): ... these.
22980 ("fixuns_trunc<mode>si2_emu"): New expander.
22982 ("*fixuns_truncdfdi2_z13"): Rename to ...
22983 ("*fixuns_truncdfdi2_vx"): ... this.
22985 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22987 * config/s390/2964.md: Remove the single element vector compare
22988 instructions which are no longer used.
22989 * config/s390/s390.c (s390_select_ccmode): Remove handling of
22991 (s390_canonicalize_comparison): Remove handling of DFmode
22993 (s390_expand_vec_compare_scalar): Remove function.
22994 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
22995 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
22997 ("*cmp<mode>_ccs"): Add wfcdb instruction.
22999 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23001 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
23003 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
23004 will anyway by matched by mov<mode>_64dfp.
23006 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23008 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
23009 vlef/vstef. Add missing operand to vleif.
23011 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23013 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
23014 pair for all vector types with 64 bit elements.
23015 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
23016 * config/s390/vector.md (V_HW_64): ... here.
23017 (V_128_NOSINGLE): New mode iterator.
23018 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
23019 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
23020 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
23021 ("*vec_load_pairv2di"): Change to ...
23022 ("*vec_load_pair<mode>"): ... this one.
23024 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23026 * config/s390/constraints.md: Add comments.
23027 (jKK): Reject element sizes > 8 bytes.
23028 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
23030 * config/s390/s390.md: Add the s_operand checks formerly in
23031 s390_split_ok_p to various splitters where they are still
23033 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
23034 for 128 bit vectors. Plus two splitters.
23036 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23038 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
23041 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23044 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
23045 error if the boundary argument is not constant.
23047 2017-03-24 Jakub Jelinek <jakub@redhat.com>
23049 PR rtl-optimization/80112
23050 * loop-doloop.c (doloop_condition_get): Don't check condition
23051 if cmp isn't SET with IF_THEN_ELSE src.
23053 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23055 PR tree-optimization/80158
23056 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
23057 replacing a candidate statement, also replace it for the
23058 candidate's alternate interpretation.
23059 (replace_rhs_if_not_dup): Likewise.
23060 (replace_one_candidate): Likewise.
23062 2017-03-24 Richard Biener <rguenther@suse.de>
23064 PR tree-optimization/80167
23065 * graphite-isl-ast-to-gimple.c
23066 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
23068 (translate_isl_ast_to_gimple::get_rename): Likewise.
23070 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
23072 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
23073 handling of certain combinations of target options, including the
23074 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
23075 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
23077 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23080 * config/arm/arm.md (*load_multiple): Add reload_completed to
23081 matching condition.
23083 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23084 Richard Biener <rguenth@suse.de>
23086 PR tree-optimization/79908
23087 PR tree-optimization/80136
23088 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
23089 been cast away, gimplify_and_add suffices.
23091 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
23093 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
23095 2017-03-23 Richard Biener <rguenther@suse.de>
23097 PR tree-optimization/80032
23098 * gimplify.c (gimple_push_cleanup): Forced unconditional
23099 cleanups still have to go to the conditional_cleanups
23102 2017-03-22 Jakub Jelinek <jakub@redhat.com>
23104 PR tree-optimization/80072
23105 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
23107 (next_operand_entry_id): Change type to unsigned int.
23108 (sort_by_operand_rank): Make sure to return the right return value
23109 even if unsigned fields are bigger than INT_MAX.
23110 (struct oecount): Change cnt and id type to unsigned int.
23111 (oecount_hasher::equal): Formatting fix.
23112 (oecount_cmp): Make sure to return the right return value
23113 even if unsigned fields are bigger than INT_MAX.
23114 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
23117 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
23118 TREE_READONLY on result if writing it more than once.
23121 * doc/invoke.texi (-fsanitize=thread): Document that with
23122 -fnon-call-exceptions atomics are not able to throw
23126 * tsan.c: Include tree-eh.h.
23127 (instrument_builtin_call): Call maybe_clean_eh_stmt or
23128 maybe_clean_or_replace_eh_stmt where needed.
23129 (instrument_memory_accesses): Add cfg_changed argument.
23130 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
23131 if it returned true.
23132 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
23134 PR rtl-optimization/63191
23135 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
23136 wrapper function, moved the whole old content into ...
23137 (ix86_delegitimize_address_1): ... this. New inline function.
23138 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
23139 true as last argument instead of ix86_delegitimize_address.
23141 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
23143 * config/aarch64/aarch64.c (generic_branch_cost): Copy
23144 cortexa57_branch_cost.
23146 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
23148 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
23150 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23153 * doc/md.texi (Constraints): Document wA constraint.
23154 * config/rs6000/constraints.md (wA): New.
23155 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
23156 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
23157 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
23158 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
23160 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
23163 * gimplify.c (is_oacc_declared): New function.
23164 (oacc_default_clause): Use it to set default flags for acc declared
23165 variables inside parallel regions.
23166 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
23167 declared variables.
23168 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
23169 declare attribute to any decl as necessary.
23171 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
23174 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
23175 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
23176 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
23177 (arm_arch_lpae): This.
23178 * config/arm/arm.c (arm_arch7ve): Rename into ...
23179 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
23180 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
23183 2017-03-22 Martin Liska <mliska@suse.cz>
23186 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
23187 error message instead of an ICE.
23189 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23191 * doc/extend.texi (6.11 Additional Floating Types): Revise.
23193 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
23195 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
23197 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
23200 2017-03-21 Martin Sebor <msebor@redhat.com>
23202 * doc/extend.texi: Use "cannot" instead of "can't."
23203 * doc/hostconfig.texi: Same.
23204 * doc/install.texi: Same.
23205 * doc/invoke.texi: Same.
23206 * doc/loop.texi: Same.
23207 * doc/md.texi: Same.
23208 * doc/objc.texi: Same.
23209 * doc/rtl.texi: Same.
23210 * doc/tm.texi: Same.
23211 * doc/tm.texi.in: Same.
23212 * doc/trouble.texi: Same.
23214 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
23217 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
23218 (collect_checksum_attributes): Set it.
23219 (die_checksum_ordered): Use it.
23221 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23223 PR tree-optimization/79908
23224 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
23225 change: For a VA_ARG whose LHS has been cast away, use
23226 force_gimple_operand to construct the side effects.
23228 2017-03-21 David Malcolm <dmalcolm@redhat.com>
23230 PR translation/80001
23231 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
23232 more amenable to translation.
23233 (oacc_loop_auto_partitions): Likewise.
23235 2017-03-21 Marek Polacek <polacek@redhat.com>
23236 Martin Sebor <msebor@redhat.com>
23238 PR tree-optimization/80109
23239 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
23240 on INTEGRAL_TYPE_P.
23242 2017-03-21 Jakub Jelinek <jakub@redhat.com>
23243 Segher Boessenkool <segher@kernel.crashing.org>
23246 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
23247 check reg_used_between_p between insn and one of succ or succ2
23248 depending on if succ is artificial insn not inserted into insn
23251 2017-03-21 Martin Liska <mliska@suse.cz>
23253 PR gcov-profile/80081
23254 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
23255 * doc/gcc.texi: Include gcov-dump stuff.
23256 * doc/gcov-dump.texi: New file.
23258 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
23260 PR rtl-optimization/79150
23261 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
23262 conditional jump, if the jump is the last insn of the loop.
23264 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23265 Richard Biener <rguenth@suse.de>
23267 PR tree-optimization/79908
23268 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
23269 been cast away, use force_gimple_operand to construct the side
23272 2017-03-21 Martin Liska <mliska@suse.cz>
23274 PR libfortran/79956
23275 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
23278 2017-03-21 Brad Spengler <spender@grsecurity.net>
23281 * plugin.c (htab_hash_plugin): New function.
23282 (add_new_plugin): Use it and adjust.
23283 (parse_plugin_arg_opt): Adjust.
23284 (init_one_plugin): Likewise.
23286 2017-03-21 Richard Biener <rguenther@suse.de>
23288 PR tree-optimization/80032
23289 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
23290 if set force the cleanup to happen unconditionally.
23291 (gimplify_target_expr): Push inserted clobbers with force_uncond
23292 to avoid them being removed by control-dependent DCE.
23294 2017-03-21 Richard Biener <rguenther@suse.de>
23296 PR tree-optimization/80122
23297 * tree-inline.c (copy_bb): Do not expans va-arg packs or
23298 va_arg_pack_len when the inlined call stmt requires pack
23300 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
23302 2017-03-21 Jakub Jelinek <jakub@redhat.com>
23305 * tsan.c (instrument_builtin_call): If the memory model argument
23306 is not a constant, assume it is valid.
23309 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
23312 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
23314 PR rtl-optimization/79910
23315 * combine.c (can_combine_p): Do not allow combining an I0 or I1
23316 if its dest is used by an insn before I2 (other than the combined
23317 insns themselves, which are properly handled already).
23319 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
23322 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
23324 * combine.c (record_used_regs): New static function.
23325 (try_combine): Handle situations where there is an additional
23326 instruction between I2 and I3 which needs to have a LOG_LINK
23330 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
23332 * combine.c (try_combine): Delete redundant i1 test. Call
23333 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
23335 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23338 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
23339 alternatives 13/14.
23341 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23343 PR tree-optimization/80054
23344 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
23345 the optimization if a PHI or any of its arguments is not dominated
23346 by the candidate's basis. Use gphi* rather than gimple* as
23348 (replace_profitable_candidates): Clean up a gimple* variable that
23349 should be a gphi* variable.
23351 2017-03-20 Martin Sebor <msebor@redhat.com>
23354 * doc/extend.texi (attribute constructor): Document present limitation.
23356 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
23359 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
23360 __POWER9_VECTOR__ #ifdef control, change template definition to
23361 use Power9-specific built-in function.
23362 (vec_any_eq): Likewise.
23363 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
23364 to control outcomes from this test.
23365 (vector_ae_<mode>p): For VEC_F modes, likewise.
23367 2017-03-20 Ian Lance Taylor <iant@google.com>
23369 * config/i386/i386.c (ix86_function_regparm): Save an extra
23370 register for -fsplit-stack with DECL_STATIC_CHAIN.
23372 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
23375 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
23376 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
23378 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
23380 * config/riscv/riscv.c (riscv_print_operand): Use "fence
23382 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
23385 2017-03-20 Marek Polacek <polacek@redhat.com>
23388 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
23390 2017-03-20 Richard Biener <rguenther@suse.de>
23392 PR tree-optimization/80113
23393 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
23394 allocate extra SSA name for PHI def.
23395 (add_close_phis_to_outer_loops): Likewise.
23396 (add_close_phis_to_merge_points): Likewise.
23397 (copy_loop_close_phi_args): Likewise.
23398 (copy_cond_phi_nodes): Likewise.
23400 2017-03-20 Martin Liska <mliska@suse.cz>
23402 PR middle-end/79753
23403 * tree-chkp.c (chkp_build_returned_bound): Do not build
23404 returned bounds for a LHS that's not a BOUNDED_P type.
23406 2017-03-20 Martin Liska <mliska@suse.cz>
23410 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
23411 COMPLEX_CST and VECTOR_CST.
23413 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23416 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
23417 target operand. A new splitter adds the clobber statement in case
23418 the target operand is dead anyway.
23420 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
23422 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
23423 to age-old versions of binutils and glibc.
23425 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
23427 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
23429 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
23431 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
23433 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
23435 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
23436 requirement for binutils 2.13.
23438 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
23440 * combine.c (try_combine): Delete redundant i1 test. Call
23441 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
23443 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
23445 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
23446 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
23448 <riscv64-*-elf>: Re-arrange section
23449 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
23450 <riscv32-*-linux>: Likewise.
23451 <riscv64-*-elf>: Likewise
23452 <riscv64-*-linux>: Likewise.
23454 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
23457 * aarch64.opt(verbose-cost-dump): Fix typo.
23459 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
23462 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
23463 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
23465 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
23467 * reload.c (find_reloads): When reloading a nonoffsettable address,
23468 use RELOAD_OTHER for it and its address reloads.
23470 PR rtl-optimization/79910
23471 * combine.c (record_used_regs): New static function.
23472 (try_combine): Handle situations where there is an additional
23473 instruction between I2 and I3 which needs to have a LOG_LINK
23476 2017-03-17 Jeff Law <law@redhat.com>
23478 PR tree-optimization/71437
23479 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
23480 conditional in the hash table first.
23481 (vrp_dom_walker::before_dom_children): Extract condition from
23482 ASSERT_EXPR. Record condition, its inverion and any implied
23483 conditions as well.
23485 2017-03-17 Marek Polacek <polacek@redhat.com>
23486 Markus Trippelsdorf <markus@trippelsdorf.de>
23488 PR tree-optimization/80079
23489 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
23492 2017-03-17 Richard Biener <rguenther@suse.de>
23494 PR middle-end/80075
23495 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
23496 Properly verify the LHS before the RHS possibly claims to be
23498 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
23501 2017-03-17 Martin Jambor <mjambor@suse.cz>
23503 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
23504 (List of -O2 options): Likewise.
23505 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
23508 2017-03-17 Tom de Vries <tom@codesourcery.com>
23510 * gcov-dump.c (print_usage): Print bug_report_url.
23512 2017-03-17 Richard Biener <rguenther@suse.de>
23514 PR middle-end/80050
23515 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
23516 (parser::peek): Likewise.
23518 2017-03-17 Richard Biener <rguenther@suse.de>
23520 PR tree-optimization/80048
23521 * sese.c (free_sese_info): Properly release rename_map and
23522 copied_bb_map elements.
23524 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
23526 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
23527 Add linked-list forward and backlinks. Insert on
23528 construction, remove on destruction.
23529 (class pass_store_merging): Add m_stores_head field.
23530 (pass_store_merging::terminate_and_process_all_chains):
23531 Iterate over m_stores_head list.
23532 (pass_store_merging::terminate_all_aliasing_chains):
23534 (pass_store_merging::execute): Check for debug stmts first.
23535 Push new chains onto the m_stores_head stack.
23537 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
23540 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
23541 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
23542 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
23544 2017-03-16 Jeff Law <law@redhat.com>
23546 PR tree-optimization/71437
23547 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
23548 member function. Implementation moved into after_dom_children
23549 member function and into the threader's thread_outgoing_edges
23551 (dom_opt_dom_walker::after_dom_children): Simplify by moving
23552 some code into new thread_outgoing_edges.
23553 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
23554 definition. Simplify marker handling (do it here). Assume we always
23555 have the available expression and the const/copies tables.
23556 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
23558 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
23559 * tree-vrp.c (equiv_stack): No longer file scoped.
23560 (vrp_dom_walker): New class.
23561 (vrp_dom_walker::before_dom_children): New member function.
23562 (vrp_dom_walker::after_dom_children): Likewise.
23563 (identify_jump_threads): Setup domwalker. Use it rather than
23564 walking edges in a random order by hand. Simplify setup/finalization.
23565 (finalize_jump_threads): Remove.
23566 (vrp_finalize): Do not call identify_jump_threads here.
23567 (execute_vrp): Do it here instead and call thread_through_all_blocks
23570 PR tree-optimization/71437
23571 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
23573 (simplify_stmt_for_jump_threading): Add basic_block argument. All
23575 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
23576 (dom_opt_dom_walker::thread_across_edge): Remove
23577 handle_dominating_asserts argument. All callers changed.
23578 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
23579 changes. Remove calls to lhs_of_dominating_assert. Other
23580 uses of handle_dominating_asserts turn into unconditional code
23581 (simplify_control_stmt_condition_1): Likewise.
23582 (simplify_control_stmt_condition): Likewise.
23583 (thread_through_normal_block, thread_across_edge): Likewise.
23584 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
23585 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
23586 object if it is not an SSA_NAME.
23587 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
23588 before calling into the VRP specific simplifiers.
23589 (identify_jump_threads): Remove handle_dominating_asserts
23592 2017-03-16 Jakub Jelinek <jakub@redhat.com>
23595 * tree-diagnostic.c (default_tree_printer): No longer static.
23596 * tree-diagnostic.h (default_tree_printer): New prototype.
23598 2017-03-16 Tamar Christina <tamar.christina@arm.com>
23600 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
23601 Change ins into fmov.
23603 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23605 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
23606 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
23607 Use h_con constraint for operand 1.
23608 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
23609 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
23611 2017-03-15 Jeff Law <law@redhat.com>
23613 PR tree-optimization/71437
23614 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
23615 (record_temporary_equivalences): Use it.
23617 PR tree-optimization/71437
23618 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
23619 tree-ssa-scopedtables.
23620 (lookup_avail_expr, build_and_record_new_cond): Likewise.
23621 (record_conditions, record_cond, vuse_eq): Likewise.
23622 (record_edge_info): Adjust to API tweak of record_conditions.
23623 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
23624 (record_temporary_equivalences, optimize_stmt): Likewise.
23625 (eliminate_redundant_computations): Likewise.
23626 (record_equivalences_from_stmt): Likewise.
23627 * tree-ssa-scopedtables.c: Include options.h and params.h.
23628 (vuse_eq): New function, moved from tree-ssa-dom.c
23629 (build_and_record_new_cond): Likewise.
23630 (record_conditions): Likewise. Accept vector of conditions rather
23631 than edge_equivalence structure for first argument.
23632 for the first argument.
23633 (avail_exprs_stack::lookup_avail_expr): New member function, moved
23634 from tree-ssa-dom.c.
23635 (avail_exprs_stack::record_cond): Likewise.
23636 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
23637 from tree-ssa-dom.c.
23638 (avail_exprs_stack): Add new member functions lookup_avail_expr
23640 (record_conditions): Declare.
23642 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
23645 * lra-constraints.c (process_alt_operands): Increase reject for
23646 reloading an input/output operand.
23648 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
23651 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
23652 insns to convert from signed/unsigned char/short to IEEE 128-bit
23654 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
23656 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
23659 * config/i386/i386.c (ix86_vector_duplicate_value): Create
23660 subreg of inner mode for values already in registers.
23662 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
23664 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
23665 iteration reg is used after the loop.
23667 2017-03-14 Martin Sebor <msebor@redhat.com>
23669 PR tree-optimization/79800
23670 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
23671 precision in negative-positive range.
23672 (format_floating): Call non-const overload with adjusted precision.
23674 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
23677 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
23680 2017-03-14 Martin Sebor <msebor@redhat.com>
23682 PR middle-end/80020
23683 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
23684 * builtins.def (aligned_alloc): Use it.
23687 * Makefile.in (GTFILES): Add calls.c.
23688 * calls.c: Include "gt-calls.h".
23690 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
23692 PR rtl-optimization/79728
23693 * regs.h (struct target_regs): New field
23694 x_contains_allocatable_regs_of_mode.
23695 (contains_allocatable_regs_of_mode): New macro.
23696 * reginfo.c (init_reg_sets_1): Initialize it, and change
23697 contains_reg_of_mode so it includes global regs as well.
23698 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
23699 rather than contains_regs_of_mode.
23701 2017-03-14 Martin Liska <mliska@suse.cz>
23703 * doc/invoke.texi: Document options that can't be combined with
23704 -fcheck-pointer-bounds.
23706 2017-03-14 Martin Liska <mliska@suse.cz>
23708 PR middle-end/79831
23709 * doc/invoke.texi (-Wchkp): Document the option.
23711 2017-03-14 Martin Liska <mliska@suse.cz>
23713 * Makefile.in: Install gcov-dump.
23715 2017-03-14 Martin Liska <mliska@suse.cz>
23717 * multiple_target.c (expand_target_clones): Bail out for
23718 an invalid attribute.
23720 2017-03-14 Richard Biener <rguenther@suse.de>
23722 * alias.c (struct alias_set_entry): Pack properly.
23723 * cfgloop.h (struct loop): Likewise.
23724 * cse.c (struct set): Likewise.
23725 * ipa-utils.c (struct searchc_env): Likewise.
23726 * loop-invariant.c (struct invariant): Likewise.
23727 * lra-remat.c (struct cand): Likewise.
23728 * recog.c (struct change_t): Likewise.
23729 * rtl.h (struct address_info): Likewise.
23730 * symbol-summary.h (function_summary): Likewise.
23731 * tree-loop-distribution.c (struct partition): Likewise.
23732 * tree-object-size.c (struct object_size_info): Likewise.
23733 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
23734 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
23735 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
23736 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
23737 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
23738 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
23739 (struct _stmt_vec_info): Likewise.
23741 2017-03-14 Martin Liska <mliska@suse.cz>
23744 * multiple_target.c (create_dispatcher_calls): Check that
23745 a target can create a function dispatcher.
23747 2017-03-14 Martin Liska <mliska@suse.cz>
23750 * multiple_target.c (expand_target_clones): Drop local.local
23751 flag for default implementation.
23753 2017-03-14 Richard Biener <rguenther@suse.de>
23755 PR tree-optimization/80030
23756 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
23758 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
23760 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
23761 gcc_fallthrough() instead of __attribute__((fallthrough));
23763 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
23765 * doc/gcc.texi: Remove "up" link to (DIR).
23766 * doc/gccint.texi: Ditto.
23768 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
23770 * doc/install.texi (Specific) <avr>: Remove reference to
23773 2017-03-13 Jeff Law <law@redhat.com>
23775 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
23776 attribute rather than comments.
23778 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
23779 match_scratch operand is highest.
23781 2017-03-13 Martin Liska <mliska@suse.cz>
23783 PR middle-end/78339
23784 * ipa-pure-const.c (warn_function_noreturn): If the declarations
23785 is a CHKP clone, use original declaration.
23787 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23789 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
23790 (arc_conditional_register_usage): Use a different allocation order
23791 when optimizing for size.
23792 * common/config/arc/arc-common.c (arc_option_optimization_table):
23793 Section anchors default on when optimizing for size.
23795 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23797 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
23799 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23801 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
23802 * config/arc/arc.md (cpu_facility): Add cd variant.
23803 (*movqi_insn): Add code density variant.
23804 (*movhi_insn): Likewise.
23805 (*movqi_insn): Likewise.
23806 (*addsi3_mixed): Likewise.
23807 (subsi3_insn): Likewise.
23809 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23811 * config/arc/arc.md (movsi_cond_exec): Update constraint.
23813 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23815 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
23816 expressions with MINUS and UNARY ops.
23818 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23821 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
23823 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
23824 between vec_select and vector argument.
23825 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
23826 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
23827 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
23828 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
23829 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
23830 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
23832 2017-03-13 Richard Biener <rguenther@suse.de>
23835 * params.def (vect-max-peeling-for-alignment): Fix typo.
23837 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23839 * doc/install.texi (Specific) <mips-*-*>: Remove description of
23840 issue that only occurred with binutils below 2.18.
23842 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23844 * doc/install.texi (Specific) <cris-axis-elf>: No longer
23845 refer to binutils 2.11/2.12 minimum.
23847 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23849 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
23850 ftp.kernel.org and simplify binutils requirement.
23852 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
23854 * doc/invoke.texi (Warning Options): Fix spelling of link-time
23856 (Optimize Options): Ditto. Also remove redundancy.
23858 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23860 PR translation/79848
23861 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
23863 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
23864 to G_ to avoid double translation.
23866 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23868 PR translation/79923
23869 * auto-profile.c (get_combined_location): Convert leading
23870 character of diagnostics to lower case and remove trailing period.
23871 (read_profile): Likewise for various diagnostics.
23872 * config/arm/arm.c (arm_option_override): Remove trailing period
23873 from various diagnostics.
23874 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
23875 (msp430_expand_delay_cycles): Likewise.
23877 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23880 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
23881 full command-line argument, rather than just "str".
23882 (aarch64_validate_march): Likewise.
23883 (aarch64_validate_mtune): Likewise.
23885 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
23887 PR rtl-optimization/78911
23888 * lra-assigns.c (must_not_spill_p): New function.
23889 (spill_for): Use it.
23891 2017-03-10 Jakub Jelinek <jakub@redhat.com>
23893 PR tree-optimization/79981
23894 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
23895 ATOMIC_COMPARE_EXCHANGE ifn result.
23896 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
23897 IFN_ATOMIC_COMPARE_EXCHANGE.
23899 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23902 * opts.c (parse_sanitizer_options): Add missing question mark to
23903 "did you mean" message.
23905 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23907 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
23909 (VMULEUH_UNS): Likewise.
23910 (VMULOUB_UNS): Likewise.
23911 (VMULOUH_UNS): Likewise.
23912 * config/rs6000/rs6000.c (builtin_function_type): Remove
23913 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
23915 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23918 * read-rtl-function.c (function_reader::read_rtx_operand): Update
23919 x with result of extra_parsing_for_operand_code_0.
23920 (function_reader::extra_parsing_for_operand_code_0): Convert
23921 return type from void to rtx, returning x. When reading
23922 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
23923 larger size containing struct block_symbol.
23925 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
23927 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
23928 -mfloat128-hardware without -m64.
23930 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
23933 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
23934 entries to the case statement that marks unsigned arguments to
23935 overloaded functions.
23937 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
23939 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
23940 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
23942 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
23945 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
23946 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
23948 2017-03-10 Martin Liska <mliska@suse.cz>
23952 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
23953 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
23956 2017-03-10 Olivier Hainque <hainque@adacore.com>
23958 * tree-switch-conversion (array_value_type): Start by resetting
23959 candidate type to it's main variant.
23961 2017-03-10 Jakub Jelinek <jakub@redhat.com>
23963 PR rtl-optimization/79909
23964 * combine.c (try_combine): Use simplify_replace_rtx on individual
23965 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
23966 of the whole CALL_INSN_FUNCTION_USAGE.
23968 PR tree-optimization/79972
23969 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
23970 get_range_info on SSA_NAMEs. Formatting fixes.
23972 2017-03-10 Richard Biener <rguenther@suse.de>
23973 Jakub Jelinek <jakub@redhat.com>
23975 PR tree-optimization/77975
23976 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
23977 edge to be constant.
23978 (get_val_for): For constant x return it. Formatting fix.
23979 (loop_niter_by_eval): Avoid pointless looping if the next iteration
23980 would use the same bases as the current one.
23982 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23984 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
23985 instead of vec_select for V1TImode.
23986 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
23988 (VSX_LE_128): Add V1TI to this mode iterator.
23989 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
23990 (*vsx_le_perm_store_<mode>): Likewise.
23991 (pre-reload splitter for VSX stores): Likewise.
23992 (post-reload splitter for VSX stores): Likewise.
23993 (*vsx_xxpermdi2_le_<mode>): Likewise.
23994 (*vsx_lxvd2x2_le_<mode>): Likewise.
23995 (*vsx_stxvd2x2_le_<mode>): Likewise.
23997 2017-03-09 Michael Eager <eager@eagercon.com>
23999 Correct failures with --enable-checking=yes,rtl.
24001 * config/microblaze/microblaze.c (microblaze_expand_shift):
24002 Replace GET_CODE test with CONST_INT_P and INTVAL test with
24003 test for const0_rtx.
24004 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
24005 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
24007 2017-03-09 Richard Biener <rguenther@suse.de>
24009 PR tree-optimization/79977
24010 * graphite-scop-detection.c (scop_detection::merge_sese):
24011 Handle the case of extra exits to blocks dominating the entry.
24013 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
24015 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
24018 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
24020 PR rtl-optimization/79949
24021 * lra-constraints.c (process_alt_operands): Check memory when
24022 trying to predict a cycle. Print about the overall increase.
24024 2017-03-09 Richard Biener <rguenther@suse.de>
24026 PR middle-end/79971
24027 * gimple-expr.c (useless_type_conversion_p): Preserve
24028 TYPE_SATURATING for fixed-point types.
24030 2017-03-09 Richard Biener <rguenther@suse.de>
24033 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
24034 alignment of BLKmode params.
24036 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24039 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
24040 (VALL_NO_V2Q): Likewise.
24042 * config/aarch64/aarch64-simd.md
24043 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
24045 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
24046 VALL_NO_V2Q mode iterator.
24047 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
24049 2017-03-09 Martin Liska <mliska@suse.cz>
24051 PR tree-optimization/79631
24052 * tree-chkp-opt.c (chkp_is_constant_addr): Call
24053 tree_int_cst_sign_bit just for INTEGER constants.
24055 2017-03-09 Martin Liska <mliska@suse.cz>
24059 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
24062 2017-03-09 Marek Polacek <polacek@redhat.com>
24065 * tree.c (inchash::add_expr): Handle TREE_VEC.
24067 2017-03-09 Martin Liska <mliska@suse.cz>
24070 (chkp_narrow_size_and_offset): New function.
24071 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
24072 (void chkp_parse_bit_field_ref): New function.
24073 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
24074 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
24076 2017-03-09 Martin Liska <mliska@suse.cz>
24079 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
24080 (chkp_find_bounds_1): Remove gcc_unreachable.
24082 2017-03-09 Jakub Jelinek <jakub@redhat.com>
24085 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
24086 BUILT_IN_SYNC*, determine the access type from the size suffix and
24087 always build a MEM_REF with that type. Handle forgotten
24088 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
24091 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
24092 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
24093 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
24094 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
24095 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
24096 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
24097 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
24098 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
24099 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
24100 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
24101 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
24102 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
24103 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
24104 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
24105 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
24106 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
24107 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
24108 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
24109 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
24110 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
24111 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
24112 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
24113 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
24114 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
24115 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
24116 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
24117 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
24118 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
24119 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
24120 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
24121 definitions outside of __OPTIMIZE__ guarded section.
24124 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
24125 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
24126 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
24127 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
24130 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24132 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
24133 ("vfenez<mode>"): Add missing constraints.
24135 2017-03-08 Martin Sebor <msebor@redhat.com>
24138 * config/nds32/nds32.c (nds32_option_override):
24139 Fix misspelled diagnostic.
24141 2017-03-08 Jakub Jelinek <jakub@redhat.com>
24144 * gimplify.c (gimplify_omp_for): Replace index var in outer
24145 taskloop statement with an artificial variable and add
24146 OMP_CLAUSE_PRIVATE clause for it.
24148 2017-03-08 Richard Biener <rguenther@suse.de>
24150 PR tree-optimization/79955
24151 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
24152 for accesses that are completely outside of the variable.
24154 2017-03-08 Andrew Haley <aph@redhat.com>
24156 PR tree-optimization/79943
24157 * tree-ssa-loop-split.c (compute_new_first_bound): When
24158 calculating the new upper bound, (END-BEG) should be added, not
24161 2017-03-08 Jakub Jelinek <jakub@redhat.com>
24163 * config/avr/avr.md (setmemhi): Make sure match_dup
24164 operand number comes before match_scratch.
24166 2017-03-08 Richard Biener <rguenther@suse.de>
24168 PR tree-optimization/79920
24169 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
24170 with ncopies == 1 to ...
24171 (vect_transform_slp_perm_load): ... here. Properly compute
24172 all element loads by iterating VF times over the group. Do
24173 not handle ncopies (computed in a broken way) in
24174 vect_create_mask_and_perm.
24176 2017-03-08 Jakub Jelinek <jakub@redhat.com>
24179 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
24180 is a uniform vector, use uniform_vector_p return value instead of
24181 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
24183 2017-03-07 Marek Polacek <polacek@redhat.com>
24185 PR middle-end/79809
24186 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
24187 (alloca_call_type): Likewise.
24189 2017-03-07 Martin Liska <mliska@suse.cz>
24191 * gcov.c (process_args): Put comment to correct location.
24193 2017-03-07 Martin Liska <mliska@suse.cz>
24195 PR middle-end/68270
24196 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
24197 Use array_at_struct_end_p instead of DECL_CHAIN (field).
24198 (chkp_narrow_bounds_for_field): Likewise.
24199 (chkp_parse_array_and_component_ref): Pass one more argument to
24202 2017-03-07 Richard Biener <rguenther@suse.de>
24204 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
24207 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
24209 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
24210 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
24212 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24215 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
24216 to end of description.
24217 (PARAM_MAX_STORES_TO_MERGE): Likewise.
24219 2017-03-07 Jakub Jelinek <jakub@redhat.com>
24221 PR rtl-optimization/79901
24222 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
24224 (*avx512f_<code><mode>3<mask_name>): ... this.
24225 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
24226 iterator instead of VI8_AVX2_AVX512BW.
24228 PR rtl-optimization/79901
24229 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
24230 min/max expander, expand it using expand_vec_cond_expr.
24233 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
24236 2017-03-06 Jakub Jelinek <jakub@redhat.com>
24239 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
24240 to void * for PCH reasons.
24241 * dwarf2out.c (output_loc_operands, output_die): Cast
24242 v.val_vec.array to unsigned char *.
24244 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
24247 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
24250 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
24252 PR rtl-optimization/79571
24253 * lra-constraints.c (process_alt_operands): Calculate static
24254 reject and subtract it from overall when only addresses will be
24257 2017-03-06 Julia Koval <julia.koval@intel.com>
24260 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
24261 incoming stack boundary to 128 for 64-bit targets.
24263 2017-03-06 Richard Biener <rguenther@suse.de>
24265 PR tree-optimization/79894
24266 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
24267 to NULL after folding it.
24269 2017-03-06 Richard Biener <rguenther@suse.de>
24271 PR tree-optimization/79824
24272 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
24273 check disabling peeling for gaps.
24275 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
24277 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
24278 attributes): Document gettimeofday.
24280 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
24282 * config/s390/s390.c (s390_option_override_internal): Set
24283 PARAM_MIN_VECT_LOOP_BOUND
24285 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
24287 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
24288 * config/s390/s390.md: Likewise.
24290 2017-03-06 Jakub Jelinek <jakub@redhat.com>
24293 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
24294 (<avx2_avx512>_perm<mode>): Rename to ...
24295 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
24297 (<avx512>_perm<mode>_mask): Rename to ...
24298 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
24300 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
24301 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
24302 instead of VI8F_256_512.
24303 (avx512f_perm<mode>): New define_expand.
24304 (avx512f_perm<mode>_mask): Likewise.
24305 (avx512f_perm<mode>_1<mask_name>): New define_insn.
24306 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
24308 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
24310 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
24311 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
24313 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
24315 2017-03-06 Martin Liska <mliska@suse.cz>
24318 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
24319 when having a SSA NAME w/o VAR_DECL assigned to it.
24321 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
24323 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
24324 msa_dpsub_<su>_d): Fix MODE for vec_select.
24326 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
24328 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
24330 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
24332 2017-03-06 Richard Biener <rguenther@suse.de>
24334 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
24335 * plugin.c (register_plugin_info): Likewise.
24336 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
24338 2017-03-05 Jakub Jelinek <jakub@redhat.com>
24340 * config/i386/sse.md (sse_storehps, sse_storelps,
24341 avx_<castmode><avxsizesuffix>_<castmode>,
24342 avx512f_<castmode><avxsizesuffix>_<castmode>,
24343 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
24344 in condition that at least one operand is not a MEM.
24346 2017-03-03 Jakub Jelinek <jakub@redhat.com>
24348 PR middle-end/79805
24349 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
24350 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
24352 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
24353 gimple_call_nothrow_p flag based on whether original builtin can throw.
24354 If it can, emit following stmts on the fallthrough edge.
24355 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
24356 don't create new bb if inserting just debug stmts on the edge, try to
24357 insert them on the fallthru bb or just reset debug stmts.
24359 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
24362 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
24363 restore recog_data (including the operand rtxes inside it) around
24364 the call to get_insn_template.
24366 2017-03-03 Martin Sebor <msebor@redhat.com>
24368 PR tree-optimization/79699
24369 * context.c (context::~context): Free MPFR caches to avoid
24370 a memory leak on program exit.
24372 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24374 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
24375 Use wide_int::ulow () instead of .elt (0).
24377 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
24379 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
24380 (*pushxf): Limit oF constraint to 32bit targets and add oC
24381 constraint for 64bit targets.
24382 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
24383 (*pushdf): Change rmF constraint to rmC.
24385 2017-03-03 Martin Liska <mliska@suse.cz>
24387 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
24388 Remove unused variable.
24390 2017-03-03 Jakub Jelinek <jakub@redhat.com>
24393 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
24394 is a memory operand, increase num_memory.
24395 (ix86_expand_args_builtin): Likewise.
24397 2017-03-03 Jan Hubicka <jh@suse.cz>
24400 * ipa-devirt.c (maybe_record_node): Properly handle
24401 __cxa_pure_virtual visibility.
24403 2017-03-03 Martin Liska <mliska@suse.cz>
24405 PR tree-optimization/79803
24406 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
24408 (pass_loop_prefetch::execute): Disabled optimization if an
24409 assumption about L1 cache size is not met.
24411 2017-03-03 Martin Liska <mliska@suse.cz>
24413 PR rtl-optimization/79574
24414 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
24415 (hash_scan_set): Likewise.
24416 (dump_hash_table): Likewise.
24417 (hoist_code): Likewise.
24419 2017-03-03 Richard Biener <rguenther@suse.de>
24421 * fixed-value.c (fixed_from_string): Restore use of elt (1)
24422 in place of uhigh ().
24423 (fixed_convert_from_real): Likewise.
24425 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
24428 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
24430 2017-03-03 Richard Biener <rguenther@suse.de>
24432 PR middle-end/79818
24433 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
24434 TYPE_OVERFLOW_UNDEFINED check.
24436 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24438 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
24440 (vector_ae_<mode>_p): Likewise.
24441 (vector_nez_<mode>_p): Likewise.
24442 (vector_ne_v2di_p): Likewise.
24443 (vector_ae_v2di_p): Likewise.
24444 (vector_ne_<mode>_p): Likewise.
24445 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
24447 (vsx_tsqrt<mode>2_fe): Likewise.
24449 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
24452 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
24454 2017-03-02 Jakub Jelinek <jakub@redhat.com>
24456 PR rtl-optimization/79780
24457 * cprop.c (one_cprop_pass): When second and further conditional trap
24458 in a single basic block is turned into an unconditional trap, turn it
24459 into a deleted note to avoid RTL verification failures.
24461 2017-03-02 Richard Biener <rguenther@suse.de>
24463 * fold-const.c (const_binop): Use ulow () instead of elt (0).
24465 2017-03-02 Richard Biener <rguenther@suse.de>
24467 PR tree-optimization/79345
24469 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
24470 param and abort the walk, returning -1 if it is hit.
24471 (walk_aliased_vdefs): Take a limit param and pass it on.
24472 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
24473 defaulting to 0 and return a signed int.
24474 * tree-ssa-uninit.c (struct check_defs_data): New struct.
24475 (check_defs): New helper.
24476 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
24477 about uninitialized memory.
24478 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
24479 bogus uninitialized warning.
24480 (fixed_convert_from_real): Likewise.
24482 2017-03-02 Bin Cheng <bin.cheng@arm.com>
24484 PR tree-optimization/66768
24485 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
24486 iv_use if base object can't be determined.
24488 2017-03-02 Jakub Jelinek <jakub@redhat.com>
24490 PR tree-optimization/79345
24491 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
24492 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
24493 (get_pattern_stats): Initialize it.
24494 * genemit.c (gen_expand): Verify match_scratch numbers come after
24495 match_operand/match_dup numbers.
24496 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
24497 match_scratch numbers.
24498 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
24500 * config/s390/s390.md (trunctdsd2): Likewise.
24502 2017-03-02 Richard Biener <rguenther@suse.de>
24504 * wide-int.h (wide_int_storage::operator=): Implement in terms
24507 2017-03-02 Richard Biener <rguenther@suse.de>
24509 PR tree-optimization/79777
24510 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
24511 the to insert expression to sth existing.
24513 2017-03-01 Martin Sebor <msebor@redhat.com>
24515 PR middle-end/79692
24516 * gimple-ssa-sprintf.c
24517 (directive::known_width_and_precision): New function.
24518 (format_integer): Use it.
24519 (get_mpfr_format_length): Consider the full range of precision
24520 when computing %g output with the # flag. Set the likely byte
24521 count to 3 rather than 1 when precision is indeterminate.
24522 (format_floating): Correct the lower bound of precision.
24524 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24526 * doc/invoke.texi: Document default code model for 64-bit Linux.
24528 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24531 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
24532 udiv rather than div since input pattern is unsigned.
24534 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
24536 * config/i386/i386.c (print_reg): Warn for values of
24537 unsupported size in integer register.
24539 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
24542 * config/rs6000/predicates.md (current_file_function_operand): Do
24543 not allow self calls to be local if the function is replaceable.
24545 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
24548 * config/rs6000/altivec.h (vec_ctz and others): Change the
24549 preprocessor macro that controls conditional compilation from
24550 _ARCH_PWR9 to __POWER9_VECTOR__.
24551 (vec_all_ne): Change parameterization of __altivec_scalar_pred
24552 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
24553 control (instead of _ARCH_PWR9 control) so that template
24554 definition uses power9-specific function.
24555 (vec_any_eq): Likewise.
24556 (vec_all_ne): Change macro definition to use a power9-specific
24557 expansion under #ifdef __POWER9_VECTOR__ control (instead of
24558 _ARCH_PWR9 control).
24559 (vec_any_eq) Likewise.
24560 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
24561 expansion for CMPNEF to remove support for xvcmpnesp instruction.
24562 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
24563 support for xvcmpnedp instruction.
24564 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
24565 macro expansion so that Power9 implementation of vec_all_ne does
24566 not use the AltiVec predicate framework.
24567 (VCMPNEH_P): Likewise.
24568 (VCMPNEW_P): Likewise.
24569 (VCMPNED_P): Likewise.
24570 (VCMPNEFP_P): Likewise.
24571 (VCMPNEDP_P): Likewise.
24572 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
24573 implementation of vec_any_eq to not use AltiVec predicate
24575 (VCMPAEH_P): Likewise.
24576 (VCMPAEW_P): Likewise.
24577 (VCMPAED_P): Likewise.
24578 (VCMPAEFP_P): Likewise.
24579 (VCMPAEDP_P): Likewise.
24580 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
24581 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
24582 not use the AltiVec predicate framework.
24583 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
24584 of vec_any_eq to not use AltiVec predicate framework.
24585 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
24586 support for predefined __POWER9_VECTOR__ macro to indicate that
24587 Power9 instruction selection is enabled.
24588 (altivec_overloaded_builtins): Remove extraneous
24589 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
24590 function argument types RS6000_BTI_bool_V16QI and
24591 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
24592 entry for overloaded function argument types RS6000_BTI_bool_V4SI
24593 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
24594 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
24595 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
24596 Power9 for implementations of vec_cmpne. Change the signature for
24597 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
24598 (representing vec_all_ne) to remove the previously described first
24599 argument of type RS6000_BTI_INTSI, as this was an artifact of
24600 reliance on the AltiVec predicate framework, which is no longer
24601 used in the implementation of these functions. Add
24602 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
24603 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
24604 since, unlike the AltiVec predicate framework implementation, we
24605 do not share function descriptors between vec_alle and vec_anyeq.
24606 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
24607 set of modes that receive special treatment even when
24608 TARGET_P9_VECTOR is true. The special treatment emits code that
24609 does not depend on Power9 instructions.
24610 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
24611 define_expand to not rely on AltiVec predicate framework.
24612 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
24614 (vector_ne_v2di_p): Change this define_expand to not rely on
24615 AltiVec predicate framework.
24616 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
24618 (vector_ne_<mode>_p): Change this define_expand to not rely on
24619 AltiVec predicate framework.
24620 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
24622 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
24623 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
24624 define_insn pattern.
24625 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
24626 define_insn pattern because the xvcmpne<VSs>. instruction is not
24628 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
24629 instruction is not supported.
24631 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24633 * config/nvptx/nvptx.c: Include intl.h.
24635 2017-03-01 Martin Jambor <mjambor@suse.cz>
24638 * ipa-prop.h (ipa_bits): Removed field known.
24639 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
24640 to pointers. Adjusted their comments to warn about their sharing.
24641 (ipcp_transformation_summary): Change bits to a vector of pointers.
24642 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
24643 (ipa_get_ipa_bits_for_value): Declare.
24644 * tree-vrp.h (value_range): Mark as GTY((for_user)).
24645 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
24646 (ipa_bits_hash_table): Likewise.
24647 (ipa_vr_ggc_hash_traits): Likewise.
24648 (ipa_vr_hash_table): Likewise.
24649 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
24650 being pointers and vr_known being removed.
24651 (ipa_set_jf_unknown): Likewise.
24652 (ipa_get_ipa_bits_for_value): New function.
24653 (ipa_set_jfunc_bits): Likewise.
24654 (ipa_get_value_range): New overloaded functions.
24655 (ipa_set_jfunc_vr): Likewise.
24656 (ipa_compute_jump_functions_for_edge): Use the above functions to
24657 construct bits and vr parts of jump functions.
24658 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
24659 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
24661 (ipcp_grow_transformations_if_necessary): Also allocate
24662 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
24664 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
24665 them. Fix too long lines.
24666 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
24667 vr_known being removed.
24668 (ipa_read_jump_function): Use new setter functions to construct bits
24669 and vr parts of jump functions or set them to NULL.
24670 (write_ipcp_transformation_info): Adjust for bits being pointers.
24671 (read_ipcp_transformation_info): Likewise.
24672 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
24674 Include gt-ipa-prop.h.
24675 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
24677 (ipcp_store_bits_results): Likewise.
24678 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
24679 Do not write to existing jump functions but use a temporary instead.
24681 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24684 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
24685 attempt to use its first operand as BIT_FIELD_REF base.
24687 2017-03-01 Richard Biener <rguenther@suse.de>
24689 PR middle-end/79721
24690 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
24691 interpolating formula in wrapping arithmetic.
24692 (chrec_apply): Convert chrec_evaluate return value to wanted type.
24694 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24696 PR tree-optimization/79734
24697 * tree-vect-generic.c (expand_vector_condition): Optimize
24698 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
24699 Handle VEC_COND_EXPR where comparison has different inner width from
24700 type's inner width.
24702 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
24704 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
24705 markup, and similar issues. Remove @opindex entries for things
24706 that aren't options. Add missing -mmpy-option entries.
24708 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24710 PR tree-optimization/79737
24711 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
24712 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
24713 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
24714 instead of byte_size. Formatting fix.
24715 (shift_bytes_in_array_right): Formatting fix.
24717 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
24720 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
24721 condition on optimize for the leaf function test.
24723 2017-02-28 Martin Liska <mliska@suse.cz>
24726 * read-rtl-function.c (function_reader::handle_unknown_directive):
24727 Bail out when one uses -flto.
24729 2017-02-28 Martin Liska <mliska@suse.cz>
24731 * common.opt: Replace space with tabular for options of <number>
24733 * config/i386/i386.opt: Show <number> value for
24734 -mlarge-data-threshold.
24735 * opts.c (print_filtered_help): Do not display number in hexadecimal
24738 2017-02-28 Martin Liska <mliska@suse.cz>
24740 * common.opt: Fix --help=option -Q for options which are of
24743 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
24745 * config/i386/i386.c (print_reg): Error out for values
24746 of 8-bit size in invalid integer register.
24748 2017-02-28 Martin Sebor <msebor@redhat.com>
24750 PR tree-optimization/79691
24751 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
24753 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24756 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
24757 gcc_unreachable with output_operand_lossage.
24759 2017-02-28 Richard Biener <rguenther@suse.de>
24761 PR tree-optimization/79740
24762 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
24764 (visit_nary_op): Insert the nary into the hashtable if we
24765 pattern-matched sth.
24766 * tree-ssa-pre.c (eliminate_insert): Robustify.
24768 2017-02-28 Richard Biener <rguenther@suse.de>
24770 PR middle-end/79731
24771 * fold-const.c (decode_field_reference): Reject out-of-bound
24774 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24776 * config/i386/i386.c: Include intl.h.
24777 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
24778 instead of just cond ? "..." : "...".
24779 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
24780 * coverage.c (read_counts_file): Likewise.
24781 * omp-offload.c: Include intl.h.
24782 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
24783 of just cond ? "..." : "...".
24784 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
24785 of just cond ? "..." : "...".
24787 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
24790 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
24792 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
24793 'tune for' CPU name.
24794 * config/arm/arm-cpu-data.h: Regenerated.
24796 2017-02-28 Richard Biener <rguenther@suse.de>
24798 PR tree-optimization/79732
24799 * tree-inline.c (expand_call_inline): Do not shadow var.
24801 2017-02-28 Richard Biener <rguenther@suse.de>
24803 PR tree-optimization/79723
24804 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
24805 address-space properly.
24807 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
24809 * doc/optinfo.texi (Optimization groups): Fix option used for
24811 * doc/invoke.texi (-fopt-info): Document "omp".
24812 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
24813 (OPTGROUP_ALL): Add OPTGROUP_OMP.
24814 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
24815 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
24816 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
24818 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
24820 * dumpfile.c (optgroup_options): Instead of "openmp", associate
24821 OPTGROUP_OMP with "omp".
24823 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
24826 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
24827 for arithmetic shift of unsigned V2DI.
24829 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
24831 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
24832 arc/linux.h headers.
24833 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
24834 (LINK_SPEC): Likewise.
24835 (ARC_TLS_EXTRA_START_SPEC): Likewise.
24836 (EXTRA_SPECS): Likewise.
24837 (STARTFILE_SPEC): Likewise.
24838 (ENDFILE_SPEC): Likewise.
24839 (LIB_SPEC): Likewise.
24840 (TARGET_SDATA_DEFAULT): Likewise.
24841 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
24842 (MULTILIB_DEFAULTS): Likewise.
24843 (DWARF2_UNWIND_INFO): Likewise.
24844 * config/arc/big.h: New file.
24845 * config/arc/elf.h: Likewise.
24846 * config/arc/linux.h: Likewise.
24847 * config/arc/t-uClibc: Remove.
24849 2017-02-27 Bin Cheng <bin.cheng@arm.com>
24851 PR tree-optimization/77536
24852 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
24853 (tree_transform_and_unroll_loop): Use above function to compute the
24854 estimated niter of unrolled loop and use it when scaling profile.
24855 Also use count info rather than frequency if it's non-zero.
24856 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
24857 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
24858 (vect_transform_loop): Call above function.
24860 2017-02-27 Richard Biener <rguenther@suse.de>
24862 PR tree-optimization/45397
24863 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
24864 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
24865 (visit_nary_op): Add pattern matching for CSEing sign-changed
24866 or truncated operations with wider ones.
24868 2017-02-27 Richard Biener <rguenther@suse.de>
24870 PR tree-optimization/79690
24871 * tree-vect-stmts.c (vectorizable_store): Use vector type
24872 built from the DR with address-space.
24874 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
24876 * doc/invoke.texi (Optimize Options): Refine the description
24877 of asan-use-after-return.
24879 2017-02-25 Alan Modra <amodra@gmail.com>
24881 PR rtl-optimization/79584
24882 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
24883 base, not ad->base_term, the reg within base. Remove assertion
24884 that ad->base == ad->base_term. Replace gen_int_mode using
24885 bogus mode with const0_rtx.
24887 2017-02-25 Jakub Jelinek <jakub@redhat.com>
24889 PR middle-end/79396
24890 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
24891 FMA_EXPR like tcc_binary or tcc_unary.
24893 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
24896 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
24898 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
24899 (output_loc_operands): Handle DW_OP_call_ref and
24900 DW_OP_GNU_variable_value.
24901 (struct variable_value_struct): New type.
24902 (struct variable_value_hasher): Likewise.
24903 (variable_value_hash): New variable.
24904 (string_types): Remove.
24905 (copy_loc_descr): New function.
24906 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
24907 (prepend_loc_descr_to_each): New function.
24908 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
24909 instead of add_loc_descr_to_each if the first argument is single
24910 location list and the second has multiple.
24911 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
24912 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
24913 when looking for variable value which doesn't have other location info.
24914 (loc_list_from_tree): Formatting fix.
24915 (gen_array_type_die): Simplify DW_AT_string_length handling.
24916 (adjust_string_types): Remove.
24917 (gen_subprogram_die): Don't call adjust_string_types nor test/set
24918 string_types. Call resolve_variable_values.
24919 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
24920 (resolve_addr_in_expr): Likewise. Add A argument.
24921 (copy_deref_exprloc): Remove deref argument. Adjust for the
24922 original expression being DW_OP_GNU_variable_value with optionally
24923 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
24924 optionally after it.
24925 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
24926 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
24927 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
24928 (variable_value_hasher::hash, variable_value_hasher::equal): New
24930 (resolve_variable_value_in_expr, resolve_variable_value,
24931 resolve_variable_values, note_variable_value_in_expr,
24932 note_variable_value): New functions.
24933 (dwarf2out_early_finish): Call note_variable_value on all toplevel
24936 2017-02-24 Jakub Jelinek <jakub@redhat.com>
24939 * opts.h (handle_generated_option): Add GENERATED_P argument.
24940 * opts-common.c (handle_option): Adjust function comment.
24941 (handle_generated_option): Add GENERATED_P argument, pass it to
24943 (control_warning_option): Pass false to handle_generated_option
24945 * opts.c (maybe_default_option): Pass true to handle_generated_option
24947 * optc-gen.awk: Likewise.
24949 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24951 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
24952 a REG, look at the REG it is a SUBREG of.
24953 (splitter for cmpeqsi_t): Ditto.
24955 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24957 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
24958 the special USEs with the pattern of the insn, not the insn itself.
24960 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
24963 * doc/invoke.texi: Document -mload-store-pairs.
24965 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24966 Sandra Loosemore <sandra@codesourcery.com>
24968 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
24969 argument isn't a CONST_INT.
24970 (nios2_alternate_compare_const): Assert op is a CONST_INT.
24971 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
24972 (nios2_validate_compare): Bypass alternate compare logic if *op2
24973 is not a CONST_INT.
24974 (ldstwm_operation_p): Return false if first_base is not a REG or
24975 if first_offset is not a CONST_INT.
24977 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24979 * config/cris/cris.md: Use correct operand in a define_peephole2.
24981 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24983 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
24985 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24987 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
24988 this_insn if it is an INSN or JUMP_INSN.
24989 (force_offsettable): Look at base, not at addr.
24990 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
24991 on things that aren't necessarily CONST_INTs.
24993 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
24995 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
24996 -mfpmath=sse is the default also for x86-32 targets with SSE2
24997 instruction set when @option{-ffast-math} is enabled
24999 2017-02-24 Jeff Law <law@redhat.com>
25001 PR rtl-optimizatoin/79286
25002 * ira.c (update_equiv_regs): Drop may_trap_p exception to
25005 2017-02-24 Richard Biener <rguenther@suse.de>
25007 PR tree-optimization/79389
25008 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
25011 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
25013 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
25014 function comment to reflect reality.
25015 (loop_exits_before_overflow): Fix typo in function description.
25017 2017-02-24 Richard Biener <rguenther@suse.de>
25019 PR tree-optimization/79389
25020 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
25021 properly that a threading opportunity exists. Detect conditional
25022 copy/constant propagation opportunities.
25024 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
25026 * config/visium/visium.md (type): Add trap.
25027 (b): New mode attribute.
25028 (*btst): Rename into...
25029 (*btst<mode>): ...this and adjust.
25030 (*cbranchsi4_btst_insn): Rename into...
25031 (*cbranch<mode>4_btst_insn): ...this and adjust.
25032 (trap): New define_insn.
25034 2017-02-23 Jakub Jelinek <jakub@redhat.com>
25036 PR tree-optimization/79389
25037 * ifcvt.c (struct noce_if_info): Add rev_cond field.
25038 (noce_reversed_cond_code): New function.
25039 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
25040 reversed_comparison_code. Formatting fix.
25041 (noce_try_store_flag): Test rev_cond != NULL in addition to
25042 reversed_comparison_code.
25043 (noce_try_store_flag_constants): Likewise.
25044 (noce_try_store_flag_mask): Likewise.
25045 (noce_try_addcc): Use rev_cond if non-NULL instead of
25046 reversed_comparison_code.
25047 (noce_try_cmove_arith): Likewise. Formatting fixes.
25048 (noce_try_minmax, noce_try_abs): Clear rev_cond.
25049 (noce_find_if_block): Initialize rev_cond.
25050 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
25051 instead of false as last argument never attempt to reverse it
25054 2017-02-23 Bin Cheng <bin.cheng@arm.com>
25056 PR tree-optimization/79663
25057 * tree-predcom.c (combine_chains): Process refs in reverse order
25058 only for ZERO length chains, and add explaining comment.
25060 2017-02-23 Jeff Law <law@redhat.com>
25062 PR tree-optimization/79578
25063 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
25064 in call to operand_equal_p.
25066 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
25069 * config/i386/cpuid.h: Fix another undefined behavior.
25071 2017-02-23 Richard Biener <rguenther@suse.de>
25073 PR tree-optimization/79683
25074 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
25075 vector types for data-refs.
25077 2017-02-23 Martin Liska <mliska@suse.cz>
25079 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
25081 2017-02-23 Jakub Jelinek <jakub@redhat.com>
25083 PR middle-end/79665
25084 * internal-fn.c (get_range_pos_neg): Moved to ...
25085 * tree.c (get_range_pos_neg): ... here. No longer static.
25086 * tree.h (get_range_pos_neg): New prototype.
25087 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
25088 are known to be in between 0 and signed maximum inclusive, try to
25089 expand both unsigned and signed divmod and use the cheaper one from
25092 2017-02-22 Jeff Law <law@redhat.com>
25094 PR tree-optimization/79578
25095 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
25096 to compare base operands.
25098 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
25101 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
25102 gpc_reg_operand instead of fpr_reg_operand.
25104 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
25106 * config/mips/mips.c (mips_return_in_memory): Force FP
25107 vector types to be returned in memory for o32 ABI.
25109 2017-02-22 Jakub Jelinek <jakub@redhat.com>
25111 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
25112 instead of DW_TAG_member for static data member declarations and don't
25113 set no_linkage_name for static inline data members.
25114 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
25117 2017-02-22 Martin Liska <mliska@suse.cz>
25119 * doc/invoke.texi: Replace inequality signs with square brackets
25122 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25124 PR tree-optimization/68644
25125 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
25127 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
25130 * lra-constraints.c (simplify_operand_subreg): Handle
25131 WORD_REGISTER_OPERATIONS targets.
25133 2017-02-22 Jakub Jelinek <jakub@redhat.com>
25136 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
25137 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
25138 elimination by swapping fld*.
25140 2017-02-22 Richard Biener <rguenther@suse.de>
25142 PR tree-optimization/79673
25143 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
25144 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
25145 irrelevant address-space qualifiers and avoiding a
25146 ADDR_SPACE_CONVERT_EXPR from fold_convert.
25148 2017-02-22 Richard Biener <rguenther@suse.de>
25150 PR tree-optimization/79666
25151 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
25152 to not symbolically negate if that may introduce undefined
25155 2017-02-22 Martin Liska <mliska@suse.cz>
25158 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
25159 * data-streamer-out.c (streamer_write_gcov_count_stream):
25161 * value-prof.c (stream_out_histogram_value): Make assert more
25162 precise based on type of counter.
25164 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
25167 * config/i386/i386.md (standard_x87sse_constant_load splitter):
25168 Use nonimmediate_operand instead of memory_operand for operand 1.
25169 (float-extend standard_x87sse_constant_load splitter): Ditto.
25171 2017-02-21 Jeff Law <law@redhat.com>
25173 PR tree-optimization/79621
25174 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
25175 blocks with edges to themselves.
25177 2017-02-21 Jakub Jelinek <jakub@redhat.com>
25180 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
25181 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
25182 Use gimple_call_builtin_p.
25185 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
25186 on temporarily removed DEBUG_INSNs.
25188 PR tree-optimization/79649
25189 * tree-loop-distribution.c (classify_partition): Give up on
25190 non-generic address space loads/stores.
25192 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
25194 * doc/loop.texi (Loop manipulation): Remove nonexistent
25195 tree_ssa_loop_version from the documentation.
25196 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
25198 2017-02-21 Jakub Jelinek <jakub@redhat.com>
25201 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
25202 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
25203 * config/rs6000/rs6000.c: Include except.h.
25204 (rs6000_expand_split_stack_prologue): Call
25205 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
25207 2017-02-21 Martin Jambor <mjambor@suse.cz>
25210 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
25211 have been analyzed.
25213 2017-02-21 Martin Jambor <mjambor@suse.cz>
25215 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
25216 for backward compatibility only.
25217 * doc/invoke.texi (Option Summary): Remove all references to
25218 -fipa-cp-alignment.
25220 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
25224 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25226 * lra-constraints.c (curr_insn_transform): Handle
25227 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
25229 2017-02-21 Martin Liska <mliska@suse.cz>
25231 * config/i386/i386.opt: Replace -masm-dialect with -masm.
25233 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
25235 PR translation/79638
25236 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
25238 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
25241 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
25242 (arm_function_ok_for_sibcall): Return false for an indirect call by
25243 descriptor if all the argument registers are used.
25244 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
25245 alignment of the function.
25247 2017-02-21 Jakub Jelinek <jakub@redhat.com>
25249 PR tree-optimization/61441
25250 * simplify-rtx.c (simplify_const_unary_operation): For
25251 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
25252 the sNaN unmodified.
25254 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
25256 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25257 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25258 instead of SYSTEM_HEADER_DIR.
25260 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
25261 Martin Liška <mliska@suse.cz>
25263 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
25264 Fix typos and grammar, use active voice, and clarify.
25266 2017-02-20 Marek Polacek <polacek@redhat.com>
25268 PR middle-end/79537
25269 * gimplify.c (gimplify_expr): Handle unused *&&L;.
25272 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
25274 2017-02-20 Jakub Jelinek <jakub@redhat.com>
25277 * config/i386/i386.c (ix86_expand_builtin): Handle
25278 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
25279 ix86_builtins_isa[fcode].isa as a requirement of those
25280 flags and any other flag in the bitmask.
25281 (ix86_init_mmx_sse_builtins): Use 0 instead of
25282 ~OPTION_MASK_ISA_64BIT as mask.
25283 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
25284 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
25285 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
25286 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
25288 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25291 * lra-constraints.c (split_reg): Check requested split mode
25292 is supported by the register.
25294 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25296 * lra-constraints.c (simplify_operand_subreg): Remove early
25299 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25302 * lra-constraints.c (curr_insn_transform): Tighten condition
25303 for converting SUBREG reloads from OP_OUT to OP_INOUT.
25305 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25308 * lra-constraints.c (curr_insn_transform): Handle
25309 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
25311 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
25314 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
25316 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
25318 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
25321 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
25324 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25326 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
25329 2017-02-18 Jakub Jelinek <jakub@redhat.com>
25331 * final.c (last_columnnum, override_columnnum): New variables.
25332 (final_start_function): Set last_columnnum, pass it to begin_prologue
25333 hook and pass 0 to dwarf2out_begin_prologue.
25334 (final_scan_insn): Update override_columnnum. Pass last_columnnum
25335 to source_line debug hook.
25336 (notice_source_line): Compute last_columnnum and for debug_column_info
25337 return true on column changes.
25338 * debug.h (struct gcc_debug_hooks): Add column argument to
25339 source_line and begin_prologue hooks.
25340 (debug_nothing_int_charstar_int_bool): Remove prototype.
25341 (debug_nothing_int_int_charstar,
25342 debug_nothing_int_int_charstar_int_bool): New prototypes.
25343 (dwarf2out_begin_prologue): Add column argument.
25344 * debug.c (do_nothing_debug_hooks): Adjust source_line and
25345 begin_prologue hooks.
25346 (debug_nothing_int_charstar_int_bool): Remove.
25347 (debug_nothing_int_int_charstar,
25348 debug_nothing_int_int_charstar_int_bool): New functions.
25349 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
25350 through to dwarf2out_source_line.
25351 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
25352 (dwarf2out_source_line): Add column argument, emit it if requested.
25353 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
25355 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
25356 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
25357 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
25358 through to dwarf2out_begin_prologue.
25359 (vmsdbgout_source_line): Add column argument, pass it through to
25360 dwarf2out_source_line.
25361 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
25362 dbxout_source_line caller.
25363 (dbxout_source_line): Add column argument.
25365 * common.opt (gno-column-info, gcolumn-info): New options.
25366 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
25367 (check_die): Also test for multiple DW_AT_decl_column attributes.
25368 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
25369 DW_AT_decl_column if requested.
25370 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
25372 (gen_variable_die): Likewise.
25373 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
25374 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
25377 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
25378 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
25379 (ix86_handle_option): Handle OPT_m3dnowa.
25380 * doc/invoke.texi (-m3dnowa): Document.
25381 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
25382 -m3dnowa instead of -m3dnow -march=athlon.
25385 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
25386 instead of gcc_assert for K, r and R code checks. Formatting fixes.
25388 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25391 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
25392 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
25393 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
25394 generator for vsx_xxpermdi_<mode>_be.
25395 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
25396 force big-endian semantics.
25397 (vsx_xxpermdi_<mode>_be): New define_expand with same
25398 implementation as previous version of vsx_xxpermdi_<mode>.
25400 2017-02-17 Jakub Jelinek <jakub@redhat.com>
25402 PR tree-optimization/79327
25403 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
25404 variable, its initialization and use.
25406 2017-02-17 Julia Koval <julia.koval@intel.com>
25408 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
25409 (OPTION_MASK_ISA_PKU_UNSET): New.
25410 (ix86_handle_option): Handle -mrdpid.
25411 * config/i386/cpuid.h (bit_RDPID): New.
25412 * config/i386/driver-i386.c (host_detect_local_cpu):
25413 Detect RDPID feature.
25414 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
25415 * config/i386/i386-c.c (ix86_target_macros_internal):
25417 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
25418 (ix86_valid_target_attribute_inner_p): Add "rdpid".
25419 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
25420 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
25421 * config/i386/i386.md (define_insn "rdpid"): New.
25422 * config/i386/i386.opt Add -mrdpid.
25423 * config/i386/immintrin.h (_rdpid_u32): New.
25425 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
25427 PR rtl-optimization/79541
25428 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
25429 instead of transforming it into USE.
25431 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
25433 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
25434 If HONOR_SNANS (SFmode) force the input to a register.
25435 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
25436 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
25437 an frsp or similar insn.
25439 2017-02-17 Martin Liska <mliska@suse.cz>
25441 PR rtl-optimization/79577
25442 * params.def (selsched-max-sched-times): Increase minimum to 1.
25444 2017-02-17 Martin Liska <mliska@suse.cz>
25446 PR rtl-optimization/79574
25447 * gcse.c (want_to_gcse_p): Prevent integer overflow.
25449 2017-02-17 Martin Liska <mliska@suse.cz>
25451 PR tree-optimization/79529
25452 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
25453 ssa_defined_default_def_p to handle cases which are implicitly
25455 * tree-ssa.c (ssa_defined_default_def_p): New function.
25456 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
25457 which are implicitly defined.
25458 * tree-ssa.h (ssa_defined_default_def_p): Declare.
25460 2017-02-17 Richard Biener <rguenther@suse.de>
25462 PR middle-end/79576
25463 * params.def (max-ssa-name-query-depth): Limit to 10.
25465 2017-02-17 Richard Biener <rguenther@suse.de>
25467 PR tree-optimization/79552
25468 * tree-ssa-structalias.c (visit_loadstore): Properly verify
25471 2017-02-17 Richard Biener <rguenther@suse.de>
25474 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
25476 2017-02-17 Marek Polacek <polacek@redhat.com>
25478 PR middle-end/79536
25479 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
25480 (fold_negate_expr): New wrapper.
25482 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
25484 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
25485 Correct terminology and de-emphasize pre-standard behavior.
25487 2017-02-16 Alan Modra <amodra@gmail.com>
25489 PR rtl-optimization/79286
25490 * ira.c (def_dominates_uses): New function.
25491 (update_equiv_regs): Don't create an equivalence for insns that
25492 may trap where the register def does not dominate the use.
25494 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
25496 PR rtl-optimization/78127
25497 * lra.c (lra): Call lra_eliminate before finish the loop after
25500 2017-02-16 Richard Biener <rguenther@suse.de>
25502 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
25504 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
25505 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
25506 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
25507 (isl_val_int_from_wi): New function.
25508 (extract_affine_gmp): Rename to ...
25509 (extract_affine_wi): ... this, take a widest_int.
25510 (extract_affine_int): Just wrap extract_affine_wi.
25511 (add_param_constraints): Use isl_val_int_from_wi.
25512 (add_loop_constraints): Likewise, and extract_affine_wi.
25514 2017-02-15 Jeff Law <law@redhat.com>
25516 PR middle-end/79521
25517 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
25518 ira_init_register_move_cost_if_necessary.
25520 2017-02-15 Martin Sebor <msebor@redhat.com>
25522 PR middle-end/32003
25523 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
25524 removed in a prior commit.
25526 2017-02-15 Bin Cheng <bin.cheng@arm.com>
25528 PR tree-optimization/79347
25529 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
25530 counters during peeling.
25532 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
25534 * Makefile.in (site.exp): Remove "set ISLVER".
25536 2017-02-15 Jakub Jelinek <jakub@redhat.com>
25539 * real.c (real_from_integer): Call real_convert even for decimal.
25541 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
25544 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
25546 2017-02-14 Andrew Pinski <apinski@cavium.com>
25548 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
25549 cores and change the partno/implementer to be correct.
25550 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
25551 the 'B" as the implementer.
25552 * config/aarch64/aarch64-tune.md: Regenerate.
25554 2017-02-14 Carl Love <cel@us.ibm.com>
25556 * config/rs6000/rs6000.c: Add case statement entry to make the
25557 xvcvuxdsp built-in argument unsigned.
25558 * config/rs6000/vsx.md: Fix the source and return operand types so they
25559 match the instruction definitions from the ISA document. Fix typo
25560 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
25563 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
25566 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
25567 member early_clobber_alts.
25568 * lra-lives.c (reg_early_clobber_p): New.
25569 (process_bb_lives): Use it.
25570 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
25571 (debug_operand_data): Initialize early_clobber_alts.
25572 (setup_operand_alternative): Set up early_clobber_alts.
25573 (collect_non_operand_hard_regs): Ditto. Pass early clobber
25574 alternatives to new_insn_reg.
25575 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
25577 (lra_update_insn_regno_info): Pass the new arg.
25579 2017-02-14 Jakub Jelinek <jakub@redhat.com>
25581 PR middle-end/79505
25582 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
25583 (new_oacc_loop_raw): Don't clear already cleared fields.
25586 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
25587 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
25588 _mm512_prefetch_i64gather_ps): New inline functions and macros.
25590 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
25593 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
25595 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
25598 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
25599 the extra instruction to the right place to store 128-bit constant
25602 2017-02-14 Martin Sebor <msebor@redhat.com>
25604 PR middle-end/79448
25605 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
25606 warning for strings of unknown length.
25608 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
25610 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
25612 2017-02-14 Jeff Law <law@redhat.com>
25615 * ira-costs.c (scan_one_insn): Initialize register move costs
25616 for pseudos seen in USE/CLOBBER insns.
25618 PR tree-optimization/79095
25619 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
25620 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
25621 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
25622 if the operands are known to be not equal, then the resulting range
25624 (intersect_ranges): If the new range is ~[0,0] and the old range is
25625 wide, then prefer ~[0,0].
25626 * tree-vrp.c (overflow_comparison_p_1): New function.
25627 (overflow_comparison_p): New function.
25628 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
25629 if NAME is used in an overflow test.
25630 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
25631 overflow check that can be expressed as an equality test, then adjust
25632 ops to be that equality test.
25634 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25636 * config/s390/s390-builtin-types.def: Remove flags argument.
25637 * config/s390/s390.c (s390_init_builtins): Likewise.
25639 2017-02-14 Martin Liska <mliska@suse.cz>
25641 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
25642 vector. Fix trailing white spaces.
25644 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
25646 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
25649 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25651 PR rtl-optimization/68664
25652 * config/arm/arm.c (arm_sched_can_speculate_insn):
25653 New function. Declare prototype.
25654 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
25656 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25658 PR rtl-optimization/68664
25659 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
25661 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
25663 2017-02-14 Amit Pawar <amit.pawar@amd.com>
25665 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
25666 max skip bytes for function, loop and jump.
25668 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25670 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
25671 ABS_EXPR for gimple dump.
25673 2017-02-14 Jakub Jelinek <jakub@redhat.com>
25676 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
25678 PR tree-optimization/79408
25679 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
25680 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
25681 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
25682 also if rhs1 is INTEGER_CST.
25684 2017-02-14 Richard Biener <rguenther@suse.de>
25686 PR middle-end/79432
25687 * tree-into-ssa.c (insert_phi_nodes): When the function can
25688 have abnormal edges rewrite SSA names with broken use-def
25689 dominance out of SSA and register them for PHI insertion.
25691 2017-02-13 Martin Sebor <msebor@redhat.com>
25693 PR middle-end/79496
25694 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
25695 clearing info.nowrite flag when snprintf size argument is a range.
25697 2017-02-13 Jakub Jelinek <jakub@redhat.com>
25699 * cprop.c (cprop_jump): Add missing space in string literal.
25700 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
25701 (get_constraint_for_component_ref): Likewise.
25702 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
25703 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
25704 * lra-constraints.c (process_alt_operands): Likewise.
25705 * ipa-inline.c (inline_small_functions): Likewise.
25706 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
25707 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
25708 * trans-mem.c (diagnose_tm_1_op): Likewise.
25709 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
25710 (grid_parallel_clauses_gridifiable): Likewise.
25712 * config/nvptx/mkoffload.c (process): Add space in between
25715 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
25716 "MOD4_SSE_REGS" and "ALL_REGS".
25718 * spellcheck.c (test_data): Add , in between "foo" and "food".
25720 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25723 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
25724 boundary crossing check and subsequent code generation agree.
25726 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25728 * config/aarch64/aarch64.c (has_memory_op): Delete.
25729 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
25732 2017-02-13 Jakub Jelinek <jakub@redhat.com>
25734 PR rtl-optimization/79388
25735 PR rtl-optimization/79450
25736 * combine.c (distribute_notes): When removing TEM_INSN for which
25737 corresponding dest has last value recorded, invalidate that last
25740 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25742 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
25743 of explicit '@'. Add missing assembly comment marker on branch costs
25746 2017-02-13 Nathan Sidwell <nathan@acm.org>
25748 * gengtype-lex.l (<in_struct>): Add '/'.
25750 2017-02-13 Martin Liska <mliska@suse.cz>
25753 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
25755 2017-02-13 Richard Biener <rguenther@suse.de>
25757 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
25759 * configure: Re-generate.
25760 * config.in: Likewise.
25761 * graphite-dependences.c: Simplify as if
25762 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
25763 * graphite-isl-ast-to-gimple.c: Likewise.
25764 * graphite-optimize-isl.c: Likewise.
25765 * graphite-poly.c: Likewise.
25766 * graphite-sese-to-poly.c: Likewise.
25767 * graphite.h: Likewise.
25768 * toplev.c: Include isl/version.h and use isl_version () for
25769 printing the ISL version.
25770 * doc/install.texi: Update ISL requirement.
25772 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
25774 * doc/standards.texi (Standards): Update reference to
25777 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
25779 * doc/extend.texi (Named Address Spaces): sourceware.org now
25781 * doc/install.texi (Binaries): Ditto.
25784 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
25786 * doc/cpp.texi: Replace "stringify"/"stringification" with C
25787 standard terminology "stringize"/"stringizing" throughout.
25788 * doc/cppinternals.texi: Likewise.
25790 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
25792 * doc/extend.texi: Fix some spelling mistakes and typos.
25793 * doc/invoke.texi: Likewise.
25795 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
25798 * params.def (inline-min-speedup) Change from 10 to 8.
25800 2017-02-11 Jakub Jelinek <jakub@redhat.com>
25802 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
25805 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
25808 * ipa-inline-analysis.c (get_minimal_bb): New function.
25809 (record_modified): Use it.
25810 (remap_edge_change_prob): Handle also ancestor functions.
25812 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
25814 * doc/contrib.texi (Contributors): Remove broken link into
25815 the Mauve CVS repository.
25817 2017-02-11 Jakub Jelinek <jakub@redhat.com>
25819 PR middle-end/79454
25820 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
25821 result computation whenever lhs doesn't have vector mode, not
25822 just when it has BLKmode.
25824 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
25826 * doc/makefile.texi (profiledbootstrap): Refer to the
25827 installation instructions only in textual form.
25829 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25832 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
25834 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
25836 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
25837 (Specific): Update mingw-w64 reference.
25839 (Specific): Remove broken link to Renesas RX processor.
25841 2017-02-10 Richard Biener <rguenther@suse.de>
25843 * toplev.c (process_options): Do not mention obsolete graphite
25844 options when printing sorry message about missing graphite support.
25845 Mention -floop-nest-optimize.
25847 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
25849 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
25850 (vtst_p16): Likewise.
25851 (vtstq_p8): Likewise.
25852 (vtstq_p16): Likewise.
25854 (vtstq_p64): Likewise.
25855 * config/arm/arm_neon.h (vgetq_lane_p64): New.
25856 (vset_lane_p64): New.
25857 (vsetq_lane_p64): New.
25859 2017-02-10 Jakub Jelinek <jakub@redhat.com>
25861 PR tree-optimization/79411
25862 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
25863 stmt operands are SSA_NAMEs used in abnormal phis.
25864 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
25867 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
25870 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
25873 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
25875 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
25877 2017-02-09 Jakub Jelinek <jakub@redhat.com>
25879 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
25883 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
25884 not arbitrary TREE_CONSTANT.
25887 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
25888 "omp declare target link" attribute unless is_global_var.
25889 * omp-offload.c (find_link_var_op): Likewise.
25891 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
25892 Chung-Lin Tang <cltang@codesourcery.com>
25894 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
25896 (gimplify_adjust_omp_clauses): Don't delete TILE.
25897 (gimplify_omp_for): Deal with TILE.
25898 * internal-fn.c (expand_GOACC_TILE): New function.
25899 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
25901 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
25902 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
25904 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
25905 avoid DIV for outermost collapse var.
25906 (expand_oacc_for): Insert tile element loop as needed. Adjust.
25907 Remove out of date comments, fix whitespace.
25908 * omp-general.c (omp_extract_for_data): Deal with tiling.
25909 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
25910 adjust OLF_DIM_BASE value.
25911 (struct omp_for_data): Add tiling field.
25912 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
25913 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
25914 for auto loops. Remove default auto determining, moved to
25915 oacc_loop_fixed_partitions.
25916 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
25917 stmts, add e_mask field.
25918 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
25919 (oacc_thread_numbers): Use oacc_dim_call.
25920 (oacc_xform_tile): New.
25921 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
25922 (finish_oacc_loop): Adjust for ifns vector.
25923 (oacc_loop_discover_walk): Append loop abstraction sites to list,
25924 add case for GOACC_TILE fns.
25925 (oacc_loop_xform_loop): Delete.
25926 (oacc_loop_process): Iterate over call list directly, and add
25927 handling for GOACC_TILE fns.
25928 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
25930 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
25931 vector partitioning to outer loops. Assign 2 partitions to loops
25932 when available. Add TILE handling.
25933 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
25934 (execite_oacc_device_lower): Process GOACC_TILE fns,
25935 ignore unknown specs.
25936 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
25937 * tree.c (omp_clause_num_ops): Adjust TILE ops.
25938 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
25940 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
25942 * configure.ac (ACX_BUGURL): Update.
25943 * configure: Regenerate.
25945 2017-02-09 Richard Biener <rguenther@suse.de>
25947 PR tree-optimization/69823
25948 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
25949 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
25951 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
25953 * config/arc/arc-c.def: Add __NPS400__ definition.
25954 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
25955 (TARGET_NPS400): Define.
25957 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
25959 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
25961 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
25962 pointer, arch_info.
25963 (arc_cpu_types): Fill the arch_info field with a pointer into the
25964 arc_arch_types table.
25965 (arc_selected_cpu): Declare.
25966 * config/arc/arc.c (arc_selected_cpu): Make global.
25967 (arc_selected_arch): Delete.
25968 (arc_base_cpu): Delete.
25969 (arc_override_options): Remove references to deleted variables,
25970 update access to arch information.
25971 (ARC_OPT): Update access to arch information.
25972 (ARC_OPTX): Likewise.
25973 * config/arc/arc.h (arc_base_cpu): Remove declaration.
25974 (TARGET_ARC600): Update access to arch information.
25975 (TARGET_ARC601): Likewise.
25976 (TARGET_ARC700): Likewise.
25977 (TARGET_EM): Likewise.
25978 (TARGET_HS): Likewise.
25979 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
25982 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
25985 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
25986 condition/operands for integer GE/LE/GEU/LEU operations.
25988 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
25990 PR translation/79397
25991 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
25994 2017-02-08 Martin Jambor <mjambor@suse.cz>
25997 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
25998 whether allocation happened.
25999 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
26000 nothing was allocated.
26002 2017-02-08 Jakub Jelinek <jakub@redhat.com>
26004 PR tree-optimization/79408
26005 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
26006 constant, but SSA_NAME with a known integer range, use the minimum
26007 of that range instead of op1 to determine if modulo can be replaced
26008 with its first operand.
26010 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26012 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
26014 2017-02-08 Richard Biener <rguenther@suse.de>
26016 PR tree-optimization/71824
26017 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
26018 Check all loops contained in the merged region.
26020 2017-02-07 Andrew Pinski <apinski@cavium.com>
26022 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
26024 2017-02-07 Andrew Pinski <apinski@cavium.com>
26026 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
26027 (thunderxt88): Likewise.
26028 (thunderxt81): Disable LSE and change v8.1 to v8.
26029 (thunderxt83): Likewise.
26031 2017-02-07 Jakub Jelinek <jakub@redhat.com>
26032 Richard Biener <rguenther@suse.de>
26034 PR middle-end/79399
26035 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
26036 type from int to size_t.
26037 * ira-costs.c (struct_costs_size): Change type from int to size_t.
26039 2017-02-07 Jakub Jelinek <jakub@redhat.com>
26041 PR rtl-optimization/79386
26042 * cprop.c (bypass_conditional_jumps): Initialize
26043 bypass_last_basic_block already before splitting bbs after
26044 unconditional traps...
26045 (bypass_conditional_jumps): ... rather than here.
26048 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
26049 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
26050 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
26051 fix -masm=intel patterns.
26053 2017-02-07 Richard Biener <rguenther@suse.de>
26055 PR tree-optimization/79256
26056 PR middle-end/79278
26057 * builtins.c (get_object_alignment_2): Use min_align_of_type
26058 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
26059 and ADJUST_FIELD_ALIGN.
26061 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
26063 * doc/tm.texi: Regenerate.
26064 * stor-layout.c (layout_decl): Adjust.
26065 (update_alignment_for_field): Likewise.
26066 (place_field): Likewise.
26067 (min_align_of_type): Likewise.
26068 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
26069 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
26070 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
26071 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
26072 * config/frv/frv.c (frv_adjust_field_align): Likewise.
26073 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
26074 * config/i386/i386.c (x86_field_alignment): Likewise.
26075 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
26076 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
26077 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
26078 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
26079 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
26080 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
26084 2017-01-30 Richard Biener <rguenther@suse.de>
26086 PR tree-optimization/79256
26087 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
26088 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
26091 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
26093 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
26094 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
26095 builtins to SImode and emit a zero-extend, if necessary.
26097 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
26099 * docs/invoke.texi (RISC-V Options): Alphabetize.
26101 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
26103 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
26106 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
26108 * config/riscv/riscv.c: New file.
26109 * common/config/riscv/riscv-common.c: Likewise.
26110 * config.gcc: Likewise.
26111 * config/riscv/constraints.md: Likewise.
26112 * config/riscv/elf.h: Likewise.
26113 * config/riscv/generic.md: Likewise.
26114 * config/riscv/linux.h: Likewise.
26115 * config/riscv/multilib-generator: Likewise.
26116 * config/riscv/peephole.md: Likewise.
26117 * config/riscv/pic.md: Likewise.
26118 * config/riscv/predicates.md: Likewise.
26119 * config/riscv/riscv-builtins.c: Likewise.
26120 * config/riscv/riscv-c.c: Likewise.
26121 * config/riscv/riscv-ftypes.def: Likewise.
26122 * config/riscv/riscv-modes.def: Likewise.
26123 * config/riscv/riscv-opts.h: Likewise.
26124 * config/riscv/riscv-protos.h: Likewise.
26125 * config/riscv/riscv.h: Likewise.
26126 * config/riscv/riscv.md: Likewise.
26127 * config/riscv/riscv.opt: Likewise.
26128 * config/riscv/sync.md: Likewise.
26129 * config/riscv/t-elf-multilib: Likewise.
26130 * config/riscv/t-linux: Likewise.
26131 * config/riscv/t-linux-multilib: Likewise.
26132 * config/riscv/t-riscv: Likewise.
26133 * configure.ac: Likewise.
26134 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
26135 Waterman as RISC-V maintainers.
26136 * doc/install.texi: Add RISC-V entries.
26137 * doc/invoke.texi: Add RISC-V options section.
26138 * doc/md.texi: Add RISC-V constraints section.
26139 * configure: Regenerated.
26141 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
26144 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
26145 false values to be constant vectors with all 0 or all 1 bits set.
26146 (vcondu<mode><mode>): Likewise.
26147 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
26149 (fpmask_comparison_operator): Update comment.
26150 (vecint_comparison_operator): New predicate.
26151 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
26152 vector conditionals when the true and false values are constant
26153 vectors with all 0 bits or all 1 bits set.
26155 2017-02-06 Martin Sebor <msebor@redhat.com>
26157 PR tree-optimization/79376
26158 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
26160 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
26162 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
26163 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
26164 to simplify split condition.
26166 2017-02-06 Jakub Jelinek <jakub@redhat.com>
26168 * omp-expand.c (oxpand_omp_atomic_fetch_op,
26169 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
26172 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
26174 PR rtl-optimization/68664
26175 * target.def (can_speculate_insn): New hook.
26176 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
26177 * doc/tm.texi: Regenerate.
26178 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
26179 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
26180 (rs6000_sched_can_speculate_insn): New function.
26182 2017-02-06 Jakub Jelinek <jakub@redhat.com>
26184 PR tree-optimization/79284
26185 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
26186 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
26187 vectorizable_mask_load_store, vectorizable_operation,
26188 vect_is_simple_cond, get_same_sized_vectype): Use it instead
26189 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
26190 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
26191 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
26192 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
26193 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
26194 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
26195 is_gimple_assign (stmt). Replace another such test with
26196 is_gimple_assign (stmt).
26198 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
26201 * config/avr/avr.c (rtl-iter.h): Include it.
26202 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
26203 (avr_legitimate_combined_insn): ...and implementation.
26205 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
26207 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
26208 * config/s390/s390.c (s390_const_operand_ok)
26209 (s390_canonicalize_comparison, s390_extract_part)
26210 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
26211 (s390_contiguous_bitmask_p, s390_rtx_costs)
26212 (legitimize_pic_address): Likewise.
26213 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
26214 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
26215 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
26216 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
26217 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
26219 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
26221 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
26222 REGNO($0) == REGNO($1).
26224 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26226 * config/s390/linux.h(SIZE_TYPE): Add comment.
26228 2017-02-06 Julian Brown <julian@codesourcery.com>
26229 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26230 Virendra Pathak <virendra.pathak@broadcom.com>
26232 * config/aarch64/aarch64-cores.def: Change the scheduler
26234 * config/aarch64/aarch64.md: Include thunderx2t99.md.
26235 * config/aarch64/thunderx2t99.md: New file.
26237 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
26239 * doc/standards.texi (Go Language): Update link to language
26242 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
26244 * tree-eh.c (lower_resx): Sanitize profile.
26245 (cleanup_empty_eh_move_lp): Likewise.
26247 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
26250 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
26252 * cfgloopmanip.h (loop_version): Update prototype.
26253 * modulo-sched.c (sms_schedule): Update call of loop_version.
26254 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
26255 * tree-parloops.c (gen_parallel_loop): Likewise.
26256 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
26257 * tree-ssa-loop-split.c (split_loop): Likewise.
26258 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
26259 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
26261 2017-02-05 Martin Liska <mliska@suse.cz>
26264 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
26266 (print_operand_address): Initialize a struct to zero.
26268 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
26270 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
26271 garbage collector only in textual form.
26273 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
26275 * doc/extend.texi (x86 specific memory model extensions for
26276 transactional memory): Simplify a phrase.
26278 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
26281 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
26282 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
26283 (atomic_storedi_1): Likewise.
26285 2017-02-04 Jakub Jelinek <jakub@redhat.com>
26287 PR tree-optimization/79338
26288 * tree-parloops.c (gather_scalar_reductions): Don't call
26289 vect_analyze_loop_form for loop->inner before destroying loop's
26292 2017-02-03 Martin Sebor <msebor@redhat.com>
26294 PR tree-optimization/79327
26295 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
26296 when precision has resulted in leading zeros.
26297 (format_integer): Adjust the likely counter to assume an unknown
26298 argument that may be zero is non-zero.
26300 2017-02-03 Jason Merrill <jason@redhat.com>
26303 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
26304 avoid copying non-taken branch.
26306 2017-02-03 Jakub Jelinek <jakub@redhat.com>
26308 PR tree-optimization/79340
26309 * tree-vect-loop.c (vectorizable_reduction): Release
26310 vec_defs elements after safe_splicing them into other vectors.
26313 PR tree-optimization/79327
26314 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
26315 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
26317 (format_integer): Use wide_int_to_tree instead of build_int_cst
26318 + to_?hwi. If argmin is NULL, just set argmin and argmax to
26319 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
26320 of shortest and longest sequence.
26322 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
26324 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
26325 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
26327 2017-02-03 Walter Lee <walt@tilera.com>
26330 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
26331 after initial stackframe link reg save.
26332 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
26334 2017-02-03 Jakub Jelinek <jakub@redhat.com>
26337 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
26338 wu for stxssp alternative.
26340 2017-02-03 Martin Sebor <msebor@redhat.com>
26342 PR tree-optimization/79352
26343 * gimple-fold.c (get_range_strlen): Add argument.
26344 (get_range_strlen): Change return type to bool.
26345 (get_maxval_strlen): Pass in a dummy argument.
26346 * gimple-fold.h (get_range_strlen): Change return type to bool.
26347 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
26348 * tree.h (array_at_struct_end_p): Add argument.
26349 * tree.c (array_at_struct_end_p): Handle it.
26351 2017-02-03 Martin Liska <mliska@suse.cz>
26354 * multiple_target.c (create_dispatcher_calls): Redirect edge
26355 from a caller of a dispatcher.
26356 (expand_target_clones): Make the clones local.
26357 (ipa_target_clone): Do both target clones and resolvers.
26358 (ipa_dispatcher_calls): Remove the pass.
26359 (pass_dispatcher_calls::gate): Likewise.
26360 (make_pass_dispatcher_calls): Likewise.
26361 * passes.def (pass_target_clone): Put as very first IPA early
26364 2017-02-03 Martin Liska <mliska@suse.cz>
26366 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
26367 in case of a function with ifunc attribute.
26369 2017-02-03 Martin Liska <mliska@suse.cz>
26371 * cgraph.c (cgraph_node::dump): Dump function version info.
26372 * symtab.c (symtab_node::dump_base): Add missing new line.
26374 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
26376 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
26377 (ifcombine_ifandif): Use it.
26379 2017-02-03 Martin Liska <mliska@suse.cz>
26381 * doc/invoke.texi: Document default value for
26382 use-after-scope-direct-emission-threshold.
26384 2017-02-03 Martin Liska <mliska@suse.cz>
26386 PR tree-optimization/79339
26387 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
26388 (format_floating): Likewise.
26390 2017-02-03 Martin Liska <mliska@suse.cz>
26393 * ipa-prop.c (ipa_node_params_t::insert): Remove current
26395 (ipa_node_params_t::remove): Likewise.
26396 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
26397 initialization from removed ipa_node_params_t::insert.
26398 (ipa_node_params::~ipa_node_params): Move from removed
26399 ipa_node_params_t::release.
26400 * symbol-summary.h (symbol_summary::m_released): New member.
26401 Do not release a summary twice. Do not allow to call finalizer
26402 for types of a summary that live in GGC memory.
26404 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26406 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
26409 2017-02-02 Martin Sebor <msebor@redhat.com>
26411 PR middle-end/79275
26412 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
26413 (format_string): Tighten up the range of output for non-constant
26414 strings and correct the expected range for wide non-constant strings.
26416 2017-02-02 Martin Sebor <msebor@redhat.com>
26418 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
26420 PR middle-end/32003
26421 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
26423 (-fdump-tree-@var): Add to index and document how to come up
26424 with pass-specific option and dump file names.
26425 (-fdump-passes): Clarify where to look for output.
26427 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
26429 PR middle-end/77445
26430 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
26431 statistics of the analyzed path; allow threading for speed when
26432 any of BBs along the path are optimized for speed.
26434 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
26436 PR middle-end/78468
26437 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
26438 settings of the virtual registers.
26441 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
26443 * explow.c (get_dynamic_stack_size): Take known alignment of stack
26444 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
26447 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26449 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
26450 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
26452 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26454 * config/s390/s390.md: Add missing comments with the expanded
26456 * config/s390/vector.md: Likewise.
26457 * config/s390/vx-builtins.md: Likewise.
26459 2017-02-02 Jakub Jelinek <jakub@redhat.com>
26462 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
26463 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
26464 conditions on a single line.
26466 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26468 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
26469 __S390_VX__ to __VX__.
26471 2017-02-01 Andrew Pinski <apinski@cavium.com>
26473 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
26474 stmt_info to record_stmt_cost.
26475 (vect_get_known_peeling_cost): Pass stmt_info if known to
26477 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
26478 cpu_vector_cost field into
26479 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
26480 field into vec_int_stmt_cost and vec_fp_stmt_cost.
26481 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
26482 splitting of scalar_stmt_cost and vec_stmt_cost.
26483 (thunderx_vector_cost): Likewise.
26484 (cortexa57_vector_cost): LIkewise.
26485 (exynosm1_vector_cost): Likewise.
26486 (xgene1_vector_cost): Likewise.
26487 (thunderx2t99_vector_cost): Improve after the splitting of the two
26489 (aarch64_builtin_vectorization_cost): Update for the splitting of
26490 scalar_stmt_cost and vec_stmt_cost.
26492 2017-02-01 Torvald Riegel <triegel@redhat.com>
26493 Richard Henderson <rth@redhat.com>
26495 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
26496 conditional on existance of a fast atomic load.
26497 * optabs-query.c (can_atomic_load_p): New function.
26498 * optabs-query.h (can_atomic_load_p): Declare it.
26499 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
26500 no fast atomic load is available for the particular size of access.
26501 (expand_atomic_compare_and_swap): Likewise.
26502 (expand_atomic_load): Likewise.
26503 (expand_atomic_store): Likewise.
26504 (expand_atomic_fetch_op): Likewise.
26505 * testsuite/lib/target-supports.exp
26506 (check_effective_target_sync_int_128): Remove x86 because it provides
26507 no fast atomic load.
26508 (check_effective_target_sync_int_128_runtime): Likewise.
26510 2017-02-01 Richard Biener <rguenther@suse.de>
26512 * graphite.c: Include tree-vectorizer.h for find_loop_location.
26513 (graphite_transform_loops): Provide opt-info for optimized nests.
26514 * tree-parloop.c (parallelize_loops): Provide opt-info for
26515 parallelized loops.
26517 2017-02-01 Richard Biener <rguenther@suse.de>
26519 PR middle-end/79315
26520 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
26521 was not set before.
26523 2017-02-01 Richard Biener <rguenther@suse.de>
26525 PR tree-optimization/71824
26526 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
26527 Verify the loops are valid in the merged SESE region.
26528 (scop_detection::can_represent_loop_1): Check analyzing the
26529 evolution of the number of iterations in the region succeeds.
26531 2017-01-31 Ian Lance Taylor <iant@golang.org>
26533 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
26534 REG_ARGS_SIZE note to 32-bit push insns and call insn.
26536 2017-01-31 David Malcolm <dmalcolm@redhat.com>
26538 PR preprocessor/79210
26539 * input.c (get_substring_ranges_for_loc): Replace line_width
26540 assertion with error-handling.
26542 2017-01-31 Richard Biener <rguenther@suse.de>
26544 PR tree-optimization/77318
26545 * graphite-sese-to-poly.c (extract_affine): Fix assert.
26546 (create_pw_aff_from_tree): Take loop parameter.
26547 (add_condition_to_pbb): Pass loop of the condition to
26548 create_pw_aff_from_tree.
26550 2017-01-31 Jakub Jelinek <jakub@redhat.com>
26552 * config/s390/s390.c (s390_asan_shadow_offset): New function.
26553 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
26555 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
26559 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
26561 (convert_int_to_float128): Likewise.
26562 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
26563 (convert_int_to_float128): Likewise.
26564 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
26565 (UNSPEC_IEEE128_CONVERT): Likewise.
26566 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
26567 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
26568 Use local variables for IBM extended format.
26569 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
26570 (fix_trunc<mode>si2_fprs): Likewise.
26571 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
26572 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
26573 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
26574 to know that we can now have integers of all sizes in vector
26576 (fix<uns>_<mode>di2_hw): Likewise.
26577 (float<uns>_<mode>si2_hw): Likewise.
26578 (fix_<mode>si2_hw): Likewise.
26579 (fixuns_<mode>si2_hw): Likewise.
26580 (float<uns>_<mode>di2_hw): Likewise.
26581 (float_<mode>di2_hw): Likewise.
26582 (float_<mode>si2_hw): Likewise.
26583 (floatuns_<mode>di2_hw): Likewise.
26584 (floatuns_<mode>si2_hw): Likewise.
26585 (xscvqp<su>wz_<mode>): Delete, no longer used.
26586 (xscvqp<su>dz_<mode>): Likewise.
26587 (xscv<su>dqp_<mode>): Likewise.
26588 (ieee128_mfvsrd_64bit): Likewise.
26589 (ieee128_mfvsrd_32bit): Likewise.
26590 (ieee128_mfvsrwz): Likewise.
26591 (ieee128_mtvsrw): Likewise.
26592 (ieee128_mtvsrd_64bit): Likewise.
26593 (ieee128_mtvsrd_32bit): Likewise.
26595 2017-01-31 Martin Liska <mliska@suse.cz>
26598 * ipa-prop.c (ipa_free_all_node_params): Call release method
26599 instead of ~sumbol_summary to not to trigger double times
26602 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
26604 PR tree-optimization/71691
26605 * bitmap.h (class auto_bitmap): New.
26606 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
26607 is_maybe_undefined instead of ssa_undefined_value_p.
26609 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26611 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
26612 __S390_ARCH_LEVEL__ to __ARCH__.
26614 2017-01-31 Jakub Jelinek <jakub@redhat.com>
26616 PR tree-optimization/79267
26617 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
26618 if should_remove_lhs_p is true.
26620 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
26623 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
26624 (add_alignment_attribute): New.
26625 (base_type_die): Add alignment attribute.
26626 (subrange_type_die): Likewise.
26627 (modified_type_die): Likewise.
26628 (gen_array_type_die): Likewise.
26629 (gen_descr_array_type_die: Likewise.
26630 (gen_enumeration_type_die): Likewise.
26631 (gen_subprogram_die): Likewise.
26632 (gen_variable_die): Likewise.
26633 (gen_field_die): Likewise.
26634 (gen_ptr_to_mbr_type_die): Likewise.
26635 (gen_struct_or_union_type_die): Likewise.
26636 (gen_subroutine_type_die): Likewise.
26637 (gen_typedef_die): Likewise.
26638 (base_type_cmp): Compare alignment attribute.
26640 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26643 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
26644 (setb_unsigned) New pattern for setb with CCUNS.
26645 * config/rs6000/rs6000.c (expand_block_compare): Use a different
26646 subfc./subfe sequence to avoid overflow problems. Generate a
26647 shorter sequence with cmpld/setb for power9.
26648 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
26649 for generating subfc. instruction.
26650 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
26651 now uses this instruction.
26653 2017-01-30 Ian Lance Taylor <iant@google.com>
26656 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
26657 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
26659 2017-01-30 Martin Sebor <msebor@redhat.com>
26661 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
26662 Move constant to the right of a relational operator.
26663 (get_mpfr_format_length, format_character, format_string): Ditto.
26664 (should_warn_p, maybe_warn): Same.
26666 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
26668 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
26671 * asan.c (get_translation_unit_decl): Remove function.
26672 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
26674 2017-01-30 Martin Liska <mliska@suse.cz>
26676 PR gcov-profile/79259
26677 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
26678 -fprofile-generate.
26680 2017-01-30 Martin Liska <mliska@suse.cz>
26683 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
26684 Initialize variables with NULL value.
26686 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
26689 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
26691 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
26693 2017-01-30 Richard Biener <rguenther@suse.de>
26695 PR tree-optimization/79276
26696 * tree-vrp.c (process_assert_insertions): Properly adjust common
26697 when removing a duplicate.
26699 * gcc.dg/torture/pr79276.c: New testcase.
26701 2017-01-30 Richard Biener <rguenther@suse.de>
26703 PR tree-optimization/79256
26704 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
26705 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
26707 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
26709 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
26712 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
26713 ("*r<noxa>sbg_<mode>_sll_bitmask")
26714 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
26715 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
26716 Use contiguous_bitmask_nowrap_operand.
26718 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26721 * config/rs6000/altivec.h (vec_xl): Revise #define.
26722 (vec_xst): Likewise.
26724 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
26726 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
26728 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
26730 PR rtl-optimization/79194
26731 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
26732 traps before call to bypass_conditional_jumps.
26734 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
26736 PR tree-optimization/71374
26737 * lra-constraints.c (check_conflict_input_operands): New.
26738 (match_reload): Use it.
26740 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
26743 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
26744 account to calculate conflict_set.
26746 2017-01-27 Bin Cheng <bin.cheng@arm.com>
26748 PR rtl-optimization/78559
26749 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
26750 other_insn in combine.
26752 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26754 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
26755 uint16_type_node for BT_UINT16.
26757 2017-01-27 David Malcolm <dmalcolm@redhat.com>
26759 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
26760 "RTL Tests" to menu.
26761 (GIMPLE Tests): New node.
26762 (RTL Tests): New node.
26764 2017-01-27 Richard Biener <rguenther@suse.de>
26766 PR tree-optimization/79245
26767 * tree-loop-distribution.c (distribute_loop): Apply cost
26768 modeling also to detected patterns.
26770 2017-01-27 Richard Biener <rguenther@suse.de>
26772 PR tree-optimization/71433
26773 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
26774 (compare_assert_loc): New function.
26775 (process_assert_insertions): Sort and optimize assert locations
26776 to remove duplicates and push down identical assertions on
26777 edges to their destination block.
26779 2017-01-27 Richard Biener <rguenther@suse.de>
26781 PR tree-optimization/79244
26782 * tree-vrp.c (remove_range_assertions): Forcefully propagate
26783 out SSA names even if abnormal.
26785 2017-01-27 Jakub Jelinek <jakub@redhat.com>
26787 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
26788 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
26789 instead of MPFR_RNDN.
26791 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
26794 * arm.c (arm_option_override): Don't call build_target_option_node
26795 until after doing all option overrides.
26796 (arm_valid_target_attribute_tree): Likewise.
26798 2017-01-27 Martin Liska <mliska@suse.cz>
26800 * doc/invoke.texi (-fprofile-arcs): Document profiling support
26801 for {cd}tors and C++ {cd}tors.
26803 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
26805 * config/s390/s390.md ("*setmem_long_and")
26806 ("*setmem_long_and_31z"): Use zero_extend instead of and.
26808 2017-01-26 Martin Sebor <msebor@redhat.com>
26810 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
26813 2017-01-26 Martin Sebor <msebor@redhat.com>
26815 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
26816 HAVE_DFmode before using XFmode or DFmode.
26817 (parse_directive): Avoid using the z length modifier to avoid
26818 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
26820 PR middle-end/78703
26821 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
26822 to accept adjustment as an array.
26823 (get_int_range): New function.
26824 (struct directive): Make width and prec arrays.
26825 (directive::set_width, directive::set_precision): Call get_int_range.
26826 (format_integer, format_floating): Handle width and precision ranges.
26827 (format_string, parse_directive): Same.
26829 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26832 * dwarf2out.c (generate_skeleton_bottom_up): For children with
26833 comdat_type_p set, just clone them, but keep the children in the
26837 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
26838 which have direct callers with -fvar-tracking-assignments enabled
26840 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
26841 inside of type units.
26843 2017-01-26 Martin Sebor <msebor@redhat.com>
26845 PR middle-end/78703
26846 * gimple-ssa-sprintf.c (struct result_range): Add likely and
26848 (struct format_result): Replace number_chars, number_chars_min,
26849 and number_chars_max with a single member of struct result_range.
26851 (format_result::operator+=): Adjust.
26852 (struct fmtresult): Remove bounded. Handle likely and unlikely
26854 (fmtresult::adjust_for_width_or_precision): New function.
26855 (fmtresult:type_max_digits): New function.
26856 (bytes_remaining): Handle likely and unlikely counters.
26857 (min_bytes_remaining): Remove.
26858 (format_percent): Simplify.
26859 (format_integer, format_floating): Set likely and unlikely counters.
26860 (get_string_length, format_character, format_string): Same.
26861 (format_plain, should_warn_p): New function.
26862 (maybe_warn): Call should_warn_p. Update diagnostic messages
26863 and handle those for all directives, including plain strings.
26864 (format_directive): Handle likely and unlikely counters.
26865 Remove unnecessary quoting from diagnostics. Add an informational
26867 (add_bytes): Remove.
26868 (pass_sprintf_length::compute_format_length): Simplify.
26869 (try_substitute_return_value): Handle likely and unlikely counters.
26871 2017-01-26 Carl Love <cel@us.ibm.com>
26873 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
26874 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
26876 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
26879 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
26880 endianess for subregs into account.
26881 * lra-constraints.c (lra_constraints): Do risky transformations
26882 always on the first iteration.
26883 * lra-lives.c (check_pseudos_live_through_calls): Add arg
26884 last_call_used_reg_set.
26885 (process_bb_lives): Define and use last_call_used_reg_set.
26886 * lra.c (lra): Always continue after lra_constraints on the first
26889 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
26891 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
26893 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
26895 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26897 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
26898 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
26899 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
26900 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
26901 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
26902 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
26903 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
26904 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
26905 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
26907 2017-01-26 Marek Polacek <polacek@redhat.com>
26910 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
26911 for the third operand.
26913 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26915 PR middle-end/79236
26916 * omp-low.c (struct omp_context): Add simt_stmt field.
26917 (scan_omp_for): Return omp_context *.
26918 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
26919 context to the _simt_ SIMD stmt.
26920 (lower_omp_for): For combined SIMD with sibling _simt_
26921 SIMD, make sure to use the same decls in _looptemp_
26922 clauses as in the sibling.
26924 2017-01-26 David Sherwood <david.sherwood@arm.com>
26926 PR middle-end/79212
26927 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
26930 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26933 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
26934 emit fld b; fld a; if possible.
26936 * brig-builtins.def: Update copyright years.
26937 * config/arm/arm_acle_builtins.def: Update copyright years.
26939 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
26942 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
26943 constraint instead of o for the stxsd instruction.
26945 2017-01-25 Carl Love <cel@us.ibm.com>
26947 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
26948 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
26950 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
26952 * doc/invoke.texi (C++ Dialect Options): Fix typo.
26954 2017-01-25 Richard Biener <rguenther@suse.de>
26956 PR tree-optimization/69264
26957 * target.def (vector_alignment_reachable): Improve documentation.
26958 * doc/tm.texi: Regenerate.
26959 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
26961 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
26962 earlier changes with respect to TYPE_USER_ALIGN.
26963 (vector_alignment_reachable_p): Likewise. Improve dumping.
26965 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26968 * config/arm/arm.md (xordi3): Force constant operand into a register
26971 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26973 * doc/invoke.texi (-fstore-merging): Correct default optimization
26974 levels at which it is enabled.
26975 (-O): Move -fstore-merging from list to...
26978 2017-01-25 Richard Biener <rguenther@suse.de>
26981 * omp-expand.c: Include debug.h.
26982 (expand_omp_taskreg): Make sure to generate early debug before
26983 outlining anything from a function.
26984 (expand_omp_target): Likewise.
26985 (grid_expand_target_grid_body): Likewise.
26987 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
26990 * asan.c (get_translation_unit_decl): New function.
26991 (asan_add_global): Extract modules file name from globals
26992 TRANSLATION_UNIT_DECL name.
26994 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
26997 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
26998 for long calls with APCS frame and VFP.
27000 2017-01-24 David Malcolm <dmalcolm@redhat.com>
27002 * cfg.c (original_copy_tables_initialized_p): New function.
27003 * cfg.h (original_copy_tables_initialized_p): New decl.
27004 * cfgrtl.c (relink_block_chain): Guard the call to
27005 free_original_copy_tables with a call to
27006 original_copy_tables_initialized_p.
27007 * cgraph.h (symtab_node::native_rtl_p): New decl.
27008 * cgraphunit.c (symtab_node::native_rtl_p): New function.
27009 (symtab_node::needed_p): Don't assert for early assembly output
27010 for __RTL functions.
27011 (cgraph_node::finalize_function): Set "force_output" for __RTL
27013 (cgraph_node::analyze): Bail out early for __RTL functions.
27014 (analyze_functions): Update assertion to support __RTL functions.
27015 (cgraph_node::expand): Bail out early for __RTL functions.
27016 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
27018 * function.h (struct function): Update comment for field
27020 * gimple-expr.c: Include "tree-pass.h".
27021 (gimple_has_body_p): Return false for __RTL functions.
27022 * Makefile.in (OBJS): Add run-rtl-passes.o.
27023 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
27025 (gcc::pass_manager::get_clean_slate): New accessor.
27026 * passes.c: Include "insn-addr.h".
27027 (should_skip_pass_p): Add logging. Update logic for running
27028 "expand" to be compatible with both __GIMPLE and __RTL. Guard
27029 property-provider override so it is only done for gimple passes.
27031 (skip_pass): New function.
27032 (execute_one_pass): Call skip_pass when skipping passes.
27033 * read-md.c (md_reader::read_char): Support filtering
27034 the input to a subset of line numbers.
27035 (md_reader::md_reader): Initialize fields
27036 m_first_line and m_last_line.
27037 (md_reader::read_file_fragment): New function.
27038 * read-md.h (md_reader::read_file_fragment): New decl.
27039 (md_reader::m_first_line): New field.
27040 (md_reader::m_last_line): New field.
27041 * read-rtl-function.c (function_reader::create_function): Only
27042 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
27043 curr_properties. Set DECL_INITIAL to a dummy block.
27044 (read_rtl_function_body_from_file_range): New function.
27045 * read-rtl-function.h (read_rtl_function_body_from_file_range):
27047 * run-rtl-passes.c: New file.
27048 * run-rtl-passes.h: New file.
27050 2017-01-24 Jeff Law <law@redhat.com>
27052 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
27055 2017-01-24 Bin Cheng <bin.cheng@arm.com>
27057 PR tree-optimization/79159
27058 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
27059 (record_nonwrapping_iv): Improve boundary using above function if no
27060 value range information.
27062 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
27063 Martin Jambor <mjambor@suse.cz>
27065 * brig-builtins.def: New file.
27066 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
27067 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
27068 (DEF_HSAIL_SAT_BUILTIN): Likewise.
27069 (DEF_HSAIL_INTR_BUILTIN): Likewise.
27070 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
27071 * builtin-types.def (BT_INT8): New.
27072 (BT_INT16): Likewise.
27073 (BT_UINT8): Likewise.
27074 (BT_UINT16): Likewise.
27075 (BT_FN_ULONG): Likewise.
27076 (BT_FN_UINT_INT): Likewise.
27077 (BT_FN_UINT_ULONG): Likewise.
27078 (BT_FN_UINT_LONG): Likewise.
27079 (BT_FN_UINT_PTR): Likewise.
27080 (BT_FN_ULONG_PTR): Likewise.
27081 (BT_FN_INT8_FLOAT): Likewise.
27082 (BT_FN_INT16_FLOAT): Likewise.
27083 (BT_FN_UINT32_FLOAT): Likewise.
27084 (BT_FN_UINT16_FLOAT): Likewise.
27085 (BT_FN_UINT8_FLOAT): Likewise.
27086 (BT_FN_UINT64_FLOAT): Likewise.
27087 (BT_FN_UINT16_UINT32): Likewise.
27088 (BT_FN_UINT32_UINT16): Likewise.
27089 (BT_FN_UINT16_UINT16_UINT16): Likewise.
27090 (BT_FN_INT_PTR_INT): Likewise.
27091 (BT_FN_UINT_PTR_UINT): Likewise.
27092 (BT_FN_LONG_PTR_LONG): Likewise.
27093 (BT_FN_ULONG_PTR_ULONG): Likewise.
27094 (BT_FN_VOID_UINT64_UINT64): Likewise.
27095 (BT_FN_UINT8_UINT8_UINT8): Likewise.
27096 (BT_FN_INT8_INT8_INT8): Likewise.
27097 (BT_FN_INT16_INT16_INT16): Likewise.
27098 (BT_FN_INT_INT_INT): Likewise.
27099 (BT_FN_UINT_FLOAT_UINT): Likewise.
27100 (BT_FN_FLOAT_UINT_UINT): Likewise.
27101 (BT_FN_ULONG_UINT_UINT): Likewise.
27102 (BT_FN_ULONG_UINT_PTR): Likewise.
27103 (BT_FN_ULONG_ULONG_ULONG): Likewise.
27104 (BT_FN_UINT_UINT_UINT): Likewise.
27105 (BT_FN_VOID_UINT_PTR): Likewise.
27106 (BT_FN_UINT_UINT_PTR: Likewise.
27107 (BT_FN_UINT32_UINT64_PTR): Likewise.
27108 (BT_FN_INT_INT_UINT_UINT): Likewise.
27109 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
27110 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
27111 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
27112 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
27113 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
27114 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
27115 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
27116 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
27117 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
27118 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
27119 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
27120 * doc/frontends.texi: List BRIG FE.
27121 * doc/install.texi (Testing): Add BRIG tesring requirements.
27122 * doc/invoke.texi (Overall Options): Mention BRIG.
27123 * doc/standards.texi (Standards): Doucment BRIG HSA version.
27125 2017-01-24 Richard Biener <rguenther@suse.de>
27127 PR translation/79208
27128 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
27130 2017-01-24 Martin Jambor <mjambor@suse.cz>
27133 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
27134 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
27135 and known_contexts.
27137 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
27139 PR middle-end/79123
27140 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
27141 casts from signed to unsigned really don't have a range.
27143 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
27145 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
27146 GMP_RNDx for compatiblity.
27148 2017-01-24 Martin Liska <mliska@suse.cz>
27151 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
27152 that would prevent us to call alloca with -1 as argument.
27154 2017-01-24 Jakub Jelinek <jakub@redhat.com>
27156 * dwarf2out.c (output_compilation_unit_header, output_file_names):
27157 Avoid -Wformat-security warning.
27159 2017-01-23 Andrew Pinski <apinski@cavium.com>
27161 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
27164 2017-01-23 Martin Sebor <msebor@redhat.com>
27166 PR middle-end/78703
27167 * gimple-ssa-sprintf.c (warn_level): New global.
27168 (format_integer): Use it here and throughout the rest of the file.
27169 Use the same switch to compute sign as base.
27170 (maybe_warn): New function.
27171 (format_directive): Factor out warnings into maybe_warn.
27172 Add debugging output. Use warn_level.
27173 (add_bytes): Use warn_level.
27174 (pass_sprintf_length::compute_format_length): Add debugging output.
27175 (try_substitute_return_value): Same.
27176 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
27178 PR middle-end/78703
27179 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
27180 (struct fmtresult, format_integer, format_floating): Adjust.
27181 (fmtresult::fmtresult): Set max correctly in two argument ctor.
27182 (get_string_length, format_string,format_directive): Same.
27183 (pass_sprintf_length::compute_format_length): Same.
27184 (try_substitute_return_value): Simplify slightly.
27186 PR middle-end/78703
27187 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
27188 (fmtresult::operator+=): Outlined.
27189 (struct fmtresult): Add ctors.
27190 (struct conversion_spec): Rename...
27191 (struct directive): ...to this. Add and remove data members.
27192 (directive::set_width, directive::set_precision): New functions.
27193 (format_percent): Use fmtresult ctor.
27194 (get_width_and_precision): Remove.
27195 (format_integer): Make naming changes. Avoid computing width and
27197 (format_floating): Same. Adjust indentation.
27198 (format_character, format_none): New functions.
27199 (format_string): Moved character handling to format_character.
27200 (format_directive): Remove arguments, change return type.
27201 (parse_directive): New function.
27202 (pass_sprintf_length::compute_format_length): Move directive
27203 parsing to parse_directive.
27205 2017-01-23 Jakub Jelinek <jakub@redhat.com>
27207 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
27208 (assign_assembler_name_if_needed): ... this.
27209 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
27210 (assign_assembler_name_if_needed): ... this.
27211 (free_lang_data_in_cgraph): Adjust callers.
27212 * cgraphunit.c (cgraph_node::analyze): Likewise.
27213 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
27216 2017-01-23 Richard Biener <rguenther@suse.de>
27218 PR tree-optimization/79088
27219 PR tree-optimization/79188
27220 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
27221 resetting loop bounds after last path deletion. Reset loop
27222 bounds of the target loop, make code match the comments.
27223 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
27224 Make sure loops need no fixups.
27226 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
27228 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
27229 exponent support with double type for first argument.
27230 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
27231 type returned by __builtin_vec_extract_sig,
27232 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
27233 functions from "vector int" to "vector unsigned int" or from
27234 "vector long long int" to "vector unsigned long long int".
27235 Changed type returned by __builtin_vec_extract_exp,
27236 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
27237 functions from "vector int" to "vector unsigned int" or from
27238 "vector long long int" to "vector unsigned long long int".
27239 Changed return type of __builtin_vec_test_data_class,
27240 __builtin_vec_test_data_class_sp, and
27241 __builtin_vec_test_data_class_dp from "vector int" to
27242 "vector bool int" or from "vector long long int" to "vector bool
27243 long long int" and changed second argument type from "unsigned
27244 int" to "int". Added new overloaded function forms "vector float
27245 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
27246 "vector float __builtin_vec_insert_exp_sp (vector float, vector
27247 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
27248 double, vector unsigned long long int)" and "vector double
27249 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
27250 long int)". Changed return type of
27251 __builtin_scalar_test_data_class and
27252 __builtin_scalar_test_data_class_sp and
27253 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
27254 int" and changed second argument from "unsigned int" to "int".
27255 Changed type returned by __builtin_scalar_test_neg,
27256 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
27257 from "int" to "bool int". Added new overloaded function form
27258 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
27259 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
27260 exponent double-precision with floating point first argument.
27261 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
27262 documentation of scalar_test_data_class, scalar_test_neg,
27263 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
27264 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
27265 vec_test_data_class built-in functions to reflect refinements in
27266 their type signatures.
27268 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
27270 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
27272 (aarch64_elf_asm_destructor): Likewise.
27274 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
27276 PR rtl-optimization/78634
27277 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
27278 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
27279 * ifcvt.c (noce_try_cmove): Add missing cost check.
27281 PR rtl-optimization/71724
27282 * combine.c (if_then_else_cond): Look for situations where it is
27283 beneficial to undo the work of one of the recursive calls.
27285 2017-01-23 Bin Cheng <bin.cheng@arm.com>
27287 PR tree-optimization/70754
27288 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
27289 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
27290 combined stmt before it if not NULL.
27291 (combine_chains): Process refs reversely and compute dominance point
27294 2017-01-23 Martin Liska <mliska@suse.cz>
27296 PR tree-optimization/79196
27297 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
27298 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
27300 (strlen_optimize_stmt): Call the renamed function.
27302 2017-01-23 Michael Matz <matz@suse.de>
27304 PR tree-optimization/78384
27305 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
27307 2017-01-23 Richard Biener <rguenther@suse.de>
27309 PR tree-optimization/79186
27310 * tree-vrp.c (register_new_assert_for): Make sure we've seen
27311 both incoming edges before moving an assert.
27313 2017-01-23 Martin Jambor <mjambor@suse.cz>
27315 * ipa-prop.c (load_from_param_1): Removed.
27316 (load_from_unmodified_param): Bits from load_from_param_1 put back
27318 (load_from_param): Removed.
27319 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
27320 with stmt. Reverted back to use of load_from_unmodified_param.
27322 2017-01-23 Martin Jambor <mjambor@suse.cz>
27325 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
27326 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
27327 field a pointer to garbage collected vector, mark lattices and
27328 ipcp_orig_node with GTY((skip)).
27329 (ipa_get_param_count): Adjust to descriptors being a pointer.
27330 (ipa_get_param): Likewise.
27331 (ipa_get_type): Likewise.
27332 (ipa_get_param_move_cost): Likewise.
27333 (ipa_set_param_used): Likewise.
27334 (ipa_get_controlled_uses): Likewise.
27335 (ipa_set_controlled_uses): Likewise.
27336 (ipa_is_param_used): Likewise.
27337 (ipa_node_params_t): Move into garbage collector. New methods insert
27339 (ipa_node_params_sum): Annotate wth GTY(()).
27340 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
27342 (ipa_load_from_parm_agg): Adjust declaration.
27343 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
27344 * ipa-profile.c (ipa_profile): Likewise.
27345 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
27346 (ipa_populate_param_decls): Make descriptors parameter garbage
27348 (ipa_dump_param): Adjust to descriptors being a pointer.
27349 (ipa_alloc_node_params): Likewise.
27350 (ipa_initialize_node_params): Likewise.
27351 (load_from_param_1): Make descriptors parameter garbage collected.
27352 (load_from_unmodified_param): Likewise.
27353 (load_from_param): Likewise.
27354 (ipa_load_from_parm_agg): Likewise.
27355 (ipa_node_params::~ipa_node_params): Removed.
27356 (ipa_free_all_node_params): Remove call to delete operator.
27357 (ipa_node_params_t::insert): New.
27358 (ipa_node_params_t::remove): Likewise.
27359 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
27360 copy known_csts and known_contexts vectors.
27361 (ipa_read_node_info): Adjust to descriptors being a pointer.
27362 (ipcp_modif_dom_walker): Make m_descriptors field garbage
27364 (ipcp_transform_function): Make descriptors variable garbage
27367 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
27369 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
27370 * config/i386/avx512dqintrin.h: Ditto.
27371 * config/i386/avx512fintrin.h: Ditto.
27372 * config/i386/i386.c: Handle new builtins.
27373 * config/i386/i386-builtin.def: Add new builtins.
27374 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
27375 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
27377 2017-01-23 Jakub Jelinek <jakub@redhat.com>
27378 Martin Liska <mliska@suse.cz>
27380 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
27381 * asan.c (asan_expand_poison_ifn): Support stores and use
27382 appropriate ASAN report function.
27383 * internal-fn.c (expand_ASAN_POISON_USE): New function.
27384 * internal-fn.def (ASAN_POISON_USE): Declare.
27385 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
27386 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
27387 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
27388 ASAN_POISON calls w/o LHS.
27389 * tree-ssa.c (execute_update_addresses_taken): Create clobber
27390 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
27391 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
27392 * gimplify.c (asan_poison_variables): Add attribute
27393 use_after_scope_memory to variables that really needs to live
27395 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
27396 having the attribute.
27398 2017-01-23 Martin Liska <mliska@suse.cz>
27400 * asan.c (create_asan_shadow_var): New function.
27401 (asan_expand_poison_ifn): Likewise.
27402 * asan.h (asan_expand_poison_ifn): New declaration.
27403 * internal-fn.c (expand_ASAN_POISON): Likewise.
27404 * internal-fn.def (ASAN_POISON): New builtin.
27405 * sanopt.c (pass_sanopt::execute): Expand
27406 asan_expand_poison_ifn.
27407 * tree-inline.c (copy_decl_for_dup_finish): Make function
27409 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
27410 * tree-ssa.c (is_asan_mark_p): New function.
27411 (execute_update_addresses_taken): Rewrite local variables
27412 (identified just by use-after-scope as addressable) into SSA.
27414 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
27416 * doc/install.texi (Specific): opensource.apple.com uses https
27417 now. Remove trailing slash.
27419 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
27421 * README.Portability: Remove note on an Irix compatibility issue.
27423 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
27425 * gcov.c (INCLUDE_ALGORITHM): Define.
27426 (INCLUDE_VECTOR): Define.
27427 No longer include <vector> and <algorithm> directly.
27429 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
27431 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
27433 * doc/invoke.texi (Code Gen Options): Ditto.
27435 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
27438 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
27440 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
27442 rtl-optimization/79125
27443 * cprop.c (local_cprop_pass): Handle cases where we make an
27444 unconditional trap.
27446 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
27450 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
27451 read from, for big endian.
27453 2017-01-20 Jiong Wang <jiong.wang@arm.com>
27455 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
27456 register pauth builtins for LP64 only.
27458 2017-01-20 Marek Polacek <polacek@redhat.com>
27461 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
27464 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
27466 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
27468 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
27469 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
27470 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
27472 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27475 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
27476 in big-endian mode when they are not a single duplicated value.
27478 2017-01-20 Richard Biener <rguenther@suse.de>
27480 * BASE-VER: Bump to 7.0.1.
27482 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
27484 * omp-low.c (omplow_simd_context): New struct. Use it...
27485 (lower_rec_simd_input_clauses): ...here and...
27486 (lower_rec_input_clauses): ...here to hold common data. Adjust all
27487 references to idx, lane, max_vf, is_simt.
27489 2017-01-20 Graham Markall <graham.markall@embecosm.com>
27491 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
27494 2017-01-20 Martin Jambor <mjambor@suse.cz>
27496 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
27497 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
27499 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
27500 (GTFILES): Rename hsa.c to hsa-common.c.
27501 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
27502 * hsa-dump.c: Likewise.
27503 * hsa-gen.c: Likewise.
27504 * hsa-regalloc.c: Likewise.
27505 * ipa-hsa.c: Likewise.
27506 * omp-expand.c: Likewise.
27507 * omp-low.c: Likewise.
27508 * toplev.c: Likewise.
27510 2017-01-20 Marek Polacek <polacek@redhat.com>
27513 * doc/invoke.texi: Document -Wduplicated-branches.
27514 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
27515 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
27516 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
27517 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
27518 return 0 only when not OEP_LEXICOGRAPHIC.
27519 (fold_build_cleanup_point_expr): Use the expression
27520 location when building CLEANUP_POINT_EXPR.
27521 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
27522 * tree.c (add_expr): Handle error_mark_node.
27524 2017-01-20 Martin Liska <mliska@suse.cz>
27527 * tree-profile.c (init_ic_make_global_vars): Do not call
27529 (gimple_init_gcov_profiler): Likewise.
27531 2017-01-20 Martin Liska <mliska@suse.cz>
27534 * cgraph.h (maybe_create_reference): Remove argument and
27536 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
27538 * ipa-cp.c (create_specialized_node): Likewise.
27539 * symtab.c (symtab_node::maybe_create_reference): Handle
27540 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
27542 2017-01-20 Martin Liska <mliska@suse.cz>
27544 * read-rtl-function.c (function_reader::create_function): Use
27545 build_decl instread of build_decl_stat.
27547 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
27549 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
27550 * config/i386/avx512dqintrin.h: Ditto.
27551 * config/i386/avx512fintrin.h: Ditto.
27552 * config/i386/i386-builtin-types.def: Add new types.
27553 * config/i386/i386.c: Handle new types.
27554 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
27555 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
27556 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
27557 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
27558 (__builtin_ia32_kshiftridi): New.
27559 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
27561 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
27565 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
27566 define to rs6000_init_stack_protect_guard.
27567 (rs6000_init_stack_protect_guard): New function.
27569 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
27570 Yunqiang Su <yunqiang.su@imgtec.com>
27572 * config.gcc (supported_defaults): Add madd4.
27573 (with_madd4): Add validation.
27574 (all_defaults): Add madd4.
27575 * config/mips/mips.opt (mmadd4): New option.
27576 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
27578 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
27580 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
27581 (ISA_HAS_FUSED_MADD4): Likewise.
27582 * doc/invoke.texi (-mmadd4): Document the new option.
27583 * doc/install.texi (--with-madd4): Document the new option.
27585 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27587 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
27588 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
27589 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
27590 (aarch64_init_pauth_hint_builtins): New.
27591 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
27592 (aarch64_expand_builtin): Expand new builtins.
27594 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27596 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
27597 * combine-stack-adj.c (no_unhandled_cfa): Handle
27598 REG_CFA_TOGGLE_RA_MANGLE.
27599 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
27600 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
27601 info for return address signing.
27602 (aarch64_expand_epilogue): Likewise.
27604 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27606 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
27607 * config/aarch64/aarch64-protos.h
27608 (aarch64_return_address_signing_enabled): New declaration.
27609 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
27611 (aarch64_expand_prologue): Sign return address before it's pushed onto
27613 (aarch64_expand_epilogue): Authenticate return address fetched from
27615 (aarch64_override_options): Sanity check for ILP32 and ISA level.
27616 (aarch64_attributes): New function attributes for "sign-return-address".
27617 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
27618 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
27619 ("*do_return"): Generate combined instructions according to key index.
27620 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
27621 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
27623 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
27624 * config/aarch64/aarch64.opt (msign-return-address=): New.
27625 * doc/extend.texi (AArch64 Function Attributes): Documents
27626 "sign-return-address=".
27627 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
27629 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
27631 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
27632 overall option summary.
27634 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27636 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
27637 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
27638 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
27639 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
27641 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
27643 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
27644 -mpower9-minmax by default for -mcpu=power9.
27645 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
27646 128-bit floating point.
27648 2017-01-20 Alan Modra <amodra@gmail.com>
27650 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
27651 optimizing for size.
27653 2017-01-20 Alan Modra <amodra@gmail.com>
27656 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
27657 for strcmp and strncmp from corresponding builtin decl.
27659 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
27661 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
27662 instead of i386/rtems-64.h.
27663 * config/i386/rtems-64.h: Remove.
27665 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
27669 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
27671 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
27673 2017-01-19 Tamar Christina <tamar.christina@arm.com>
27675 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
27676 Change int to HOST_WIDE_INT.
27677 * config/aarch64/aarch64-protos.h
27678 (aarch64_simd_gen_const_vector_dup): Likewise.
27679 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
27681 2017-01-19 David Malcolm <dmalcolm@redhat.com>
27683 * langhooks-def.h (lhd_type_for_size): New decl.
27684 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
27685 * langhooks.c (lhd_type_for_size): New function, taken from
27688 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
27690 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
27691 define_bypass for CR latency.
27692 (power9-cracked-alu): Update bypass latency and remove power9-branch.
27693 (power9-alu2): Add define_bypass for CR latency.
27695 (power9-mul): Update insn latency.
27696 (power9-mul-compare): Update insn latency, bypass latency and remove
27699 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27701 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
27703 * config/aarch64/aarch64.md
27704 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
27705 aarch64_nopcrelative_literal_loads.
27706 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
27708 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
27710 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
27711 TARGET_LOONGSON_3A.
27712 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
27714 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
27717 * config.gcc (supported_defaults): Add lxc1-sxc1.
27718 (with_lxc1_sxc1): Add validation.
27719 (all_defaults): Add lxc1-sxc1.
27720 * config/mips/mips.opt (mlxc1-sxc1): New option.
27721 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
27723 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
27724 __mips_no_lxc1_sxc1.
27725 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
27726 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
27727 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
27729 2017-01-19 Richard Biener <rguenther@suse.de>
27731 PR tree-optimization/72488
27732 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
27733 sure to restore SSA info.
27734 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
27736 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
27738 PR rtl-optimization/79121
27739 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
27740 of the inner type when shifting an extended value.
27742 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
27745 * symtab.c (symtab_node::equal_address_to): Fix comparing of
27746 interposable aliases.
27748 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
27751 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
27752 Use the evmergelohi instruction.
27753 (mov_si<mode>_e500_subreg4_2_le): Likewise.
27754 (mov_sitf_e500_subreg8_2_be): Likewise.
27755 (mov_sitf_e500_subreg12_2_le): Likewise.
27756 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
27757 (mov_si<mode>_e500_subreg4_2_be): Likewise.
27758 (mov_sitf_e500_subreg8_2_le): Likewise.
27759 (mov_sitf_e500_subreg12_2_be): Likewise.
27761 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27763 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
27764 attribute from vecsimple to vecperm.
27765 (altivec_vbpermq2): Likewise.
27767 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27770 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
27772 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27773 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
27774 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
27775 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
27776 case where N arg is SIZE_MAX.
27777 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
27778 (cmpstrsi): Add pattern.
27780 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27782 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27783 __builtin_vec_revb builtins.
27784 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
27785 built-in functions to support generation of the ISA 3.0 XXBR<x>
27786 vector byte reverse instructions.
27787 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
27788 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
27789 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
27790 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
27791 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
27792 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
27793 (P9V_BUILTIN_VEC_REVB): Likewise.
27794 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
27795 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
27796 (p9_xxbrq_v16qi): Likewise.
27797 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
27798 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
27799 (p9_xxbrh_v8hi): Likewise.
27800 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
27801 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
27802 vec_revb built-in functions.
27804 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
27806 PR rtl-optimization/78952
27807 * config/i386/i386.md (any_extract): New code iterator.
27808 (*insvqi_2): Use any_extract for source operand.
27809 (*insvqi_3): Use any_shiftrt for source operand.
27811 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
27813 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
27815 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
27817 2017-01-18 Matthias Klose <doko@ubuntu.com>
27819 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
27821 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27823 * config/rs6000/altivec.h (vec_bperm): Change #define.
27824 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
27825 (altivec_vbpermq2): New define_insn.
27826 (altivec_vbpermd): Likewise.
27827 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
27828 function interface.
27829 (VBPERMD): Likewise.
27830 (VBPERM): New polymorphic function interface.
27831 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
27832 Add entries for P9V_BUILTIN_VEC_VBPERM.
27833 * doc/extend.texi: Add interfaces for vec_bperm.
27835 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27837 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
27838 first letter of error messages.
27839 (s390_resolve_overloaded_builtin): Likewise.
27840 * config/s390/s390.c (s390_expand_builtin): Likewise.
27841 (s390_invalid_arg_for_unprototyped_fn): Likewise.
27842 (s390_valid_target_attribute_inner_p): Likewise.
27843 * config/s390/s390.md ("tabort"): Likewise.
27845 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
27847 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
27848 (ISA_AVOID_DIV_HILO): New macro.
27849 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
27850 (ISA_HAS_DDIV): Likewise.
27852 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27854 * doc/invoke.texi (fabi-version): Correct number of occurrences.
27856 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27858 * doc/invoke.texi (fabi-version): Spelling fix.
27860 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27863 * doc/invoke.texi (fabi-version): Mention mangling fix for
27866 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27869 * doc/invoke.texi (fabi-version): Document discriminator mangling.
27871 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
27874 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
27875 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
27877 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
27879 (stack_protect_test): Ditto.
27880 * config/rs6000/rs6000.opt (mstack-protector-guard=,
27881 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
27883 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
27884 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
27885 -mstack-protector-guard-offset=.
27886 (RS/6000 and PowerPC Options): Ditto.
27888 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
27890 * config/i386/i386.h (MASK_CLASS_P): New define.
27891 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
27892 there are no registers from different register sets also when
27893 mask registers are used. Update function comment.
27894 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
27895 to (*k/*r) and (*k/*km) alternatives.
27897 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
27899 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
27900 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
27901 (EH_RETURN_HANDLER_RTX): New define.
27902 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
27903 Force frame pointer in EH return functions.
27904 (aarch64_expand_epilogue): Add barrier for eh_return.
27905 (aarch64_final_eh_return_addr): Remove.
27906 (aarch64_eh_return_handler_rtx): New function.
27907 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
27909 (aarch64_eh_return_handler_rtx): New prototype.
27911 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27913 * config/rs6000/altivec.h (vec_rlmi): New #define.
27914 (vec_vrlnm): Likewise.
27915 (vec_rlnm): Likewise.
27916 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
27917 (UNSPEC_VRLNM): Likewise.
27918 (VIlong): New mode iterator.
27919 (altivec_vrl<VI_char>mi): New define_insn.
27920 (altivec_vrl<VI_char>nm): Likewise.
27921 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
27923 (VRLDNM): Likewise.
27924 (RLNM): New polymorphic function entry.
27925 (VRLWMI): New monomorphic function entry.
27926 (VRLDMI): Likewise.
27927 (RLMI): New polymorphic function entry.
27928 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
27929 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
27930 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
27933 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27936 * dwarf2out.c (field_byte_offset): Restore the
27937 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
27938 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
27939 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
27942 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
27945 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
27947 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27950 * dwarf2out.c (add_data_member_location_attribute): For constant
27951 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
27952 instead of DW_AT_data_member_location, DW_AT_bit_offset and
27953 DW_AT_byte_size attributes.
27955 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
27957 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
27958 after forcing to constant memory when the code model is medium.
27960 2017-01-17 Julia Koval <julia.koval@intel.com>
27963 * config/i386/avx512fintrin.h
27964 (_mm512_i32gather_ps): Change __addr type to void const*.
27965 (_mm512_mask_i32gather_ps): Ditto.
27966 (_mm512_i32gather_pd): Ditto.
27967 (_mm512_mask_i32gather_pd): Ditto.
27968 (_mm512_i64gather_ps): Ditto.
27969 (_mm512_mask_i64gather_ps): Ditto.
27970 (_mm512_i64gather_pd): Ditto.
27971 (_mm512_mask_i64gather_pd): Ditto.
27972 (_mm512_i32gather_epi32): Ditto.
27973 (_mm512_mask_i32gather_epi32): Ditto.
27974 (_mm512_i32gather_epi64): Ditto.
27975 (_mm512_mask_i32gather_epi64): Ditto.
27976 (_mm512_i64gather_epi32): Ditto.
27977 (_mm512_mask_i64gather_epi32): Ditto.
27978 (_mm512_i64gather_epi64): Ditto.
27979 (_mm512_mask_i64gather_epi64): Ditto.
27980 (_mm512_i32scatter_ps): Change __addr type to void*.
27981 (_mm512_mask_i32scatter_ps): Ditto.
27982 (_mm512_i32scatter_pd): Ditto.
27983 (_mm512_mask_i32scatter_pd): Ditto.
27984 (_mm512_i64scatter_ps): Ditto.
27985 (_mm512_mask_i64scatter_ps): Ditto.
27986 (_mm512_i64scatter_pd): Ditto.
27987 (_mm512_mask_i64scatter_pd): Ditto.
27988 (_mm512_i32scatter_epi32): Ditto.
27989 (_mm512_mask_i32scatter_epi32): Ditto.
27990 (_mm512_i32scatter_epi64): Ditto.
27991 (_mm512_mask_i32scatter_epi64): Ditto.
27992 (_mm512_i64scatter_epi32): Ditto.
27993 (_mm512_mask_i64scatter_epi32): Ditto.
27994 (_mm512_i64scatter_epi64): Ditto.
27995 (_mm512_mask_i64scatter_epi64): Ditto.
27996 * config/i386/avx512pfintrin.h
27997 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
27998 (_mm512_mask_prefetch_i32gather_ps): Ditto.
27999 (_mm512_mask_prefetch_i64gather_pd): Ditto.
28000 (_mm512_mask_prefetch_i64gather_ps): Ditto.
28001 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
28002 (_mm512_prefetch_i32scatter_ps): Ditto.
28003 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
28004 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
28005 (_mm512_prefetch_i64scatter_pd): Ditto.
28006 (_mm512_prefetch_i64scatter_ps): Ditto.
28007 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
28008 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
28009 * config/i386/avx512vlintrin.h
28010 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
28011 (_mm_mmask_i32gather_ps): Ditto.
28012 (_mm256_mmask_i32gather_pd): Ditto.
28013 (_mm_mmask_i32gather_pd): Ditto.
28014 (_mm256_mmask_i64gather_ps): Ditto.
28015 (_mm_mmask_i64gather_ps): Ditto.
28016 (_mm256_mmask_i64gather_pd): Ditto.
28017 (_mm_mmask_i64gather_pd): Ditto.
28018 (_mm256_mmask_i32gather_epi32): Ditto.
28019 (_mm_mmask_i32gather_epi32): Ditto.
28020 (_mm256_mmask_i32gather_epi64): Ditto.
28021 (_mm_mmask_i32gather_epi64): Ditto.
28022 (_mm256_mmask_i64gather_epi32): Ditto.
28023 (_mm_mmask_i64gather_epi32): Ditto.
28024 (_mm256_mmask_i64gather_epi64): Ditto.
28025 (_mm_mmask_i64gather_epi64): Ditto.
28026 (_mm256_i32scatter_ps): Change __addr type to void*.
28027 (_mm256_mask_i32scatter_ps): Ditto.
28028 (_mm_i32scatter_ps): Ditto.
28029 (_mm_mask_i32scatter_ps): Ditto.
28030 (_mm256_i32scatter_pd): Ditto.
28031 (_mm256_mask_i32scatter_pd): Ditto.
28032 (_mm_i32scatter_pd): Ditto.
28033 (_mm_mask_i32scatter_pd): Ditto.
28034 (_mm256_i64scatter_ps): Ditto.
28035 (_mm256_mask_i64scatter_ps): Ditto.
28036 (_mm_i64scatter_ps): Ditto.
28037 (_mm_mask_i64scatter_ps): Ditto.
28038 (_mm256_i64scatter_pd): Ditto.
28039 (_mm256_mask_i64scatter_pd): Ditto.
28040 (_mm_i64scatter_pd): Ditto.
28041 (_mm_mask_i64scatter_pd): Ditto.
28042 (_mm256_i32scatter_epi32): Ditto.
28043 (_mm256_mask_i32scatter_epi32): Ditto.
28044 (_mm_i32scatter_epi32): Ditto.
28045 (_mm_mask_i32scatter_epi32): Ditto.
28046 (_mm256_i32scatter_epi64): Ditto.
28047 (_mm256_mask_i32scatter_epi64): Ditto.
28048 (_mm_i32scatter_epi64): Ditto.
28049 (_mm_mask_i32scatter_epi64): Ditto.
28050 (_mm256_i64scatter_epi32): Ditto.
28051 (_mm256_mask_i64scatter_epi32): Ditto.
28052 (_mm_i64scatter_epi32): Ditto.
28053 (_mm_mask_i64scatter_epi32): Ditto.
28054 (_mm256_i64scatter_epi64): Ditto.
28055 (_mm256_mask_i64scatter_epi64): Ditto.
28056 (_mm_i64scatter_epi64): Ditto.
28057 (_mm_mask_i64scatter_epi64): Ditto.
28058 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
28059 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
28060 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
28061 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
28062 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
28063 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
28064 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
28065 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
28066 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
28067 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
28068 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
28069 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
28070 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
28071 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
28072 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
28073 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
28074 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
28075 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
28076 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
28077 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
28078 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
28079 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
28080 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
28081 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
28082 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
28083 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
28084 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
28085 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
28086 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
28087 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
28088 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
28089 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
28090 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
28091 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
28092 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
28093 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
28094 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
28095 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
28096 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
28097 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
28098 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
28099 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
28100 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
28101 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
28102 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
28103 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
28104 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
28105 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
28106 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
28107 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
28108 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
28109 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
28110 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
28111 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
28112 definitions accordingly.
28114 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
28115 Kuan-Lin Chen <kuanlinchentw@gmail.com>
28118 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
28121 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
28124 * ira-conflicts.c (ira_build_conflicts): Update total conflict
28125 hard regs for inner regno.
28127 2017-01-17 Martin Liska <mliska@suse.cz>
28130 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
28131 assumption and add comment.
28133 2017-01-17 Nathan Sidwell <nathan@acm.org>
28135 * ipa-visibility.c (localize_node): New function, broken out of ...
28136 (function_and_variable_visibility): ... here. Call it.
28138 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
28140 PR middle-end/77445
28141 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
28142 correctly set frequency of oudgoing edge.
28143 (duplicate_thread_path): Fix profile updating.
28145 2017-01-17 Jakub Jelinek <jakub@redhat.com>
28148 * configure.ac: Add GCC_BASE_VER.
28149 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
28150 version from BASE-VER file.
28151 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
28152 (gcc.o): Depend on $(BASEVER).
28153 * common.opt (dumpfullversion): New option.
28154 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
28155 * doc/invoke.texi: Document -dumpfullversion.
28156 * doc/install.texi: Document --with-gcc-major-version-only.
28157 * configure: Regenerated.
28159 2017-01-17 Richard Biener <rguenther@suse.de>
28161 PR tree-optimization/71433
28162 * tree-vrp.c (register_new_assert_for): Merge same asserts
28163 on all incoming edges.
28164 (process_assert_insertions_for): Handle insertions at the
28167 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
28169 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
28170 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
28172 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
28175 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
28178 2017-01-17 Alan Modra <amodra@gmail.com>
28181 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
28182 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
28183 symbolic stack limit when pic.
28185 2017-01-16 Martin Sebor <msebor@redhat.com>
28187 PR tree-optimization/78608
28188 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
28190 2017-01-16 Jeff Law <law@redhat.com>
28193 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
28194 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
28195 for several include directories that may be relative to sysroot.
28196 * config/i386/x-mingw32 (gplus_includedir): Define.
28197 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
28198 (native_system_includedir): Likewise.
28199 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
28200 override if TARGET_SYSTEM_ROOT is defined.
28201 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
28203 PR tree-optimization/79090
28204 PR tree-optimization/33562
28205 PR tree-optimization/61912
28206 PR tree-optimization/77485
28207 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
28208 and computed trims into the dump file.
28210 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
28212 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
28214 2017-01-16 Jakub Jelinek <jakub@redhat.com>
28217 * gimplify.c (gimplify_init_constructor): If want_value and
28218 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
28222 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
28223 sequence. Formatting fixes.
28224 (doloop_optimize): Formatting fixes.
28227 * gcc.c (debug_level_greater_than_spec_func): New function.
28228 (static_spec_functions): Add debug-level-gt spec function.
28229 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
28231 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
28232 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
28233 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
28234 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
28235 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
28236 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
28238 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
28240 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
28241 QImode fixups to general and mask registers only.
28243 2017-01-16 Carl Love <cel@us.ibm.com>
28245 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
28246 for built-in functions
28247 vector signed char vec_nabs (vector signed char)
28248 vector signed short vec_nabs (vector signed short)
28249 vector signed int vec_nabs (vector signed int)
28250 vector signed long long vec_nabs (vector signed long long)
28251 vector float vec_nabs (vector float)
28252 vector double vec_nabs (vector double)
28253 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
28255 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
28256 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
28257 * doc/extend.texi: Update the documentation file for the new built-in
28260 2017-01-16 Martin Sebor <msebor@redhat.com>
28262 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
28265 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28267 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
28268 UNSPEC_VSX__XXSPLTD to require special splat handling.
28270 2017-01-16 David Malcolm <dmalcolm@redhat.com>
28273 * system.h: Poison strndup.
28275 2017-01-16 Alan Modra <amodra@gmail.com>
28278 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
28281 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
28283 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
28285 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
28287 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
28288 call recog here. Assert that INSN_CODE (insn) is non-negative.
28290 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
28293 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
28295 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
28296 in the currently scheduled RTL fragment.
28298 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
28300 PR rtl-optimization/78751
28301 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
28304 2017-01-14 Jeff Law <law@redhat.com>
28306 PR tree-optimization/79090
28307 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
28308 variable length stores.
28309 (compute_trims): Delete dead assignment to *trim_tail.
28310 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
28313 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
28315 PR rtl-optimization/78626
28316 PR rtl-optimization/78727
28317 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
28318 of a block, and split such blocks after everything else is finished.
28320 2017-01-14 Alan Modra <amodra@gmail.com>
28323 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
28324 target legitimate_combined_insn.
28325 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
28326 (rs6000_legitimate_combined_insn): New function.
28327 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
28329 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
28330 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
28331 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
28333 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
28335 * doc/frontends.texi (G++ and GCC): Remove references to Java.
28337 2017-01-13 Jeff Law <law@redhat.com>
28339 PR tree-optimization/33562
28340 PR tree-optimization/61912
28341 PR tree-optimization/77485
28342 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
28344 (delete_dead_assignment): Likewise.
28345 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
28346 statement to delete_dead_call and delete_dead_assignment.
28348 2017-01-13 David Malcolm <dmalcolm@redhat.com>
28351 * substring-locations.c (format_warning_va): Strengthen case 1 so
28352 that both endpoints of the substring must be within the format
28353 range for just the substring to be printed.
28355 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
28357 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
28358 * config/i386/i386.c (ix86_target_string): Add missing options
28359 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
28360 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
28361 flags_other and ix86_target_other to flags2_other. Display unknown
28363 (ix86_valid_target_attribute_inner_p): Add missing options and
28364 reorder options by implied ISAs, as in ix86_target_string.
28366 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
28368 * hash-table.h (hash_table::too_empty_p): New function.
28369 (hash_table::expand): Use it.
28370 (hash_table::traverse): Likewise.
28371 (hash_table::empty_slot): Use sizeof (value_type) instead of
28372 sizeof (PTR) to convert bytes to elements. Shrink the table
28373 if the current size is excessive for the current number of
28376 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
28378 * ira-costs.c (record_reg_classes): Break from the inner loop
28379 early once alt_fail is known to be true. Update outer loop
28380 handling accordingly.
28382 2017-01-13 Jeff Law <law@redhat.com>
28384 * tree-ssa-dse.c (decrement_count): New function.
28385 (increment_start_addr, maybe_trim_memstar_call): Likewise.
28386 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
28387 when we know the partially dead statement is a mem* function.
28389 PR tree-optimization/61912
28390 PR tree-optimization/77485
28391 * tree-ssa-dse.c: Include expr.h.
28392 (maybe_trim_constructor_store): New function.
28393 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
28395 PR tree-optimization/33562
28396 PR tree-optimization/61912
28397 PR tree-optimization/77485
28398 * doc/invoke.texi: Document new dse-max-object-size param.
28399 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
28400 * tree-ssa-dse.c: Include params.h.
28401 (dse_store_status): New enum.
28402 (initialize_ao_ref_for_dse): New, partially extracted from
28404 (valid_ao_ref_for_dse, normalize_ref): New.
28405 (setup_live_bytes_from_ref, compute_trims): Likewise.
28406 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
28407 (maybe_trim_partially_dead_store): Likewise.
28408 (maybe_trim_complex_store): Likewise.
28409 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
28410 Track what bytes live from the original store. Return tri-state
28411 for dead, partially dead or live.
28412 (dse_dom_walker): Add constructor, destructor and new private members.
28413 (delete_dead_call, delete_dead_assignment): New extracted from
28415 (dse_optimize_stmt): Make a member of dse_dom_walker.
28416 Use initialize_ao_ref_for_dse.
28418 PR tree-optimization/33562
28419 PR tree-optimization/61912
28420 PR tree-optimization/77485
28421 * sbitmap.h (bitmap_count_bits): Prototype.
28422 (bitmap_clear_range, bitmap_set_range): Likewise.
28423 * sbitmap.c (bitmap_clear_range): New function.
28424 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
28426 2017-01-13 Martin Liska <mliska@suse.cz>
28429 * function.c (set_cfun): Add new argument force.
28430 * function.h (set_cfun): Likewise.
28431 * ipa-inline-transform.c (inline_call): Use the function when
28432 strict alising from is dropped for function we inline to.
28434 2017-01-13 Richard Biener <rguenther@suse.de>
28436 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
28437 for dumping GIMPLE INTEGER_CSTs.
28439 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28441 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
28442 to 201112L since C++17.
28444 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
28447 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
28448 if -fsanitize=kernel-address is present.
28450 2017-01-13 Richard Biener <rguenther@suse.de>
28452 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
28453 as _Literal ( type ) number in case usual suffixes do not
28454 preserve all information.
28456 2017-01-13 Richard Biener <rguenther@suse.de>
28458 PR tree-optimization/77283
28459 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
28460 and ssa-iterators.h.
28461 (is_feasible_trace): Implement a cost model based on joiner
28464 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
28467 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
28468 char or short to __float128/_Float128 directly.
28470 2017-01-12 Martin Sebor <msebor@redhat.com>
28472 to -Wformat-overflow.
28473 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
28474 (min_bytes_remaining): Same.
28475 (get_string_length): Same.
28476 (format_string): Same.
28477 (format_directive): Same.
28479 (pass_sprintf_length::handle_gimple_call): Same.
28481 2017-01-12 Jakub Jelinek <jakub@redhat.com>
28483 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
28484 info.nowrite calls with no lhs that can't throw. Return bool
28485 whether gsi_remove has been called or not.
28486 (pass_sprintf_length::handle_gimple_call): Return bool whether
28487 try_substitute_return_value called gsi_remove. Formatting fix.
28488 (pass_sprintf_length::execute): Don't use gsi_remove if
28489 handle_gimple_call returned true.
28492 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
28493 be removed due to side-effects, don't remove following barrier nor
28494 turn the successor edge into fallthru edge.
28496 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28499 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
28500 element-reversing loads and stores as not swappable.
28502 2017-01-12 Nathan Sidwell <nathan@acm.org>
28503 Nicolai Stange <nicstange@gmail.com>
28505 * combine.c (try_combine): Don't ignore result of overlap checking
28506 loop. Combine overlap & asm check into single loop.
28508 2017-01-12 Richard Biener <rguenther@suse.de>
28510 * tree-pretty-print.c (dump_generic_node): Provide -gimple
28511 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
28513 2017-01-12 Richard Biener <rguenther@suse.de>
28515 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
28516 and TS_TARGET_OPTION directly derive from TS_BASE.
28517 * tree-core.h (tree_optimization_option): Derive from tree_base.
28518 (tree_target_option): Likewise.
28520 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
28522 * config/i386/i386.c (memory_address_length): Increase len
28523 only when rip_relative_addr_p returns false.
28525 2017-01-11 Julia Koval <julia.koval@intel.com>
28527 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
28528 (OPTION_MASK_ISA_SGX_SET): New.
28529 (ix86_handle_option): Handle OPT_msgx.
28530 * config.gcc: Added sgxintrin.h.
28531 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
28532 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
28533 * config/i386/i386.c (ix86_target_string): Add -msgx.
28535 (ix86_option_override_internal): Handle new options.
28536 (ix86_valid_target_attribute_inner_p): Add sgx.
28537 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
28538 * config/i386/i386.opt: Add msgx.
28539 * config/i386/sgxintrin.h: New file.
28540 * config/i386/x86intrin.h: Add sgxintrin.h.
28542 2017-01-11 Jakub Jelinek <jakub@redhat.com>
28545 * fold-const.c (maybe_nonzero_address): Return 1 for function
28547 (tree_single_nonzero_warnv_p): Don't handle function local objects
28551 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
28554 2017-01-11 David Malcolm <dmalcolm@redhat.com>
28557 * opts.c: Include "spellcheck.h"
28558 (struct string_fragment): New struct.
28559 (struct edit_distance_traits<const string_fragment &>): New
28561 (get_closest_sanitizer_option): New function.
28562 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
28564 2017-01-11 Jakub Jelinek <jakub@redhat.com>
28566 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
28568 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
28569 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
28570 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
28571 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
28572 for initial die_offset if dwarf_split_debug_info.
28573 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
28574 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
28576 (output_skeleton_debug_sections): Formatting fix. Use
28577 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
28578 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
28580 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
28582 * config/arm/cortex-a53.md: Add bypasses for
28583 cortex_a53_r2f_cvt.
28584 (cortex_a53_r2f): Only use for transfers.
28585 (cortex_a53_f2r): Likewise.
28586 (cortex_a53_r2f_cvt): Add reservation for conversions.
28587 (cortex_a53_f2r_cvt): Likewise.
28589 2017-01-11 Tamar Christina <tamar.christina@arm.com>
28591 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
28592 to all inlined functions, change static to extern.
28594 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
28597 * config/arm/arm.c (legitimize_pic_address): Handle reference to
28599 (arm_assemble_integer): Likewise.
28601 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
28603 * config.gcc: Use new awk script to check CPU, FPU and architecture
28604 parameters for --with-... options.
28605 * config/arm/parsecpu.awk: New file
28606 * config/arm/arm-cpus.in: New file.
28607 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
28609 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
28611 * config/arm/t-arm: Update dependency rules.
28612 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
28613 of processing .def files.
28614 * config/arm/genopt.sh: Deleted.
28615 * config/arm/gentune.sh: Deleted.
28616 * config/arm/arm-cores.def: Deleted.
28617 * config/arm/arm-arches.def: Deleted.
28618 * config/arm/arm-fpus.def: Deleted.
28619 * config/arm/arm-tune.md: Regenerated.
28620 * config/arm/arm-tables.opt: Regenerated.
28621 * config/arm/arm-cpu.h: New generated file.
28622 * config/arm/arm-cpu-data.h: New generated file.
28623 * config/arm/arm-cpu-cdata.h: New generated file.
28625 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
28628 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
28630 (input_varpool_node): Unpack dynamically_initialized bit.
28632 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
28634 PR rtl-optimization/79032
28635 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
28636 the alignment of the adjusted memory reference against that of MODE,
28637 instead of the alignment of the original memory reference.
28639 2017-01-11 Martin Jambor <mjambor@suse.cz>
28641 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
28643 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
28644 decorated functions.
28646 2017-01-11 Richard Biener <rguenther@suse.de>
28648 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
28649 set range/nonnull info for PHI results. Do not set it on
28650 stmts marked for removal.
28652 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
28654 * expr.c (store_field): In the bitfield case, fetch the return value
28655 from the registers before applying a single big-endian adjustment.
28656 Always do a final load for a BLKmode value not larger than a word.
28658 2017-01-10 David Malcolm <dmalcolm@redhat.com>
28661 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
28662 that we correctly handle column numbers greater than
28663 LINE_MAP_MAX_COLUMN_NUMBER.
28665 2017-01-10 Martin Sebor <msebor@redhat.com>
28667 PR middle-end/78245
28668 * gimple-ssa-sprintf.c (get_destination_size): Call
28669 {init,fini}object_sizes.
28670 * tree-object-size.c (addr_object_size): Adjust.
28671 (pass_through_call): Adjust.
28672 (pass_object_sizes::execute): Adjust.
28673 * tree-object-size.h (fini_object_sizes): Declare.
28675 2017-01-10 Martin Sebor <msebor@redhat.com>
28677 PR tree-optimization/78775
28678 * builtins.c (get_size_range): Move...
28679 * calls.c: ...to here.
28680 (alloc_max_size): Accept zero argument.
28681 (operand_signed_p): Remove.
28682 (maybe_warn_alloc_args_overflow): Call get_size_range.
28683 * calls.h (get_size_range): Declare.
28685 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
28687 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
28688 from TI's devices.csv file as of September 2016.
28689 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
28691 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
28693 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
28694 * doc/invoke.texi: Likewise.
28695 * doc/md.texi: Likewise.
28696 * doc/objc.texi: Likewise.
28698 2017-01-10 Joshua Conner <joshconner@google.com>
28700 * config/arm/fuchsia-elf.h: New file.
28701 * config/fuchsia.h: New file.
28702 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
28703 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
28705 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
28707 2016-01-10 Richard Biener <rguenther@suse.de>
28709 PR tree-optimization/79034
28710 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28711 Propagate out degenerate PHIs in the joiner.
28713 2017-01-10 Martin Liska <mliska@suse.cz>
28715 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
28716 (sort_congruence_classes_by_decl_uid): Likewise.
28717 (sort_congruence_class_groups_by_decl_uid): Likewise.
28718 (sem_item_optimizer::merge_classes): Sort class, groups in these
28719 classes and members in the groups by DECL_UID of declarations.
28720 This would make merge operations stable.
28722 2017-01-10 Martin Liska <mliska@suse.cz>
28724 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
28725 usage of m_classes_vec.
28726 (sem_item_optimizer::~sem_item_optimizer): Likewise.
28727 (sem_item_optimizer::get_group_by_hash): Likewise.
28728 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
28729 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
28730 (sem_item_optimizer::verify_classes): Likewise.
28731 (sem_item_optimizer::process_cong_reduction): Likewise.
28732 (sem_item_optimizer::dump_cong_classes): Likewise.
28733 (sem_item_optimizer::merge_classes): Likewise.
28734 * ipa-icf.h (congruence_class_hash): Rename from
28735 congruence_class_group_hash. Remove declaration of m_classes_vec.
28737 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
28739 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
28740 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
28741 * config.gcc: Add avx512vpopcntdqintrin.h.
28742 * config/i386/avx512vpopcntdqintrin.h: New.
28743 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
28744 * config/i386/i386-builtin-types.def: Add new types.
28745 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
28746 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
28747 __builtin_ia32_vpopcountq_v8di_mask): New.
28748 * config/i386/i386-c.c (ix86_target_macros_internal): Define
28749 __AVX512VPOPCNTDQ__.
28750 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
28751 (PTA_AVX512VPOPCNTDQ): Define.
28752 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
28753 TARGET_AVX512VPOPCNTDQ_P): Define.
28754 * config/i386/i386.opt: Add mavx512vpopcntdq.
28755 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
28756 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
28758 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28760 PR middle-end/77484
28761 * predict.def (PRED_CALL): Set to 67.
28763 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
28765 * expr.c (store_field): In the bitfield case, if the value comes from
28766 a function call and is of an aggregate type returned in registers, do
28767 not modify the field mode; extract the value in all cases if the mode
28768 is BLKmode and the size is not larger than a word.
28770 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
28773 * config/i386/cpuid.h: Fix undefined behavior.
28775 2017-01-04 Jeff Law <law@redhat.com>
28777 PR tree-optimization/79007
28778 PR tree-optimization/67955
28779 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
28780 conservative for pt.null when flag_non_call_exceptions is on.
28782 2017-01-09 Jakub Jelinek <jakub@redhat.com>
28784 PR translation/79019
28785 PR translation/79020
28786 * params.def (PARAM_INLINE_MIN_SPEEDUP,
28787 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
28788 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
28790 * config/avr/avr.opt (maccumulate-args): Likewise.
28791 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
28792 * common.opt (freport-bug): Likewise.
28793 * cif-code.def (CIF_FINAL_ERROR): Likewise.
28794 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
28795 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
28796 translatable string.
28797 * config/i386/i386.c (function_value_32): Likewise.
28798 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
28799 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
28801 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
28802 * common/config/msp430/msp430-common.c (msp430_handle_option):
28804 * symtab.c (symtab_node::verify_base): Likewise.
28805 * opts.c (set_debug_level): Likewise.
28806 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
28807 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
28808 missing whitespace to translatable strings.
28809 * config/avr/avr.md (bswapsi2): Fix typo in comment.
28810 * config/sh/superh.h: Likewise.
28811 * config/i386/xopintrin.h: Likewise.
28812 * config/i386/znver1.md: Likewise.
28813 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
28814 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
28815 * double-int.h (struct double_int): Likewise.
28816 * double-int.c (div_and_round_double): Likewise.
28817 * wide-int.cc: Likewise.
28818 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
28819 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
28820 * cfgcleanup.c (crossjumps_occured): Renamed to ...
28821 (crossjumps_occurred): ... this.
28822 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
28825 PR tree-optimization/78899
28826 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
28827 returning bool return struct loop *, NULL for failure and the new
28829 (versionable_outer_loop_p): Don't version outer loop if it has
28830 dont_vectorized bit set.
28831 (tree_if_conversion): When versioning outer loop, ensure
28832 tree_if_conversion is performed also on the inner loop of the
28833 non-vectorizable outer loop copy.
28834 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
28835 LOOP_VECTORIZED in inner loop of the scalar outer loop and
28836 prevent vectorization of it.
28837 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
28838 the outer loop vectorization of the non-scalar version is attempted
28839 before vectorization of the inner loop in scalar version. If
28840 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
28841 vectorization of its inner loop.
28842 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
28843 has 2 inner loops, rename also on edges from bb whose single pred
28844 is outer_loop->header. Fix typo in function comment.
28846 2017-01-09 Martin Sebor <msebor@redhat.com>
28849 * asan.c (asan_emit_stack_protection): Increase local buffer size
28850 to avoid snprintf truncation warning.
28852 2017-01-09 Andrew Pinski <apinski@cavium.com>
28854 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
28855 to reference thunderx2t99 for the tuning structure
28856 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
28858 (thunderx2t99_extra_costs): This.
28859 * config/aarch64/aarch64-tune.md: Regenerate.
28860 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
28861 (vulcan_addrcost_table): This.
28862 (vulcan_regmove_cost): Rename to ...
28863 (thunderx2t99_regmove_cost): This.
28864 (vulcan_vector_cost): Rename to ...
28865 (thunderx2t99_vector_cost): this.
28866 (vulcan_branch_cost): Rename to ...
28867 (thunderx2t99_branch_cost): This.
28868 (vulcan_tunings): Rename to ...
28869 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
28870 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
28872 2017-01-09 Martin Jambor <mjambor@suse.cz>
28876 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
28877 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
28878 (propagate_vr_accross_jump_function): Use the above function for all
28879 value range computations for pass-through jump functions and type
28880 converasion from explicit value range values.
28881 (ipcp_propagate_stage): Do not attempt to deduce types of formal
28882 parameters from TYPE_ARG_TYPES.
28883 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
28884 (ipa_write_node_info): Stream type of the actual argument.
28885 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
28887 2017-01-09 Martin Liska <mliska@suse.cz>
28890 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
28891 (lookup_compiler): Do not show error message with have_E.
28893 2017-01-09 Jakub Jelinek <jakub@redhat.com>
28895 PR tree-optimization/78938
28896 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
28897 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
28898 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
28899 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
28902 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28904 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
28907 2017-01-09 Richard Biener <rguenther@suse.de>
28909 PR tree-optimization/78997
28910 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
28911 name condition properly.
28913 2017-01-09 Richard Biener <rguenther@suse.de>
28916 * dwarf2out.c (is_cxx): New overload with context.
28917 (is_naming_typedef_decl): Use it.
28919 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28921 * invoke.texi (Option Summary): Correct spacing in option lists
28922 and add line breaks to fix over-long lines.
28924 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28926 PR middle-end/17660
28928 * extend.texi (Common Variable Attributes): Add xref to GCC
28929 Internals manual to explain mode attribute keywords.
28931 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28934 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
28935 and Preprocessor Options.
28936 (Options for Linking): Document -pthread here....
28937 (RS/6000 and PowerPC Options): ...not here.
28938 (Solaris 2 Options): ...or here.
28939 * doc/cppopts.texi: Document -pthread.
28941 2017-01-08 Martin Sebor <msebor@redhat.com>
28943 PR middle-end/77708
28944 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
28945 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
28946 New member functions.
28947 (format_directive): Used them.
28949 (pass_sprintf_length::handle_gimple_call): Same.
28950 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
28951 to avoid truncation for any argument.
28952 (extract_affine_mul): Same.
28953 * tree.c (get_file_function_name): Same.
28955 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28957 PR middle-end/77484
28958 * predict.def (PRED_INDIR_CALL): Set to 86.
28960 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28962 PR preprocessor/54124
28963 * doc/cppopts.texi: Reformat -d subtable to list the full name
28964 of the options. Add cross-reference to the docs for the general
28965 compiler -d options.
28966 * doc/invoke.texi (Developer Options): Add cross-reference to the
28967 preprocessor-specific -d option documentation.
28969 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28971 PR preprocessor/13498
28972 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
28973 redudant material, and reflect new command-line options.
28974 (System Headers): Likewise.
28976 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28978 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
28979 -isystem, and -idirafter. Copy-edit.
28980 * doc/cppopts.texi: Copy-edit. Remove contradiction about
28981 default for -ftrack-macro-expansion. Delete obsolete and
28982 badly-formatted implementation details about -fdebug-cpp output.
28983 * doc/cppwarnopts.texi: Copy-edit.
28985 2017-01-07 David Malcolm <dmalcolm@redhat.com>
28988 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
28989 that the transition from a max line width >= 1<<10 to narrower
28990 lines works correctly.
28992 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
28994 * doc/options.texi (PerFunction): New.
28995 * opt-functions.awk (switch_flags): Map both Optimization and
28996 PerFunction to CL_OPTIMIZATION.
28997 * opth-gen.awk: Test for PerFunction flag along with
28999 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
29000 it only when the latter is present. Skip those that don't in
29001 the hash function generator.
29002 * common.opt (fvar-tracking): Mark as PerFunction instead of
29004 (fvar-tracking-assignments): Likewise.
29005 (fvar-tracking-assignments-toggle): Likewise.
29006 (fvar-tracking-uninit): Likewise.
29008 2017-01-07 Jakub Jelinek <jakub@redhat.com>
29010 PR translation/79018
29011 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
29014 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
29017 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
29018 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
29019 TARGET_LONG_JUMP_TABLE_OFFSETS.
29020 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
29021 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
29022 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
29023 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
29024 * config/m68k/m68k.md (tablejump expander): Likewise.
29025 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
29026 TARGET_LONG_JUMP_TABLE_OFFSETS.
29027 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
29028 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
29030 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
29031 David Holsgrove <david.holsgrove@xilinx.com>
29033 * common/config/microblaze/microblaze-common.c
29034 (TARGET_EXCEPT_UNWIND_INFO): Remove.
29035 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
29037 * config/microblaze/microblaze.c (microblaze_must_save_register)
29038 (microblaze_expand_epilogue, microblaze_return_addr): Handle
29040 (microblaze_eh_return): New function.
29041 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
29042 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
29043 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
29044 * config/microblaze/microblaze.md (eh_return): New pattern.
29046 2017-01-06 Jakub Jelinek <jakub@redhat.com>
29048 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
29049 GCC_DIAGNOSTIC_STRINGIFY): Define.
29051 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
29053 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29055 * config/arm/arm.md (<mcrr>): New.
29057 * config/arm/arm.c (arm_arch5te): New.
29058 (arm_option_override): Set arm_arch5te.
29059 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
29061 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
29062 (arm_mcrr_qualifiers): ... this. New.
29063 (MRRC_QUALIFIERS): Define to...
29064 (arm_mrrc_qualifiers): ... this. New.
29065 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
29067 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
29068 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
29069 (MRRCI, mrrc, MRRC): New.
29070 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
29071 VUNSPEC_MRRC2): New.
29073 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29075 * config/arm/arm.md (<mcr>): New.
29077 * config/arm/arm.c (arm_coproc_builtin_available): Add
29078 support for mcr, mrc, mcr2 and mrc2.
29079 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
29080 (arm_mcr_qualifiers): ... this. New.
29081 (MRC_QUALIFIERS): Define to ...
29082 (arm_mrc_qualifiers): ... this. New.
29083 (MCR_QUALIFIERS): Define to ...
29084 (arm_mcr_qualifiers): ... this. New.
29085 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
29087 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
29088 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
29089 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
29090 VUNSPEC_MRC2): New.
29092 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29094 * config/arm/arm.md (*ldc): New.
29098 * config/arm/arm.c (arm_coproc_builtin_available): Add
29099 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
29100 (arm_coproc_ldc_stc_legitimate_address): New.
29101 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
29102 'qualifier_const_pointer'.
29103 (LDC_QUALIFIERS): Define to...
29104 (arm_ldc_qualifiers): ... this. New.
29105 (STC_QUALIFIERS): Define to...
29106 (arm_stc_qualifiers): ... this. New.
29107 * config/arm/arm-protos.h
29108 (arm_coproc_ldc_stc_legitimate_address): New.
29109 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
29110 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
29111 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
29112 stc2, stcl, stc2l): New.
29113 * config/arm/constraints.md (Uz): New.
29114 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
29115 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
29116 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
29117 VUNSPEC_STC2L): New.
29119 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29121 * config/arm/arm.md (<cdp>): New.
29122 * config/arm/arm.c (neon_const_bounds): Rename this ...
29123 (arm_const_bounds): ... this.
29124 (arm_coproc_builtin_available): New.
29125 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
29126 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
29127 (CDP_QUALIFIERS): Define to...
29128 (arm_cdp_qualifiers): ... this. New.
29130 (arm_expand_builtin_args): Add case for 6 arguments.
29131 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
29132 (arm_const_bounds): ... this.
29133 (arm_coproc_builtin_available): New.
29134 * config/arm/arm_acle.h (__arm_cdp): New.
29136 * config/arm/arm_acle_builtins.def (cdp): New.
29138 * config/arm/iterators.md (CDPI,CDP,cdp): New.
29139 * config/arm/neon.md: Rename all 'neon_const_bounds' to
29140 'arm_const_bounds'.
29141 * config/arm/types.md (coproc): New.
29142 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
29143 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
29144 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
29145 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
29147 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29149 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
29150 (UBINOP_QUALIFIERS): New.
29152 (acle_builtin_data): New. Change comment.
29153 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
29154 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
29155 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
29156 arm_acle_builtins.def.
29157 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
29158 (arm_init_acle_builtins): New.
29159 (CRC32_BUILTIN): Remove.
29160 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
29161 crc32cb, crc32ch and crc32cw.
29162 (arm_init_crc32_builtins): Remove.
29163 (arm_init_builtins): Use arm_init_acle_builtins rather
29164 than arm_init_crc32_builtins.
29165 (arm_expand_acle_builtin): New.
29166 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
29167 * config/arm/arm_acle_builtins.def: New.
29169 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29171 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
29172 (arm_builtin_datum): ... this.
29173 (arm_init_neon_builtin): Rename to ...
29174 (arm_init_builtin): ... this. Add a new parameters PREFIX
29175 and USE_SIG_IN_NAME.
29176 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
29177 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
29178 'arm_builtin_datum'.
29179 (arm_init_vfp_builtins): Likewise.
29180 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
29181 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
29182 (arm_expand_neon_args): Rename to ...
29183 (arm_expand_builtin_args): ... this. Rename builtin_arg
29184 enum values and differentiate between ARG_BUILTIN_MEMORY
29185 and ARG_BUILTIN_NEON_MEMORY.
29186 (arm_expand_neon_builtin_1): Rename to ...
29187 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
29188 values, arm_expand_builtin_args and add bool parameter NEON.
29189 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
29190 (arm_expand_vfp_builtin): Likewise.
29191 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
29193 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29195 PR middle-end/77484
29196 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
29197 * predict.c (tree_estimate_probability_bb): Reverse direction of
29198 polymorphic call predictor.
29200 2017-01-06 David Malcolm <dmalcolm@redhat.com>
29202 * passes.c (execute_one_pass): Split out pass-skipping logic into...
29203 (determine_pass_name_match): ...this new function and...
29204 (should_skip_pass_p): ...this new function.
29206 2017-01-06 Nathan Sidwell <nathan@acm.org>
29208 * ipa-visibility.c (function_and_variable_visibility): Reformat
29209 comments and long lines. Remove extrneous if.
29210 * symtab.c (symtab_node::make_decl_local): Fix code format.
29211 (symtab_node::set_section_for_node): Fix comment typo.
29213 2017-01-06 Martin Liska <mliska@suse.cz>
29216 * lra-constraints.c: Rename invariant to lra_invariant.
29217 * predict.c (set_even_probabilities): Initialize e to NULL.
29219 2017-01-05 Martin Sebor <msebor@redhat.com>
29221 PR tree-optimization/78910
29222 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
29223 (format_integer): Correct off-by-one error in the handling
29224 of precision with negative numbers in signed conversions..
29226 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
29228 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
29230 2017-01-05 Jakub Jelinek <jakub@redhat.com>
29232 PR tree-optimization/71016
29233 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
29234 factor_out_conditional_conversion. Formatting fix.
29235 (factor_out_conditional_conversion): Add cond_stmt argument.
29236 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
29237 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
29240 2017-01-05 David Malcolm <dmalcolm@redhat.com>
29242 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
29243 read-rtl-function.o, and selftest-rtl.o.
29244 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
29245 (selftest::aarch64_test_loading_full_dump): New function.
29246 (selftest::aarch64_run_selftests): New function.
29247 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
29248 selftest::aarch64_run_selftests.
29249 * config/i386/i386.c
29250 (selftest::ix86_test_loading_dump_fragment_1): New function.
29251 (selftest::ix86_test_loading_call_insn): New function.
29252 (selftest::ix86_test_loading_full_dump): New function.
29253 (selftest::ix86_test_loading_unspec): New function.
29254 (selftest::ix86_run_selftests): Call the new functions.
29255 * emit-rtl.c (maybe_set_max_label_num): New function.
29256 * emit-rtl.h (maybe_set_max_label_num): New decl.
29257 * function.c (instantiate_decls): Guard call to
29258 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
29259 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
29261 * gensupport.c (gen_reader::gen_reader): Pass "false"
29262 for new "compact" param of rtx_reader.
29263 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
29264 rather than an empty string for NULL strings.
29265 * read-md.c: Potentially include config.h rather than bconfig.h.
29266 Wrap include of errors.h with #ifdef GENERATOR_FILE.
29267 (have_error): New global, copied from errors.c.
29268 (md_reader::read_name): Rename to...
29269 (md_reader::read_name_1): ...this, adding "out_loc" param,
29270 and converting "missing name or number" to returning false, rather
29272 (md_reader::read_name): Reimplement in terms of read_name_1.
29273 (md_reader::read_name_or_nil): New function.
29274 (md_reader::read_string): Handle "(nil)" by returning NULL.
29275 (md_reader::md_reader): Add new param "compact".
29276 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
29277 (md_reader::read_file): New method.
29278 * read-md.h (md_reader::md_reader): Add new param "compact".
29279 (md_reader::read_file): New method.
29280 (md_reader::is_compact): New accessor.
29281 (md_reader::read_name): Convert return type from void to file_location.
29282 (md_reader::read_name_or_nil): New decl.
29283 (md_reader::read_name_1): New decl.
29284 (md_reader::m_compact): New field.
29285 (noop_reader::noop_reader): Pass "false" for new "compact" param
29287 (rtx_reader::rtx_reader): Add new "compact" param.
29288 (rtx_reader::read_rtx_operand): Make virtual and convert return
29289 type from void to rtx.
29290 (rtx_reader::read_until): New decl.
29291 (rtx_reader::handle_any_trailing_information): New virtual function.
29292 (rtx_reader::postprocess): New virtual function.
29293 (rtx_reader::finalize_string): New virtual function.
29294 (rtx_reader::m_in_call_function_usage): New field.
29295 (rtx_reader::m_reuse_rtx_by_id): New field.
29296 * read-rtl-function.c: New file.
29297 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
29298 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
29299 (selftest::verify_three_block_rtl_cfg): New decl.
29300 * read-rtl-function.h: New file.
29301 * read-rtl.c: Potentially include config.h rather than bconfig.h.
29302 For host, include function.h, memmodel.h, and emit-rtl.h.
29303 (one_time_initialization): New function.
29304 (struct compact_insn_name): New struct.
29305 (compact_insn_names): New array.
29306 (find_code): Handle insn codes in compact dumps.
29307 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
29308 (bind_subst_iter_and_attr): Likewise.
29309 (add_condition_to_string): Likewise.
29310 (add_condition_to_rtx): Likewise.
29311 (apply_attribute_uses): Likewise.
29312 (add_current_iterators): Likewise.
29313 (apply_iterators): Likewise.
29314 (initialize_iterators): Guard usage of apply_subst_iterator with
29315 #ifdef GENERATOR_FILE.
29316 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
29317 (md_reader::read_mapping): Likewise.
29318 (add_define_attr_for_define_subst): Likewise.
29319 (add_define_subst_attr): Likewise.
29320 (read_subst_mapping): Likewise.
29321 (check_code_iterator): Likewise.
29322 (rtx_reader::read_rtx): Likewise. Move one-time initialization
29324 (one_time_initialization): New function.
29325 (rtx_reader::read_until): New method.
29326 (read_flags): New function.
29327 (parse_reg_note_name): New function.
29328 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
29329 Handle reuse_rtx ids.
29330 Wrap iterator lookup within #ifdef GENERATOR_FILE.
29331 Add parsing support for RTL dumps, mirroring the special-cases in
29332 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
29333 values, and calling handle_any_trailing_information.
29334 (rtx_reader::read_rtx_operand): Convert return type from void
29335 to rtx, returning return_rtx. Handle case 'e'. Call
29336 finalize_string on XSTR and XTMPL fields.
29337 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
29338 "(nil)" values were omitted. Call the postprocess vfunc on the
29340 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
29341 class ctor. Initialize m_in_call_function_usage. Call
29342 one_time_initialization.
29343 * rtl-tests.c (selftest::test_uncond_jump): Call
29344 set_new_first_and_last_insn.
29345 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
29346 * selftest-rtl.c: New file.
29347 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
29348 (selftest::get_insn_by_uid): New decl.
29349 * selftest-run-tests.c (selftest::run_tests): Call
29350 read_rtl_function_c_tests.
29351 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
29352 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
29355 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
29357 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
29358 operands in a special way. Assert that pos+len <= mode precision.
29360 2017-01-05 Jakub Jelinek <jakub@redhat.com>
29362 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
29363 3 argument Alias with unlimited for the negative form.
29364 (fno-vect-cost-model): Removed.
29366 2017-01-05 Martin Liska <mliska@suse.cz>
29368 * hsa-gen.c (gen_hsa_divmod): New function.
29369 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
29371 2017-01-05 Martin Liska <mliska@suse.cz>
29374 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
29377 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29379 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
29380 small constant length operands.
29382 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29384 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
29385 between loop iterations.
29387 2017-01-05 Martin Liska <mliska@suse.cz>
29390 * gimplify.c (gimplify_decl_expr): Compare to
29391 asan_poisoned_variables instread of checking flags.
29392 (gimplify_target_expr): Likewise.
29393 (gimplify_expr): Likewise.
29394 (gimplify_function_tree): Conditionally initialize
29395 asan_poisoned_variables.
29397 2017-01-04 Jeff Law <law@redhat.com>
29399 PR tree-optimizatin/78812
29400 * rtl.h (contains_mem_rtx_p): Prototype.
29401 * ifcvt.c (containts_mem_rtx_p): Move from here to...
29402 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
29403 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
29404 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
29405 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
29407 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29409 * input.c (assert_char_at_range): Default-initialize actual_range.
29411 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29413 * df-scan.c (df_ref_create_structure): Make regno unsigned,
29414 to match the caller.
29416 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29418 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
29419 insns after final jump in test to emit dummy move.
29421 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29423 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
29424 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
29426 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29428 * multiple_target.c (create_dispatcher_calls): Init e_next.
29429 * tree-ssa-loop-split.c (split_loop): Init border.
29430 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
29433 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
29438 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
29439 (altivec_register_operand): Do not return true if the operand
29440 contains a SUBREG mixing SImode and SFmode.
29441 (vsx_register_operand): Likewise.
29442 (vsx_reg_sfsubreg_ok): New predicate.
29443 (vfloat_operand): Do not return true if the operand contains a
29444 SUBREG mixing SImode and SFmode.
29445 (vint_operand): Likewise.
29446 (vlogical_operand): Likewise.
29447 (gpc_reg_operand): Likewise.
29448 (int_reg_operand): Likewise.
29449 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
29450 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
29451 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
29453 (rs6000_emit_move_si_sf_subreg): New helper function.
29454 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
29455 fixup SUBREGs involving SImode and SFmode.
29456 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
29457 numbers for the new peephole2 optimization.
29458 (peephole2 for SFmode unions): New peephole2 to optimize cases in
29459 the GLIBC math library that do AND/IOR/XOR operations on single
29460 precision floating point.
29461 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
29462 target macros to say whether we need to avoid SUBREGs mixing
29464 (TARGET_ALLOW_SF_SUBREG): Likewise.
29465 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
29466 (UNSPEC_SI_FROM_SF): Likewise.
29467 (iorxor): Change spacing.
29468 (and_ior_xor): New iterator for AND, IOR, and XOR.
29469 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
29470 (movdi_from_sf_zero_ext): Likewise.
29471 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
29472 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
29473 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
29474 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
29475 (fms<mode>4): Likewise.
29476 (fnma<mode>4): Likewise.
29477 (fnms<mode>4): Likewise.
29478 (nfma<mode>4): Likewise.
29479 (nfms<mode>4): Likewise.
29481 2017-01-04 Marek Polacek <polacek@redhat.com>
29484 * doc/invoke.texi: Document -Wpointer-compare.
29486 2017-01-04 Jakub Jelinek <jakub@redhat.com>
29488 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
29491 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
29492 descriptions for -gdwarf-5 and emit them as uleb128 instead of
29495 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
29498 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
29499 documentation of the powerpc_popcntb_ok attribute.
29500 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29501 code to issue warning messages if a requested CPU configuration is
29502 not supported by the binary (assembler and loader) toolchain.
29503 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
29504 made to define a built-in function that has been disabled.
29505 (paired_init_builtins): Add assertion to prevent ICE if attempt is
29506 made to define a built-in function that has been disabled.
29507 (altivec_init_builtins): Add comment explaining why definition
29508 of the DST built-in functions is not preceded by an assertion
29509 check. Add assertions to prevent ICE if attempts are made to
29510 define an altivec predicate or an abs* built-in function that has
29512 (htm_init_builtins): Add comment explaining why definition of the
29513 htm built-in functions is not preceded by an assertion check.
29515 2017-01-04 Jeff Law <law@redhat.com>
29517 PR tree-optimizatin/67955
29518 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
29519 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
29520 the points-to solution does not include pt_null. Use DECL_PT_UID
29523 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
29525 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
29526 Use gen_int_mode instead of gen_lopwart for const_int operands.
29528 2017-01-04 Jakub Jelinek <jakub@redhat.com>
29530 PR tree-optimization/71563
29531 * match.pd: Simplify X << Y into X if Y is known to be 0 or
29532 out of range value - has low bits known to be zero.
29534 2017-01-04 Alan Modra <amodra@gmail.com>
29536 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
29537 * configure: Regenerate.
29538 * config.in: Regenerate.
29540 2017-01-04 Jakub Jelinek <jakub@redhat.com>
29543 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
29544 a substring of the message, but strcmp with the whole message. Ifdef
29545 ENABLE_NLS, translate the message first using dgettext.
29547 2017-01-03 Jeff Law <law@redhat.com>
29549 PR tree-optimizatin/78856
29550 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
29551 (mark_threaded_blocks): Remove code to truncate thread paths that
29552 cross multiple loop headers. Instead invalidate the cached loop
29553 iteration information and handle case of a thread path walking
29554 into an irreducible region.
29556 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
29559 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
29560 assertions. Add support for doing the signbit if the IEEE 128-bit
29561 floating point value is in a GPR.
29562 * config/rs6000/rs6000.md (Fsignbit): Delete.
29563 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
29564 Update the length attribute if the value is in a GPR.
29565 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
29566 the sign or zero extension instruction, since the value is always 0/1.
29567 (signbit<mode>2_dm2): Delete using <Fsignbit>.
29570 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
29571 extracting SImode to a GPR register so that we can generate a
29572 store, limit the vector to be in a traditional Altivec register
29573 for the vextuwrx instruction.
29575 2017-01-03 Ian Lance Taylor <iant@google.com>
29577 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
29579 2017-01-03 Martin Sebor <msebor@redhat.com>
29581 PR tree-optimization/78696
29582 * gimple-ssa-sprintf.c (format_floating): Correct handling of
29583 precision. Use MPFR for %f for greater fidelity. Correct handling
29585 (pass_sprintf_length::compute_format_length): Set width and precision
29586 specified by asrerisk to void_node for vararg functions.
29587 (try_substitute_return_value): Adjust dump output.
29589 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
29591 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
29593 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
29595 * doc/invoke.texi (SPARC options): Document -mlra as the default.
29596 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
29597 -mlra/-mno-lra was passed to the compiler.
29599 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
29601 PR rtl-optimization/65618
29602 * emit-rtl.c (try_split): Move initialization of "before" and
29603 "after" to just before the call to emit_insn_after_setloc.
29605 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
29607 * doc/md.texi (Standard Names): Remove reference to Java frontend.
29609 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
29611 * dwarf2out.c (gen_enumeration_type_die): When
29612 -gno-strict-dwarf, add a DW_AT_encoding attribute.
29614 2017-01-03 Jakub Jelinek <jakub@redhat.com>
29616 PR tree-optimization/78965
29617 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
29618 Change first argument from const call_info & to call_info &. For %n
29619 set info.nowrite to false.
29621 PR middle-end/78901
29622 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
29623 possibly throwing calls.
29625 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
29626 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
29627 and fns handling, rather than in a separate case SSA_NAME.
29629 2017-01-02 Jeff Law <law@redhat.com>
29631 * config/darwin-driver.c (darwin_driver_init): Const-correctness
29632 fixes for first_period and second_period variables.
29634 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
29637 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
29638 (*insvqi_1): New insn pattern.
29639 (*insvqi_1_mem_rex64): Ditto.
29640 (*insvqi_2): Ditto.
29641 (*insvqi_3): Rename from *insvqi.
29643 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
29645 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
29647 * doc/cfg.texi (Edges): Remove reference to Java.
29648 (Maintaining the CFG): Ditto.
29650 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29652 PR middle-end/77674
29653 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
29654 transparent aliases.
29656 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29658 PR middle-end/77484
29659 * predict.def (PRED_CALL): Update hitrate.
29660 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
29661 * predict.c (tree_estimate_probability_bb): Split CALL predictor
29662 into direct/indirect/polymorphic variants.
29664 2017-01-01 Jakub Jelinek <jakub@redhat.com>
29666 Update copyright years.
29668 * gcc.c (process_command): Update copyright notice dates.
29669 * gcov-dump.c (print_version): Ditto.
29670 * gcov.c (print_version): Ditto.
29671 * gcov-tool.c (print_version): Ditto.
29672 * gengtype.c (create_file): Ditto.
29673 * doc/cpp.texi: Bump @copying's copyright year.
29674 * doc/cppinternals.texi: Ditto.
29675 * doc/gcc.texi: Ditto.
29676 * doc/gccint.texi: Ditto.
29677 * doc/gcov.texi: Ditto.
29678 * doc/install.texi: Ditto.
29679 * doc/invoke.texi: Ditto.
29681 Copyright (C) 2017 Free Software Foundation, Inc.
29683 Copying and distribution of this file, with or without modification,
29684 are permitted in any medium without royalty provided the copyright
29685 notice and this notice are preserved.